«'
Environmental Protection
Ager
Health httects
Research Labors
Rest ogle Park
NC 27711
Research and Development
&EPA
Methodologies and Protocols
in Clinical Research:
Evaluating Environmental
Effects in Man
-------
DISCLAIMER
This report has been reviewed by the Health Effects
Research Laboratory, U.S. Environmental Protection
Agency, and approved for publication. Mention of
trade names or commercial products does not consti-
tute endorsement or recommendation for use.
-------
EPA-600/9-78-008
Methodologies and Protocols
in Clinical Research:
Evaluating Environmental
Effects in Man
Symposium Proceedings
May 1978
United States Environmental Protection Agency
Health Effects Research Laboratory
Clinical Studies Division
Research Triangle Park, North Carolina 27711
-------
Preface
The relative affluence of the industrial age has been attain-
ed at the cost of degrading the quality of some aspects of life.
Conditions of the sort described by Dickens have, in large measure,
been corrected. But there remains the pervasive and insidious danger
of man's increasing ability to alter his environment. In the final
analysis, mankind is concerned with the repercussions of environ-
mental changes on his health and happiness. Both of these quali-
ties are very difficult to measure in a quantitative way. This
symposium, however, addresses some aspects of the quantitative
assessment of health status.
In some respects, the proper study of environmental influence
on human health is the study of humans whose environment is altered
in a controlled way. The methodologic, ethical, legal, and social
aspects of such research are the major topics of this symposium.
Because of the inherent problems and restrictions in these major
topics, there are many environmental health questions that will
always remain beyond the purview of clinical research. Because of
the unique advantages of clinical research, there are some environ-
mental health questions that are answered best by clinical research.
One of the objectives of this symposium is to formulate criteria
for identifying which kinds of environmental health questions are
particularly suited to the clinical research approach and which
definitely are not.
The levels of sophistication involved in most aspects of clini-
cal environmental research is increasing rapidly. Those who are
relatively new to this field may assume that space age technology
reflects the recent advent of this kind of research. This is not
III
-------
true. Modern clinical environmental research draws on a rich
heritage of simple, elegant studies that have provided the foun-
dation of human stress physiology. I hope we are successful in
using the tools of modern technology in responding to the needs of
modern society. To accomplish this, we must continue to perform
well in the tradition of those whose work has made ours possible
so that we can improve the quality of life for those who follow
us on this planet.
John H. Knelson, M.D.
Clinical Studdies Division
IV
-------
Contents
PART ONE - PHILOSOPHY OF CLINICAL RESEARCH
Ethical Considerations in Research
Involving Human Subjects,
Harmon L. Smith
Discussion Summary
Legal Aspects of Using Human Subjects
in Environmental Research,
Michael V. Mclntire ..................... 19
Discussion Summary
Informed Consent — Its Function
and Limitations,
Charles E. Daye ....................... 31
Discussion Summary ......... ..... ....... 49
Role and Function of Committees on
Protection of Human Subjects in Research,
Edward Bishop ........................ 51
Discussion Summary ..................... 59
-------
Developing Methodologies — Environmental Studies
Steven Horvath '
63
Discussion Summary ....
69
Rationale for Experimental Design,
John H. Knelson
71
Discussion Summary «...
79
Subject Selection, Investigator Interactions,
Informed Consent in Clinical and
Environmental Research,
David A. Otto and Jeanne T. Hernandez .
81
Discussion Summary
91
Acute Versus Chronic Studies,
David Bates
93
Discussion Summary ....
101
Sensitive Populations in Environmental
Studies,
Carl Shy*
Role of Automatic Data Processing in
Clinical Research,
Frank Starmer ........
103
Discussion Summary
109
PART THREE ENVIRONMENTAL AND PHYSICAL SAFETY
CONSIDERATIONS IN HUMAN EXPOSURE FACILITIES
Environmental Controls and Safeguards,
Morton Lippmann ......
113
Discussion Summary ...
Electrical Surveillance and Integrity,
G. Guy Knickerbocker
*Paper not available at time of publication.
vi
131
133
-------
Discussion Summary 145
PART FOUR EPA HUMAN STUDIES PROGRAMS
CLEANS/CLEVER System Approach,
John H. Knelson 149
PART FIVE SPECIAL CONSIDERATIONS AND APPROACHES
IN ENVIRONMENTAL CLINICAL RESEARCH
Introduction to Panel Discussion,
Philip Bromberg
Statement, Robert Frank 165
Discussion • 171
Statement, Bernard E. Statland 175
Discussion »
Statement, L. David Pengelly
Discussion • »«• 203
Statement, Mario C. Battigelli 209
Discussion ...... 213
Appendix
Program Participants 223
VII
-------
PHILOSOPHY OF CLINICAL RESEARCH
Moderator: Milan Hazucha, M.D.
-------
Ethical Considerations in Research
Involving Human Subjects
Harmon L Smith, Ph.D.
Divinity School
Duke University
There is, I dare to think, broader consensus among physicians
and ethicists on the meaning of "clinical research" than on "ethi-
cal issues," so I want to introduce these observations on ethical
issues in clinical research by commenting on what constitutes an
ethical issue. No doubt, you have noticed that the words "ethics"
and "morals" are heir to many interpretations; indeed, they are
frequently employed as synonyms in medical literature. Therefore,
my initial focus will be to discriminate between these terms.
ETHICS AND MORALS
In the history of Western philosophical and theological re-
flection, ethics (or moral philosophy) is typically characterized
by a spirit of radical inquiry; it does not attempt to supply so-
lutions for moral dilemmas, but it does undertake to provide a ra-
tional framework for comprehending the complexities of moral judg-
ment. Put simply, the moral question is a "what" question—what
ought I do, what good should I seek, what is the right action in
this situation, what end should I pursue? Correspondingly/ the
ethical question is a "why" question—why should I do this, why
do I seek this good action rather than some other action, why is
this action right or appropriate?
In other words, ethical questions attempt to reference action
-------
to an affirmative warrant/ they ask whether there is a proper rea-
son for making a particular choice, and whether there is coherence
and congruency between what one believes and how one behaves. On
these terms, a moral action is one that expresses an antecedent
value commitment.
In this classic study of American racism, Gunnar Myrdal argues
that the roots of American racism lie in a contradiction between
American political and social ideology. In his book, An American
Dilemma, Myrdal shows that racism is the product of a distancing,
a gap, a discontinuity between the "American creed" and the "Amer-
ican deed." The "creed" declares that all persons are created
equal; the "deed" treats some persons as superior and others as
inferior. The Biblical word for this disparity is hypocrisy (from
the Greek hypokrisis, literally/ to act a part, to pretend to be
what one is not), and immorality is the name we give to conduct
that is contradictory of, or in conflict with, character. Thus,
one useful reason for distinguishing ethics from morals is to
undertake a rational analysis of the value predicates of actions
and to test their expression in behavior.
DESCRIPTIVE/PRESCRIPTIVE ANALYSIS
The analysis of the value predicates of action and their ex-
pression in behavior is principally of two sorts: descriptive and
prescriptive, or operational and normative, in descriptive analy-
sis, the values that appear to be informing action are inferred
from the actual conduct of individuals or groups. Thus, if a per-
son or group steals, it is not unreasonable to suppose that (at
least in those situations in which stealing is practiced) there are
no absolute property rights. It would appear further that property
(at least the property in this situation) appropriately belongs to
whoever can possess it, by fair means or foul. Conversely, if a
person or group does not steal/ it is reasonable to suppose that
the mores of this group prohibit it and, moreover, that property
is somehow thought to be inviolate. Descriptive analysis identifies
the ethics of persons or groups inductively from observation of
actual conduct.
On the other hand, prescriptive or normative analysis takes
its cue from an articulated system of values—usually a formal
statement that affirms what is true, beautiful, and good—to ask
two kinds of questions: (1) In view of a given statement of what
is valuable, what actions are derivatively appropriate? If this
statement represents what we believe, how ought we to behave? (2)
Given a statement of what is valuable, are the actions we observe
appropriate? Are these modes of conduct coherent with this charac-
ter, are these actions congruent with these affirmations?
-------
I tell my classes about the case of an avowed and apparently
devout cannibal who is a practicing vegetarian! Descriptive analy-
sis says simply that this is an immoral person—he does not prac-
tice what he preaches, there is no congruency or coherence between
what he affirms and how he acts. Prescriptive analysis, on the
other hand, is more concerned with the predicates of the cannibal's
behavior; that is, whether cannibalism qua cannibalism is an appro-
priate or right or good philosophy. Depending on the answer to
this question, prescriptive analysis will tend to make one of two
judgments—either that cannibalism is a licit philosophy, in which
case, this "vegetarian cannibal" is confused or cannot be responsi-
ble to his conviction and therefore is not a dependable moral agent;
or that cannibalism is an illicit philosophy, and while this person
may be doing the right thing, he's doing it for the wrong reason.
In either case, a negative judgment is rendered in the absence of
clarity and continuity between creed and deed.
ETHICAL ISSUES AS MORAL DILEMMAS
To speak about ethical issues in clinical research is to engage
this same genre of questions and judgments. But in this context,
our attention is drawn to physicians (and possibly others) who (a)
are simultaneously acting under the governance of both personal and
professional ethical sensibilities, which suggests that conflict
may sometimes occur between these two, and who (b) are carrying out
'scientific investigations on patients who have presented themselves
for care and treatment, which also suggests that a disturbance may
be experienced, this time by patients who learn that they are ex-
perimental subjects. In addition, physicians may find it difficult
to reconcile their responsibilities as clinicians with their respon-
sibilities as investigators.
So, an "ethical issue" can be either or both of two sorts: it
can be identified as an incongruence or discontinuity between
affirmation and actions, or it can be identified as an inappropriate
affirmation or a mistaken value system. In both instances, an
"ethical issue" ordinarily presents as a moral dilemma.
In today's society, the ethical issues in clinical research
seem to be basically of the first sort, that is, as a tension or
distancing between belief and behavior. As one reads the literature
of clinical investigations—say, from Nuremberg to the present—it
is evident that the principal ethical questions ask what research
is appropriate under already established guidelines, and whether
investigators are, in fact, honoring those principles in both proto-
col and practice. Of course, innovations in therapy together with
curiosity generated out of previous studies constantly push against
the boundaries of those principles, and occasionally, there is talk
-------
about reformulating the goals and goods to which medicine is com-
mitted.
It has been suggested, for example, that cloning and cyborg-
androids offer the prospect of larger and less restricted pools
for "human" experimentation, and that we should move ahead with
technologies in anticipation of the benefits that would result
from their use. To cite only one instance: The care with which
the current National Commission for the Protection of Human Sub-
jects has proposed and promulgated guidelines for studies using
abortuses, prisoners, the mentally infirm, and others as potential
research subjects makes it fair to say, I think, that there is a
formal consensus in the U.S. concerning the values and obligations
of clinical investigations. The ethical standards for clinical
investigations appear to be settled if one takes these documents
to be normative statements.
THE ETHICS OF INVESTIGATION
What are the standard values and obligations of the medical
profession that govern clinical investigations? Their documentary
expression can be found (among other places) in the Nuremberg Code,3
in the Declaration of Helsinki," in the policies of DHEW regarding
the protection of human subjects,5 and in the AMA's "Ethical Guide-
lines for Clinical Investigation.»< A careful reading and content
analysis of these statements shows that their substantial points
focus on two principal interests: the study and the subject. More-
over, three constituent heads can be denominated under each of these
two governing titles: the sections that comment on the nature of
the study typically address its purpose, its design and conduct,
and its results; and the sections that speak to the role of parti-
cipants in the study accentuate the physician-patient relationship,
consent, and risk. All of these sections together constitute, in
this literature, the ethical issues appropriate to clinical investi-
gations, at least as perceived by those who formulated these state-
ments .
Study Guidelines
The purpose of clinical studies, according to the Nuremberg
Code, "should be such as to yield fruitful results for the good of
society, unprocurable by other methods or means of study, and not
random or unnecessary in nature." In Helsinki, the World Medical
Association declared that, "It is essential that the results of
laboratory experiments be applied to human beings to further scien-
tific knowledge and to help suffering humanity." The AMA guidelines
for clinical investigation, adopted in 1966, endorse these same
ethical principles. In sum, they seem to state that what initially
-------
legitimates clinical research is the need to test experimental
hypotheses on human subjects to extend scientific knowledge and to
alleviate human suffering.
In addition, this purpose must be acknowledged in the design
and conduct of the study. "A physician may participate in clinical
investigation," according to the AMA guidelines, "only to the ex-
tent that his activities are a part of a systematic program com-
petently designed, under accepted standards of scientific research,
to produce data which is scientifically valid and significant."
Moreover, both the Nuremberg and Helsinki statements insist that
the research should be designed and based on animal experiments and
the natural history of the disease, and conducted only by scienti-
fically qualified persons from whom "the highest degree of skill
and care should be required." The point of these admonitions is
that a study must be ethical in its design and conduct if it
intends to be ethical at all; that is, rather than judge the ethical
aspects of a study solely on a post hoc basis, there are possibili-
ties for ethical evaluation already present in a study's intention
and methodology.
Finally, these ethical guidelines hold that the results of the
study should coinhere with its purpose. If the goal of clinical
research is to test experimental hypotheses on human subjects to
extend scientific knowledge and alleviate human suffering, the
results of the investigation ought to show this. Sometimes con-
structive advances in knowledge and innovative therapies are
achieved, and sometimes it is discovered that the positive accom-
plishment of a trial is to show that a given hypothesis is incorrect.
In either case, the ethical burden is placed upon the investigator
to show that the results of the study are in keeping with its
purpose.
The Subject's Rights
The second principal interest that these documents address con-
cerns the participants in clinical research and the two kinds of
special consideration due them. Presuppositional to concerns for
consent and risk is the physician-patient relationship. Without
this, and without its being conceived in a certain way, there would
be little or no point in going on to talk about consents and risks.
The AMA guidelines state unambiguously that the investigator "should
demonstrate the same concern and caution for the welfare, safety
and comfort" of the experimental subject "as is required of a physi-
cian who is furnishing medical care to a patient independently of
any clinical investigations." The Declaration of Helsinki is, if
anything, more rigorous: "The doctor can combine clinical research
with professional care. • .only to the extent that clinical research
-------
is justified by its therapeutic value for the patient" because
"it is the duty of the doctor to remain the protector of the life
and health of that person on whom clinical research is being
carried out."
These notions of subjects' rights and physicians' duties are
reinforced in both the Nuremberg and Helsinki documents, with pro-
visions for investigators to terminate, and subjects to withdraw
from, experiments at any stage. What appears to be clearly at
stake, and what is clearly affirmed in all these documents, is a
sensitive and generous humanitarianism that imposes equally upon
the investigator and the subject. Neither can treat or regard the
other as merely a means to an end, because there is a fiduciary
relationship between them, which is an inviolable end and which
transcends the immediacy of any enterprise in which they may be
jointly engaged.
In this context, concerns for consent-getting and risk-assess-
ment make sense as logical extensions of the primary commitment to
humanitarian ideals. Thus, the Nuremberg Code states that "voluntary
consent of the human subject is absolutely essential," and the De-
claration of Helsinki asserts categorically that "clinical research
on a human being cannot be undertaken without his free consent,
after he has been fully informed."
Each of us could probably cite too many examples of the fail-
ure to honor this fiduciary relationship. I recently experienced
an especially poignant demonstration of this. I was on an airplane,
comfortably settled into an aisle seat and reading an issue of The
Lancet when, almost immediately after we were airborne, a woman
who was sitting across the aisle leaned toward me and asked "Are
you a doctor?" "Yes," I said, "but probably not the kind you mean."
"Then what kind of doctor are you?" she inquired. "A Ph D a
doctor of philosophy." "Well," she said, "maybe you can help me
anyhow. And then she proceeded to tell me her story.
in early childhood her husband had received a diagnosis of
cerebral palsy ; in manhood he had experienced grand mal seizures;
and now more recently, he had undergone episodes of uncontrollable
violent behavior during which he had hurt both himself and his wife.
She was now returning home after having admitted her husband to a
C6nter ^^ th Care Of a "eurosurgeon to whom
had b ** ~ * ~' ** ^ -ting
Thereafter, she was contacted by one of the neuro surgeon's
residents, and it was with him that she had a conversation about her
husband's situation. She did not learn very much from that confer-
8
-------
ence because, she said, "He talked funny and I couldn't understand
him." In response to my questions, it developed that the resident
also had "different eyes." We concluded that he was probably not
a native American. In any event, this woman did not understand
what she was told by the resident. She had no opinion about fault-
ing the young doctor for failing to communicate with her, and she
turned aside my question about the neurosurgeon1s responsibility
to discuss her husband's case by saying that "he is so busy...he is
very important and works very hard...lots of people like my hus-
band depend on him...he's just too busy."
This woman is a registered pharmacist, and I would have guessed
her to be in her early thirties. She had signed the admissions
forms and her plans were to return to the hospital in ten days, when
her husband's initial work-up and perhaps other preliminary or
exploratory procedures would be completed, to be present at the time
of his surgery. She planned to return to the medical center in the
early morning of the day when surgery was scheduled for noon; she
did not know anything about any consent forms, but did know that
she would be expected to sign some papers when she went back to
the hospital.
She could not describe what was going to be done to her husband
over the next ten days, nor did she know what surgical procedure
was scheduled at the end of that period. When I inquired whether
her husband's treatments, whatever they were, would be directed to
his cerebral palsy or his epilepsy, the wife's response was, "No,
it's for his violence." "Well," I asked, "what surgery is going
to be done?" "I don't know," she said, "but that's what I wanted
to ask you about. I saw something about 'thalamus1 and something
about 'frontal.1 Do you know what those mean?" I avoided answering
that question, and asked whether she was satisfied with what she
knew about her husband's treatment, and whether her questions to
me indicated that she needed to know considerably more than she did.
She acknowledged the force of both those questions, then added, I
thought somewhat plaintively, "But they wouldn't hurt him, would
they?"
This was an uncomfortable conversation for me—uncomfortable for
all the conventional reasons, to be sure; but more than that, it was
uncomfortable because I am familiar with the symptoms she described
and some aspects of the "innovative therapy" of "psychiatric neuro-
surgery" being undertaken in that medical center for the control of
violence. I did not want to alarm this woman, not only because I
am not a medical doctor, but also because I did not have all the
relevant facts before me. Nevertheless, I was disturbed by what I
suspected was likely to happen to her husband, and my uneasiness was
escalated by the knowledge that this woman and her husband lacked
-------
the most rudimentary information and understanding of the husband's
treatment. Without the necessary information and understanding,
they could not participate significantly in the decision-making
process. Preconditional to that, it was plain that the presumption
of a convenantal bond, a fiduciary relationship between patient and
physician, had been violated.
By now the airplane was beginning its descent, and I wanted
to let her know, without adding to her anxiety, that I thought
the way her husband's case was being handled was very inappropriate
and unsatisfactory. I suggested that she contact a local neuosur-
geon, explain the situation, ask him to contact the medical center
neurosurgeon on her behalf, and then have him explain to her her
husband's treatment. I also suggested that she make arrangements
to talk with the attending surgeon prior to surgery and before
signing any further forms. Even with these suggestions, I felt
frustrated and powerless. As she left the airplane, she expressed
her disappointment that I would not or could not say any more than
I had about what was likely to be in store for her husband.
THE PATIENT'S CONSENT—AN ETHICAL ISSUE OF THE 20TH CENTURY
I find it interesting that, while the moral traditions of
Western medicine contain provisions for a consent mechanism, the
prominent role of consent as an explicit issue in medical ethics
is a relatively recent one. In fact, there is no mention of con-
sent in the Hippocratic Oath, or by Maimonides in the 12th century,
or by other pre-19th century medical authorities. Indeed, there
is no apparent concern among classical authorities for a special
ethical obligation with respect to consent—and this is the case
for both experimental and established therapies.
The most persuasive explanation for this state of affairs is
two-fold: (1) The beginning of human experimentation is typically
identified with William Harvey's research in human circulation in
the early 17th century (1628); however, the great commitments to
research as a predominating direction of scientific medicine were
not expressed until the mid-20th century. So, while there has
always been curiosity in the clinical setting, it has only been in
relatively recent times that clinical investigation has achieved
such high priority and institutionalization. (2) Concurrently
with this phenomenon of the socialization and politicization of
medicine, an intellectual attitude developed in Western culture,
which had as its primary objective the yielding of information
through systematically designed experiments. This attitude, in
turn, provoked conflict and competition with an older principle
of primary patient benefit from medical intervention.
10
-------
World War II, and the Nazi medical experiments in particular,
exposed the weakness of experimental ethics that were based either
solely or largely on social utility or scientific advance; and
Nuremberg, although it focuses chiefly on experimentation, has
become the landmark in the evolution of consent as a prominent
issue in the ethics of health care. (I want to add, parenthetically,
that the force of the Nuremberg Code is unquestionably directed
toward the protection of patients and subjects, although it was
propounded to guide physicians in carrying out studies on human
subjects. In recent years, however, and in the wake of wide
exposure of unethical studies and malpractice litigation, we are
observing a subtle shift in the intention of consent-mechanisms
from patient-subject protection to physician-investigator protec-
tion. )
Since Nuremberg, four types of consent situations have been
plainly identified: (1) the therapeutic setting, in which treatment
is specifically and exclusively directed to the benefit of the pri-
m patient (e.g., appendectomy); (2) yet another therapeutic
setting, but one in which the benefit of treatment is specifically
and exclusively directed toward another recipient (e.g., tissue
and/or organ transplantation involving a living donor); (3) the
experimental setting, in which general or specific information is
sought by procedures that are unrelated to a primary patient's
(or subject's) care, but which may benefit others and/or increase
medical knowledge (e.g., mass screenings for hypertension or trans-
mission of hepatitis); and finally, (4) permutations or combinations
of the preceding situations, in which a treatment of unknown efficacy
and safety is administered both for possible benefit to patients
(or subjects) and for potential extension of medical information
(e.g., new chemotherapeutic agents for carcinoma).
In all of these settings, the implementation of adequate con-
sent-getting has been left primarily to physicians. Indeed, the
Nuremberg Code states explicitly that consent procurement "rests
upon each individual who initiates, directs, or engages in the
experiment. It is a personal duty and responsibility which may not
be delegated with impunity." The literal meaning of that last
phrase, "with impunity," is "with freedom from punishment or pen-
alty." It is my experience, by and large, that consent-getting is
nevertheless regularly delegated to either nurses or house officers.
This practice presents the kind of ethical issue, which in my judg-
ment, is our foremost concern—a mode of behavior that contradicts
the stated ethical principle.
The consent process is complex enough in itself without this
kind of obviously faulty procedure that could be easily remedied.
Since Nuremberg, it has been generally acknowledged that a valid
11
-------
consent rests upon and consists of three elements: information,
freedom, and competency. Thus, a consent is valid if it is
secured from a patient or subject who is knowledgeable, who
agrees voluntarily, and who is compos mentis (or who, in the case
of legal or mental incompetency, is represented by a guardian).
In this context, I would like to plead for abolition of the phrase,
"informed consent," because it misrepresents (both intentionally
and grammatically) what is required in a valid consent.
But even if this misanthropic phrase should disappear from
the medical lexicon, consents will remain problematic because it
is difficult (perhaps, in an absolute sense, impossible) to assess
precisely and accurately the extent to which any given patient or
subject is informed, free, and competent in the consent situation.
I have not found any consent situation (except, perhaps, for a
genuinely elective procedure) which unambiguously satisfies pro
forma the full range of interests signified by these elements?"
Moreover, I do not know of any single rule or protocol which in
itself is comprehensive enough to guard all the contingencies and
to guarantee adequacy.
I do not, however, conclude that we should abandon the concept
of valid consent or blunt our moral sensibilities sufficiently to
enable us to accept a more imperfect model. Rather— and presuming
the subject's competency, which among the three elements of valid
consent is customarily the easiest to certify— i believe that it
ought to remain the physician's personal and professional struggle
to 111 ti f Crite!ia f°r VaUd consent' *>y heightened sensitivity
to all the elements that comprise voluntary consent, and by strenuous
- -formation^ the
Ideally, the consent situation is a covenant between persons,
a fiduciary relationship, which is intended to guard and protect
while simultaneously opening to larger no^iHHi^ *-u Protect
humanity of the parties to I share^r^^ ~g the
consent situation is easily forgotten or neglected in the day-to-
day routinization of research, or in the enthusiasm ror a study?
But acknowledgement of the consent situation as a fiduciary
relationship, by both physicians and patients alike Jm
go farther than any formal requirement p
spirit and the letter of ^
RISK-ASSESSMENT
of the problem to be solve. by theee Ration
12
-------
of Helsinki similarly expresses its formal interest in risk-assess-
ment by stating in its "Basic Principles" that "clinical research
cannot legitimately be carried out unless the J"f r*J"°V?* *?•,,
objective is in proportion to the inherent risk to the subject.
However, none of the documents venture to define precisely
what constitutes an ordinate or inordinate risk, although the
Nuremberg statement does provide that "proper preparation should
be made and adequate facilities provided to protect the experimen-
tal subject against even remote possibilities of in3ury, ^ability,
or death." This commentary is not insignificant. Unlike the FDA
requirement, which states only that drugs must be "safe" and
"effective," but does not further explain or define these catego-
ries, the Nuremberg statement offers some clues about which conditions
constitute inordinate risk.
Diagnostic precision must be developed in the applicability of
clinical studies, just as there is need for technical and scientific
excellence in research design. Those human trials that have treated
indiscriminately the application of an innovative therapy have
tended to be the most disastrous. For example, the use of a special
high protein diet in the treatment of liver disease seemed to have
a sound theoretical basis and to be administratively innocuous, but
in the absence of the demonstrated benefit of this regimen, a decade
passed before it became apparent that many patients were dying in
hepatic coma as a result of this diet. That kind of risk appears
to be inherent in any experimental study, but my point in citing
this instance is to show that the risk need not be made inordinate
by accepting a study as a success before adequate evidence is in to
prove its validity.
Withal, three things seem clear: (1) the risk factor will vary
from study to study, depending on the nature of the disease, the
condition of the patient-subject, and on other factors that are
relatively unique to the situation; (2) the physician's initial
obligation—primum non nocere—is his minimal duty toward subjects,
and their protection "against even remote possibilities of injury,
disability, or death" is an appropriate extension of that basic
desideratum; and (3) a good research design will identify both
potential and/or anticipated risks and make provisions for unpre-
dictable and unexpected risks.
While these aspects of the ethics of human experimentation
deserve careful consideration in every study that involves human
subjects, they impact with special emphasis on research in the
clinical setting. Anyone with experience in both kinds of investi-
gations, i.e., when combined with professional care and when con-
ducted in a non-therapeutic setting, knows that the personal and
13
-------
interpersonal dynamics of these two settings can be, and usually
are, remarkably different. For example, there is a heightened vul-
nerability among subjects who are also patients. This calls for
uncommon sensitivity on the part of primary care physicians who are
also, in these cases, clinical investigators.
ETHICAL AND MORAL DILEMMAS IN THE RESEARCH SETTING
There are ethical issues in clinical research that, like the
poor, seem destined to be with us always. This is so largely be-
cause it is the task of ethics to probe behind the "what" and the
"why," and to reference action to an affirmative warrant by estab-
lishing the truth and value claims of that warrant. So long as
research behaviors probe and push against settled ethical bounda-
ries, so long will it be incumbent upon us to reexamine and reassess
our moral predicates. I know, of course, that remarkable and
gigantic strides have been made in scientific and technical achieve-
ment, and that most if not all of these achievements owe their
genesis to research and experimentation. But I also know that the
ethical dilemmas of society are not rooted there, and that the
moral crises that confront us do not emerge from the risks that
attend our scientific accomplishment or the perils that accompany
the promise of our technological triumphs.
Our ethical dilemma is actually a set of questions: Can we
agree upon a common set of values? Is there a firm consensus
among us as to what is true and good and beautiful? Can we
assume that we share common notions of duties and rights? And
our moral dilemma, correspondingly, interrogates our conduct: If
we can agree upon the principles of virtue, can we then be clear
and unconfused about the practice of virtue? I ventured to observe
at the beginning that the ethical issues that confront us today are
derived principally from incongruence or discontinuity between for-
mally adopted professional affirmations and the embodiment of these
ideals in the research setting, if that is 8Of our ethical predica-
ment represents something of an inversion of the experimental hypo-
thesis; in this instance our problem is not that "we need to know
more in order to do better" but that "we don't do as well as we
know."
14
-------
REFERENCES
1. Myrdal, G.: An American Dilemma. (2 vols.) Harper, New York,
1944.
2. Cf. 39 Federal Register No. 165 (23 August 1974), 40 Federal
Register No. 154 {8 August 1975), et seq.
3. The Nuremberg Code. _In Trials of War Criminals before the
Nuremberg Military Tribunals under Control Council Law No. 10.
(Vol. 2) Washington, D.C.: U.S. Government Printing Office,
1949, pp. 181-2.
4. World Medical Association: Declaration of Helsinki: Recommen-
dations guiding medical doctors in biomedical research involv-
ing human subjects. In Ethics in Medicine: Historical Per-
spectives and Contemporary Concerns (Reiser, S.J., Dyck, A.J.,
and Curran, W.J., eds.). Cambridge: The MIT Press, 1977,
pp. 328-9.
5. Cf. 39 Federal Register No. 165 (23 August 1974), 40 Federal
Register No. 154 (8 August 1975), et seq.
6. Ethical Guidelines for Clinical Investigation, adopted by the
House of Delegates, American Medical Association, Proceedings
of the House of Delegates, pp. 189-190, November 30, 1966.
15
-------
Discussion Summary
Conference participants raised three principal questions fol-
lowing the presentation of Dr. Smith's paper.
First, who should an investigator consult (besides himself)
to assess accurately the degree of risk and the scientific merit
of a research proposal? It was suggested that there is no set
formula to determine what the risks of a particular study might be
to the research subject.
In essence, it is the principal investigator who must assume
responsibility for selecting the appropriate panel to assess a
proposed study's risks. The principal investigator himself is
certainly among those to assess the study's more scientific or
technical risks, in addition to its scientific merit; but other
experts also typically are invited to assess both the degree of
risk and the importance of the study.
The second question dealt with the subject's motivation to
participate in a study. Historically, one of the most important
thrusts in clinical research has been that the benefit of research
accrues to the subject. Today, the research subject is frequently
less able to benefit directly from the research in which he is
involved. Thus, whatever benefit the research subject gains is via
his membership in society. Given these factors, what motivates an
individual to become a research subject?
In large-scale studies, the question of personal benefit usu-
ally is not raised between the research subject and the investiga-
tor. Rather, the questions that are typically discussed deal with
the issue of personal safety and whether certain people ought to
be allowed to participate in certain studies. Thus, the consent
mechanism looms large.
17
-------
For example, on the one hand, consent forms should protect
subjects from poorly designed, unsafe experiments. And on the other
hand, consent forms should ensure that the subject is well aware
of his or her involvement in the study. Thus, from one subject's
perspective, the emphasis tends to be on protecting the study sub-
ject, rather than on assessing the personal benefits to be gained
from participating in a research study.
A final question focused on the dividing line between social
utility as a justification for clinical research, and the subject's
rights to protection and safety. Discussants agreed that there is
no formal prescription with which to determine where this imaginary
dividing line lies. It was noted, though, that some countries em-
phasize the social benefit to be gained from research using human
subjects over the individual's rights.
18
-------
Legal Aspects of Using Human Subjects
in Environmental Research
Michael V. Mclntire, J.D.
Santa Monica, California
In this presentation, I will discuss the principles of law
that govern liability of one person to another when something goes
wrong, as those principles might apply to persons using human sub-
jects for environmental research. At the outset, I must tell you
that there is no series of legal rules, no "do's and don'ts" for
you to take note of and follow so that you can live happily and
judgment-free ever after. The legislatures of the state and fed-
eral governments have not addressed themselves to your legal prob-
lems concerned with using human subjects in environmental research,
nor have the appellate courts, as far as I know. Your problems,
to use a hackneyed phrase, aren't exactly a household word. They
are not problems that have excited the press or the public. The
result of this absence of pre-established legal rules and guidelines
means that you are on a frontier of the law. Like the pioneers who
pushed westward beyond the reach of the settled communities with
settled laws, your scientific endeavors have likewise moved you in-
to an area where the law has not yet reached. So, I have this
warning for those who came with pencils poised to record a series
of rules which, if followed, will enable you to pursue your research
free of significant legal concerns: You are about to be disappointed!
Instead, my message to you is this: As regards the legal implica-
tions of your activities, you are going to have to adjust to the
fact that you will be living with a great deal of uncertainty for
many years to come*
19
-------
When I say the law is uncertain, I don't mean it is nonexis-
tent. It is not as though you are out in trackless/ weightless
space, starting from scratch. The past, it is said, is the pro-
logue to the future. The frontier communities were guided by
the laws and the rules of the communities they left behind when they
drew up the new laws and rules that would shape their future. So,
we too can get some understanding of the legal problems that lie
before you by examining the development of the law in areas that
are related to your own legal concerns. What I propose to do is
review some developments of the law in areas that the courts and
legislatures would be likely to look to for guidance. In doing
so, I hope I can help you to understand how and why you, as
scientists, may become involved with the law and with lawyers.
A RESEARCH SUBJECT IS INJURED—WHO'S RESPONSIBLE?
Let's assume a hypothetical situation as the framework from
which to discuss your legal rights and responsibilities in the
event that a human subject of your research suffers injuries dur-
ing the course of, or as a result of your experimentation. A man
in his late thirties volunteers to be a subject in your research
program. The man is interviewed carefully by your staff. He is
told that he will be exposed to sulfur oxide gases in the air he
breathes, and that he will be asked to perform certain physical
activities, during which certain bodily functions will be electri-
cally monitored. The volunteer is told that the combination of
sulfur oxides that he will be exposed to is 10 percent higher than
the amount normally present in the ambient air in a highly pollut-
ed urban area, say, Gary, Indiana.
The volunteer is told that the effect of this amount of gas
in the air on humans has not been studied, and, as a result, the
effect on the volunteer cannot be predicted. The volunteer is
told that he is taking a risk for the benefit of scientific
research. He indicates that he understands, and is willing to do
so. The volunteer agrees to accept $4.50 per hour for a 48-hour
period, during which he will be exposed continuously to control-
led amounts of sulfur oxides. When the experiment has been under
way for about 36 hours, the volunteer becomes violently ill and
collapses.
We've said that there is little law to tell us who is respon-
sible for what in this precise situation. But what lessons are
taught by our historical review of the law in related fields?
The first lesson is one that you don't have to be a lawyer to
understand and that is, the law is a great Monday morning quarter-
back. The absence of definitive legal guidelines does not prevent
the law from second-guessing your judgments. In fact, hindsight
20
-------
is one of the mainstays upon which the law is based. So, lesson
number one is: "The absence of an existing legal standard does not
prevent you from being judged against a standard that has developed
after the fact." A second general lesson is that, even if there
were some court decisions relating to your case, they would not
remove the uncertainty as to the legal consequences arising from
a specific situation. Generalization is difficult because the
rights and responsibilities depend so much on the peculiar facts
of each case. Furthermore, different courts may decide the same
issue in opposite ways.
For example, in 1950, the New York Court of Appeals (which
is what New York calls its Supreme Court) held that the manufactur-
er of an onion-topping machine was not liable to a worker whose
fingers were cut off when his hands caught on the machine's revolv-
ina steel rollers. Although the injury could have been prevented
by an inexpensive guard or a shut-off device within the operator's
reach, the Court ruled for the manufacturer and against the worker,
because the hazard was obvious. The Court said that the manufactur-
er had no duty to protect the worker from obvious hazards. How-
ever, in 1966, an Illinois Appellate Court held that a manufactur-
er was liable for injuries caused when the operator's hand and arm
were~drawn into the unguarded rollers of a corn-picking machine.
The Court said that the design of the machine,fwithout guards against
obvious hazards, was "unreasonably dangerous."
In 1970 the California Supreme Court held a manufacturer of
an earth-moving machine liable for the death of a worker who was
run down by the machine because the machine's engine box prevented
the machine operator from seeing clearly, despite the fact that
the "blind spot" was obvious to the machine owner. I suspect
that if the New York Court were now to decide the onion-topping
case, its decision would be in line with the decisions of the
cases in California and Illinois. So, lesson number two is: "Court
decisions won't necessarily remove the legal uncertainty that sur-
rounds your actions."
THE LEGAL BASIS FOR LIABILITY
I want to discuss some of the specific problems created in
our hypothetical situation, but before I do, I have another dis-
claimer. I am going to talk about substantive issues, not the
procedural issues lawyers often raise to prevent the real issues
from ever getting decided. For example, I am no£ going to discuss
the statutes of limitations, which enable a court to avoid deciding
the substantive issues because the lawsuit wasn't filed soon enough.
Nor will I discuss sovereign immunity—the rule that prevents some
governments from being sued without their consent.
21
-------
Returning to our hypothetical example, let's assume that the
equipment functioned perfectly, but the operator carelessly
allowed too high a concentration of sulfur dioxide to build up, or
that he failed to observe warning signals, or otherwise "goofed,"
and that the operator's negligence caused the volunteer's injury.
Under those assumed facts, the operator's employer and the opera-
tor are liable to the volunteer. Liability is based on the con-
cept, as old as American jurisprudence, that a person is respon-
sible for his mistakes that cause injury to others, if he knew,
or should have known, that others might be injured. This is the
so-called "fault concept" which, although under fire, is strongly
entrenched in American law.
Now, let's assume that the operator was alert and following
instructions but because of inadequate instruction, he did not
know that the readings on the monitors measuring the volunteer's
vital signs had exceeded safe limits. Under the fault concept,
the laboratory would probably be liable. The laboratory was
negligent in failing to properly instruct the operator. So,
under the fault concept, before the laboratory can be liable, the
laboratory, or its employee, must have been negligent, and the
negligence had to be a substantial contributing factor to the
injury suffered.
However, the fault concept is not the only basis for the
laboratory's liability. Another basis for liability is the con-
cept that is often called "strict liability" or "liability with-
out fault." Let's assume that the volunteer was injured because
a monitoring device which would have warned the operator to reduce
the supply of sulfur oxide gases did not operate properly. Assume,
for discussion, that there was no way that the laboratory could
have known about the defect in the monitoring device. In this
case, the laboratory is probably liable to the volunteer, even
though the laboratory is totally without fault. The basis for
liability is public policy.
LIABILITY AS THE CONSUMER'S SAFEGUARD
In the last two decades, the courts, with the acquiescence
of state legislatures, have been attempting to protect individuals
from physical in3uries caused by the ever-increasing number of
SSStf^uiS^ te?hr109iCal society. The rules of strict
liability, or liability without fault, were developed to protect
consumers from goods that have a defect that could cause injury.
But these rules have been expanded rapidly into other situations.
For example, xn most states, it is now law that a person who sells
a defective product that is unreasonably dangerous to the consumer
is liable for physical harm caused to the ultimate user, even
22
-------
though the seller exercises all possible care in the manufacture
and sale of the product. I strongly suspect that these principles
of law will be applied as well to determine the liability of an
environmental research laboratory to its volunteer human subjects.
The trend of the law has been to extend the rules of strict
liability to allow recovery by persons who are not consumers or
users of a product, and also to hold liable persons who are not the
manufacturers or sellers of a product. The California Supreme
Court allowed a person who was injured in a head-on collision with
another car, to recover from the manufacturer of the other car
whose defect caused the accident. Similarly, the Michigan Supreme
Court allowed a bystander injured by an exploding shotgun shell to
recover from the ammunition manufacturer.
On the other end, the courts have been allowing recovery against
persons who dispense commercial services, as well as against per-
sons who sell or manufacture a product. An example is the designer
and constructor of a plant, who was held liable for injuries caused
by an explosion that resulted from an improper repair of a tube in
a heat exchanger manufactured by someone else.
Whether the rules of strict liability apply to persons render-
ing "professional" services is still an open question, as is illus-
trated by three cases arising out of New Jersey in the three-year
period between 1967 and 1969. In 1967, in a leading case, a New
Jersey Appellate Court held that the strict liability doctrine did
not apply to a dentist sued by a patient when a hypodermic needle
broke off in the patient's jaw. The Court held that strict liabil-
ity principles do not apply to persons rendering "professional"
services.8 A year later, in 1968, a federal court applying New
Jersey law held that strict liability was not applicable to a company
that designed, engineered, and supervised the initial operations
of a chemical plant, when an employee died after inhaling lethal
dust generated by the plant's operation. The Court characterized
the company's acts as "professional services," and on that basis,
held that strict liability did not extend to the company.
Then, in 1969, the New Jersey Supreme Court held a beauty
parlor strictly liable for services that caused burns to the scalp
and hair of a patron of the shop. Distinguishing the case of the
dentist decided just two years earlier, the Supreme Court said the
dentist was rendering "professional" services, while the beauty
shop was rendering "commercial" services. In my opinion, this
purported distinction between "commercial" and "professional" ser-
vices is an illusory one that is difficult to justify. I suspect
that in this age of consumerism, the case will soon come along
that will cause some court to inter that distinction in the legal
graveyard, along with other discredited legal fictions.
23
-------
LIABILITY IN THE LABORATORY
A strong argument can be made that the rules of strict liabil-
ity should be extended to apply to an environmental testing lab-
oratory. The reasons the courts give for imposing liability on a
manufacturer or seller of a defective product apply equally as
well to the operator of a laboratory conducting tests using human
subjects. Let's look at the reasons for imposing liability with-
out fault/ and then apply them to the environmental research lab-
oratory .
In a pioneer product liability case, the California Supreme
Court held that a manufacturer of a power lathe was liable for
injuries caused to the user of the machine because the machine's
defectively designed set screws worked loose while the machine
was operating. The Court said, "The purpose of such liability is
to insure that the costs of injuries resulting from defective prod-
ucts are borne by the manufacturers that put such products on the
market rather the injured persons who are powerless to protect
themselves."1l
Two years later, the Illinois Supreme Court held the manufac-
turer of a truck's brake system liable for damages suffered by the
occupants of a bus that collided with the truck when its brakes
failed. The Court thought that imposition of liability without
fault was justified by the following arguments:
• Public interest in human life and health call for
the utmost legal protection.
• The manufacturer who solicits and invites the use
of his product, by advertising and otherwise,
should bear the loss caused by such use.
• The loss caused by a defective product should
be borne by those who created the risk and
reaped the profit from it.12
Other reasons advanced in support of the rule imposing liability
without fault on the manufacturer or seller of a defective product
are:
• Liability is an incentive to make safer products.
• The manufacturer and seller are in a better position
to discover the defect than is the consumer.
• The manufacturer and seller can insure against the
24
-------
risk, and recover the cost of insurance through
the price structure.
On examination, all of these reasons apply as well to the
environmental research laboratory that injures its human subject,
as they do to the manufacturer of a product that injures a user*
The human subject is generally powerless to prevent an injury sus-
tained in a laboratory. The laboratory solicited the subject and
invited him to engage in the activity that caused the injury. The
laboratory, not the volunteer, profits from the activity that
caused the injury. The laboratory, not the volunteer, profits
from the activity. The laboratory is in a superior position to
discover the defects in the system and to insure against them.
And the rule imposing liability on the laboratory will promote
safety-consciousness and safer experiments.
Thus, a strong argument can be made that the liability rules
that govern manufacturers and sellers of products in the market-
place should apply to research laboratories whose research involves
human subjects. That being so, it seems appropriate to explain
what is meant by strict liability, or liability without fault.
THE OBVIOUS HAZARD: WHO'S TAKING THE RISK?
Liability without fault does not mean that liability arises
every time an injury occurs. For liability to arise, the injury
must have been caused by a "defect" and the concept of "defect" is
a complex legal issue. In general, a product is said to be defec-
tive if it fails to meet the reasonable expectations of the consumer.
Thus, the law does not impose liability when a consumer is injured
by ordinary risks that can be expected. For example, a restaurant
was not liable to a native New Englander for injuries caused by a
small fish bone in a bowl of chowder.1' Similarly, a shoe manu-
facturer was not liable to a customer who, while wearing the manu-
facturer's shoe, slipped on a wet floor in a laundromat, because
consumers know that shoes tend to become slippery when wet.1*
However, the fact that a danger is obvious does not necessarily pre-
vent the imposition of liability. A South Carolina court held a
seller liable for injuries caused to a child whose hand came in
contact with the unguarded blade of a power mower, even though
the danger was obvious.1*The Illinois case that I mentioned ear-
lier, involving the unguarded cornpicker, is another example of an
obvious hazard still subject to liability.
Thus, there is no clear rule for determining when liability
may be imposed in situations where the harm was caused by an obvi-
ous risk. The test seems to be whether the risk, or danger, was
unreasonable, considering the seriousness of the harm, the chances
25
-------
of its occurrence, and the ease with which it could have been pre-
vented. For example, a railroad may have no legal duty, and there-
fore, no liability to install crossing gates on a little used coun-
try road that crosses a single railroad track carrying three trains
daily. The risk of an accident is slim because there are few cars
and few trains. But the risk of an accident may be "unreasonable"
if the road is a highway, heavily travelled by high-speed automobile
traffic, and the gate could be installed relatively inexpensively.
So, in our hypothetical example, the human subject knows that
he is taking some risks. Therefore, the laboratory would probably
not be liable for a temporary asthmatic condition, or for minor
skin irritation, or for a sore throat caused by the experiments.
These risks seem to be the kinds of risks a subject should expect.
However, if the injury was permanent, i.e., lung damage, heart
damage, or even death, the laboratory may well be liable because
the risk of permanent health damage would probably be judged to be
an "unreasonable risk."..
HOW MUCH RISK DOES THE SUBJECT ASSUME?
Related to this problem of strict liability is the issue of
assumption of risk," which is often raised as a defense to liabil-
ity. The rule is that a person who voluntarily and unreasonably
proceeds to encounter a known danger cannot hold another liable for
the injuries he suffered as a result of his actions. "Assumption
of risk" means that the injured party knew of the defect creating
the danger, and appreciated the significance" of the danger to which
he was exposed.
«„.* ^°nSiderl for e***Ple, the case of the appliance repairman who
sued the manufacturer of a pressure bottle that exploded, injuring
the repairman. Despite the warning imprinted on the bottle, which
iss ^rr -rit-r SSL
the risk Tne !«„ £ t0 Pr°Te that <*• "P'lrman assumed
e££or°
0
in the refilled container. <"****•"
-------
experiment. But the same low payment also supports the volunteer's
argument that he did not intend to subject himself to the kind of
serious risks that injured him. Who among us would voluntarily risk
permanent health damage for $4.50 an hour?
Assumption of risk in products liability cases is tough to
prove. I think that in the research situation, assumption of risk
will be almost impossible to prove. A laboratory that injures a
human subject while doing research on that subject to learn the
extent of the danger of a substance will have some difficulty con-
vincing a jury that the subject knew, understood, and voluntarily
accepted that danger.
There is one other aspect of the law that should interest you.
If the defect that caused the harm was a defect in machinery that
the laboratory purchased elsewhere, then the laboratory can proba-
bly recoup the money it must pay the injured volunteer from the
manufacturer and the seller of the defective equipment.
In summary, then, a laboratory is liable for the injuries suf-
fered by a human research subject if the injury is caused by the
negligence of a laboratory's employees. Furthermore, the labora-
tory is probably liable, even in the absence of negligence, if the
injury was caused by a defect in the laboratory system, or in its
operation. How can you avoid this liability? The answer is, you
can't. You must learn to live with it. In the long run, the best
defense is to keep the safety of your human subject uppermost in
your concern. Fully disclose the risks as you see them, do your
best, and buy a good insurance policy.
27
-------
REFERENCES
1. Campo vs. Scofield (1950) 301 N.Y. 468, 95 N.E. 2nd, 802.
2' Wright vs. Massey-Harris, Inc. (1966) 68, Illinois Appeals
2nd 70, 215 N.E. 2nd, 465.
3- Pike vs. Frank G. Hough Company (1970) 2 Cal. 3rd 465, 467
Pacific 2nd 229.
4. Restatement 2d, Torts, Sec. 402A.
5> Elmore vs. American Motors Corp. (1969) 70 Cal. 2d 578, 451
P2 84 •
6> piercefield vs. Remington Arms Corn. (1965) 375 Mich. 85, 133
N.W. 2d 129.
7> yexas Metal Fabricating Co. vs. Northern Gas Products Corp.
(Kan., 10th Cir. 1968) 404 Fed 2921.
8- Maqrine vs. Krasnira (1967) 94 N.J. Super. 228, 227 A.2d 539,
*°m **aqrine vs- apaetor (1968) 100 N.J. Super 223,
o37»
u-
. (3a Cir 1968) 402
, 210
14.
(X967)
256 S.C. 490.
28
-------
Discussion Summary
Participants considered a situation in which neither the in-
vestigator nor the subject was aware of the subject's hypersensi-
tivity to a given substance, and subsequently, the subject collapsed
in the middle of the experiment. Who would be held liable in this
case? This question is one that has not yet really been answered.
Some state supreme courts have held that if there was no way that
the subject's hypersensitivity could have been predicted, the lab-
oratory cannot be held liable.
However, other courts, based on the information that one per-
cent of the U.S. population is hypersensitive, held that the experi-
menters should have issued some sort of warning to the subject prior
to the experiment, and thus, they may be held liable.
If the situation is such that the subject would not have been
injured had he not participated in the experiment, the laboratory
will probably be held liable.
Participants questioned whether review board members have ever
been held liable for quoting a subject's test data. In general, re-
view boards are not held liable in this situation unless the review
board itself is the entity conducting the experiment, as sometimes
happens•
Other discussion focused on the extent to which a funding agen-
cy can be held liable for a subject's injury. Is a funding agen-
cy e.g., the EPA, responsible if a subject in one of its funded
experiments is injured? A funding agency can be sued, but whether
the case will be won depends on the extent to which the funding
agency was involved in the experiment. A number of cases in other
areas of liability have exonerated funding institutions from liabil-
ity. For example, in the case where a bridge contracted by the
29
-------
Federal Department of Transportation collapsed, the Department was
not held liable because it had little to do with the construction
of the bridge, aside from a review of its size and location.
In a case where equipment furnished by the federal government
to the grantee is the cause of injury, the government could easily
be held liable, assuming the government agrees to be sued. In a
case in New York, a sailor was injured by an aircraft carrier cata-
pult constructed by a contractor to the U.S. Navy. When the sailor
attempted to sue the federal government, the Department of Defense
claimed sovereign immunity.
Although the court backed the Department's decision not to
be sued, it felt that the sailor should be able to recover damages
from somebody. So, even though there was no proven defect in the
catapult, the manufacturer was held liable because it was the only
source with money enough to pay damages. This is called the "deep
pocket" theory; that is, the court recovers damages from the party
with money enough to pay for the damages.
A final discussion was concerned with whether a laboratory
can be held liable for long-term adverse effects suffered by re-
search subjects. Whether a laboratory is found liable depends on
the severity of the experiment's aftereffect, and whether the sub-
ject was fully aware of all the risks to which he would be exposed.
If the long-term effect is not serious, chances are that the labo-
ratory W111 not be held liable because the subject knew he was ex-
posing himself to some element of risk. However, if a subject suf-
fers serzous consequences as a result of an experiment, the labo-
ratory may be held responsible.
the lonrt^16^' ^ ChanC6S °f a lab°^tory being held liable for
of time b~r !!°
-------
Informed Consent—Its Function and
Limitations
Charles E. Daye, J.D.
School of Law
University of North Carolina
Space limitations make it impossible to fully discuss all the
potential ramifications of the issue of informed consent as it re-
lates, or might relate, to environmental research. In truth, I
feel an even more distinct limitation in that I do not have a full
or detailed understanding of the kinds of research being conducted.
Accordingly, I had to make certain assumptions.
ASSUMPTIONS AND CAUTIONARY NOTES
General Assumptions
I understand that certain tests or experiments, which have to
do with the effects on human beings of certain kinds of environmental
conditions, are conducted under clinical and controlled conditions.
For example, I understand that individuals participating in your
research might be exposed to various levels of carbon monoxide in
order to determine certain human responses or responses of the human
body to such exposure. I fully understand that this is only an
example of the kind of research that is being conducted, but that
it is, in a general way, reasonably representative of the kinds of
research being done.
In addition, I made certain assumptions about why the medical
profession is concerned about the question of informed consent.
31
-------
In general, my assumption is that the informed consent question is
important because of the ethical considerations that environmental
research might raise. However, I make no pretense in that direction
and I will leave these issues to those more qualified to speak on
the subject of ethics. I do assume that this research is perform-
ing a valuable service for society and that researchers are inter-
ested in continuing research without incurring the societal dis-
approval that might come about if individuals participated without
having granted their informed consent. I further assume that doc-
tors are interested in the basic legal principles regarding informed
consent so that they may avoid, as nearly as possible, exposure to
legal liability for money damages to research participants. Addi-
tionally, I assume that to the extent that legal principles may be
unclear so that one cannot advise absolutely how to avoid exposure
to liability, that doctors nevertheless wish to have pointed up the
best ways to avoid any potential liability.
Even with these assumptions, I think it would be improper and
unwise for me to offer "legal advice" in its true sense about any
specific procedures or problems that have been or may be encountered
°meal ^^ First' * *> not consider myself an
»exDer»n
search ? ^ DeCt °f informed consent in the scientific re-
anv na ? T ^^^ ™ * haVe ind"^ed, I have not evaluated
any particular areas of research nor any methods employed to secure
have a £T ! ^formed consent. My perspective, then, is that I
can Jlir T Acquaintance with the legal doctrine, if we can
L lnf°rmed con^nt in general. I emphasize that I
C
do not orono nera. I empasze tat
Sms but w?it T C Solutions to «y ^nds of particular prob-
S'ma'v «iii russ general issues' &<*****. ™* prints as
ular ^at mv r ^""mental research. I emphasize, in partic-
suSs^te fL T^-f °Uld n0t ln any way be d^emed ^ adequate
by legal coun« t ^** ^^^ °f the informed ^sent problems
by legal counsel employed for that express purpose.
Some Further Cautj™.arv Notea
in thpaculacon ^ ^^ d°ctrine did not ariS
trine a^in^conSxt^f9^^ r— h' «» ^°
most particularly, the relations^ ^ Profession and involved,
The context out of whLJ III T J ? ^tW66n a doctor and his Patie
cations for our analysis. doctrine developed has certain impli-
First, since the doctrine aY-r,«,a ^ ^
patient relationship, generallv ^h u^ context of the doctor-
procedures. For our twraoaa* Jr Problem concerned therapeutic
as those procedures concerned' with^r^10 procedures ^ be defined
for "^ attemts medi
oaa*
as those procedures concerned' with^r^10 procedures ^ be defined
for perceived ills of a doctoJ l"^ attemPts to provide remedies
tor S Patient. The doctor-patient
32
-------
relationship has traditionally been one which we lawyers speak of
as "fiduciary." That word is simply a fancy way of saying that a
patient places a trust and a faith in his doctor.
Second, the doctor-patient relationship, in a therapeutic con-
text, may permit a doctor certain kinds of latitude that may not
be available to a scientist/researcher conducting research in a
nontherapeutic context. I refer specifically to a doctor's "thera-
peutic privilege" which might permit the doctor to withhold certain
kinds of information from the patient in certain kinds of situations.
For example, a doctor is not required to disclose a risk to a pa-
tient when disclosure of the risk to the patient would be contrain-
dicated in light of the patient's medical condition. For the pur-
poses of this paper, I have assumed that a scientist/researcher
would not have such a privilege.
The third consequence of the nontherapeutic situation relates
to the fact that persons participating in nontherapeutic research
do not have an expectation of a cure or other remedy to some per-
ceived medical ill as an inducement to participate in research nor
as a justification for his participation. Accordingly, I have
assumed that in a nontherapeutic context, the inducement to a per-
son to participate in research is some reason other than an expec-
tation of a direct, personal, medical benefit. This may have cer-
tain consequences when we speak in terms of factors that might in-
duce one to participate in an experiment. This, in turn, may bear
on issues such as voluntariness and free choice.
Also in the therapeutic situations there are various kinds of
exceptions to a physician's or medical person's duty to fully dis-
close relevant information, for example, in the treatment of mental-
ly incompetent persons, insane persons, persons under the influence
of drugs, persons subject to a "medical emergency," and the like.
Additionally, legal issues may be raised if participants are chil-
dren or minors. Since these are specialized issues I have omitted
them from my discussion.
To recapitulate, I have made the following assumptions:
1 that environmental research is nontherapeutic research in
which no benefit to a participant in the research in the form
of a direct medical remedy is expected,
2 that this environmental research is directed toward the
pursuit of important societal objectives, but that these are
benefits that would flow only indirectly to a research
participant, and
33
-------
3. that research subjects are in all respects legally
competent to grant consent, i.e., persons of sound mind
who are of sufficient age to personally grant consent.
Finally, I must caution that the law of informed consent
that I shall discuss is not uniform in all particulars. It is
not uniform because it is being developed simultaneously in 50
different states by state courts applying what we lawyers call
the "common law," and in several instances, by state legislatures.
Thus, to speak of "the law" of informed consent is inaccurate.
Nevertheless, there are certain general principles to which we
can refer with fair accuracy for our purposes. But complete
accuracy would require analysis of the cases and laws of the
states where the research is conducted.
HISTORICAL PERSPECTIVES ON THE CONSENT ISSUE
With these assumptions and cautions in view, let us now turn
to certain historical perspectives on the consent question, since
this may "inform" our consideration of its application in modern
contexts, such as the scientific research context.
No doubt, the foundation from which we derive our current doc-
trine of informed consent can be traced rather directly to the law's
very early recognition that "every human being of adult years and
sound mind has a right to determine what shall be done with his own
body." This statement, which is general in the law, recognizes the
human being's right to self-determination and personal autonomy.
Accordingly, it was very early in its development that the law
recogmzed that any interference with personal integrity or with
the inviolability" of a person's body is a violation that the law
is prepared to remedy by assessing damages. As a corollary of this
right, there naturally had to be developed a doctrine stating that
it a person consented to an invasion of his personal integrity,
otherwise would constitute a legal wrong, the very existence
it CnnR^t precluded the invasion from being wrongful. As
off,™ <- ,wewcan quickly see that a punch in the nose is an
offense to one's bodily dignity if it occurs between strangers on
l*ll III C°rne^. c°nversely, we can quickly recognize that precisely
i-Jlcl t. SclIuG PUTlCn in t" n*» nnea T*V,M.H.JJ-J_I - . j^y«
ma* v, j when it takes place in a lawful boxing
It was nrS n°tiexp°se the *ers°n doing the punching to liability
inteari?v t^t Y ^ f^" C°nCept °f **™°™1 ^^ and ^^
knowiJ «lII ^^ t0 d d°ctrine in the iaw of torts that is
"assault"} T!rY..J°ruS?metimeS inaccurately referred to as
•Mrt«=,i +. vi black letter" definition of battery is "any
**• tovic ri.ii\cr of 3 v^rt^v^A *• T »v* 4 v». j i
sonable - ° harmful, or offensive to
34
-------
The ingenuity of human beings to vary their conduct naturally
required further development of this legal doctrine. For example,
consider the case of a person who went to the hospital for a check-
up, who was approached by a doctor, placed under anesthesia, and
then operated on. For example, let us say that the person's appen-
dix was removed. Now clearly this is distinguishable from a punch
in the nose between strangers on a street corner. But is it never-
theless an invasion of the person's bodily integrity if the person
was not in the hospital for the purpose of having his appendix re-
moved and had authorized no one to undertake such a procedure? The
problem is that the doctor in question was not actuated by any evil
intent or desire to do harm. That, however, was not essential.
The courts had no trouble finding that the operation in those cir-
cumstances was an invasion of the person's bodily dignity, because
the "invasion" occurred without the consent of the person on whom
the operation was performed. That person could sue the doctor in
a torts action for battery. In other words, because the touching
was not consented to and because in the absence of consent it was
wrongful, it exposed the one who did the touching to liability.
Similarly, we can imagine instances in which a man represented
himself to a woman as "a doctor," secured her consent to a physical
examination, and proceeded to perform that examination. The woman
subsequently discovered that his statement that he was a doctor was
true, but that in fact he was a Ph.D. and not an M.D. When this
situation arose, the courts had little trouble in finding that,
of course, consent to the touching had been granted, but that it
had been granted pursuant to fraud or misrepresentation and was
completely ineffective. Naturally, that would lead to the liability.
It may be instructive to point out, based on the preceding hypothe-
tical^ that it is immaterial whether the patient operated on in
fact needed his appendix removed or whether the doctor who removed
the appendix had done so with every degree of skill and care.
Similarly, it is irrelevant whether the Ph.D. knew how to conduct a
physical examination and did so with every possible skill. The
nonconsensual removal of a bodily organ, even diseased, would vio-
late the very principle that without consent the touching was un-
authorized, and without any consent to the touching, it was unlawful.
So too would the physical examination be considered wrongful where
the consent had been obtained under false pretenses.
Over the years, other situations developed. The classic case,
perhaps, is that of a woman who went to the hospital and consented
to an operation on her left ear. When placed under anesthesia for
that purpose, the physician discovered that her right ear needed
the operation more than her left, and proceeded to ignore the left
ear and operatfon the right ear. Without any allegation that the
operation was not performed skillfully in every respect, the
35
-------
plaintiff sued the doctor and contended that he had no consent to
operate on her right ear. The court agreed. It held that the
operation on the right ear was indeed an unlawful invasion of the
plaintiff ' s body , and was a battery because it was done without
the plaintiff's consent. The question naturally could not be
avoided whether indeed the plaintiff had not consented to an opera-
tion of the precise character as the one performed, except on another
part of her body. It could be suggested that the plaintiff had
given consent to an operation, which was true. The consent, how-
ever, was limited to an operation on the left ear. Out of this we
derived the legal principle that consent may be granted but that
the consent must not be exceeded.
Then cases arose in which, for example, a patient had consent-
ed to an "exploratory operation," and that in the course of the
operation the doctor removed a diseased organ. Having consented to
the "exploratory operation," the issue naturally arose as to whether
the plaintiff also had not consented to the removal of the diseased
organ. The doctor thought "exploratory" meant "if you find a disease
do what in your medical judgment seems indicated." The plaintiff
thought "exploratory" had its common meaning of "look and see."
Since patients are medically untrained, the courts came to the posi-
tion that it was up to the doctor to make sure the patient under-
stood what was meant if the terms used had meanings other than those
that untrained lay persons ordinarily would accord to them. In many
cases, except those in which life or serious consequences might be
threatened by failure to remove the organ discovered to be diseased,
the courts did not hesitate to find that the removal of the organ
without the patient's consent did indeed violate the principle that
every person had a right to determine his bodily integrity. Gener-
ally speaking, the courts that looked at the problem in its early
stages considered this matter to be a battery.
Then a series of cases arose in which the patient consented to
a medical procedure, and that procedure was carried out with skill
and competence, but, for example, the incision did not heal, some
S^ r9^3 ^^ dUring the 'Potion, or the patient con-
tracted a disease as a result of the operation. The courts had
some trouble with this problem in that the patient had consented
resulL thatSb0?h°^dUre undertaken< ^ * had not produced the
wnile each Pftti
-------
he was aware, vitiated the consent granted by the patient. Upon
holding that the consent was not effective, it followed that a
battery had occurred.
Beginning around 1960, however, courts began to recognize
that the problem did not really involve an invasion of the bodily
integrity without consent. Nevertheless, they were mindful of the
notion that originally gave rise to tort of battery—that an
individual ought to be able to exercise self-determination with
respect to his own body. Accordingly, the view began to be devel-
oped that the real problem was not the absence of consent but that
that consent had been granted upon inadequate or insufficient in-
formation about the risks or potential complications of the treat-
ment. The courts found implicitly in these situations that the
patients had relied on their doctors for information and the doctors
had failed to provide enough information. In this light, the prob-
lem did not look like battery, which we saw earlier as a punch in
the nose between strangers on a street corner, but more like a prob-
lem arising out of the nature of the relationship between the doctor
and the patient.
When the doctor failed in that relationship, in areas not
involving an absence of consent, the failure was treated as "mal-
practice." Malpractice was in another branch of the tort law called
"negligence." This treatment of the problem had certain legal con-
sequences which I will omit since they are not directly germane to
our discussion. Suffice it to say that the courts began to impose
on doctors a duty, premised on their relationship to the patients,
to inform patients of all the material information that would be
necessary to enable patients to make an informed judgment as to
whether or not to undergo a medical procedure. While- not every
court would articulate the analysis this way, in general, we may say
that courts understood that the physician was an expert and the
patient was not, that patients continued to have an exclusive right
to exercise control over their own bodies, that consent premised
on inadequate information was not fully informed, and that the
patient was in complete dependence upon, and must trust in, the
physician to provide information necessary to an intelligent and
informed decision.
Based on these considerations, one of the leading cases on
the subject decided that the law should impose a requirement upon a
physician to present to his patient all information relevant to a
meaningful decisional process. Of course, the physician, by reason
of his training and experience, could make an evaluation satisfactory
to himself or herself. But the court's judgment was that the deci-
sion was not the physician's to make. It was the patient's!
37
-------
It was in this context that the failure to disclose relevant
information was treated as a malpractice problem, that is, a
breach of a physician's duty to his patient.
DEVELOPMENT OF THE INFORMED CONSENT ISSUES
General Statement of the Doctrine
A general statement of the doctrine that has come to be known
as "informed consent" may be attempted. The informed consent doc-
trine provides that a doctor must provide a patient with sufficient
information to enable that patient to make an informed and intelli-
gent choice as to whether to undergo a particular medical procedure.
When necessary to an informed and intelligent choice the information
provided, generally speaking, must include the nature and magnitude
of the direct risks attendant to the procedure, the methods and
likely results of the procedure, the alternatives, if any, to the
procedure contemplated, and the nature and magnitude of any collat-
eral risks associated with the procedure. Failure to provide such
information will make the doctor subject to liability if a risk
which was not disclosed, but which could have been disclosed, results
in harm to the patient, if the risk were such that if disclosed the
patient would not have consented to the procedure.
Elements of the Doctrine
Having once made a general statement such as that above, it
nevertheless becomes necessary to identify the precise parts or
elements involved in the general statement and then we must attempt
to apply those elements in the context of scientific research.
Upon analysis, we will notice that the doctrine requires (1)
that certain information be provided, (2) that the person granting
the consent have an ability to understand and appreciate the nature
of the information provided, and (3) that a willingness to undergo
the procedure be voluntarily expressed. Accordingly, we may speak
of the elements as "information providing," »use of information,"
and "voluntariness" of consent.
With respect to information providing, in the context of medi-
cal research, it would appear clear that information regarding the
procedure and methods of the research must be provided. But that I*
4, doctrine of ^formed consent relates particularly to
ab°Ut the known ri**s to which one may
4,
bfexoo^dT °f in!f mati°n ab°Ut the known ri**s to which one m
be exposed upon participating in the research. Any known risk of
™ Y **** BU8t be Pro^d robabl in every
instance Y *** BU8t be Pro^d, Probably in every
38
-------
However, it appears that more information needs to be provided.
Probably every potential adverse consequence of significance ought
to be disclosed. However, in the context of environmental research,
it is precisely the purpose of the research to identify risks that
may be involved in a particular kind of environmental exposure.
Thus, it seems that a clear and positive statement should be made
of the extent to which risks of the research are unknown, together
with a statement of the researcher's best judgment as to the nature
of such unknown risks, as well as the magnitude of the possibility
of any of the unknown risks coming to fruition. It clearly seems
possible that scientific research may be involved with risks that
are unknown and that, in the exercise of ordinary care on the part
of the researcher, cannot be known. Thus it seems very likely
that a failure to inform any participant in such research that there
may be risks that are unknown would constitute a failure to provide
adequate information.
The second element dealing with the subject's ability to use
the information provided relates simply to making the information
available in a form that is understandable and comprehensible to
the subject participating in the research. For example, if the
subjects are lay persons, the use of scientific terms that might
not be understandable ought to be avoided, or if not avoided,
scientific terms ought to be defined in lay language. For example,
there is a case involving a doctor who informed his patient that
he wanted to do certain kinds of exploratory surgery for cancer of
the breast. At the hospital the patient was given a standard form
which indicated the procedure to be a mastectomy. The patient,
remembering the doctor's description of exploratory surgery and not
understanding the definition of mastectomy, signed the form. Dur-
ing the operation the doctor discovered that indeed the breast was
cancerous and removed it. The court held that the doctor could not
rely on the patient's signing of the standard form of consent, not-
withstanding its use of the term mastectomy, because the patient
had no understanding of that particular term. It went on to hold,
therefore, that her consent was not an informed consent.
This is instructive in two respects. First, as indicated, the
information must be provided in a form that is usable in light of
the researcher's knowledge of the subject's education, training,
language limitations, and similar factors. Second, the mere signing
of a form will not necessarily protect the doctor, although the form
itself in all respects may be adequate. This may be because the
subject does not understand the terms in the form, or perhaps,
because the subject did not properly or attentively study the form.
In such a case, a serious question will be raised as to whether
the consent granted was informed, notwithstanding that the subject
signed a standard form of consent.
39
-------
At this point I would emphasize that a form is a limited device.
Generally speaking, a form is nothing more than a memorial of the
meeting of the minds of the researcher and the subject. It is no
substitute for the direct, personal provision of information. This
does not mean that signed forms are not important. They are. They
may be evidence of the information provided as well as the consent
granted. They should not, however, in all cases, be treated as a
substitute for a more effective method of providing information.
The third element of the doctrine of informed consent relates
to voluntariness. Voluntariness connotes, of course, a free-willed
determination, on the part of the subject, to participate in the
research. Naturally, coerced consent would not be effective. Coer-
cion, of course, would exist in the case of a person who simply
overpowered a person's free-willed exercise of determination by
threats or force. But coercion may come about in many, many other
ways. However, the greatest concern ordinarily in a research con-
text would not be about direct coercion, but about, what I shall
call, "coercion in the circumstances." This relates to who the par-
ticipant is and the circumstances that surround a granting of consent.
Some studies have investigated why people agree to be research
subjects. In the case of doctor-patient relationships, in a thera-
peutic context, participation may come about because of the trust
and confidence the patient has in the doctor. I have assumed no
such motivation would exist on the part of the participants in envi-
ronmental research. I have assumed that these participants may be
public spirited persons, needy persons, or persons who for some other
reason are influenced to become subjects. I imagine that the induce-
ment to participate in environmental research is probably monetary.
While I think significant problems of voluntariness are not likely,
researchers might want to keep- in mind the nature of any inducement
they hold out in light of the circumstances of the persons who might
choose to participate, it is not inconceivable, although I think
the possibility is remote, that the voluntariness of a consent might
be questioned if subjects are persons who are in such dire financial
needs that the exercise of free will may be precluded by the very
circumstance of their need.
Application of the Doctrine in Research Situations
As I interpret the informed consent doctrine in the context of
environmental research my advice would be that researchers provide
their subjects with "all the information a reasonable person would
want and need to know in order to make an intelligent and understand-
ing determination of whether to participate as a .subject or not."
Who is a "reasonable person"? while that inquiry seems simple.
40
-------
its explication is somewhat.more difficult. For the purposes of
environmental research, I would think there is a soundness in the
assumption that a "reasonable person" is a person of ordinary intel-
ligence, who possesses all the data that ordinary persons in the
community possess. If you are dealing with other kinds of persons,
the problem would become more significant. The concept of the
reasonable person has a direct relevance to the first element of
the doctrine of informed consent dealing with information providing.
It is probably sound to suggest that information need not be provid-
ed about things that a person in the community who has a sound mind,
ordinary intelligence, and the average capacity for understanding
would already know. I am not sufficiently knowledgeable about
environmental research to know the exact implications of that state-
ment. I would stress, however, that researchers should provide any
information that a person of ordinary intelligence and common under-
standing in the community would be unaware of. I would generally
include in this any risks of the research, the procedures, the meth-
ods, and any other information about the research that would be use-
ful to a subject in determining whether to participate in a research
project.
In connection with the precise content and detail of the infor-
mation that needs to be provided, the courts have formulated a doc-
trine that all "material" information should be disclosed. What is
"material" information? In an oversimplified way, it is any infor-
mation that the reasonable person would consider important and want
to evaluate in making a determination whether to become a subject
in your research. For example, I think a court would be likely to
hold that the possibility of a risk arising, a risk as low as one
percent, if that risk would expose a person to death or serious bod-
ily injury, is a "material" risk that ought to be disclosed. When
the risk is one that does not involve a risk of death or serious
bodily injury, the question is somewhat less clear. However, anjj
risk that is known and would be significant should be disclosed,
even if the risk is one that threatens only short-term discomfort of
even a nonserious nature. I have already indicated that the very
fact that certain risks may be unknown is itself a risk that ought
to be disclosed.
I think most courts would apply the standard that a risk, to
be material, has to be a risk that would be material to a reason-
able person. In thoroughness, however, I should mention that some
courts might hold that it is not a risk that would be material to a
reasonable person, but a risk that would be material to the person
undergoing the participation in the experiment. There is, of course,
a problem here in that after a risk has come to fruition and injured
a participant, that participant has every incentive to say that if
the risk had been known, he would not have participated in the
research.
41
-------
In the medically therapeutic treatment context, there is a
division of authority as to whether the standards that govern what
information should be disclosed is a standard to be determined by
the court as a matter of law, or a standard that is determined by
what the medical profession would regard as the professional stan-
dard for disclosure. The way the problem practically arises is as
follows: Let us assume that in a research procedure a certain risk
was not disclosed. The plaintiff would claim that the risk was a
material risk. The researcher would counter that it is not the
practice in the research profession to disclose such a risk. The
question then is whether the standards of the profession should
control what disclosure should be made or whether the court, as an
independent matter, will determine what information should be dis-
closed.
In my judgment, the sounder and safer alternative for a re-
searcher is to disclose all information that would be material to
a reasonable person. By disclosing all the information that the
researcher reasonably can disclose, the researcher avoids the
possibility of failing to disclose something even competent re-
searchers would not generally disclose, but which later might be
determined by a court to have been something that should have been
disclosed. -
thA U1^mateiY' ifc aPPears that there are three resolutions to
the problem of what should be disclosed. There may exist an ethi-
leaal 1 'f dl*clos^' a practical limit of disclosure, and a
reauLeT f ^closure. Presumably, the ethical ideal would
to mak! dlSC"°Te °f everythin<>- ^ may be practically impossible
bably re^tedef ^^^ *~ »»y reasons, one of which L Pro-
cSsure SStS ^ researcher's time and capacity for making dis-
i» the iea^ ^*imum that c^ld practically be disclosed
would be mv JvJ^Vlthat mUSt be Closed is undetermined. It
'
* r6Searcher ou^ to strive to provide as much
PraCtlca11* Possible in view of all the circum-
-t.r-in.tion
t
have done the cir"6 ^—^ ""asonably could
result ywm
there has been a failure ^7 "°W assume that in a particular case
son would regard as a m^eria r isT ^fS ^f & reaS°nable ^r
the researcher to disclose that risk L v "** *°*sL*ie f°r
every failure to make a
42
-------
failure to disclose a risk will result in liability only if an
injury is caused to a person by the risk that was not disclosed.
The problem is that in advance the researcher cannot determine
whether a particular risk will result in injury to the research
subject. By failing to disclose a risk a researcher would be
taking a gamble.
One other situation in the research context that might arise
involves the following situation: Assume a researcher makes a
disclosure of all the risks that he knows about, discloses the
fact that certain risks are unknown/ and after the research is
under way, gains additional information that the subject, as a
reasonable person, would desire to know. Is there a duty to dis-
close information discovered after the experiment is under way?
I am not aware of a case that has addressed this question. How-
ever, it would be wise, I think, for the researcher to regard him-
self as under a duty to make a disclosure at any point during the
process of the research if he learns of risks that were previously
unknown or learns of risks that are greater than, or different
from those disclosed.
One final matter that I think should be touched on deals with
the question of whether a person must be informed at the outset,
as part of the researcher's duty to make a full disclosure, that
he may at any time withdraw from the experiment or the research
program. No definitive answer can probably be given based on
previous cases. However, it is reasonable to assume that if a
person has an initial right to participate or not to participate
after full disclosure is made, it seems likely that the right of
non-participation continues throughout the process of the research,
and that a person may withdraw at any time and must be informed of
the right to do so.
Situations are imaginable, however, where the withdrawal of
the subject will have serious consequences to the validity of the
study being undertaken and may impose a serious burden on the re-
searcher. My best judgment is that these are burdens and disad-
vantages that researchers must be prepared to tolerate.
SOME PRACTICAL DIMENSIONS
Without intending to suggest that the case law would require
the following things that I shall mention, I would raise these
additional matters for consideration.
Who should make disclosure in a particular research program?
It would probably be wise to have a person designated as responsi-
ble for providing information to participants. The practical reason
43
-------
is that such a procedure will avoid the possibility that one per-
son will think another has given the appropriate information and
that other person will be thinking someone else has done it. The
consequence, as you can see, will be that no one would have given
the information. This situation has arisen in instances, for
example, in which the doctor thought the hospital would provide
certain information, and the hospital, in turn, relied on the
doctor to provide the information. The consequence was that the
patient was not informed. A breakdown in the procedure for pro-
viding information will not justify a failure to provide it.
sure !hnulf hUld f SCl°SUre be "**" It is not clear when disclo-
sure should be made except to say that it must be made before the
thrSubiect91nS' T6Ver' the Pr°ViSi0n °f the -formation after
the subjects are all assembled and are ready to undergo the re-
search experiment may constitute a coercive circumstance in itself.
uat'Tol thr rT6Cti0n' —Cation with others, or sober eva -
of the rLearch foliatl°; ^^ ^ be avai1*^ when the onset
informal * ^ ° ^^ after the Provision of the
provide a rLsorbl917' * "^ ^^ ** SOUnd P^ice to
is provided an^H V ****** ^^ tte time when th* information
II Salted P.,fe tim%When the P«»on is asked whether consent
all^rcuitalcf *** ^ ?** ^ limlt Can be stated- *he over-
ail circumstances would obviously be relovani- BC,
patlng? ° "^ * S°ber ^ detached Judgment about partici-
ful tST thexinc'"1^ ^^^ *** it is doubt-
in circumstances where nre'T1™8 "^ aVOid liabilit^
practically speaking it would dlSCl°Se information has occurred,
review panels! pee^review Soun?*" ^ existence of i^ouse
would be helpfu? in tw^ysT Tsti^r* 6Valuati°n teams
would make it more likely that all thf T reV16W pr°cedures
information can be provided »rp J relevant areas in which
act as a check on thrresearch^,^11' S6C°nd' reV±ew P-cedures
research that might cauTthf ".Lr^."^^ ********* ^
the
pate in the search ^^ °f sub^ts to partici-
44
-------
Accordingly, as a safeguard against the possibility of the
researchers taking shortcuts, which may result in a failure to dis-
close, it would be good practice to have such review procedures.
It can be seen that the review procedures relate to the relation-
ship between the researcher and the research institution. Arrange-
ments between the institution in which the research is undertaken
and the researcher cannot ordinarily be thought to affect the rela-
tionship between the researcher and the participants.
What is the relevance of federal, state, or other legislation?
Compliance with statutory requirements may well have a direct bear-
ing on the reasonableness of the researcher's conduct in disclosing
or not disclosing certain kinds of information. It would be sound
practice to scrupulously comply with any statutory requisites. It
is possible that failure to comply with statutory requisites may
give rise to an argument that the compliance failure itself estab-
lishes the unreasonableness of the researcher's actions.
What about federal agency regulations? The same thing that can
be said about compliance with statutory requirements can be said of
compliance with various applicable regulations of agencies. In
addition, it may well be valuable to follow requirements of agen-
cies that are administering the various kinds of research, because
it may be that such guidelines or regulations will have been drafted
by those who are thoroughly conversant with the issues of informed
consent, and that by following them, the researcher will render
less likely any liability for failure to meet the case law require-
ments regarding the informed consent.
45
-------
BIBLIOGRAPHY
1. Annas, G.J., et al.: Informed Consent To Human Experimentation:
The Subject's Dilemma. Ballinger Pub. Co., Massachusetts, 1977.
2. Bogomolny, R.L.: Human Experimentation. Southern Methodist
Univ. Press, Texas/ 1976.
3. Fried, C.: Medical Experimentation: Personal Integrity and
Social Policy. American Elsevior Pub. Co., New York, 1974.
4. Gray, B.H.: Human Subjects in Medical Experimentation. John
Wiley & Sons, New York 1975.
5. Hershey, N. , et al.: Human experimentation and the Law. Aspen
Publishers, Maryland 1976.
6. Katz, J.: Experimentation with Human Beings. Russell Sage
Foundation, New York, 1972, pp. 521-674.
7. Katz, J,, et al.: Catastrophic Diseases: Who Decides What?
Russell Sage Foundation, New York, 1975, pp. 79-115.
8. Posser, W.: .Torts. West Pub. Co., Minnesota, 1971, pp. 104-
105; 165-166. (Fourth ed.)
9. Shannon, T.A.: Bioethics. Paulist Press, New York, 1976,
pp. 209-291.
10. McCoid: The care required of medical practitioners.
12 Vanderbilt Law Review 549, 1969.
11. Plant: An analysis of "informed consent." 36 Fordam Law Review
639, 1968.
12. Riskin: Informed consent: Looking for the action. 1975 Univ.
Illinois Law Forum 580.
13. WaltZ and Scheunmeman: Informed consent to therapy. 64 North-
western Univ. Law Review 628, 1969.
14. Mohr v. Williams , 95 Minn. 261, 104 N.W. 12 (1905).
15. Tabor v. Scobee, 254 S.w.2d 474 (Kentucky 1952).
Mill«r *"T«-L 251 Minn. 427, 88 N.W.2d 186
46
-------
17. Nathanson v. Kline, 186 Kan. 393, 350 P.2d 1093 (1960).
18. Nishi v. Hartwell, 52 Haw. 188, 473 P.2d 116 (1970).
19. Canterbury v. Spence, 464 F.2d 772 (1972).
20. zeBarth v. Swedish Hospital Medical Center, 499 P.2d 1 (Wash.
1973).
21. Cobbs v. Grant, 8 Cal.3d 229, 502 P.2d 1 (1972).
22. Wilkinson v. Vesey, 110 R.I. 606, 295 A.2d 676 (1972).
23. Foqal v. Genessee Hospital, 344 N.Y. Supp.2d 552 (1973).
47
-------
Discussion Summary
Participants discussed whether increasing a subject's pay as
a project progresses would be construed as a way of pressuring a
subject to stay with a study until its completion. Discussants
said that this procedure generally would not be considered
coercive.
49
-------
Role and Function of Committees on
Protection of Human Subjects in Research
Edward Bishop, M.D.
School of Medicine
University of North Carolina
As chairman of the Committee on Protection of the Rights of
Human Subjects, I can assure you that membership is associated
with problems. Therefore, I would like to discuss the problems
of such committees as ours.
I would like to quote from a recent issue of the American Bar
Foundation Research Journal, in which Benjamin Duval states quite
clearly one of the problems that a committee has, in that, "...an
institutional review board is in itself a legal institution." The
act that called for the establishment of these institutional review
boards stated that the Secretary of Health, Education and Welfare
must require the establishment of an institutional review board
for every entity that receives funding. Therefore, although review
boards are instructed not to approve proposals that will invade
any subject's legal rights, the interests protected under such
regulations are significantly broader than the legal rights of the
subject. Benjamin Duval makes an interesting comparison when he
says that institutional review boards combine the elements of an
administrative agency and a jury. They also act as a legal author-
ity to carry out laws and also to make laws.
With that introduction, I can describe our committee's problems
and methods by giving a brief outline of the methods we use at the
School of Medicine at the University of North Carolina. We began
our committee about five and a half years ago. Before that time,
51
-------
the functions involved in institutional review had been carried
out by a committee that was in charge of the Clinical Research Unit.
Because of an increasing workload, there was an obvious need for a
specific committee to act as an institutional review board. Compa-
rable committees exist in five other schools within the University.
So, our committee was formed to function specifically as a medical
school committee.
Our committee's purpose and principles are best described in
our Document of Assurance to HEW and in some words from the Arti-
cles of Implementation. We feel that our function, as specified
in our Document of Assurance, is to "fully protect the rights and
the welfare of all subjects who participate in research, in inves-
tigations, in study, in development, in demonstrations, or in other
projects." Obviously, this includes many direct involvements, such
as the action of drugs, medical and surgical procedures, and so
on. But our committee has extended its responsibilities further
than this, for example, to include non-invasive observations and
data collection. We are concerned that there not be undue invasion
of the subject's rights to privacy, and that there is no undue risk
to the subject, at least not physical risk. The same is true of
moral rights.
One of the first obligations we had was to inform the faculty
that if they plan to do human research, they must have their work
reviewed in advance. Oftentimes, the investigator is reminded of
this obligation when his funding organization will not consider
his application until it has been reviewed by a board.
Our primary obligation is to look at human rights, and not at
the scientific merit of a research proposal. This is impossible
because usually we must consider both aspects. If someone submits
a frxvolous scientific experiment, we would not accept it if there
was any type of risk to the subject. On the other hand, there are
times that the scientific merit and the amount of medical informa-
tion to be gained do warrant some risks, and the experiment thus
,
two ™~-JCCef Sble" /e mUSt in a11 cases have a Balance between the
two
, although the question of scientific merit is
not our specific concern. Finally, we feel that one of our serious
obligations is to make sure there is no undue enticement to persuade
a patient to continue an experiment against his wishes.
°*e of th« °ther principles that we work under is that we feel
that this is a serious committee, and therefore, it should have the
representation of the most experienced and the senior faculty.
Presently, the people who are on the committee, with only a couple
exceptions, are people in senior positions at the University. We
use these people, not because their age necessarily implies wisdom
52
-------
or good judgment, but because they are familiar with the University.
They are also familiar with the investigators and have their re-
spect, so that if we on the committee disagree or need to advise
the investigator, this guidance will come from a body that speaks
with a certain amount of authority.
Other important members of our committee are the representa-
tives of the public. I think, probably, we should have a greater
representation from the community, because the people who work on
a voluntary basis, the public representatives, do an exceedingly
good job. In reviewing research proposals using human subjects,
the public representatives approach this task with a completely
different point of view than we do, and oftentimes, they make
very important judgments or find problems that we with our bias of
a medical education, with our bias of seeing patients and subjects
all the time, would have missed. We are not legally required to
include public representatives on the committee, but including
representation from the public or from the community is very
strongly recommended.
We differ from other institutional review boards across the
country. For example, our committee meets only on a few occasions
a year, and we do most of our review by mail. We do most of our
voting by ballot. One of the reasons why we operate this way is
because of the volume of our work. In a three-year period, we
reviewed 297 new proposals; we reviewed for re-review 398 contin-
uing proposals, or 695 proposals duVing this period of time. This
comes down to four or five proposals a week, 19 to 20 a month. I
could not keep people on the committee if we convened twice a week
to discuss these proposals, as we would have to, or once a week.
I could not ask the representatives of the community to meet with
us that often. This is the first reason why we do most of our
work by mail review and by balloting.
I think there are some advantages to this work procedure. In
general, a proposal comes to my office when it has been reviewed
by the principal investigator's departmental chairman. It is then
reviewed by my staff for completeness to see that all the forms are
in. There is a consent form and there is a review of the proposal.
At some time this proposal may be given administrative approval.
This is a right that has been given to me by a local committee.
An illustration of administrative approval would be a proposal that
has been approved, but has not been funded. Thus, the proposal
has not been active, and the investigator wants to continue the
approval process. The committee allows me to give administrative
approval in other similarly benign or uncomplicated situations,
but they in turn review at the next meeting of the committee all
of my administrative approvals and sanctions, and in some cases,
they deny approval.
53
-------
Once a proposal is reviewed for completeness, it is sent to a
subcommittee, and all the information that is sent to me, all the
documents that come in for the research proposal, go to a subcommit-
tee that goes through this package in depth. Then they report back
to us. If the subcommittee reports a unanimous vote of approval,
there are no problems, and a simplified form is sent to the full
committee, who in turn ballot back again. If, however, the sub-
committee says there is some problem; for example, if they are con-
cerned about the use of human subjects in the project, then an ad
hoc committee of experts in the appropriate field is appointed to
review the proposal with the principal investigator.
Now, I, as an obstetrician, cannot be expert on cardiac prob-
lems. Our surgeon is not an expert on pharmacology. Therefore,
we use members outside of our committee, and at times, outside of
this University, to serve on the ad hoc committee to review with
the principal investigator the proposal and to try to decide about
the sticky problems. The subsequent report in turn comes back to
the full committee, who now ballot by mail. They have the right
to either accept the proposal, to request modification, or to
reject the proposal.
It has been decided that no research activity will be rejected
without a majority vote, and that no research activity will be
accepted without a majority vote. Actually we usually work on a
unanimous vote. If I see that 12 members on a committee have
accepted a research proposal, but one person picks up a very
important but overlooked infringement of the rights of a human
sublet, then we will turn that proposal down, because I think we
must lean over backwards to protect the subjects, even though the
vote may have been leaning the opposite way.
One reason why I like the idea of the mail ballot is because
instead of having to make up your mind in a committee meeting at
five o clock on Friday, you have the opportunity to make up your
own mind when you want to sit .down and review these proposals when
you feel like it, when you can give the time to it, and when you
reach an adequate decision. So, i do feel that the balloting
woulf h1S d^enSlble' and ™ Probably get better reviews than we
would by setting in a committee as a group. We do sit in a commit-
whLa^a gr°UP at regularlv stipulated intervals, for example,
when there is any division of opinion about a proposal, or when we
need to speak to the principal investigator.
«™. ^ ^lai^iS±0n °f the committee is final and it is not
questioned by higher authorities within the University. I thin
54
-------
it is stipulated by certain regulations that if we turn down some
research activity, there cannot be pressures from the Dean, the
Chancellor, the Vice-chancellor, or from anyone else to change
our decision*
PROPOSAL REVIEW
Out of 297 research proposals, we rejected only three. I
think this is to the credit of the faculty at the University.
They now recognize, not just because of the threat of legal lia-
bility, the importance of human life. They do not submit research
proposals that infringe upon the rights of human subjects or endan-
ger their health in any way. When we did request modification of
31 proposals, we were primarily asking someone to take out a risky
portion, or what we considered to be a risky portion of their
research proposal, or we asked them to modify their consent form,
or, in some cases, an application was incomplete. If we had not
requested and given the principal investigator opportunity for
modification, we would have rejected more proposals.
One of the problems that the committee has is the amount of
work. Another problem we have is overlapping with jurisdictional
problems between one school of the University with another. What
if a proposal is a combined research activity between the Dental
School and the Medical School; between the School of Nursing and
the Dental School; who has jurisdiction? That question has not
been solved yet and it has been a sticky problem, but I think it
is an administrative problem.
One of the other problems is, and I am not going to quote
this directly, that the HEW regulations regarding review committees
state that no matter how distant the research work may be from the
granting institution or from the institution receiving the grant,
that institution is responsible for protecting the rights of these
research subjects. This regulation became very important to me
a couple of years ago when we were funded for some research activi-
ties in Chapel Hill. Those monies were being spent in India, in
Pakistan, in Iran, and in a number of European and Asian countries.
We were legally responsible to those subjects in Charez. It was
up to us to make sure they understood the consent forms they were
signing an almost impossible thing to do with any guarantee of
completeness.
This is a problem that is not only concerned with research
conducted out of the continental United States; it is also a prob-
lem in cases where someone is doing research where part of the
research subjects are located in a distant city, or when they are
at a distant hospital. Do they use what we think is a consent form,
55
-------
or what they think is a consent form? So far we have insisted that
our consent form be used since we are responsible for research sub-
jects. It is a problem for other institutions, other schools in
the University that want to use patients of the medical school
hospital. A professor of anthropology would like to talk to cancer
patients to examine some aspect of their psychological reaction to
their disease— ^an he do this? Who reviews his proposal? How about
those people who just want to use record room reviews? Is this,
even though they do not identify the subject, appropriate to do?
ETHICAL PROBLEMS AND POLICIES
Another difficult problem that is unsolved is the use of excess
tissues and fluids. Someone says/ "Well, when you send a tube of
blood to the laboratory, they are going to use just 1 ml. I would
like to use the other four. I do not want to get consent from the
:U.Want to look at the blood. I do not have to know
oatient w - a< to the
patient. The blood is going to be thrown away anyhow." At first
the idea sounds like "why not?" But you find that soon, the same
person says, "A little extra blood is all right, but why don't you
get another tube when you collect it? it won't hurt to" get more
° ""
subiec °£u ar"" ^ *** *** *** «« -Dinging on a human
become four or a 1^°™* tO *"" *** ™y much' but tw° tubes
ecome four, or a little more cerebrospinal
are
faculty of the Medical Schoo " research d=">e * ">«
on the patients ofthi h l\ " ab°ut the »B«rch that is done
that isPnot Sone by £:oSri,aSS°°iated "^ ^ Medi°al S°t>°°i'
the '
=0
and we have kno»n no in»«i ^ ^"^ Io°ked into this
been held liable? Bu^e hf "he" me°be« »* the committee have
whose faculty £ ^If™"^"'*™ ""^ersity one school
L" ?
by sayinc, that they will lo* T£ Cha«=ell°r.s office has responded
se,ve on
A final problem is, what is research. We say that we review
56
-------
research on human subjects/ but where is the dividing line between
clinical care and research? We change our methods of doing a medi-
cal procedure—and I have to use an obstetrical example. If we
decide that all breeches should be delivered by Caesarean section
rather than vaginally, and we keep a record to see how it works
out/ is that considered research/ or is that just observation of
what we hope is improved clinical practice? This is a difficult
decision. Do you have to go to that patient and say, "You have
a choice between a vaginal delivery and a Caesarean section"?
Obviously, she cannot make that choice.
Another problem we have is, when does research become accepted
practice? Again, for obvious reasons, I use examples in the repro-
ductive field. One faculty member developed a clip for steriliza-
tion—obviously this was research. He wanted to see how successful
the clip was, how many pregnancies resulted when this clip was used/
was it better or was it worse than other contraceptive methods.
After a period of time, he looked at a thousand or so subjects and
said, "Well, my next thousand subjects—will they be considered as
research subjects or have I already done the research? I know
there were three pregnancies in the thousand, and the clip is avail-
able clinically now." He says, "I am still going to collect data
on the clip because I am not sure how it is going to work out. It
has not been tested long enough yet, but it is commercially avail-
able." When does research become accepted practice? What about
informing patients who have unexpected results or abnormal findings?
Is it the investigator's responsibility to tell that patient?
ADVICE AND CONSENT
Finally/ our biggest problem is with consent forms. First, I
dislike the term "consent form." I do not think there should be
a form for giving informed consent. There cannot be a standard
form for an institution. There cannot be a standard form between
different research activities. I think each consent form should
be written in a language that the patient understands. This is
where we can use our community members on the committee. If they
cannot understand the consent form, then I am sure the patient
cannot. The patient should understand what is being done, the
benefits, the risks, the discomforts; whether there could be an
alternative procedure other than the research procedure.
Then, we ask that it be included in the consent form that the
subject is free to withdraw from the experiment without penalty.
We try to serve as the research subject's advocate by saying that
in the consent the subject should be told that if he feels his
rights are being threatened, he can contact the Committee on the
Protection of Rights of Human Subjects. The consent forms have to
57
-------
comply with the laws of this state, which may not be the laws of
another state.
Yesterday I received two research proposals. One consent form
had three lines. It said the investigators were going to withdraw
blood, 5 ml, and the patient is supposed to sign. That is about
all the form said. It was not written in the first person. I think
if you are going to consent to something, the consent form has to
be written in the first person.
The other consent form, believe it or not, was fourteen pages
long. I think this is equally bad as the three-line consent form.
I am not sure how many subjects will wade through fourteen pages
to give consent to some research activities. I am afraid the
committee may be at fault for pushing investigators to make sure
that all the essential elements are on the consent form—^ve have
made them make it quite voluminous.
Finally, we do not like to act as a threat to the investigator,
but rather, we like to act as a protector of the research subject.
We are also protecting the investigator, not necessarily from any
legal liability, but we are protecting him by helping him realize
the rights of human subjects.
in spite of regulations made up by committees, I am sure that
research is going to continue. I am sure that research will contin-
ue wlth human subjects, no matter how much basic or ani^l work can
at someVmeW ^l??1"8' "^ ^^ mt * tried on huma" subjects
as ours wSJ*™ ^ C°ntinuation of "view mechanisms such
as ours will remain, and must remain, in existence for as long as
aCC6*table- Fi-^ the most important
consent
58
-------
Discussion Summary
Participants discussed the advantages and disadvantages of
having obstreperous members on review committees. On the positive
side, participants felt that those who are apt to be meticulous
about every point discussed in a committee meeting may bring up an
issue that has been overlooked. A disadvantage is that this type
of committee member may tend to disagree simply for the sake of
disagreeing. In this case/ some committee chairmen have the
prerogative of deeming a committee member's objection as
"unacceptable."
Participants also discussed the need to keep bureaucratic
overload in committees to a minimum. Discussants felt that staff
review of research proposals is justified in that these reviews
reduce the amount of time spent considering proposals that are
obviously unacceptable or missing required forms.
59
-------
METHODOLOGIES AND PROTOCOLS IN
ENVIRONMENTAL CLINICAL RESEARCH
Moderator: Ralph Stacy, Ph.D.
-------
Developing Methodologies—Environmental
Studies
Steven Horvath, Ph.D.
Institute of Environmental Stress
Santa Barbara, California
I would like to discuss a number of problems associated with
the development of methodology, primarily from the non-invasive
standpoint of investigating man under conditions of stress, whether
the stresses are combined or single stresses. I would like to
emphasize that most of our new procedures have been designed to
work in multiple environments and under multiple stresses.
CHARACTERIZATION: A PRE-TEST PROCESS
I would like to start off by reviewing some of the basic things
you do before developing a methodology. The researcher must effec-
tively characterize the subject so he knows just exactly what he is
going to do, what he is going to work with, and what he is going
to do for the research subject from the standpoint of assuring
that he has adequate safety in all of the procedures that will be
employed.
Sensitive Subjects
Researchers have to contend with the problem of sensitive sub-
jects and how you determine whether sensitive subjects are going to
be in your pool, and what you are going to do with those subjects
after you have them in your pool, or, in some cases, whether the
researcher should set out to select a group of sensitive subjects
63
-------
among a so-called normal population—not to look at individuals
with disease, but individuals who are essentially considered to
be normal in physiological terms, in both their pulmonary and
their cardiovascular systems.
Actually, we have had a problem with sensitive research
subjects, because in almost all of our studies we have found one
or two subjects who were sensitive to the particular pollutant
that we were studying. In retrospect, it appears that we could
have determined the subjects' sensitivity by using a question-
naire. If you are going to study sensitive subjects, you
should have those subjects available to you as sensitive sub-
jects, and not have to include them as part and parcel of the
overall population.
Work capacity
A number of techniques of carefully categorizing an indi-
vidual have been developed over the years. One of these tech-
niques has to do, especially in air pollution studies or in
toxicological studies, with evaluating the potential capacity of
the subject to perform work. It has become apparent that one of
the major factors in the response of an individual to a pollutant
is whether or not he is active or inactive. Many of the studies
that have been done on environmental pollution have used resting
subjects, and have consequently missed the value of or the
importance of the role the pollutant will play in the response of
the subject, because most subjects exposed in a natural envi-
ronment do have to work.
The first evaluation of any subject has to do with his
potential work capacity, and there are several ways that this
work capacity can be measured. The best way, since almost
everybody can walk—and this applies to individuals of all ages,
anywhere from six to eighty odd years of age—is to measure their
maximal aerobic capacity by having them walk on a treadmill that
changes its rate. The treadmill can maintain a constant speed or
it can change its speed, as well. YOU can perform this test very
simply by a relatively straightforward technique. A computer
gives the test results out immediately, so that you know not
only the subject's ventilation, but also his oxygen uptake, his
C02 production, his body temperature, his heart rate, and his
electrocardiogram as it is measured.
Another procedure that we use to measure potential work
capacity is the bicycle ergometer. There is an important dif-
ference between the measurements made on a bicycle ergometer and
those that are made on a treadmill. The difference is in the
64
-------
neighborhood of 10 percent in terms of maximum aerobic capacity.
When an investigator studies an individual using either a bicycle
or a treadmill, he must take this into account. Furthermore, the
position of the individual on the bicycle, whether he is in the
upright position, or whether he is in a supine position, makes a
further difference in his aerobic capacity.
The question that arises here is, how are we going to determine
whether or not an individual should perform a fixed load of work,
namely, something like, say, 300 K cals per hour, or whether he
should do work on the basis of a percentage of his capacity? One
of the major factors that influences man's ability to perform is
his age.. As he gets older, past about 18 or 19 years of age, he
generally shows a relative decline in his capacity to perform
work. Therefore, if you utilize only a fixed workload, this may
represent a considerable variation of the subject's capacity to
do work. Consequently, it is almost necessary to measure the
maximum aerobic capacity of the individual, so that you can relate
the individual's performance under any stress in terms of percentage
of his maximum aerobic capacity. Maximum aerobic capacity is
influenced not only by age, but by the sex of the individual.
Besides characterizing an individual by the percent of his
capacity to perform work, and in terms of absolute fixed load, an
individual can also be studied in another way, namely, in terms of
a fixed ventilation. Some of the studies that we have done have
utilized this characterization. A test of fixed ventilation is
easily automated and very easily conducted under careful conditions.
Body Fat
Further characterization of the individual is necessary, and
one of the things that we have found extremely important is to
have some measurements of the subject's body fat, so that we can
determine the lean body mass or the effect of protoplasmic mass.
There are two basic techniques that are commonly employed to
measure body fat. One is to measure the skin folds at various
places—this is not the best technique, although it is probably
utilized much more than any other. A much better technique is
where you measure the individual under water. This technique
utilizes some of the advances in technology. Instead of measuring
the individual with an autopsy scale, we use load cells or string
gauges to make these measurements. Individuals vary in body fat
to a great extent, and some of the effects on the individual
by pollutants are determined by the amount of body fat.
We also utilize another type of technique, which is to measure
people under water while they are exposed to various pollutants.
65
-------
This is a new way of looking at individuals, and it is a fairly
well automated technique.
Blood Constituents
Another factor in characterizing individuals, especially in
terms of the effects on the individual, is blood constituents.
This is a characterization that has been ignored for a long time.
There has been a great deal of difficulty in some of the measure-
ments that have been reported; namely, blood samples have been
taken at unspecified times, and the measurements have been uncon-
trolled in terms of position of the subject. Nevertheless, blood
constituents are an important factor in characterizing research
subjects, in that this characterization takes advantage of the
fact that we know there are changes in plasma volumes, and that
these will result in marked changes in the concentration of materials
in the plasma, as a consequence of the subject being in different
positions.
NON-INVASIVE METHODOLOGY
I would like to discuss more methodology techniques, starting
off with measures of indirect methods of measuring cardiac output.
There are a number of methods available; some are practical, and
some are impractical. In general, we want to get away from the
invasive techniques, which we use so commonly, and go over to non-
invasive techniques.
Of the non-invasive techniques, there are three that are
potentially usable. One is the gas technique: the carbon dioxide
re-breathing procedure, the nitrous oxide procedure, and the
acetylene procedure. These are three fairly decent procedures,
but they have some limitations due to the fact that they cannot be
repeated at too frequent intervals.
Another approach to measuring cardiac output, which has a cer-
tain amount of value, that we have been utilizing to a great extent
is the use of the impedance device. This method is—like all of
these methods—not really as good as it should be in characterizing
the resting individual, but it is extraordinarily comparable to
direct methods, that is, the dilution techniques, if you use it in
the individual who is active to any degree whatsoever. Unfortunately,
this technique, although it is completely automated, can only be
employed in individuals performing a slight activity, and only on
the bicycle. The advantage of this technique is that you can leave
the measuring devices on the individual for long periods of time.
Thus, this technique detects rather striking changes that occur
in the individual during the night. So, this is a fairly effective
66
-------
technique for measuring cardiac output over long periods of
time, with the least amount of interference to the subject.
We have also tried using the echocardiogram to measure cardiac
output. The only difficulty with the echocardiogram at the present
time is the need for good microprocessing elements/ and more impor-
tantly, the need to be able to apply the test to an exercising sub-
ject. So far, we have not had too much success using the echocardio-
gram on exercising subjects, but I anticipate that very shortly we
will have a device that will be very effective in this regard.
Another indirect method of measuring cardiac output is to look
at another part of the cardiovascular system, and that is, to
measure the peripheral blood flow, whether you do it in the forearm,
the lower leg, or in the finger. This is a completely automated
technique over which neither the investigator nor the subject has
any control. It is all automated, and can be timed. Instead of
having to look at the old-fashioned records to derive the slope,
the slope is derived electronically, and then the only calculation
that has to be made is dividing the slope by a volume, which is also
displayed automatically on the unit. The data are also put on a
graph at the same time.
This automated technique is a very new development in terms of
measuring peripheral blood flow. It eliminates almost all experi-
menter bias, and it is a very rapid procedure. In fact, you can do
something in the neighborhood of 40 to 50 blood flows in a period
of several minutes without any real difficulty to the subject.
Investigators are faced with several path problems when they
put subjects in an environment of 48°C. Test results indicate that
people of different racial origins and different ethnic origins have
a different response to exposure to a hot environment. Any time
one studies subjects, one must characterize the subject not only in
terms of his physical and medical characteristics, but in terms of
ethnic origin, at least to some extent.
A newer development has involved an examination of nasal
airway resistance. We are now working with a device to measure
both nasal airway resistance and oral airway resistance. This is
an important new development, since it provides for separation of
two portions of the respiratory system which have not been looked
at too carefully in the past, and that is the point of time when the
individual shifts from breathing through his mouth to breathing
through his nose, or when there is a combined effect of those two
functions. Microprocessors handle the calculations and the
evaluations, and determine the pattern of the experiment.
67
-------
Another factor that I think is of some importance is the need
to develop devices that telemeter the heart rate. We measured a
cross-country coach, and every time he saw any of his runners come
by, his heart rate went up to 160, yet he gave the appearance of
having no reaction to the race. Similarly, we measured the heart
rate of a volleyball coach. His team lost the first set, won the
next four, and despite the fact that during the entire period of
time he sat very quietly, and apparently was unconcerned about the
entire event, his heart rates went as high as 160. I might point
out that his maximal heart rate was 180, so in that he is in a maxi-
mal aerobic capacity test, he is under considerable stress. These
studies indicate that many of the studies that we have been doing on
subjects where we look at them and obtain measurements of the heart
rate or using the electrocardiogram only at the fixed intervals may
be missing a great deal.
A few devices have been developed to make it possible to look
at individuals non-invasively, and at the same time, to describe
very simply the way in which a subject can be categorized by utiliz-
ing, again, non-invasive techniques. The more effectively you can
analyze the subject's response and his condition, the more effectively
you can determine what is happening to the individual.
One of the main advantages of the new methodology is not simply
that researchers arc able to utilize computer devices or new develop-
ments in electronics to help build better instrumentation and to
help organize the exposure of the individual, that is, to determine
what he is going to respond to and how he is going to respond to it.
More importantly, the new methodology has provided us with a means
by which we can take the other techniques that are not usually
available to us, from the standpoint of our own expertise, but which
are commonly used in other areas, and by proper handling and working
with electronics people, and with people who are interested in physi-
ological response, it is possible to develop new devices that will
make it easier for us to successfully invade the individual without
doing it by an invasive technique.
68
-------
Discussion Summary
Participants discussed the merits of using microprocessors
rather than minicomputers in research activities. Discussants said
that the minicomputers are more likely to break down than the micro-
processors, but the microprocessors are more adept at carrying out
different operations simultaneously. Some laboratories use mini-
computers as a supplement to the microprocessors.
Discussants observed that the use of microprocessors has enabled
investigators to automate the measurement of various bodily functions.
In one participant's laboratory, the data that is obtained is tape-
recorded to the IBM systems where it is analyzed. Tape-recording
of data is an easy process, and the eight-channel capability enables
the investigator to collect a good quantity of data.
It was noted that although the development of automated analyti-
cal procedures can be a timely and expensive process, these disad-
vantages are offset by the tremendous amount of time the investigator
saves in the long run once the automated units are set up.
Participants discussed the steps the accumulated data goes
through to be put into an analytic form. The automated units have
a data base for each research subject that includes every measurement
made on the subject* Statistical analyses are made from this data.
The computer can immediately produce data on the basic body functions.
To obtain derived functions, the base material is put on a data base,
and then the various derived functions of statistical analysis of the
computer data are carried out on the computer from the data base.
Discussants noted that it is important to develop a data base that
can be manipulated to handle large volumes of data.
In a discussion of the differences between the Bruce system and
the Balke technique, participants said that the Balke technique is
more suited to studying populations of different ages than the Bruce
system. For example, the Bruce system cannot be used to study
69
-------
children under six or seven years of age, or on adults in their
seventies and eighties.
The Bruce system is used in some laboratories, but not for
normal populations because there is no way of characterizing subjects
throughout their entire life span or in relation to their sex.
Participants also talked about the various methods of measuring
blood pressure. Measuring blood flow by using radioisotopes is a
viable technique, but it is invasive. An alternative is to use
strain gauges to measure peripheral blood flow. Discussants said
that attempts to automate the measurement of blood pressure have been
unsuccessful, and that the best measurement is to use a blood pressure
cuff and two good ears.
70
-------
Rationale for Experimental Design
John H. Knelson, M.D.
Health Effects Research Laboratory
U.S. Environmental Protection Agency
I would like to describe how we decide what clinical environ-
mental research we are going to conduct within the constraints im-
posed by ethics/ the law, and logic. I think it is very impor-
tant, in discussing the rationale of experimental design, to bear
in mind a few definitions and a few relationships.
CLINICAL ENVIRONMENTAL RESEARCH: A DEFINITION
Over the years that I have been involved in this type of
activity, I have developed a simplistic definition of clinical
environmental research. I think, as we expand our purview and
bring new tools to bear on the topic, and as we learn a little
more about the problems of environmental medicine, my simplistic
definition will merit some scrutiny, and perhaps, need to be
overhauled and expanded. But very briefly, the definition I have
used is that any time you manipulate a human being and his environ-
ment, you are doing clinical and environmental research. Now,
what does that mean? More germanely, what doesn't it mean? When
we study what is happening to a population under specific circum-
stances, or when we simply avail ourselves of the opportunity to
study a population, I do not consider that we are conducting
clinical research.
I think some might argue with that, so I will give an example
of what I am talking about, that is, what we have been doing in
Chapel Hill all along, and what most of our colleagues have been
71
-------
doing around the country. In clinical environmental research, the
subject and his environment are manipulated. The subject is
placed in an artifically-controlled environmental laboratory. You
manipulate that person by putting him on a bicycle ergometer, on a
treadmill, or by having him breathe through a pneumotachometer or
into a spirometer, or in some other way study some aspect of his
physiology. You do not allow the subject to range normally through
his environment. Then you manipulate in a positive, very carefully
thought out and contrived way, the subject's environment. You
manipulate the environment by making it clean to a standard con-
dition, putting the subject in a Class 1,000 or a Class 100 clean
room, and the investigator gets a baseline of information.
The investigator may manipulate the environment by having the
subject breathe an air pollutant, or by manipulating the temper-
ature, relative humidity, light intensity, sound levels, iso-
lation, or other characteristics of the environment. This is the
simplistic definition of clinical environmental research. I
think, though, as we familiarize ourselves with the rapidly,
exponentially growing technology that is available to us, we will
need to enlarge that definition.
Going back to one of the historical antecedents of clinical
environmental research, one of the projects of the Harvard Fatigue
Lab was to study the adaptation of workers to their new environ-
ment as they constructed what was then called Boulder Dam. That
project was certainly a form of clinical environmental research.
Clinical research was done on these workers as they adapted to the
new environment, but their environment was not manipulated. The
subjects were manipulated only to the extent that some measure-
ments were made concerning their adaptation.
As we face the newer, more challenging, and more far-reaching
questions that society will put to us with respect to our environ-
ment, we will need to adopt a broader definition of clinical
environmental research. For the most part, though, my discussion
of the rationale involved in design experiments in clinical environ-
mental research is based on the simplistic definition, where you
put a person in a controlled environmental laboratory, and mani-
pulate his environment in a predetermined, well-thought-out way.
RELEVANCE: A KEY TO EXPERIMENTAL DESIGN
The theme of my discussion is relevance. We are designing
experiments and studies that have a very specific relevance to a
question posed by society. Frequently, in other kinds of clinical
research, much activity is pursued in a somewhat descriptive way.
By that I mean that we are not really focusing before we undertake
72
-------
the research, or before we design a particular study on a very
precise question. We are exploring the results of a particular
intervention, be it surgical or medical, and we do not have all of
the consequences of that intervention completely in mind.
The objectives/ then, put in this simplistic paradigm of
clinical environmental research, are to focus on a specific question
that society is posing, and then to establish the experimental
framework within which to address that question. If we do not
keep these objectives in mind, we will conduct much clinical
environmental research that will not be relevant to the questions
that are being asked.
I think we do need to keep in mind the relationship between
clinical research, human experimentation, and probably, the other
couple of ways that we can get at the answers to the same ques-
tions. Much of the information that precedes the clinical experi-
mentation, and much of the information that is the basis for the
rationale of the experimental design, comes from two other disci-
plines—basic toxicology and epidemiology.
Basic toxicology has to precede and continue to be an inte-
gral part of the rationale for clinical experimentation. The
amount of basic animal toxicology that we have at our disposal to
design clinical research will have a profound effect on what
research is done and how it is done, and of course, on the ration-
ale for approaching a particular experimental design.
The discipline that I would like to discuss is epidemiology.
Most people think that epidemiology is not really a discipline,
that it is a state of mind with which you approach a problem. But
epidemiology is important because it is closely related to clini-
cal research, and it deals with the species of interest, human
beings.
Epidemiology differs from clinical research in that it lacks
two of the overriding characteristics included in the definition
of clinical research. In epidemiology the subject is not mani-
pulated. As a matter of fact, in epidemiology you do everything
you can to avoid interfering with the normal behavior of your
subject. Epidemiologists are concerned about a kind of Heisenberg1s
principle of biology: anything you do to study your subjects in
an epidemiologic study design will interfere with those measurements
because you specifically do not want to intrude on your subjects'
way of life, and you certainly do not want to manipulate their
environment.
I want to discuss briefly the interrelationship between
epidemiology, clinical environmental research, and environmental
73
-------
toxicology. In the latter discipline, we have an exquisite
handle on the dose, and you can do anything you wish to the subjects
because they are expendable, and they are available in large
quantities of inbred strains. You get all kinds of very precious,
very specific information from these animals, but we always have
to deal with the problem of extrapolating from these non-human
species to what is going on in people. We know that there is
often a great leap in extrapolating from animal data to the human
situation.
In epidemiology, we take advantage of the fact that we are
studying humans, but we have only an imperfect idea of what the
dose or what the extent of environmental stress is. There are
just a few parameters that we can measure, and most of them with
less than satisfactory precision.
Taken together all three disciplines give us the coherent
database from which we move toward regulatory action. With these
interrelationships and strengths and weaknesses of the various
approaches in mind, I would like to focus more on the discussion
of the rationale of designing human experimentation to study the
effects of environmental agents.
LOGISTICS CONSTRAINTS
I think it is important and absolutely necessary to take into
consideration logistic constraints. I have to face the fact that
I have only three sets of resources to work with: people, money,
and time. I cannot control the third resource, but given the
objective that targeted research has to be relevant to social
questions, you must decide how to go about choosing the right
question and answering that question with the resources at hand.
A laboratory may require $10,000 a day to operate. You have a
through-put of so many human subjects in a reasonable experimental
paradigm per unit of time, so you have to very carefully decide
how to use these scarce resources to address the question you are
trying to answer.
That is an integral part of the rationale of experimental
design. I could come up with many experimental designs that would
be appropriate, that would be scientifically meritorious, that
would supply answers to the questions in an intellectually satisfy-
ing way. But we do not have the luxury of pursuing research questions
in that particular way. We have to look at the end points, the
criteria for subject selection, to make certain that when we begin
a five-day study that all of the subjects will participate for
five full days, because the data each day are really quite ex-
pensive. These elements in the rationale of the study design are
more important than we scientists like to admit. In developing a
74
-------
rationale for an experimental design, it is important to have a
statistical design that is highly parsimonious.
I will give an example that will illustrate several points
with respect to subject selection, range of susceptibility, and a
few other points. The best example I can give is one having to do
with carbon monoxide research because it is very simple.
The carbon monoxide experiment was based on what we knew we
could do, in view of some of the material constraints. It was
based on what we knew carbon monoxide should do to people, and on
the fact that we had an ambient air quality standard for carbon
monoxide that was rather low. The automotive industry said the
standard was irrationally low, but there was no real basis for
their statement.
I discussed with a colleague what test he thought we ought to
conduct to study the effect of carbon monoxide on a particular
human function. Historically the effects of carbon monoxide on
the central nervous system had been the focus of research. My
colleague was a cardiologist who knew about exercise stress testing
and exercise electrocardiology. He said that the myocardium does
not function very well during relative hypoxia. As a matter of
fact, the blood leaving the myocardium has a partial pressure of
oxygen, a PO of 25 millimeters of mercury, which is as low as it
ever gets at rest. It does not get any lower at exercise. So, if
you do something to interfere with oxygen delivery to the myo-
cardium, you almost expect a priori to be able to measure some
effect. We could measure something going on in the myocardium at
relatively low levels of carbon monoxide. Here is how we did the
experiment. We had people get on a treadmill and do an exercise
ECG before they were subjected to a standard dose of carbon monoxide.
One of the nice things about this test is that we can measure
carboxyhemoglobin in the circulating blood, so we have an accurate
estimate of the body burden of this environmental agent. We do
not even have to depend on ambient air measurements, we just
measure carboxyhemoglobin in the blood. We gave the subjects a
certain amount of carbon monoxide to breathe, and we measured the
carboxyhemoglobin again, and then we had them go on the treadmill
again, and we looked at the electrocardiogram.
In young, healthy adult males, who were the least susceptible
because they should have the healthiest myocardium, we did not see
any changes in the electrocardiogram. We saw changes in heart
rate at various levels of activity. There was a price paid for
the carbon monoxide load, but it did not show up in that particular
objective parameter when we looked at the electrocardiogram.
75
-------
Then we postulated that a cohort of men whose mean age was 45
instead of 20 should include a few individuals who had preclinical
ischemia heart disease, and that they should show some changes
during the same experimental design. We tested that hypothesis
and found that our presumption was true, that there were a certain
number that changed their Minnesota code during the course of the
experiment. We postulated that if this were true, then people who
have well-defined, stable angina pectoris, people who have well-
characterized coronary artery disease, ought to be much more
sensitive to carbon monoxide. We tested that hypothesis, and it
was valid. This is an oversimplified presentation of this particular
set of experiments, but it serves as an example of logical progres-
sion of hypothesis testing with the results of each experiment
influencing the design of the subsequent one.
HISTORICAL ANTECEDENTS OF CLINICAL RESEARCH
Clinical research has been around for a long time in one form
or another, only recently have scientists in general, not just
physicians, begun to focus on man's relationship to his environ-
ment. I believe that the Harvard Fatigue Lab of the mid-1920's
represents the keystone of clinical environmental research. The
Harvard Fatigue Lab was housed in the Harvard School of Business
because the people who were interested in setting up the lab
wanted to investigate "the adaptation of the normal individual to
industry."
There was a certain amount of social responsibility in industry
at that time, and, as I read it, a hard-nosed desire to find out
how you could maximize human productivity under certain environ-
mental conditions, i.e., industrial conditions.
The techniques, the attitude, and the rationale for environ-
mental human clinical research were established by the investi-
gators in the Harvard Fatigue Lab around 1927. The 20-year life
of that institution takes us just past World War II. Interest-
ingly enough, many of the alumni of the Harvard Fatigue Lab are
active in clinical environmental research today. They moved on
during the course of World War II to broaden their perspectives,
and have led into what we have been doing for the last decade in
clinical environmental research. These investigators became
involved in cold and heat stress experimentation because of military
operations under those conditions. Of course, the human element
was the salient element of all that research—altitude physiology/
dividing physiology, and submarine physiology, as examples.
A modern corollary of that research is now being conducted in
the hyperbaric chamber at Duke University. Because of the quest
for new energy sources, investigators are studying ways man can
76
-------
adapt to 1,000 feet of sea water so that oil may be extracted from
the North Sea and other places.
We are not alone in developing a rationale for clinical
environmental research. We can rely on a number of years of high-
level investigator or research scientist history to help us develop
the rationale for clinical environmental research.
FURTHER CONSIDERATIONS IN DEVELOPING A RATIONALE
We must consider three further characteristics when develop-
ing our design rationale. The first characteristic is subject
selection. What is the rationale of subject selection? Perhaps
because most of my earlier research work was with animals, I
regret not being able to deal with nice inbred strains of mice
when I am conducting human experimentation. Because of that, I
suppose, I sometimes err in the direction of trying to design
clinical experiments where I have a very homogenous population. I
try to control as many of the covariants within the population as
I possibly can, to minimize both inter- and intra-subject variance.
The globe is not populated with 20-year-old robust Caucasian
males who happen to be students at the University of North Carolina.
Many inter-subject characteristics must be taken into consider-
ation—ethnicity, race, and geography, to name a few. There are
differences in cardiovascular performance and pulmonary per-
formance, and there are probably profound differences in metabolic
characteristics that we are just now learning to measure. What is
true for our experimental subjects may not be quite true for the
population at large. However, I think tests have to be based on a
highly homogeneous population so as to minimize the inter-subject
variance.
My tendency as a first step is to try to identify all of the
characteristics I can possibly measure in the subjects, and specific-
ally select for or against those characteristics. In this way, I
achieve as homogeneous a population as possible, albeit an artifical
one. Using the information from the first step, I inject that
information into the rationale for the succeeding experiments that
we do, changing the characteristics of subject selection as I
pointed out in the carbon monoxide experiments.
The second characteristic to consider in developing a rationale
is dose selection. Dose in the usual sense of the word is the
amount or level of environmental stress, or the quantification of
the environmental agent that will be used in the experiment.
There is a bit of circular logic involved in how you decide what
is going to happen with respect to your dose selection. At the
minimum, we like to make some statistical statements about dose
77
-------
selection and relevance because we are conducting the research for
a specific reason.
The third characteristic that must be considered in develop-
ing a rationale for clinical experimentation is the selection of
end points. As I indicated earlier, a cardiologist will not worry
about evoked potentials, and a neurophysiologist does not care
much about stroke volume. The person who designs research in
clinical environmental medicine has to take several backward steps
to look at what is known about the biology of the environmental
agents he is studying. He must design experiments that are con-
strained by staff and budget, and take advantage of an exposure
situation to explore a variety of end points that are judged to be
the most relevant biologically, and the most likely to yield
important information.
In summary, the rationale for designing clinical environ-
mental research is multifaceted, and has to do with research
objectives and available resources. Most importantly, because
both the objectives for the research and the resources that are
available are determined by the questions that society asks, the
rationale has to strive diligently to maintain a high degree of
relevance to the social questions that are being posed.
78
-------
Discussion Summary
The issue was raised whether experimental scientists should
concentrate on anticipating and solving problems, rather than
waiting for problems to come up and then solving them. Partici-
pants noted that some scientists currently operate according to
the former procedure. For example, scientists have been success-
ful in using biomathematical engineering to study the possible
effects of certain pollutants.
However, investigators said that they do not always have the
resources to design and conduct what would be considered antici-
patory studies. In addition, many of the topics chosen for study
are questions that are formally posed to the EPA by Congress and
that are important to the public (e.g., how safe is our environ-
ment?) Thus, the investigator has a limited amount of time in
which to conduct anticipatory studies.
Participants also discussed the drawbacks in conducting
experiments in a clinical rather than a natural setting. Is a
clinically manipulated environment representative of what goes on
in the natural environment? Discussants considered the case where
a group of factory women who, in a laboratory situation sweat less
than men, were observed on the job to be producing as much sweat
per hour as the males did in the clinical setting. Although this
points up the fact that the way a subject reacts in the laboratory
may not be the way he reacts in a natural environment, partici-
pants pointed out that no matter what group of subjects you are
working with, the investigator will always be faced with some form
of bias. For example, the factory women who disproved the theory
based on clinical research that women sweat less than men, in
themselves are not representative of women in general. They are
women who have chosen to work in a hot factory, who have chosen to
participate in the study, and so forth. Thus, any test reactions,
be they derived from a clinical or natural setting, are not
necessarily representative of what occurs in the general population.
79
-------
Finally, investigators discussed the need to allot adequate
time to completely examine and answer a study question. Partici-
pants felt that the Congress and EPA administrators appreciate the
need to plan long-range targeted research, and that money may be
set aside for anticipatory research. Anticipatory research is not
research that investigators plan to conduct in the future; rather,
it is research that will take many years to complete.
80
-------
Subject Selection, Investigator Interactions,
Informed Consent in Clinical and
Environmental Research
David A. Otto, Ph.D.
U.S. Environmental Protection Agency
Jeanne T. Hernandez
Institute for Environmental Studies*
A revolution is going on in psychology. A
different image of man is being tried as a guide to
research, theory, and application. Over the years,
theorists have conceptualized man as a machine; as
an organism comparable to rats, pigeons, and monkeys;
as a communication system; as an hydraulic system;
as a servomechanism; as a computer—in short, he has
been viewed by psychologists as an analogue of
everything but what he is; a person. Man is, indeed,
like all those things; but first of all he is a free,
intentional subject.1
—Sidney M. Jourard
PHILOSOPHICAL PERSPECTIVES: BEHAVIORISM vs. HUMANISM
During the past two decades a lively debate has ensued between
the proponents of the humanistic and the behavioristic traditions
in psychology and related disciplines in the social and clinical
sciences. This debate is directly relevant to the conduct of
human research, since many time-honored principles concerning the
relationship of the subject and experimenter have been called in
question.
* University of North Carolina
81
-------
Beneath the "sound and fury" of men like Sidney Jourard and
B.F. Skinner lies the fundamental philosophical question of man's
freedom to behave in a non-deterministic, unpredictable manner.
According to a strict behavioristic model (or Freudian model for
that matter), man's behavior is entirely determined by either
external environmental stimuli or unconscious inner motives and
drives bubbling up from the depths of the psyche. Although Skinner
and Freud may seem to be rather incompatible bedfellows, Skinnerian
and Freudian theories are both highly deterministic with respect
to human behavior: that is, both theories imply that human behavior
can be predicted and, to some extent, controlled by the appropriate
manipulation of internal or external stimulus conditions.
This deterministic view of man conflicts rather strenuously
with the traditional humanistic view of man as a rational being,
possessed of a variety of constitutional freedoms, including the
right to withdraw from any scientific investigation whenever the
spirit moves. In fact, the humanistic revolution in social psycho-
logy has spawned an entire field of research devoted to subject-
investigator interactions.2r3 The implications of the humanistic
revolution extend far beyond psychology, however, since the Federal
Government has now imposed strict regulations in accordance with
the National Research Act to safeguard the "free will" or voluntary
nature of participation in all HEW-funded human research projects."
In essence, these legal and ethical constraints are designed to
preserve the freedoms essential to the humanist view and to deny
the behavioral controls alleged to the opposing view.
Requirements of the National Research Act are difficult, in
some respects, to reconcile with scientific objectivity, as well as
specific objectives of environmental research. Miller 5 has succintly
summarized the dilemma: "The goals of meaningful scientific inquiry
may be at odds with the value of treating man in a dignified and
respectful manner." This paper will explore some of the potential
conflicts among the elements of informed consent and the conduct
of environment research.
The traditional objective of human research is to determine the
effect of some carefully controlled manipulation on behavior. In
other words, we predict that some external stimulus (such as noise)
or internal stimulus (such as food or drugs) will alter behavior in
a systematic way. If we seriously entertain the notion that man
behaves in a totally non-deterministic, unpredictable manner, then
human research would be a waste of time. We can easily extract
ourselves from this paradox by adopting a compromise philosophy
wherein some behaviors are subject to external control, while other
behaviors are mediated by internal volitional processes. Even the
most extreme humanist would be unlikely to deny the effects on
82
-------
human behavior of heat, cold, sleep deprivation, or carbon mon-
oxide poisoning.
What is the consequence of man's tendency to behave unpredict-
ably? The exercise of free will in the laboratory yields consider-
able noise in the data which the humanist calls "individual dif-
ferences" and the behaviorist calls "intersubject variability."
At this point the two investigators rapidly part company, for the
humanist is concerned with enhancing individual differences, while
the behaviorist seeks to minimize intersubject variability. The
manner in which subjects are selected determines, to a large
extent, how much heterogeneity or homogeneity the investigator may
expect. After a decade of behavioral research, however, we can
attest that the application of stringent criteria based on stand-
ard personality and medical inventories, physicals, and interviews
to secure "normal, healthy" populations has been remarkably unsuc-
cessful in reducing intersubject variability. No matter how
rigorously subjects are screened in preparation for human experi-
ments, there is little hope of ever attaining a pure laboratory
strain of Homo sapiensI
The problem of intersubject variability is particularly
severe in clinical environmental research where the objective is
to define the threshold level at which a given substance produces
significant impairment in function. Threshold effects, by defi-
nition, are "just noticeable differences." The challenge in this
field is to minimize individual differences, regardless of the
philosophical bias of the investigator.
SELECTION OF A PROTOTYPE SUBJECT
How do we go about selecting human subjects for environmental
health effects studies? Subject recruitment, though seldom pursued
as an independent profession, has evolved to an exacting science
in Chapel Hill. Since the University constitutes the primary
industry, the student population is the major, if somewhat transient,
source of subjects. That is, the normal, healthy, red-blooded
collegian represents our closest approximation to a standardized
laboratory animal.
One might object at this point that our choice of "prototype
man" is not particularly representative of the larger population
to whom we wish to generalize our findings. Dr. Shy, for in-
stance, argued at this symposium for the use of specific clinical
populations considered to be unusually sensitive or susceptible to
certain environmental insults. The rationale for using healthy
young adults is that any functional impairments observed can be
inferred to be more severe in susceptible populations. The prob-
ability of observing effects attributable to low level exposure
83
-------
may be artificially low in healthy young adults leading to Type II
errors of statistical interpretation. That is, we may conclude
that substance "X" produces no deleterious effects in the general
population when, in fact, the same exposure may actually produce a
functional impairment in an older or less healthy population. The
risk factor, however, is correspondingly lower in healthy young
adults than other populations.
How do we define our standard subject? Prior to initiating
any experiment, the principal investigator provides the recruit-
ment staff with a specific list of criteria. I will quickly
review the criteria used in selecting subjects for recent neuro-
behavioral and physiological studies of environmental insult. In
general, the criteria for selection of subjects to participate in
studies which involve at least a marginal health risk are consider-
ably more stringent than the criteria used in other studies which
do not entail any health risk. The monetary inducements also vary
relative to the associated risks or discomforts such as repeated
blood draws.
Ethical considerations preclude the exposure to risk of
children, minors, or any other population that cannot provide
informed consent. Current studies are limited to adults aged 18-
40, although middle-aged and elderly subjects have been used in
previous studies of the effects of CO on the onset of angina
pectoris.6
For pragmatic reasons, we have limited research to adult
males. The potential danger of toxicant effects on the unborn
fetus and the increased variability associated with hormonal
changes during the menstrual cycle are primary considerations. In
order to preclude the possibility of fetal damage, women must be
given a pregnancy test prior to each experimental session. Since
the observed effects of environmental toxicants, particularly at
anl osvchT T' tSnd t0 bS extremelv subtle, the physiological
as oo^ih^T °f SUbJeCtS mUSt be "^tained as constant
* E*PO™eS W0uld have ^ be carefully synchronized to
! menstrual cvcle * ™™* subjects to control
This strategy can be problematic in the present climate of
.iity. For example, three coeds once walked into our
demanding equal time and equal oav
The^i'riV!!!'1 ^ pr°gress in a transparent plastic exposure chamber,
bicycle ergometerVwhUet0 °bS8rVe * ****** ^° "** P^alin^ a
. ripped to the waist for EGG recording.
. * -^u!re 9S if they sti11 wished to participate, two
and withdrew their request. The third coed, however,
84
-------
smiled and asked when she could begin. The Institutional Review
Committee, however, had not authorized us to expose females to
ozone!
The variable of race, like sex, raises ethical and pragmatic
difficulties. Ewing et al.,7 for instance, have shown that blacks,
orientals, and whites are differentially tolerant to the effects
of alcohol. There is reason, therefore, to expect differential
racial responsivity to environmental stressors such as carbon
monoxide which is also considered to be a CNS depressant. If a
multi-racial population is studied, a sufficiently large sample
from each racial group is required to statistically evaluate
possible racial differences. The usual practice (which could
easily be construed as discriminatory) is to use subjects of a
single race in order to eliminate this source of variability from
the data. One strategy to avoid statistical and ethical conflicts
is to restrict subjects in an individual study to a single race,
but to use different races in different studies.
How do we define "psychological" normality? Prospective sub-
jects complete the Minnesota Multiphasic Personality Inventory
(MMPI) and must score below the 75th percentile, unless otherwise
specified by the principal investigator, in order to participate
in environmental studies. Scores on the MF scale are disregarded,
since most college students score high on this scale. The MF
scale reflects aesthetic interests more than the masculinity-
feminity dimension for which it was originally intended.
The use of the MMPI as a psychological screen is admittedly
arbitrary, although the test is standarized and widely used. The
MMPI is neither foolproof nor all inclusive. For instance, the
standardized MMPI profile does not indicate if a prospective
subject is claustrophobic. It is simpler to ask directly if the
individual is afraid of staying in a closed room for a long time.
Why should the environmental researcher be concerned about
the psychological normality of prospective subjects? The need is
readily apparent for physical examinations and careful scrutiny of
the medical history of prospective subjects to minimize the possibil-
ity of extreme physiological reactions to pollutant exposure. It
is also necessary to assess the emotional stability of prospective
subjects since prolonged confinement in a relatively small testing
chamber and/or exposure to a substance that is presumed to have at
least mildly toxic effects can induce considerable psychological
or emotional stress, even in psychologically normal subjects.
Individuals exhibiting any signs or history of psychological
abnormality should not be used in environmental health effects
studies.
85
-------
The potential psychological risks inherent in any clinical
environmental study are seldom considered seriously by investi-
gators and are rarely articulated to subjects during pre-experi-
ment briefings. In many studies, this oversight represents a
clear breech of informed consent! The measurement of pollutant
effects on psychological dimensions such as mood, anxiety, or
hostility has likewise been neglected and constitutes an important
area for future study in clinical environmental research.
Smokers have been excluded from experiments testing the
effects of carbon monoxide, ozone, and sulfuric acid aerosol.
This exclusion is particularly important where the test substance
is expected to alter oxygen uptake or respiratory function. How
long should prospective subjects have been non-smokers in order to
ensure "normal" lung function? Our physicians have specified a
minimum of 10 years in oxidant experiments, a requirement that has
strained the resources of the recruitment staff on occasion. In
essence, this requirement limits subjects in college-aged popu-
lations to those who have never smoked tobacco.
Should the same criteria apply to marijuana smokers? In
practice, a double standard is applied in the case of "pot" verus
"tobacco" smoking. Investigators have adopted a much more tolerant
attitude toward the occasional marijuana smoker than to the occasion-
al tobacco smoker. At least half of our undegraduate recruits
admit to the occasional use of pot, and the percentage is probably
much higher. Why the double standard? in this case, scientific
objectivity has been sacrificed to humanistic (and pragmatic)
considerations.
Other criteria used in subject selection include limitations
on the use of stimulant beverages. Subjects should not have
consumed more than two cups of coffee, tea, or coca-cola, or any
other liquid stimulant within one hour of the start of the experi-
ment. In addition, subjects cannot have consumed more than the
following limits of alcohol within the 24 hours preceding the
experiment (these limits vary for different experiments): in
neuro-physiological experiments, the limit is 24 ounces of beer,
or approximately two cans, within 24 hours of the test; 12 ounces
of wine, or approximately two glasses; three ounces of hard liquor,
two shots of whiskey, rum, or vodka, and so on.
Furthermore, the subject cannot take any drugs, prescribed or
otherwise, within 48 hours of the experiment. The term "drug-
includes antibiotics, antihistamines, barbiturates, amphetamines,
marihuana, heroin, cocaine, steroids, or other substances pre-
scribed or self-administered for health, happiness, or habit. In
some cases, the attending physician or psychologist may allow the
subject to take aspirin, bufferin, or related compounds before the
experiment.
86
-------
To avoid the problems of sleep deprivation or the converse
during the experiment, subjects are required to have six to ten
hours sleep during the previous night, and to sleep regularly
within this normal range over the course of the experiment.
Prior to exposure to any toxicants, prospective subjects
complete an exhaustive medical history/ the Duke Medical In-
ventory, and are given a physical exam by EPA physicians. Any
history of allergies or significant medical problems generally
excludes candidates from participating in an experiment.
Does the stringent selection process destine our experiments
to produce negative findings or no effects? Perhaps it does. In
other environmental laboratories in this county and in Europe
there is much feeling that an investigator would get better results
if he used a susceptible population.
INVESTIGATOR ATTRIBUTES AND INTERACTIONS
We have focussed thus far on attributes of prospective subjects
which might influence performance in an environmental health
effects study. Attributes of the experimenter can likewise affect
the performance of subjects and possibly bias experimental results.
This problem has been studied intensively by social psychologists
during the past decade. Rosenthal2 reviews a wide range of bio-
logical (e.g., sex, age, race) and psychosocial (e.g., anxiety,
hostility-warmth, authoritarianism) attributes of the experimenter
which affect subject behavior. For instance, college-aged subjects
tend to cooperate better with experimenters of their own age and
race, but of the opposite sex. The simple effects of biological
variables, however, can be counterbalanced by or interact with a
multitude of psychosocial variables. An experimenter perceived to
be anxious, hostile, or rigidly authoritarian will elicit less
cooperation from subjects than an experimenter perceived to be
confident/ warm, or tolerant.
These observations may seem banal, but researchers seldom
consider such obvious subject-experimenter interactive effects in
the design of a study or the analysis of data. Let us consider a
hypothetical example. The objective of a proposed study is to de-
termine the effect of low level oxidant exposure on treadmill en-
durance in healthy young adult males. Experimenter A is an at-
tractive, congenial young woman, and experimenter B is a quiet,
slightly over-weight, middle-aged man. Which experimenter is
likely to achieve better rapport with subjects? And how might
subject-experimenter interactions affect the results of the study?
Based on age and sex factors, we can predict that subjects
will probably establish better rapport with the female than with
87
-------
the male experimenter. In terms of performance, we can also
predict that subjects will probably exhibit greater treadmill
endurance for the female than male experimenter. The objective of
the experiment, however, is not to enhance subject-experimenter
rapport or treadmill performance, but to impartially determine the
effect of oxidant exposure on treadmill endurance. If we wish to
minimize the effects of experimenter attributes on performance we
would probably obtain less biased results from the middle-aged
male!
This example, furthermore, confronts us with a basic dilemma
in clinical environmental research. Subject-experimenter inter-
action is but one of many motivational variables which can affect
the performance of subjects. Since the effects of low level
toxicant exposure are likely to be subtle, motivational variables
of this kind can easily mask the effect of exposure and lead to
false negative results. On the other hand, the current Zeitgeist
of human research, mandated in many respects now by federal regu-
lations, is to treat subjects in a war™, open, humanistic manner.
Subjects are persons, not guinea pigs! The chaTIe^Tlfto strike
a balance between humanistic concerns and scientific objectivityT
SUBJECT EXPECTANCIES, INFORMED CONSENT, AND SCIENTIFIC OBJECTIVITY
gator™ction ^^'^V* ^ect-investi-
subject performance, what aubjLt
an experiment is an important detemin* I I ! °r tO happen in
experiment. Psychologists fre^ue prefer **£*"""*
variable as experimental or instruction^ « I . SXPec
familiar with this phenomenon '
be
other variables that the exper^Lf ^ **** ^ 6ffect8 of
studies, for instance, usSny^nclud! ^ tO t6St'
which subjects ^knowlngirreceivf f K P a°eb° conditio"
contain the active ingredients of \L*?*t°XlC* that d°SS not
receiving the "placebo" aredLef S% ^ be±ng tested' Subjects
to distinguish the genui^ eff "tl the,,in-sti9ator in order
often therapeutic,
*
. udth
from subjects the true objectives of ^ PaSt W&S Simply to conceal
under test. Although the necessitv tl ," ^ " ^ h^otheses
studies is generally accepted tn. * ^^ condit^ns in drug
other types of human research'if "**, °f dec^^ procedures in
institutional review boards? ^11^ *ly **ncti°™* any longer by
Full disclosure to
—-r o«i«^tj.onea any longer by
disclosure to subjects of the
-------
objectives/ procedures, discomforts, and risks constitute elements
of informed consent which must now be obtained prior to partici-
pation in any experiment.
We are faced again with a dilemma in which ethical consider-
ations and human rights may be at odds with scientific object-
ivity. A priori disclosure to subjects of the anticipated effects
of an experimental manipulation will obviously shape expectancies
and bias responses. Let's consider a specific example: an experi-
menter warns subjects that ozone exposure may cause temporary
irritation of the eyes, throat, and chest; headache; and nausea.
The experimenter cannot obtain a valid, unbiased measure of the
frequency or severity of these symptoms following exposure. Nor
can single- or double-blind control procedures be effectively
employed since the presence of ozone is readily apparent to both
subject and investigator.
THE ELEMENT OF RISK IN CLINICAL ENVIRONMENTAL RESEARCH
Specification of risk in environmental research also raises
contradictions. How can we adequately inform subjects of the
anticipated risks of exposure when one of the basic objectives of
environmental health effects studies is to determine empirically
what those risks are? On the other hand, we have no legal right
to expose humans to any conditions that might significantly impair
any vital function for an extended period of time. The margin of
risk in any proposed human study must be extremely small, and any
functional impairments produced by exposure must be transient and
completely reversible. The enigma of clinical environmental research,
therefore, is how to determine the threshold of human risk to
pollutant exposure without exposing humans to significant riskl
In conclusion, federal regulations to protect the rights of
human subjects severely limit the scope and conduct of clinical
environmental research. The U.S. Environmental Protection Agency
has been mandated by Congress to determine the risk to human
health of a wide gamut of potentially toxic substances. The
Department of Health, Education, and Welfare has likewise been
mandated to protect the rights of human subjects. Unfortunately,
the two mandates are not entirely congruent. Clinical environ-
mental researchers must carefully tread the tightrope between
these opposing mandates since the achievement of meaningful environ-
mental quality standards hangs in the balance.
89
-------
REFERENCES
1. Jourard, S.M.: Disclosing Man to Himself. Litton Educational
Publishing, New York, 1968.
2. Rosenthal, R.: Experimenter Effects in Behavioral Research.
John Wiley, New York, 1976.
3. Rosenthal, R., and Rosnow, R.L.: The Volunteer Subject. John
Wiley, New York, 1975.
4. PL 93-348 implemented in accordance with the Code of Federal
Regulations (45 CFR 46).
5. Miller, A.G. (ed.): The Social Psychology of Psychological
Research. The Free Press (McMillan), New York, 1972, p. 76.
6. Anderson, E.W., et al.: Effect of low-level carbon monoxide
exposure on onset and duration of angina pectoris: A study in
ten patients with ischemic heart disease. Ann. Intern. Med.
79:46-50, 1973.
7. Ewing, J.A., Rouse, B.A., and Pellizarri, E.D.: Alcohol
sensitivity and ethnic background. Am. J. Psychiat. 131:2, 1974,
90
-------
Discussion Summary
Participants discussed the importance of physical criteria in
screening subjects; for example, what is the subject's aerobic
capacity? Is the subject under or overweight? It was agreed that
all subjects should undergo a rigorous physical examination before
being allowed to participate in an experiment. It was further
noted that some investigators simplify the process of finding
healthy subjects for each experiment by forming subject pools from
which to draw subjects as they are needed.
Discussants also questioned how the investigator verifies the
information the subject tells him. For example, how can the
investigator be sure that a particular subject has never smoked a
cigarette? Or that the subject didn't drink four glasses of wine
the night before the experiment? Some investigators said they
take subjects at their word. Others devise series of questions
designed to elicit the truth from prospective subjects.
Participants discussed methods of screening middle-aged and
elderly subjects. In general, the screening method depends on the
purpose of the study. For example, middle-aged persons are regarded
as particularly sensitive to pollutants. Therefore, any tests
involving pollutants would necessitate testing to ensure that none
of the middle-aged subjects are over-sensitive to pollutants.
Discussants agreed that all subjects, regardless of age, must
be carefully tested, perhaps more intensively than they are at
present, to ensure the subjects' safety.
Assuming that there is no such thing as a "normal" man,
discussants questioned how investigators arrive at a set of criteria
to select subjects for specific experiments. The process of
determining a set of criteria was described as "evolutionary."
Thus, over the span of a number of tests, investigators gradually
91
-------
develop a list of variants that have proved to be important in
choosing subjects and preparing them for an experiment. For
example, after conducting a number of tests, investigators deter-
mined that a lack of sleep would distort a subject's test responses.
Thus, subjects are required to sleep a certain number of hours on
the night before the experiment.
In some instances, criteria are tailored to specific experi-
ments. For example, in tests involving respiratory measures,
investigators would want to know whether a subject has a history
of lung disorders.
Some discussants indicated that they have used female sub-
jects in their experiments with no problems. Other participants
urged the use of healthy middle-aged and elderly subjects in
experiments.
Finally, one participant suggested a randomization of the
double blind procedure, to avoid biasing a subject's test response
by telling him at the start of the experiment what effects to
expect. A subject would still know what test effects to expect,
but he would not know on what day he would actually be exposed to
the test substance.
Some participants felt that if a subject was sufficiently
susceptible to the powers of suggestion, it would not matter
whether or not he was exposed to the test substance because he
would react the same in either situation, m addition, partici-
pants said that if the effect of the exposure is strong enough,
responses induced by suggestion would not be critical.
92
-------
Acute Versus Chronic Studies
David Bates, M.D.
University of British Columbia
Vancouver, Canada
Acute studies are those that involve an exposure of less than
12 hours duration in a controlled environment, with observations
of changes in physiological function. Recent concern over air
pollution has undoubtedly precipitated an increase of interest in
what these studies may reveal and also in their limitations. If my
memory serves me right, in the 1890's, J.S. Haldane graphically
demonstrated the unique importance and value of human studies when
he breathed carbon monoxide when exercising and noted the effects
on himself. I recall that the experiment ended when he fell off the
bicycle ergometer. Haldane noted that these experiments gave him an
insight into the physiological effects of progressive carbon monoxide
intoxication, which he couldn't possibly have gained from animal
observations.
I believe that a few years from now, historians will be rather
surprised at how slowly the development of acute studies in a
controlled environment occurred. After all, the major pollution
episodes took place more than 20 years ago/ and it is still possible
to go into the medical literature looking for a specific answer to
an important question and find that very little experimentation has
been done. In the case of sulphur dioxide, for instance, we do not
know precisely the range of individual variability, or the association
of a high sensitivity to sulphur dioxide as measured by changes in
airway resistance, to other thresholds of sensitivity as measured
93
-------
by the inhalation of mecholyl or histamine. We do not have a
precise quantitation of the effect of mouth or nose breathing
in an atmosphere of S02 on the development of changes in airway
resistance. We do not know whether a two-hour exposure on one day
will influence the response in a given individual in succeeding
days. We do not know whether the acute response to sulphur
dioxide is any different in individuals who live and work in
relatively high concentrations of this gas and, in general, we
have very little acute laboratory data to put alongside our
guesses of its chronic effects. I mention these obvious outstand-
ing questions to indicate that it should be a matter of surprise
to us that one can ask these and find that we do not have reliable
laboratory data to answer them, when we have known for a long time
that sulphur dioxide is one of the principal constituents of the
older type of air pollution episode.
ACUTE EXPOSURES
In general, we require that acute studies mimic actual
exposure conditions and we try to isolate independent variables.
This may be a difficult task since exercise alone without any
environmental contaminants will cause a decline in expiratory
flow rate or an increase in respiratory resistance in asthmatics.
In the case of oxides of nitrogen, we have recently been given
the results of an acute exposure experiment in asthmatics,1
which indicates that this gas in relatively low concentration
enhances the response of the asthmatic to a subsequent challenge
by mecholyl. This is a good example of the value of such acute
experiments. It has been suggested that an exacerbation of
spontaneously occurring asthma is one of the earlier effects of
oxidant pollution. The acute experiments on ozone have undoubtedly
helped us to define better the levels at which significant airway
obstruction develops, and have, in addition, given us an important
idea of individual variation. We have also found evidence, I
think for the first time, that the response to this gas, when
acutely encountered, may be lower in those who customarily live
in an area of relatively high oxidant pollution.
Taken together, these and many other studies give us a good
idea of the kind of information that such acute experiments can
provide. I would tabulate these advantages as follows:
• The determination of the earliest measurable physiological
or biochemical effect.
• The study of individual variation and the modification
of effects by other pollutants or by other factors
such as exercise or heat.
94
-------
• The study of individual adaptation or modification of a
response by previous exposure.
• The study of the mechanism of action of the gas concerned
and the relationship between early symptoms and changes
in function.
Acute human studies are also important because they form a
part of the total evidence of the effect of a pollutant, fitting
between animal data and chronic exposure studies. It is when data
from all three sectors begin to show some general concordance that
we begin to be confident of the emerging picture. I do not
believe that that picture can be satisfactorily completed unless
acute human exposure data are made available.
As I have mentioned earlier, the surprise to me is that such
acute studies have taken such a long time to be developed. I
haven't much doubt that we still have a great deal to learn from
such studies and I think that it is very important that the present
tendency to discount or to make difficult such human studies
should not discourage investigators from trying to undertake them.
OCCUPATIONAL EXPOSURE
In listening to a discussion on the precautions that must be
taken by investigators to ensure that volunteers fully understand
the nature of the risks involved in such experiments, it occurred
to me to wonder whether, or how often, we apply the same standards
in relation to occupational exposure. Are the hazards of asbestos
discussed with those who are going to handle the material? Are
workers in the rubber tire industry told about the increased
incidence of bladder cancer in workers in some sections of that
industry? Interestingly enough, a recent Royal Commission in
Ontario2 has recommended that because the radiation hazard of
uranium was not explained to the miners in advance, and they were
not told of the hazard of this exposure when paired with cigarette
smoking all men in that occupation who develop lung cancer should
be considered eligible for full compensation.
I have a feeling that some people are unenthusiastic about
acute human studies because of a particular perspective on the
problems of decision-making. Their argument would run somewhat
lika this- "I am not interested in the level of a gas sufficient
to cause reversible bronchospasm in a normal individual, unless
vou are nrepared to tell me that reversible bronchospasm is a
disease." I think this argument requires discussion. I note that
tne Sorid Health Organization report on atmospheric pollutants
laid down four categories of pollutant levels, which were designed
95
-------
to be helpful to those considering exposure guidelines. Their
Level III reads as follows:
"Concentrations and exposure times at and above
which there is likely to be impairment of vital
physiological functions or changes that may lead
to chronic disease or shortening of life,"
This is a convenient definition since obviously a concentra-
tion of the gas sufficient to cause acute airway obstruction is
causing impairment of a vital physiological function. We there-
fore do not have to distinguish between that effect and the
possibility of causation of a chronic disease in setting a pollu-
tant level within the category of Level III.
I think, that this objection to acute studies is really quite
illusory. We are very likely to make a decision that concentra-
tions of gases high enough to cause easily measurable impairment
of airflow in normal people is not an environment that we should
accept as strictly necessary or desirable. Those who would argue
that reversible bronchospasm is not a "disease" are presumably
prepared to argue that reversible bronchospasm can be safely
ignored in the setting of ambient standards. As far as I am
concerned, in that discussion the burden of proof rests with them
to demonstrate that repetitive bronchospasm of that kind is not
producing adverse effects, before we should accept that the
environment that produces it should be considered acceptable for
the general public. I do not feel therefore that this objection
to acute studies, which is essentially an attitude of "we won't
know what to do with the results when we get them," is one that
should play much part in determining their appropriateness.
The other objections to acute studies have to do with the
necessary artificiality of the protocol. This is unavoidable if
you want to eliminate a good many variables present in the normal
outdoor environment. But it may be a valid objection that ethical
considerations prevent especially sensitive people frolbeing
studied and preclude studies on those whose carlo-pulmonary
acutely ust? C™1Vd' * ^ alread* "ned a recent
acute study using nitrogen dioxide at very low concentration in
in
h<^thy normal people in the
96
-------
CHRONIC EXPOSURES
In my opinion, chronic experiments on pollutants are not
really possible in the laboratory. I do not have a precise
definition of chronic, but exposures lasting weeks or months do
not seem feasible in terms of human experimentation, and there-
fore, the only "laboratory" that can be used for such studies is
one in which individuals are already exposed to higher levels of
some materials than are the rest of us. These chronic exposures
give us an opportunity to study comparative morbidity in different
occupational settings, and without doubt they are helpful in
defining the upper level of exposure permissibility.
In the case of oxides of nitrogen, the studies of tunnel
workers in New York exposed to relatively high concentrations of
automobile exhaust indicated a probable upper level beyond which
overt and obvious respiratory consequences would be detectable.
In the case of sulphur dioxide, a recent study of smelter workers
exposed chronically and intermittently to less than five parts per
million of S02*will prove very valuable in indicating the conse-
quences of exposure to sulphur dioxide at that concentration on
ventilatory function. In the case of ozone, it would be helpful
to have detailed knowledge of any deterioration or lack of it in
welders or in air crew who are intermittently exposed to ozone
levels high enough to cause characteristic symptomatology. The
recent reports of acute ozone symptoms occurring in some flights
from the United States to Alaska indicate the kind of opportunity
that could be taken advantage of once the actual levels of ozone
exposure were accurately known.
In the case of carbon monoxide, there are population groups
exposed to high concentrations who may well be used to provide
information on its long-term effects. Because of the common
dissociation between those involved in environmental work and
those involved in occupational medicine, there is a need for
bridging the gap between the two disciplines, and there is
certainly good use to be made of data secured in one kind of
experiment in relation to decisions made in another sector.
There are valid objections to this kind of chronic study that
are relevant to decision-making, particularly when they apply in
the occupational setting. The work force that is exposed to
these materials is not the same as the general public. It is
sometimes argued that, with a specially selected work force and
careful medical surveillance, higher exposure levels may be toler-
ated than you would allow for the population at large. But such
a point of view often takes for granted the existence of detailed
and precise medical surveillance, which, in my experience, is not
97
-------
««iitv In spite of these objections, every opportunity
The taken lo make measurements of individuals breathing higher
"ncentratfons of some materials than the rest of us, and I cannot
see anything wrong with that kind of scientific opportune.
The whole field of epidemiological study may be viewed as the
loaical way for us to begin to understand effects. In respect to
carcinogenesis, of course, it is the only tool open to us except
for laboratory testing of suspicious materials.
STANDARD-SETTING
Before I conclude, I wish to reflect on the relationship
between all these modes of experimentation and decision-making on
standards. It seems to me that it is the concordance of data that
provides the most convincing case on which defense of any single
number for human exposure can be successfully mounted. There are,
however, some other considerations that I feel I should draw to
your attention.
First, in reading the voluminous and expanding literature on
standard-setting, I think it is as well to remind ourselves that
the essence of the standard-setting process is to be able to
advance a hypothesis on the basis of the best information avail-
able. The hypothesis may be simply that exposure of a normal
population to not more than some level of a contaminant will not .
have any adverse effects on the physiological function of the
individual, nor any long-term effects on his health. I would
argue that this kind of hypothesis is not greatly different from
the kind of hypothesis that scientists are constantly advancing in
respect to laboratory experiments, and the deductions that may be
drawn from them. There is sometimes a tendency to blur this
essential similarity by arguing that in one case the scientist is
dealing with "hard fact," and in other cases, the information is
relatively soft.
We can also recognize that it is quite easy to frame questions
from the political sector that are beyond the scope of any kind of
scientific experiment to answer in any precise way. However, what
is being attempted in the standard-setting process is to look
carefully at some hundreds of scientific papers and to distill
from the accumulated data the best available hypothesis for protec-
tion of the public.
In that endeavor, both acute and chronic studies are neces-
sary, and there doesn't seem to me to be any reason to favor one
kind of study over the other, since they answer different kinds of
questions and provide a different sort of information. What seems
98
-------
to me to be indisputable is that both kinds of information are
needed, as is the information we get from animal toxicological
studies, and one in no sense excludes the value of the other. The
initial title I was given to address on this program perhaps
suggested that there was some kind of competition between the two
kinds of studies I have talked about, but this doesn't seem to me
to be an appropriate attitude. What seems to me more necessary to
emphasize is how little of this general work has been done in view
of the obviously pressing nature of the questions that naturally
arise from the man-made environmental deterioration to which most
of us are exposed.
Second, I feel that in the acute studies and, indeed, in some
of the chronic ones, there is a special advantage in studying
physicians. Our initial studies of 0.75 parts per million of
ozone5 were all done on physicians for several reasons. They
could be presumed to understand fully the nature of the risk;
they might be specially capable of recording early symptoms; and
we felt we could rely on them to discontinue the exposure if
untoward symptoms occurred. In some situations, therefore, there
is a special merit in adopting that kind of protocol. In chronic
studies too there are some advantages to studying physicians and
you will remember that Dr. Richard Doll's original study on ciga-
rette smoking and lung cancer was based on a prospective study of
British doctors, partly because he expected that the diagnosis
would be more accurate in them, and also because he hoped that he
would have a lower dropout rate.
Third, I think we should not only view our work as being
necessitated by questions that society is insisting should be
answered, we should accept the responsibility of generating the
questions and of working towards knowledge that we can see is
going to be needed, often in advance of any public appreciation
that the questions are indeed important. It is important to
preserve the opportunity for innovative research in advance of
public opinion.
In that connection I might end by telling the story of the
encounter in the street at night between a slightly inebriated
man who was circling around under a street lamp, and a helpful
passerby who asked him what he was doing. "I have dropped my
keys somewhere," the first man said, and the second joined him in
looking for them. After a few seconds, the second man inquired,
"Are you sure you dropped them here?" And the first man replied,
"No, but this is where the light is." In reviewing our research
effort, we must always make room for the important idea in a field
not yet illuminated by the glare of public attention.
99
-------
REFERENCES
1. orehek, J., Massari, J.P., Gayrard, P., Grimaud, C., and Charpin,
j.: Effect of short-term, low-level nitrogen dioxide exposure on
bronchial sensitivity of asthmatic patients. J. Clin. Invest.
57: 301-307, 1976.
2. Report of the Royal Commission on the Health and Safety of Workers
in Mines. Province of Ontario, Toronto, Canada, 1976.
3. Ayres, S.M., Evans, R., Licht, D., Griesbach, J., Reimold, F.,
Ferrand, E.F., and Criscitiello, A.: Health effects of exposure
to high concentrations of automotive exhaust. Arch. J5nv. Health
27: 168-178, 1973.
4. Smith, T.J., Peters, J.M., Reading, J.C., and Castle, C.H.:
Pulmonary impairment from chronic exposure to sulfur dioxide in
a smelter. Artier* Rev. Resp. Pis. 116: 31-39, 1977.
5. Bates, D.V., Bell, G.M., Burnham, C.D., Hazucha, M., Mantha, J.,
Pengelly, L.D., and Silverman, F.: Short-term effects of ozone
on the lung. J. Appl. Physiol. 32: 176-181, 1972.
100
-------
Discussion Summary
Participants discussed the problem of distinguishing defense
mechanisms from toxic responses in the lower dose acute studies.
Whether or not they trigger reversible or irreversible changes,
defense mechanisms cannot be equated with toxic responses. Some
participants questioned the exact meaning of "toxic response."
For example, sweating is a normal non-toxic response to heat, but
excessive heat, even with the sweating response, can kill a person.
Further discussion focused on the phenomenon of a repetitive
stimulus that provokes a physiological response. Participants
noted that evidence is mounting to indicate that a repetitive
stimulus that causes a physiological response may produce long-
term adverse effects. For example, a follow-up study of 18-month-
old infants afflicted with bronchiolitis revealed that at ages
seven to ten, these children showed compromised pulmonary function.
Participants agreed on the need for more follow-up data and acute
studies on repetitive stimuli.
Participants also discussed the trend away from controlled
experimental studies that last longer than twelve hours. Studies
of this duration were not thought to be useful to the researcher
for two reasons. First, most exposures occurring in the environ-
ment are episodic; for example, an exposure to a pollutant. Thus,
a study that exposes a subject to a pollutant for an extended time
period would not be a valid model for an investigator.
Second, studies that run longer than twelve hours are not
long enough to be valid models for studying long-term effects.
For example/ if a researcher were interested in studying the
effects of long-term exposure to sulfur dioxide, the appropriate
study subject would be a smelter worker who had been exposed to
the substance for the last five years. Thus, the experiment that
lasts longer than twelve hours is usually too short or too long to
be practicable.
101
-------
Role of Automatic Data Processing in
Clinical Research
Frank Starmer, Ph.D.
Duke University Medical Center
When discussing automatic data processing in the setting of
clinical research, it is helpful to have a sharp idea of what
clinical research is. For this discussion, we will define clinical
research as the activities associated with acquiring and transform-
ing clinical or patient experiences into knowledge. This knowledge
allows us then to predict or anticipate outcomes, given a character-
ization of a patient and his surrounding environment. This knowl-
edge of patient-environment interaction is thus helpful not only
in establishing effective treatment, but also in developing policies
for maintaining our surroundings that are compatible with a desired
quality of life.
The basic tool of the clinical investigator is the stimulus-
response experiment. Experimental preparations are characterized,
as well as the stimulus or change in environment with which the
preparation exists. Certain signals representing function of the
preparation are monitored over time and changes are detected.
Because signals sometimes contain variations over space or time
that seem unrelated to the experiment, a control study is usually
carried out where the environment is standardized. Thus, changes
in the monitored signals obtained from the experimental study are
considered significant only when they exceed the changes observed
in the control study.
103
-------
SOFTWARE
Data processing plays a number of roles in the support of
clinical research as described above. First, signals from patients
or experimental subjects must be acquired. These signals usually
contain much redundancy and therefore are not dealt with in their
virgin form. The signals are first transformed into a set or
sequence of primitive elements.1' *' 3 These primitives are assumed
to allow adequate description of the original signal. The
representation of a signal by a sequence of primitives is called a
message. As an example, when dealing with the electrocardiogram
we can choose several classes of primitives to construct a message.
One class contains the characters p, q, r, s/ t, where each
character represents a certain feature. A message constructed
from this set of primitives might appear as
P_qrs_t ___ P_q_rs__t ___
Another class of primitives might be the set of integers where the
integer selected is the amplitude of the signal at a
time. Thus, the sequence
1 2 3 2 1 0 0 -1 -2 -3 -3 -2 2 4 10 20 10 4 2
-2-3-2-1000012321000
n
in
in the pattern recognition.
features must be
of consistency
Ascribe the
H°WeVer' tO be useful<
and organized in a
role where
clinical research
messages and distribute «,«
derived features to 9S3'
management
such, can be
and data analysis actvities
°f
features from incoming
comP°*ents , and
*** bulk
»^«»««t and as
data
104
-------
Although feature recognition is not usually considered part
of data management activities, I would submit that it, in fact,
is. Feature extraction or recognition, such as computing heart
rate from the EKG, many times does not need to occur in real time
as does data acquisition. Therefore, it can and probably should
be deferred to minimize the overhead during data acquisition. The
logical place to put feature selection is in the activity that
deals with message and file management, i.e., the data management
activity.
Another reason for considering feature extraction as part of
data management is that it tends to be evolutionary in nature.
For instance, during the early phase of an experiment, heart rate
may be the only parameter thought to be of interest. However, as
the experiments progress, it is observed that S-T segments really
seem to reflect the stimulus and therefore should be analyzed. If
the original data were represented in sufficient detail by primi-
tives, then this new feature could be extracted from old studies
without reacquiring the primary data. This activity is clearly
data management.
Data analysis is perhaps the most difficult area in the
support of clinical research. Because data analysis tends to be
evolutionary during the life cycle of a research project, the
tools for supporting analysis must be flexible. It is here that
the flexibility of a computer can be used to a considerable extent.
Data analysis is the combination of selecting subsets of the
original database, and analyzing selected variables within these
subsets or comparing subsets. Since the degree of subsetting
cannot be easily forecast, it is important to organize a data file
in such a way that subsetting is easy. For this reason, a flat
file or matrix file is a useful data structure.1* The rows of the
matrix represent instances of the research protocol (1 row = 1
patient), while the columns represent the data values of various
parameters. The matrix can be represented in either row order
(direct file) or column order (transpose file). The transpose
representation is most useful for subsetting, while the direct
file is most useful for data analysis.
In a computing system supporting data analysis, flexibility
is achieved by avoiding the binding of variables in a user program
directly to columns in the matrix. By interposing a dictionary
that associates variable names with column positions between the
user program and the data files in the matrix, the accessing
program can avoid the need to know the exact location of variables
in the file structure. Thus, adding or changing variables simpli-
fies modifying dictionary entries, while the software driving the
105
-------
data analysis remains unchanged. Program maintenance is minimized
using such dictionary driven databases.
HARDWARE
The partitioning of research support activities into data
acquisition, data management, and data analysis leads naturally to
a distributed view of the hardware used to implement these activi-
ties. In our own laboratory, we have developed the support soft-
ware in such a way that the communication between data acquisition,
data management, and data analysis is relatively clean and well-
defined. A clean interface then makes it possible to support
various components on a variety of hardware devices. For instance,
while text based data entry (history, physical, etc.) is supported'
by the computer that primarily manages the database, some of the
graphic data entry activity is supported by remotely located
microprocessor based systems. In addition, while some data analysis
is performed by the data management machinery, we also utilize the
resources of our university computation center.
The point of all this flexibility is to allow for the known
unknown, that is, we all know that tomorrow we would like to
improve some aspect of the experiment but we don't know exactly
what. Decentralizing data acquisition in particular is very
helpful in this regard, if a small micro or mini computer
supports some phase of data acquisition, it can be changed without
much involvement of other hardware/software functions, whereas
if^h JT9 activitv d°*s not lead to easy change. Similarly,
if too data, manaf™ent ^rategy is supported on a dedicated system,
it too can be modlfled without much long-range effect. With
switchin * dr°!f lng' SU<=h Sn aPProach aPPears viable. Task
" minimized at
switchin .
cation SLTET " minimized at the ~»t of increased communi-
SYStemS* Jt W
watch h Jt WU1 certai«ly be interesting to
watch the progress over the next few years.
SUMMARY
106
-------
REFERENCES
1. Horowitz/ S.L.: A syntactic algorithm for peak detection in
waveforms with application to cardiography. Comm. A.C.M, 18:
281-285, 1975.
2* Kirsch, R.A.: Computer determination of the constituent
structure of biological images. Comp. Biomed. Res. 4:315-328,
1971.
3. Cox, J.R., Fozzard, H.A., Nolle, F.M., and Oliver, G.C.: Some
data transformations useful in electrocardiography. In
Computers in Biomedical Research. (Vol. 3) Academic Press, pp*
181-206. 1969.
4. Starmer, C.F., Rosati, R.A., and Simon, S.B.: Interactive
acquisition and analysis of discrete data. Comp. Biomed. Res.
5:505-514, 1972.
107
-------
Discussion Summary
In a discussion of the computer operator's role in processing
clinical data, participants said that the increased reliability on
the operator to review complex records as they are processed
increases the reliability of data analyses. This is true particu-
larly in those instances where the computer program does not seem
to be acting reliably or where extraneous signals are coming in.
Participants said that in some cases, the computer operator is a
laboratory clinician.
Some of the discussants prefer to use an automatic data
acquisition system that eliminates the necessity for operator
interaction. Participants said this procedure ensures that an
objective data base will be produced. These discussants felt that
using operators in the data review process introduces an element
of doubt as to the reliability of the resulting data base.
Participants also discussed the role of automatic data
processing in patient care. When the patient enters the hospital,
an intern records his medical history and chief complaints on a
check list form distributed by the data processing laboratory.
Next, the laboratory determines whether the patient's data is
accurate by checking the collected data against other hospital
test results. If there is a discrepancy between the laboratory
data and hospital test results, the patient is re-examined.
Discussants said that automatic data processing improves the
quality of patient data by double checking its accuracy.
109
-------
ENVIRONMENTAL AND PHYSICAL SAFETY
CONSIDERATIONS IN HUMAN
EXPOSURE FACILITIES
Moderator: Russell Pimmel, Ph.D.
-------
Environmental Controls and Safeguards
Morton Lippmann, Ph.D.
Department of Environmental Medicine
New York University Medical Center
The test environment, and, in particular, the exposure atmos-
phere, are of primary concern in controlled inhalation studies.
There are many physical factors in the test environment that can
affect the results of an experiment* Temperature, humidity, light
and noise levels, and many other factors need to be stabilized,
controlled, and described adequately. However, considering the
focus of this conference, it seems most appropriate to emphasize
the exposure atmosphere.
It is necessary to control the environmental variables in an
exposure atmosphere for two basic reasons* One reason is to protect
the subjects from accidental exposures and overexposures. The other
reason for controlling environmental variables is to protect the
integrity of the experiment. If, because of faulty instrumentation
or sloppy work, the subject is exposed to the wrong level of contami-
nant, the integrity of the experiment becomes questionable. Extra-
neous materials in the exposure atmosphere can also affect an experi-
ment's re sul t s •
The integrity of the exposure level must be assured in order to
interpret the experimental data accurately. For example, slug
exposures may elicit different responses than steady exposures, and
although we generally design experiments to have a constant level of
exposure, we do not always achieve it. In some of the older studies,
which were performed without the aid of modern instrumentation,
113
-------
investigators had variable exposure levels during the course of
an exposure. Sometimes the reports simply indicated the average
concentration, without any indication of how much the exposure
level varied.
DESIGNING A FAIL-SAFE EXPOSURE SYSTEM
There are some basic principles of control which are used in
designing a fail-safe exposure system. Control valves can be
either normally open or normally closed, and must be energized to
turn them to their alternate position. Such controllers will
revert to their normally open or closed mode when the power fails.
Clearly, in designing a test system involving human exposures, the
investigator would want to make sure that whenever there is a
possibility of equipment failure, the exposure would be reduced or
stopped, or at least maintained at a near constant level. The
investigator cannot have a system where the supply valve of a gas
bottle is left open at the same time that the ventilation or
dilution air closes down because of a fan failure. This could
lead to the subjects being over-exposed to the test substance,
with possible serious consequences. In1 addition, the integrity of
a whole series of experiments could be ruined.
Another principle or variable that has to be considered in
experimental design is the temporal response of the test system.
How rapidly can the exposure concentration be raised or lowered to
achieve the appropriate level of exposure? If the concentration
starts to go up or down for reasons extraneous to the experiment,
how rapidly can the investigator bring it back? How quickly can he
take action, either manually or automatically, that will keep the
concentration within prescribed tolerance limits? These are the
basic elements of feedback response that have to be considered in
the experimental design, and these variables are part of the
specifications investigators have to establish in the fabrication
or construction and installation of test equipment.
A selection must be made between manual and automatic control
of the test environment. How much automatic control is necessary
or desirable, and how much interaction should there be between the
analog output of concentrations and the individual operator or
mechanical controller who is doing the fine-tuning of the concen-
tration? Clearly, for most investigators, the answer to this
question depends not only on what method is ideal, but also on
what method is compatible with the budget. The absence of sophis-
ticated automatic controls and data processors need not limit
experimentation. With adequate means of manual control, many
kinds of experiments can be performed quite successfully.
114
-------
Redundancy is important in the generation, monitoring, and
control of the test atmosphere. We do not want an experiment
interrupted or endangered by the failure of one component of a
system. It is relatively inexpensive to provide alternate means of
at least monitoring and controlling the concentration. This way, if
one element fails, the investigator can finish the day's experiment
on the back-up system and not have to worry about losing control of
the experiment or terminating it prematurely.
Finally, the investigator must be somewhat compulsive about all
the problems that could occur in the experiment, and have contingency
plans to follow in the case of an emergency, in order to protect the
subjects and the integrity of the experiment.
FORMATION OF THE EXPOSURE ATMOSPHERE
The first consideration in a discussion of test atmospheres is
methods of atmosphere generation. This discussion will be restricted
to areas relevant to human studies, primarily concentrating on the
ambient pollutant gases and aerosols of both oxidizing and reducing
atmospheres. Current studies involve chambers and exposure equipment
for studies with sulfur dioxide (SO2 ), nitrogen dioxide (NO2 ), ozone
(0, ), carbon monoxide (CO), and sulfur oxide (SC^ ) particulates.
GENERATION OF GASEOUS ATMOSPHERES
Carbon monoxide, sulfur dioxide, and nitrogen dioxide are
readily available as bottled gases, and we do not have to worry
about extraneous contaminants being introduced in the process of
generating the gas. On the other hand, for gases formed in chemical
conversions, other products might be introduced during the reaction
process. There are well-established systems for metering compressed
gases into a dilution airstream to serve as the exposure atmosphere.
Both SCfe and N02 are highly corrosive vapors, and should only be
passed through chemically resistant lines. For NCj , which condenses
to a liquid at room temperatures, the lines must be heated.
Ozone, which is a very reactive and unstable gas, cannot be
bottled. The conventional system for generating ozone uses an
ultra-violet light source to form ozone from atmospheric oxygen.
Aerosols, of course, cover a broad range of chemical composition.
An aerosol is a suspension of particles in air. These particles can
be either liquid or solid, and they come in all different particle
sizes. One specific aerosol used in several current human studies
is sulfuric acid. Other human studies have used other SOX aerosols;
in particular, ammonium sulfate and ammonium bisulfate.
115
-------
AEROSOL GENERATION
All of the SO aerosols are soluble in water. The general
technique for generating such aerosols into human exposure environ-
ments is to use a nebulizer, which shears a liquid into a stream of
fine droplets of the appropriate size. The size of the droplets is
somewhat controllable, depending on the design of the nebulizer, and
whether it is driven by compressed air or an ultrasonic transducer.
The particle size is also affected by the concentration of the
chemical in the water being nebulized.
The nebulizer can generate a limited range of droplet sizes,
and since the equilibrium size depends on the humidity of the air,
the generators produce both water vapor and droplets, which are more
concentrated in terms of the solute than the liquid being nebulized.
Using conventional nebulizers, the investigator can readily
generate particles whose predominant size ranges from 0.1 to 1 micron.
This is usually an appropriate size in air pollution studies, because
a large part of the SOX aerosol in the atmosphere falls into this
size range.
A single particle size is not generated by any conventional
nebulizer. The investigator generates a particle size dispersion, a
so-called hetero-dispersed or poly-dispersed aerosol. This is not
necessarily a bad technique, in that it produces an integral exposure
that corresponds to the exposure people are exposed to in the
environment. On the other hand, in calibrating our systems for
their response and the response of our monitors for particles,
response is usually very much dependent on particle size.
MONO-DISPERSED AEROSOLS
Sometimes mono-dispersed (particles of the same size) laboratory
aerosols are needed. There is no uniform method of generating a
mono-dispersed calibration or test aerosol over the entire range of
particle sizes* The particle sizes that we are concerned with run
from about a few hundredths of a micron to ten microns.
The vibrating orifice generator (Figure 1) is one method of mono-
dispersed aerosol generation in general use recently for particles
of about one micron and larger. A pressurized liquid stream is
forced through a small hole, which is either three, five, or ten
microns in diamteter, depending on the desired size range of the
droplets. An aerosol will form when the liquid stream is forced
through the hole under pressure. However, such aerosols will be
relatively large and hetero-dispersed. If this technique is combined
with a high frequency electrical disturbance, which translates into
116
-------
AEROSOL
OPTIONAL—
Kr-85 NEUTRALIZE? •
(TSI MODEL 3054)
DROPLET GENERATION
AND
DISPERSION ASSEMBLY
Schematic
of particle
formation.
o
o
0
o
JL
COMPRESSED
AIR •
(15PSIG)
SYRINGF
PUMP
Figure 1. Schematic diagram of vibrating orifice aerosol generator.
Source: TSI, Inc., literature.
a mechanical disturbance, the investigator can make the jet break
up into uniform droplets. With an appropriate frequency applied to
the piezo-electric transducer, the droplet size will be about
three.
With the vibrating orifice, the investigator can produce
droplets with a geometric standard deviation as low as 1.02, or
just as mono-dispersed as possible in a laboratory, and comparable
to the dispersion nature achieves with various pollens. Residual
solid particles of a much smaller size can be obtained by using
dilute liquids. This method is preferred for forming particles
larger than one micron.
A system cannot be run reliably with an orifice smaller than
about two or three microns in diameter. So, the investigator is
limited to droplets of six to eight microns and to solid particles
of about half a micron at the smallest.
117
-------
For the larger size particles, there is an alternate to the
vibrating orifice technique/ the spinning disc, which produces
aerosols of essentially the same size range and concentration as
the vibrating orifice.
A different technique can be used to produce a narrow range of
particle sizes in the very small size range. As shown in Figure 2,
mobility classifier introduces the aerosol with unit electrical
charges at an outer annulus of a cylinder, and the particles
migrate toward the central electrode, propelled by their electrical
charge. There is a clean air sheath that all the particles have
to traverse. This is important because it means that all particles
have essentially the same radial distance to cross the central
electrode. If the particles of the same mobility came in across
the whole cross section, those that started nearer to the electrode
would reach it first.
In a poly-dispersed aerosol, where all the particles have one
charge, their migration against the aerodynamic drag of the airstream
toward the central electrode is determined by their size. If, at
this point, there is a place to withdraw the aerosol on the electrode,
Charged
clean airT06'080'
Figure 2. Schematic diagrams of electric mobility analyzer tube as
used in mobility size analyzer (left) and as particle
separator (right). Source: Fine Particles (Liu, B.Y.H.,
ed.). Academic Press, 1976, p. 599.
118
-------
only particles of a very limited range of electrical mobility and
particle size will be directed there. The aerosol drawn through
the open path will have a narrow range of particle sizes. One can
vary the particle size range drawn out at that point by varying the
air flow velocity or the voltage differential. This, in effect,
enables the investigator to go from a few hundredths of a micron up
to about half a micron in mono-dispersed aerosol generation*
This technique is inefficient in that only a fraction of the
generated poly-dispersed aerosol is used. However, if the investi-
gator is using a dye or some relatively inexpensive material as a
calibration aerosol, this disadvantage presents no great problem.
This technique cannot be used for larger particles because at
above about half a micron, the particles tend to have more than
one charge and can no longer be sorted according to their mobility.
MONITORING THE EXPOSURE ENVIRONMENT
There are various approaches to monitoring the exposure atmo-
sphere and characterizing the concentrations of the chemicals to
which subjects are exposed. The traditional and simplest monitoring
technique is manual sampling, where the investigator draws a known
volume of air at a known flow rate through a collector. For
particles, the collector can be a filter; for soluble gases, it
can be a bubbler; for organics, it can be a charcoal trap. The
appropriate collector is used to trap particles with known, and
presumably, high efficiency.
For gases and vapors/ the collection efficiency needs to be
known and constant. If it is less than 100 percent, the amount
collected can be corrected to compensate for the lack of total
collection. For aerosols, on the other hand, there must be 100
percent collection efficiency, since the collection efficiency is
particle size-dependent, and the investigator wants to know the
total amount, and not some estimate that varies with its particle
size distribution. In practice, this requirement doesn't cause
any great problem, since filters and other types of collectors
with essentially quantitative collection capabilities are readily
available.
One advantage of sample collection and subsequent analysis is
that there is almost no limit in the choice of sample processing
that can be done, or in the range of sophisticated and sensitive
laboratory instruments that can be brought to bear on analytical
problems. The investigator can get the ultimate in sensitivity
and specificity, with correction for interferences. On the other
hand, there is a basic limitation in this procedure in that there
119
-------
is a significant time lag between sample collection and the determi-
nation of what was collected. Manual sampling techniques may,
therefore, be used to back up continuous monitors, and perhaps, as
the final arbiter on a substance's exact concentration. However,
in themselves, they are not of much use in maintaining an atmosphere
at a desired level.
INTERMITTENT/CONTINUOUS INSTRUMENTATION
Generally, therefore, investigators use two basic types of
automated instrumentation. One is the intermittent type of operation,
and the other is continuous. For monitoring particles, a combination
of techniques is generally used because there is a very large
particle range to be covered, and there is no universal instrument
that can indicate the concentration of each particle size interval
over the entire size range.
One of the instruments used to measure very small particles
(0.08 ym to 0.5 pm) is a close relative of the mono-dispersed small
particle generator. As shown in Figure 2, the electrical aerosol
analyzer works on the same principle of sorting out the particles
of a poly-dispersed aerosol according to their electrical mobility.
Instead of withdrawing the particles at a specific location on the
axial electrode, they are collected on a current collecting filter
at the end of the tube, and the amount of charge they deposit on
that filter is measured incrementally. The voltage gradient and/or
the flow rate is changed sequentially to allow a different size cut
to reach the filter. In this way, there is a progression of size
increments. This technique gives the investigator an accurate size
distribution in a stable atmosphere, and, of course, attempts are
made to keep the chamber atmospheres stable. On the other hand, it
takes a finite amount of time to run through these increments and
collect the size band data to get the size analysis. This tech-
nique is an example of an intermittent operation.
In many cases, this technique is unsatisfactory, especially if
the investigator wants very tight control of the substance's concen-
tration, and needs a technique that senses the particles in a
dynamic system and continuously indicates precisely what the level
is. Investigators working with carbon monoxide can use an infrared
analytical technique, which does not collect the sample at all. As
shown in Figure 3, it simply directs the gas through the sample
tube, through which also passes infrared radiation at wavelengths
that are sensitively absorbed by carbon monoxide. There will be an
attenuation of that infrared wavelength because of the absorption
of the carbon monoxide in the sample tube, and, thus, the energy
received on the detector will vary with the substance concentration.
Operationally, this technique is best implemented by using a
reference tube of clean air, and getting the difference in attenua-
tion between clean air and the air containing the carbon monoxide.
120
-------
| LIGHT
SOURCE
H n J i,
* i (i r n i ^^« i P f r j
ROTATING (-VAU II *-*-^'
Stream C=
SAf
Sire
O FT
PLE
UBE
m 1=
1
|
1
kid
i
WINDOWS
W£C
PI I
1 1
1 1 |
c r T r\ o
U b 1 ti \v i v/ n
1
1
1 ii 4
I
L_
— ~
ows
i— TV
1—=^
^"""^^
'-*—~>
m
V
_}
REF
TUB
=1
i]
I
' . 1 ' *
1
OPT CAL
SYSTEM
RENCE READ-OUT
F
INDICATOR
And
RECORDER
OUTPUT
r
? ! i R.F. __ SIGNAL
' 1 LIMIT AMPLIFIER
i
1
ELECTRONICS
Figure 3. Schematic diagram of infra-red gas analyzer. Source:
Air Sampling Instruments (5th edition). ACGIH, Cincinati,
Ohio, 1978.
Other gaseous constituents absorb infrared energy; for example,
water vapor. However, these gases and vapors are wavelength-specific,
so by tuning to the wavelength at which carbon monoxide has prefer-
ential absorptive capacity, the effect of interference can be
removed, giving a sensitive, accurate, and specific analysis of
carbon monoxide. The only time lag is the insignificant amount of
time it takes to flush the sample through the tube. This, then, is
a commonly used method of monitoring carbon monoxide on a continuous
basis. The sensitivity is a function of the length of the sample
tube, which can be 10 or 20 meters with folded tubes.
There is practically no limit in the laboratory situation to
the sensitivity that can be achieved using this type of instrumenta-
tion. Similar instruments can also be used in measuring other
gases by choosing appropriate wavelengths of absorption free of
significant interferences.
121
-------
An alternate method of testing carbon monoxide is an intermit-
tent technique, which involves a combination of gas chromatography
and flame ionization detection. The flame ionization detector is
nonspecific, but by coupling it with the holdup time in the chroma-
tographic column, specific analyses can be obtained because the
investigator knows when the carbon monoxide will come through the
column and move into the flame ionization detector.
The flame photometric detector illustrated in Figure 4 is
commonly used in measuring the concentrations of sulfur-containing
gases. If a sulfur-bearing material is passed through a hydrogen
burner, a light photon emission will result that can be detected
with sensitive photomultipliers. It is a nonspecific technique in
that it gives the investigator roughly an equivalent response for
hydrogen sulfide, sulfur dioxide and some of the mercaptans. So,
if the investigator knows that the only sulfur gas present is
sulfur dioxide, then the test is quite specific, but if there are
other sulfur gases, the test is not specific. In this instance,
the investigator can attach the sulfur gas detector to a chromato-
graphic column that will separate out the sulfur species so as to
make the test specific. In laboratory situations, the procedure is
usually unnecessary.
ELECTROMETER
ELECTROMETER
NARROW-BAND
OPTICAL FILTER
CLEAN
AIR
SOURCE
Figure 4.
122
Schematic diagram of flame photometric analyzer for sulfur
gases, with permeation tube calibrator. Source: Air
Sampling Instruments (5th edition).
-------
In most situations, it is desirable to have a built-in calibra-
tion device on the concentration monitor to ensure that the concen-
tration indicated on the output chart or on the dial is correct.
One of the more common devices used for in-line, continuous calibra-
tion is the permeation tube. For the flame-photometric detector,
the calibration device shown in Figure 4 contains liquid sulfur
dioxide sealed into a Teflon tube. The tube is slightly porous to
the saturated sulfur dioxide vapor above the liquid in the tube.
The rate of permeation of sulfur dioxide vapor out of the tube is
very much dependent on the temperature, but is constant at a given
temperature. If the investigator holds the permeation tube within
a constant temperature bath, he can obtain a known emission rate.
If he gets an accurately calibrated dilution air flow, he can
determine the concentration of the calibration gas and direct it
periodically into the analyzer to get a span signal for the instru-
ment. This is a very convenient way of calibrating. There are
similar calibration tubes available for nitrogen dioxide, and a
number of hydrocarbons that are readily condensed into liquids at
approximately room temperature.
Instruments that measure the light emitted during gas phase
reactions of nitric oxide and ozone are used in monitoring concen-
trations in chamber and ambient atmospheres. If an excess of ozone
is mixed with the sample containing nitric oxide, as in Figure 5,
the amount of light is proportionate to the amount of nitric oxide.
SAMPLE
FILTER
FLOW REACTOR
VACUUM
GAUGE
HI LEVEL
OUTPUT
Figure 5. Schematic diagram of chemiluminescense NO/NC>x analyzer,
Source: Scott Research Labs literature.
123
-------
This technique can be used for measuring nitrogen dioxide by passing
the sampled air through a chemical converter that converts the
nitric dioxide to nitric oxide, which can then be measured on a
mole for mole basis. The investigator can differentiate between
nitric oxide and nitrogen dioxide by sequential readings with and
without the converter in-line.
The same basic type of instrument can be used as an ozone
monitor by feeding an excess of nitric oxide into the reaction
chamber. Other ozone monitors are based on other chemiluminescent
reactions of ozone, specifically, with ethylene and organic dyes.
I indicated before that the mobility analyzer could be used as
a particle size analyzer for small particles, but that even using
this technique, we could not measure particles larger than approxi-
mately a half micron in diameter. Fortunately, there are other
techniques suitable for measuring the larger particles.
The property of light scatter is used in measuring the concen-
trations and size distributions of particles of three-tenths micron
diameter and larger. When light is focused on a particle in the
instrument illustrated in Figure 6, some of that light will be
scattered. The amount of light that a given particle will scatter
depends on its size. A photomultiplier can be used to detect the
light output from each particle. The pulses can be accumulated
according to size intervals in a multichannel pulse-height analyzer,
and these intervals can be calibrated according to particle size.
However, this technique is not as simple as it sounds. There are
other properties of the particle besides its size that affect the
amount of light scattered, including the refractive index, the
color, the shape, and so forth, and reliable data depend on accurate
calibrations of the instruments used.
PHOTOMUUIPUEn
Figure 6. Schematic diagram of single particle optical particle
size analyzer. Source: Climet, Inc., literature.
124
-------
Investigators in the laboratory working with droplet aerosols
or aerosols of fairly regular shape, have fewer problems than
investigators working with particles in the ambient air, which come
in a great variety of compositions and shapes. Thus, in the
controlled laboratory experiment, a reliable calibration can gener-
ally be developed relating the optical-inferred diameter to the
real diameter. These instruments are not continuous monitors
because they accumulate and sort the pulses for a given interval,
and then display a distribution. However, the response time can be
fairly rapid.
MEASURING THE PARTICLE'S MASS CONCENTRATION
Both of the methods described for measuring particle concentra-
tions and size distributions, i.e., the small particle mobility
analyzer and the larger particle light-scatter analyzer, measure
the diameters of particles and sort them by so many numbers of
particles in each size interval. Number concentration is important
in many studies, but in studying toxicology and its effects on
people or animals, the investigator generally needs to know the
aerosol's mass concentration. Since the mass of a particle varies
with the cube of the diameter and with the density, the investigator
needs to know more than number distribution. On the other hand,
the automatic machines accumulate a very good statistical base.
The investigator comes up with fairly good approximations by using
transformations to volume distributions. If the particles' density
is determined, the investigator can calculate a mass median diameter
distribution or a concentration.
While such data transformations may be justifiable in some
cases, the investigator is not really measuring the property that
is being reported. It is, therefore, sometimes necessary to directly
determine the aerodynamic size distribution, since this is the
distribution of sizes that affects deposition in the respiratory
tract. If a system of somewhat redundant measurements can be
justified, it is best to use a variety of complementary aerosol
measurements involving light scatter, mobility, and aerodynamic
properties.
The beta attenuation technique combined with a two-stage
collection system, as illustrated in Figure 1, sorts the particles
by their aerodynamic size and measures the mass in each fraction*
Using this technique, the air enters an impaction jet at the top,
and particles above a cutoff size, depending on their aerodynamic
properties, will be collected on the back-up filter. The investi-
gator can continuously monitor the mass of accumulated particles
collected at each stage using the beta attenuation technique. A
carbon 14 source is used as a beta emitter. The amount of &-radiation
125
-------
TWOMASS
s
t OETE
CTOR^
^•••••IMBMM
6
Q
•••••••••••
•1
1Ur/
I4c
X
Tl^
^IMPACTION
/ STAGE
0
.---
S^
s^-
'0
•••••••••1
1
-Si DETECTOR
-FILTRATION
STAGE
-14C
TO PUMP
PREAMP
«.
AMP/
cue.
r \
COUNTER
t *•
TIUEN
COUNTER
-• \
AMP/
DISC
r \.
PREAMP
t
P
R
N
i
~V
PROORAMMASLEl
CALCULATOR )
1 TAPE |
Figure 7. Schematic diagram of TWOMASS mass concentration analyzer.
Source: Fine Particles, p. 551.
that reaches the detector depends on the ^-absorption in the accumu-
lated sample between the source and the detector.
There is only about a 10 percent variation in beta attenuation
with mass number (Z), with the exception of hydrogen, and hydrogen
does not usually contribute much to aerosol mass. So, by and
large, the amount of beta attenuation by the impacted particles on
the first stage tells the investigator how much mass of material
has been collected in large particles above the impactor cutoff
size, and the attenuation of the particles on the second stage
indicates the mass of small particles below the cutoff size. How-
ever, the response time of this technique is not very rapid, and
may not be adequate for feedback control of generation rates in
chamber atmospheres.
Another type of direct monitor of aerosol mass concentration
utilizes quartz-crystal oscillators as mass balances, and is illus-
trated in Figure 8. Very small sample masses can be detected as
they accumulate on the quartz crystal oscillator. The quartz
crystal is cut into a particular mode, and electrodes are attached
126
-------
Mass Sensitivity
Distribution
0.2mm
CRYSTAL
ELECTRODES
Figure 8. Schematic diagram of quartz-crystal oscillator used as
a mass balance, showing sensitivity over crystal face.
Source: Fine Particles, p. 489.
on each side* When a high frequency signal is applied, the crystal
oscillates, and its oscillation frequency depends on the mass of
the crystal.
When particles are deposited on the crystal, its mass increases
and the oscillation frequency decreases, therefore, the sample
accumulation can be measured by the change in the frequency of the
oscillator. In modern instrumentation, frequency counting is
relatively simple and precise, and very sensitive. Thus, infinites-
imal masses produce readily measurable signals*
However, this technique has its limitations. The sensitive
zone does not have a uniform sensitivity. It is greatest at the
center of the electrodes and falls off toward the periphery.
However, it does provide a sensitive indication of mass, and with
enough calibration to be sure of its performance under the given
operating conditions, can be used as a sensitive mass monitor.
Another instrument that can be used to provide an approxima-
tion of mass concentration of particles in the one-tenth to one
micron range is the integrating nephelometer, which is illustrated
in Figure 9. This technique measures the total scatter of an
127
-------
AEROSOL OPTICAL PARAMETERS
CLEAN ATP AEROSOL
PURGE NARROW T?AND OUTLET
TUNGSTEN FILAMENT
LIGHT SOURCE
,
OPTICAL FILTER
I | 1
OPAL GLASS
SCATTERING VOLUME
COLLIMATING DISKS
AFROSOL
INLET
CLEAN AIR
PURGE
AEROSOL
OUTLET
TUNGSTEN FILAMENT
LIGHT SOURCE
SCATTERING VOLUME
Figure 9. Schematic diagram of integrating nephelometer.
Fine Particles, p. 521.
Source:
aerosol. The instrument discussed earlier measures the scatter
from individual particles. The nephelometer, by contrast, measures
the scatter of a cloud, i.e., the total scatter of all of the
particles in the sensing zone.
If the investigator has the proper calibration and particles
in the one-tenth to one micron range, he can obtain a correspondence
between the total light scatter, the b-scat function, and the mass
concentration. This gives the investigator a rapid response, and
is a good indicator that nothing very drastic is happening to the
atmosphere. This test may be a good monitor of the constancy of
the atmosphere, even if it doesn't determine exactly what the
concentration is.
128
-------
RECOMMENDED BIBLIOGRAPHY
1. Drew, R.T., and Lippmann, M.: Calibration of air sampling
instruments. In Air Sampling Instruments (5th Edition).
Section I, pp. 1-1-38. American Conference of Governmental
Industrial Hygienists, P.O. Box 1937, Cincinnati, Ohio 45201
(1978).
2. Raabe, O.G.: The generation of aerosols of fine particles.
In Fine Particles (Liu, B.Y.H., ed.). New York, Academic
Press, 1976, pp. 57-110.
3. Grassel, E.E.: Aerosol generation for industrial research and
product testing. In Fine Particles (Liu, B.Y.H., ed.). New York,
Academic Press, 1976, pp. 145-172.
4. Corn, M., and Esmen, N.A.: Aerosol generation. In Handbook on
Aerosols. (Chapter 2, pp. 9-39.) TID-26608, NTIS, U.S. Depart-
ment of Commerce, Springfield, Va., 22161, 1976. $6.00.
5. Kerker, M. : Laboratory generation of aerosols. Advances in
Colloid and Interface Science 5:105-172, 1975.
6. Drew, R.T., and Laskin, S.: Environmental inhalation chambers.
In Methods of Animal Experimentation (Vol. IV). New York,
Academic Press, Chapter 1, pp. 1-41.
129
-------
Discussion Summary
Participants asked Dr. Lippman two questions: 1) had he attempt-
ed to manipulate the orifice shape and jet stream pressure as a
means of controlling particle size; and 2) had he considered
combining the electrical charge aerosol generator with a fluidic
vortex amplifier as a means of controlling power, speed, and size
in the aerosol particles.
Dr. Lippman responded that he had no personal experience with
either the vibrating orifice or the electrical charge analyzer.
Participants agreed that the lack of uniformity of shape and
diameter of fibers and other particulates present a real problem
as far as monitoring and calibration are concerned.
131
-------
Electrical Surveillance and Integrity
G. Guy Knickerbocker, Ph.D.
Emergency Care Research Institute
Plymouth Meeting, Pennsylvania
INTRODUCTION
It is my pleasure to be able to discuss with you some of
the many factors that contribute to the safe and effective use
of electrically operated devices in human studies. In my
discussion, I will draw largely on the experience of the Emergency
Care Research Institute, which evaluates and tests devices and
systems used for patient care. The Institute's testing spans
both the field of diagnostic and therapeutic devices, a somewhat
broader range than average. In general, I would judge that
your interests would most clearly coincide with the diagnostic
field in clinical medicine because as I perceive your work, you
are recording physiologic variables primarily to assess the
effect of stimuli acting on the subject.
While electrical safety in hospitals is related to two
different aspects of the electrical shock phenomenon, commonly
referred to as macroshock and microshock, it is my understanding
that for the most part your concern would center on macroshock.
Macroshock generally refers to electrical currents that occur
through contact points on the surface of the body. Microshock
occurs when currents penetrate the skin barrier through conductors
and go to organs within the body. In general, microshock
currents concern us less than macroshock currents.
133
-------
RISKS ATTENDANT TO THE USE OF ELECTRICAL DEVICES
This presentation will concentrate on electrical risks such
as electrical shock, inadvertent or unwanted stimulation of excitable
tissues, and electrocution. However, there are other risks that
must not be overlooked, and any program that intends to ensure
safe and effective use of instrumentation and systems must consider
these risks. Additional risks in using electrical devices include
the possibility of fire, especially in oxygen-enriched environments;
burns; the subtle effects of chronic exposure to otherwise subthresh-
old electrical currents; or electrical and/or magnetic fields and
mechanical trauma that may be associated with improperly designed
electrodes or cabinets that have sharp edges.
The risk of burns must be broadened to include burns that
occur because of the passage of electrical current through resistant
tissue, causing heat to be produced by joule heating, and those
burns that occur because of chemical reactions at the surface of
the body, caused by the passage of current through the tissue
electrode interface. Current-induced thermal burns are not at all
unlikely in those cases where the electrical current used is at a
frequency above that which evokes sensation, and current densities
can easily be so great as to cause severe tissue damage without
any electrical shock effect. For example, the problem of current-
induced thermal burns has been particularly acute in the use of
the electrosurgical unit. This unit employs currents of the order
of 1 mHz and levels that approach or exceed 1 amp.
The subtle effects of chronic exposure to subthreshold currents
and fields have been studied, producing much controversial data.
It is not an area that I would judge as significant to your studies,
but I feel you should be aware of the possible risks of chronic
exposure to subthreshold currents and fields.
The device that performs improperly and gives wrong information
is a risk factor that normally is not considered. The device may
be out of calibration, have an altered frequency response, or have
developed an inappropriate nonlinearity. The use of such a device
is considered a risk to the patient because it may lead to a wrong
conclusion concerning the subject's condition. In clinical medicine,
a missed diagnosis because of incorrect performance of a device is
a risk whose consequences may be every bit as great as those of
electrical shock, burn, or fire.
THE EFFECTS OF ELECTRICAL SHOCK
The host of effects that electrical current can cause cover the
whole gamut of severity. For example, a subject may experience little
134
-------
or no sensation upon exposure to an electrical current. On the
other hand, electrical current may cause a subject to lose
consciousness, to suffer severe tissue burns, or to die. Shock
sensation can be mild as a gentle tingling at the point of
contact; or, the subject may experience a much more general
shock, producing a tetany of muscles that would make it impossible
for the person to break contact with the energized circuit,
thus holding the respiratory muscles in a contracted or immovable
state so that breathing becomes impossible. The more general
shock may also cause such severe muscle contractions that bones
may be broken. Electrical current can cause fibrillation of
the heart or other irregularities of cardiac rhythm, burns of
the skin and deeper tissue/ or set off a convulsion.
The outcome of an electrical shock is dependent on a
number of factors. As Table 1 indicates, an important factor
is the intensity of the shock, usually expressed in terms of
the current (for example, in milliamperes or amperes).
Table 1
Effects
Electrical Current
Threshold of sensation
of electrical shock
Failure to let go
Interruption of respiration
Ventricular fibrillation
Defibrillation of the heart
Prolonged respiratory paralysis
and severe burns
1 milliampere
10 milliamperes
20 milliamperes
60 milliamperes
to 4 amperes
4 amperes to 10 amperes
above 10 amperes
The values in Table 1 should not be taken as clearly defined
thresholds for these phenomena, but rather as approximate
orders of magnitude at which these responses will occur. There
are many other factors that affect the subject's response to a
given current.
135
-------
Ventricular fibrillation is a state of chaotic disorganized
contraction of the heart muscle fibers that causes the blood to
stop circulating. In human beings, ventricular fibrillation
rarely spontaneously reverses itself, and is usually fatal
unless treated promptly. It may seem paradoxical that an
electrical current greater than that which causes fibrillation
can be less lethal, but in fact, this is the case in the current
ranges that are identified with defibrillation of the heart.
This fact is exploited in devices called cardiac defibrillators,
which are used to reinstitute cardiac activity.
Duration of the shock is almost equally as important as
shock intensity because it governs the amount of energy imparted
to the body. Moreover, many excitatory phenomena provoked by
electrical currents have an inverse relationship between the
intensity of the current and the duration of the current necessary
to evoke a response. For example, a short shock may cause a
muscle or a nerve to respond only if the current is significantly
higher than that current which will excite the nerve at a
longer duration. The contact area also plays a role in that it
frequently determines the magnitude of current that will result
upon contact. In general, a large area of contact implies a
lower resistance for the shock circuit, and therefore, generally,
a greater current.
The pathway of the shock through the body cannot be overlooked.
A current pathway that does not include the heart, for example,
a pathway from one leg to the other, is much less likely to
result in cardiac irregularities than a current pathway that
travels from one hand to one foot, traversing the torso and
going directly through the heart.
Frequency or waveshape of the electrical source has a
marked bearing on how a shock affects the subject. In general,
the body is less sensitive to high frequency electrical currents
than it is to low frequency currents. Studies on the sensitivity
of the heart to varying electrical frequencies show that the
frequency at which the lowest electrical current will cause
ventricular fibrillation is at approximately 60 Hz. This
frequency has been chosen for electrical power distribution,
and thus is widely available.
The temporal relationship of the shock to periodic phenomena
in the body sometimes will determine whether a detrimental
effect will result. For example, very short electrical currents,
those of the order of a 10th of a second or less, must occur at
a specific portion of the cardiac cycle, frequently referred to
as the vulnerable period, if they are to cause ventricular
136
-------
fibrillation. Undoubtedly, there are other physical or psychological
factors that influence the outcome of electrical shock. It is
not outside the realm of possibility that emotional or other
factors generally relating to the physical or mental well-being
of the subject affect whether a response to an electrical shock
occurs.
Thus, in many cases it is difficult to predict the precise
outcome resulting from an electrical shock. Time does not permit
me to go into depth in this area, but any of the effects of electri-
cal shock that seem paradoxical become much more understandable if
these complex interactions are considered in greater detail.
METHODS OF PROTECTING SUBJECT
A general philosophy for protecting the subject from electrical
risks can be summed up simply: The intention of any protective
measure is to restrict the current that can pass through the
subject or to limit the voltage that can be applied across the
subject.
There are numberous approaches for enhancing the subject's
safety, for example, because inadvertent current pathways so
frequently involve the passage of electrical current to ground,
the subject should be protected by insultation materials or
isolated from grounded surfaces. This reduces the possibility
that the subject will become a portion of an unintended current
pathway to ground.
Electrical safety is enhanced when subject-connected circuits
for either the acquisition of physiologic information from the
subject or for applying stimuli or monitoring currents to the
subject are isolated from the ground. When such circuits are not
ground referenced, the possibility of inadvertent currents to
ground through the patient are markedly diminished.
Proper electrical distribution systems also contribute to
overall safety. In this country, our electrical distribution
systems are overwhelmingly of the grounded type, that is, one of
the power-supplying conductors at each outlet is grounded. There
are valid reasons why these systems have been developed, though at
first glance it may seem that greater safety would be afforded by
a system that has no intentional ground connections. Suffice it
to say that the choice of a grounded electrical distribution
system is predicated upon factors that contribute to the reliability
of the system's performance.
Since we have to live with the grounded distribution system,
137
-------
it is necessary to follow proper procedures for installation and
maintenance. It is particularly important that receptacles be
wired consistently so that the grounded conductor can always be
identified. Let me comment parenthetically that the grounded
conductor is that power conductor that is a ground potential. It
should not be confused with the third conductor in a power cord,
the green wire, which serves as a grounding conductor to which the
cases and chassis of electrically-operated equipment are connected.
The differentiation may seem subtle, but even in devices operated
on electrical distribution systems that do not have any of the
power conductors grounded, it is still important in improving
safety to ground the cases and chassis through the green grounding
conductor. When wiring polarity of receptacles is carried out
consistently, devices can be plugged into that system with a fair
degree of confidence that the devices will be connected in a
manner that will ensure the least leakage current.*
Effective grounding may be considered an extension of the
idea that distribution systems must be properly installed. I
draw specific attention to this as an important factor in the
safe operation of a system. Devices that are properly grounded
reduce the subject's risk of exposure to adverse and inadvertent
currents. Effective grounding includes not only ensuring that
the cases or chassis of electrically operated equipment are
grounded, but also that other conductive materials in the region
in which equipment is being used are sufficiently interconnected
to one another and to ground. When these precautions are taken,
potential differences or sources of electrical current that could
become a risk are harmlessly shunted away from or around the
subject.
Good design is a keystone to building greater electrical
safety. In particular, a design that reduces leakage current and
minimizes the chance of failures leading to energization of the
accessible surfaces or subject-connected leads, reduces the risk
of shock or adverse effects.
Great strides have been made in recent years as attention
has focused on what was presumed to be an electrical shock problem
in hospitals* While studies concerning leakage current were never
adequately documented to provide convincing proof that leakage
*Leakage current is usually low current that normally occurs in
the operation of electrical devices and in pathways other than
the intended current-carrying circuits. It emanates from the
chassis of the equipment and is conveyed harmlessly to ground by
the grounding conductor in the power cord.
138
-------
current was causing the "electrical shock" problem, the attention
focused on this area nevertheless stimulated improvement in the
design of medical equipment, so that now the leakage current in
vast numbers of medical devices has been reduced to levels that
are considerably lower than those considered achievable in past
years. On the average, chassis leakage current from these devices
is less than 100 microamperes, often as low as a few tens of
microamperes, and leakage currents from subject-connected leads
are routinely well below the commonly accepted limit of 50 micro-
amperes* These values are commonly accepted as upper limits on
leakage current from patient-related devices. Other equipment is
permitted a maximum leakage current of 500 microamperes.
In addition to system and device design and implementation,
there are various specific devices that are frequently introduced
into the system with the aim of increasing electrical safety, for
example, local power distribution systems that are isolated from
any connection from ground. Such systems limit the magnitude of
currents associated with failures in plug and receptacle combinations
or in power cords. They provide an added feature in that they
permit a faulty system to operate at relatively low risk without
interruption of power until an orderly transition to alternate
equipment can be made or until completion of repairs. Many codes
and standards require that these isolated power systems be used in
anesthetizing locations, since these systems are generally not
prone to producing sparks that could ignite inflammable anesthetics.
More recently, awareness of the significance of improving electri-
cal safety for the benefit of the patient and attending personnel
has increased. This has led some to use isolated power systems in
special care areas in other hospitals and at other locations where
it is felt the need for improved electrical protection may be
necessary.
Ground fault circuit interrupters are capable of sensing
small current flows to ground that are outside of the intended
pathway, while at the same time remaining insensitive to large
magnitudes of current flowing in the intended pathway. There is
a widespread application for this device; in fact, the ground
fault circuit interrupter is required by electrical codes for use
with electrical services supplying swimming pools, outside
receptacles, and bathrooms in newly built homes. The disadvantage
of using ground fault circuit interrupters is that on detection
of a fault, power to the device is interrupted, and this interrup-
tion may be untenable in certain circumstances.
The detection techniques used in ground fault circuit
interrupters have recently been extended to another class of
devices, resulting in what is known as a ground fault monitor.
139
-------
This device can be interposed in the power line to indicate,
either visibly or audibly, when leakage current is above the
prescribed level. Thus, the ground fault monitor is able to
inform the user when limits have been exceeded, without interrup-
ting the circuit. Other protective devices, which are much less
commonly used, are those that monitor the continuity of grounding
conductors, thus assuring that adequate grounding of equipment
is present to minimize the risk of exposure to energized surfaces.
Two additional items also contribute greatly to the safe use
of electrical equipment. One is an effective regular program of
inspection and preventive maintenance of equipment. This
practice is most effective when it is directed not only towards a
product's safety and protective features, but also when it takes
into consideration the performance characteristics of a device or
system. Failure of a device to perform properly may put the
subject at risk if, for example, because of faulty equipment, the
subject must be retested in a program in which there is some
inherent risk in the testing procedure. Second, the user should
be educated in the proper use of the device and be knowledgeable
of its safety features and its risks. The importance of this
second item in enhancing the safety and integrity of electrical
devices cannot be overstated.
EXAMPLES OF TYPICAL DEVICE OR SYSTEM PROBLEMS AND SOLUTIONS
I would like to cite some problems or failures that have a
marked impact on the safety of devices used in everyday settings,
for example, in the home. Power cords and plugs often are
abused. They lie across floors and are walked on. Carts and
tables are wheeled over them. The cord is used to withdraw the
plug from the receptacle. People trip over the cord and put a
strain on both the equipment and the plug. Therefore, it is no
wonder that many safety problems arise from failures in power
cords and plugs. Problems of this nature are avoided most
easily by a regular program of preventive maintenance in which
the condition of line cords and plugs is inspected for cracked
insulation, to ensure that strain reliefs have not been abused
or are not inadequate, and to ensure that wiring at the plug is
correct and tight.
Users should insist that equipment be supplied with cords
that are strong enough to withstand abuse. It is important for
the user to know that the cord should not be used to remove the
plug from its receptacle. Plugs should be examined to ensure
that ground connections have not been cut off. Plugs molded
onto the line cord have a history of broken grounding conductors
within the plug assembly leading. There is widespread skepticism
140
-------
in the medical community concerning the integrity of molded plugs.
The siutation is changing as designs improve, but the performance
of molded plugs must be watched closely.
A class of devices has evolved within the past five years
known as hospital grade plugs and receptacles. These plugs and
receptacles have been produced especially to withstand abuse,
particularly with respect to the integrity of the grounding
connection. In general, these devices should be used when there
is great concern about the electrical safety of equipment.
These plugs and receptacles are compatible with ordinary parallel-
blade units, and are generally identified by the presence of a
green dot conspicuously placed on the surface.
The innocuous little device frequently referred to as a
cheater, an adapter that enables a two-pronged receptacle to
receive a three-pronged plug, is one of the most insidious
culprits in seriously degrading electrical safety* Frequently,
it is implemented with a wire pigtail that must be fastened
securely to a grounded point. The screw that holds the coverplate
on the receptacle is often taken as the point to which the
cheater is to be connected, but unfortunately, the coverplate is
not always connected to ground. The result is that frequently
grounding is not achieved when these devices are used. The
obvious solution is to avoid the need for these cheaters by
installing proper receptacles that have the third contact for
grounding. Any receptacle should be installed properly and then
checked immediately to ensure that it has been wired properly.
It should also be tested with one of the simple tension testers
that are now available to ensure that there is adequate holding
force for each of the plug's prongs.
Extension cords are another problem, along with their
closely related cousin, the adapter cord. When I say adapter
cord, I am referring to those "short extension cords" that
consist of a plug and a connector mounted on a short piece of
wire that permits non-standard plugs to be connected with standard
receptacles or vice versa. Extension cords and adapters introduce
another set of contacts in the electrical pathway that is subject
to failure, and at least doubles the probability of a problem
occurring.
If they are excessively long, extension cords also add
resistance to the power circuits and grounding conductors and
may reduce the safety and effectiveness of the device. The
solution is to avoid or ban the use of extension cords. The use
of extension cords could also be reduced by providing long power
cords on electrical devices. Occasionally, adapters may be
141
-------
necessary but they should be considered as electrical equipment
and be included in any inspection program to ensure that hidden
damage has not occurred.
Failures that would normally not occur occasionally show up
in devices that expose the subject or the experimenter to potentials
on subject-connected leads. Selection of well-designed equipment
is crucial to minimize this risk.
It may happen that too many grounds will occur, especially
when an effort is made to assure that everything is well-grounded.
In this situation, redundant grounding pathways may lead to a
complete circuit that might include equipment contacting the
subject. If there are strong magnetic fields in the area,
especially at power line frequency, and these magnetic fields
course through the area contained within the loop formed by this
closed ground pathway, currents known as "ground loop currents"
may be induced in the ground system. These can create, at the
very least, interference in signals being monitored and, under
extreme conditions, they can expose the subject to potentials
that would not otherwise have occurred. To avoid these problems,
grounding should be carried out carefully. In general, the
problem can be minimized if devices are singly grounded. This
is usually accomplished through the grounding conductor in the
power cord. Where it is possible, grounding conductors from
individual equipment should come to the same point or to the
same electrical branch circuit rather than going to widely
disparate points within the area. Except in unusual circum-
stances , no wire should be added to tie equipment together.
CODE STANDARDS AND REGULATORY MECHANISMS INTENDED TO ENHANCE
ELECTRICAL SAFETY
There are a number of standards and codes that users can
turn to for guidance, and which manufacturers, whose overall aim
is to raise the safety level of equipment and systems, make use
of in equipment design. Many of these standards and codes are
produced by a consensus process; that is, a process that is
intended to ensure that all parties affected by the proposed
standard will have an opportunity to review and to contribute to
its development. I will briefly mention several regulatory
bodies involved in developing standards and codes.
The Association for the Advancement of Medical Instrumentation
(AAMI) is a group that produces a number of voluntary standards
for a variety of devices that are applicable in medical and
related fields. The organization has broad representation,
including device manufacturers, clinical and biomedical engineers,
142
-------
physicians, and nurses. One standard the AAMI has created that
is probably of the greatest interest to you is the standard that
sets safe limits for leakage currents in health care devices.
Another well-known and widely respected agency that produces
codes, standards, and recommendations is the National Fire
Protection Association (NFPA). Historically, the NFPA has
concentrated on activities intended to reduce risk from fire.
However/ the NFPA has increasingly embraced other areas, such as
electrical safety. It is not surprising that the National
Electrical Code, one of the NFPA's most visible byproducts, is
an outgrowth of its objective to reduce risk from fire. The
National Electrical Code prescribes wiring methods which, among
other things, are intended to minimize fires caused by poorly
designed or implemented electrical systems.
The NFPA has numerous other standards including one presently
under development that would propose methods for the safe use of
electricity in health care. This document, commonly referred to
as NFPA 76B, when adopted, will provide a guideline likely to be
used well beyond the walls of health care institutions. Many of
the principles on which it is based can be adequately carried
over into human research facilities. The codes of the NFPA are
widely adopted, although the NFPA itself is not a code-enforcing
authority. For example, adherence to the National Electrical
Code is not monitored by the NFPA, rather, the code is adopted
by local enforcing authorities or by other governmental agencies
and is applied by force of law.
There is an intensive international effort to produce
standards. This effort is largely the work of an organization
called the International Electrotechnical Commission (IEC). The
IEC has produced a vast array of standards intended to ensure
the uniformity of products on an international level. The IEC
is presently refining a comprehensive document for medical
devices.
Underwriters Laboratories (UL) is widely recognized as an
organization that contributes to the safety of devices and
appliances. In general, the UL does not write codes and standards
for other groups, it formulates them for use in assessing equipment
submitted to the UL by manufacturers who want to obtain the
right to place the UL label on their products. Standards formulated
by the UL are reviewed by persons representing diverse interests
and a broad audience outside of the UL organization before they
are applied to devices under examination.
The UL has earned a good reputation for promoting safety in
143
-------
electrical products. Up to this time, many of the specialized
devices used in research laboratories have not carried the UL
label. There are several reasons for this. In the past, manufac-
turers of these devices have not found it necessary in the competing
marketplace to seek UL listings to promote their products. In
addition, obtaining UL listing entails a considerable expense on
the part of the manufacturer and any significant model change
means that the product must be resubmitted. In the future, we
are likely to see more devices, particularly those used in the
medical field, that carry either the UL label or display evidence
that the device has been screened by a process similar to the
Underwriters service or has been judged to be in conformity with
standards being developed by the Food and Drug Administration.
Individual investigators will have greater assurance that devices
have been built with an eye toward safety, as well as toward
efficacy or desired performance.
CONCLUSIONS
The foregoing is a brief overview of the various factors
that are important in assessing the safety and performance of
electrical systems. The topic is an immense one, which cannot
be adequately addressed in this paper. The safe use of electrically-
operated devices is the result of a combination of a number of
factors, including safe design, implementation, and knowledgeable
use of equipment. To the extent that these factors are considered
in the development and use of electrical equipment, the risks
encountered will be minimal.
144
-------
Discussion Summary
Discussants questioned whether it is necessary to use three
wires in grounding rather than two, since the neutral and the
ground are joined together in the ground system. A hypothetical
situation was constructed in response to this question/ in which
the grounding is accomplished at the service entrance point.
Next, it was assumed that there are two devices adjacent to each
other, each fed by separate branch circuits, and each having its
frame grounded to its own neutral.
One device is a high current device, with a heater element
that draws 25 amperes. The 25 amperes of load current goes back
to the neutral, and the resistance in the neutral results in a
voltage difference between the chassis of that unit and the
point where the neutral is grounded back at the transformer.
Next, it was assumed that the other device has no load
current. Hence, its chassis is essentially at the same potential
as the point where the neutrals of these two systems are grounded.
However, because of the drop in the neutral feeding of the
device with high current, there may be a significant voltage
difference between the enclosures of the two devices.
A separate grounding conductor intended to provide grounding
for the chassis is run to prevent large load currents from
flowing into those conductors, and to obtain a near equality of
the potentials of the chassis. Thus, there is a practical reason
for using three wires in the wiring system rather than two.
Participants also discussed the problem of separate power
supplies in a single area. One particular concern voiced is the
possibility of power from different service panels converging in
an area where grounding points are of different potentials.
Discussants considered the merits of equipotential grounding, a
system designed to prevent the occurrence of just such a situation.
145
-------
The concept of equipotential grounding has been a key issue
in the development of the NFPA 76B impending standard. Basically,
this standard would place limits on the potential differences
permissible within the vicinity of a patient or subject. Provisions
in the standard also call for a certain amount of interconnectedness
of the grounding systems that derive from separate panels, to
ensure that potential differences will be minimized.
Participants said that equipotential grounding should be an
important consideration in the design of any facility because it
is necessary to limit the potential differences on the grounding
system. Besides creating risks in terms of subject safety and
device performance, potential differences on the grounding
system cause interference in many physiologic recordings.
146
-------
EPA HUMAN STUDIES PROGRAMS
-------
CLEANS/CLEVER System Approach
John H. Knelson, M.D.
Health Effects Research Laboratory
U.S. Environmental Protection Agency
The concept of controlled environmental research using human
subjects is not new. I was given the mission about eight years
ago by a predecessor agency of the EPA, the National Air Pollution
Control Administration, to find out the technological accomplish-
ments in this field of research.
The result of my inquiries led to the development of the
CLEANS/CLEVER project now underway at Research Triangle Park.
CLEANS is the acronym for Clinical Evaluation and Assessment of
Noxious Substances; CLEVER stands for Clinical Evaluation and
Verification of Epidemiologic Research.
The CLEANS facility, adjacent to the medical center complex
on the campus of the University of North Carolina, consists of two
large exposure chambers with an adjoining computerized physiologic
data acquisition system. Using these computer-controlled chambers,
researchers can expose human subjects to the same conditions of
polluted air that are present in urban and rural areas*
The CLEVER project consists of two mobile laboratories, each
containing equipment identical to that housed in the stationary
CLEANS facility. The mobile units collect and evaluate clinical
data from epidemiologic field studies.
149
-------
Both CLEANS and CLEVER are capable of closely controlling
environmental conditions such as light, temperature, humidity, and
airflow, and both are computer equipped to measure a variety of
human cardiovascular and pulmonary functions. The intent of these
two types of laboratories is to link findings obtained from field
research with findings from controlled clinical studies.
To study the effects of airborne pollutants on human health
and welfare in a controlled laboratory setting, we select, with
specific criteria in mind, human research subjects and manipulate
these subjects and their environment in a predetermined way. We
define the dose, the pollutant, and the exposure regimen, and then
we decide which physiologic, metabolic, neuroendocrine, hemato-
logic, and psychophysiologic end points we intend to study. Once
these end points are defined, we study the responses of indivi-
duals who reside in a manipulated environment for a set period of
time.
By manipulated environment, I mean one that either simulates
urban or rural conditions, or one that is very clean in terms of
particulate counts and gaseous concentrations. In such an environ-
ment, factors such as temperature, humidity, light intensity, and
sound levels are controlled to predetermined values.
What kind of machinery do we use in our experiments? How do
we get our data? How do we process our data? How do we validate
it? How are we assured that we are doing what we think we are
doing, both in terms of manipulating the environment and in acquir-
ing the physiologic or biologic data that we think we are acquiring?
These questions form the subject of this discussion.
During the formative years of the CLEANS/CLEVER program, I
visited every laboratory conducting environmental research that I
could identify. One of the first places I visited was Dr. Kerr's
laboratory. Dr. Kerr impressed upon me that a study design that
depends upon external environmental conditions is subject to many
problems.
Some of the early experiments at Ranch Los Amigos were design-
ed to study and document the physiologic functions of people
residing in a clean environment. To gather data on polluted air,
the researchers depended on the ambient, supposedly polluted, air
obtained from the laboratories' urban location. Although there
were certain merits to that experimental design, the research was
subject to the vagaries of prevailing weather conditions. I
decided not to contend with the kind of problems such as experi-
mental design involves, and instead would design a laboratory that
had the capability of stimulating in a highly reliable, predicta-
ble, and well-characterized fashion the atmospheric conditions
that occur in our urban environments.
150
-------
A design of this type is a leap from the real world into the
stimulated laboratory world. You pay something for that leap, but
you gain better control of your experiment. Over the years that we
have been developing the EPA Human Studies Program, the substantial
advances made in air quality assessment have diminished the price we
pay for simulating the real environment. The advantages, however,
of being able to control our experiments have remained the same.
Another impression I received from the laboratories I visited
was that most investigators were able to acquire a tremendous
amount of data, but they really did not have the logistic capabili-
ty to thoroughly examine the data they acquired. It was clear
that if the data were computerized and preprocessed, it could be
examined as it was acquired. Moreover, preprocessed data is much
easier to analyze than thumbing through stacks of line printer
output in an attempt to make sense out of the data after the
experiment is finished. The practicality of preprocessed data plus
the advances in air quality measurement and the advances in the
technology for simulating environmental conditions helped us
decide to computerize and automate our environmental control
system.
It has become a cliche to have a complete feedback loop, in-
dependent of human intervention, to control sophisticated environ-
mental simulations. We thought we could gain better control of our
studies and construct a much more sophisticated environmental
simulation if we used high-speed computer equipment capable of
constantly adjusting the environment for us as the feedback loop
worked. To this end, we evolved an automated physiologic data
acquisition system and an automated environmental control system.
A sine qua non of clinical research is to control the research
project so effectively that the risk to the human participant in
an experiment becomes vanishingly small. If you depend on extensive
mechanical support systems to provide a simulated environment, an
interjection of many safety systems into that environmental support
machinery provides layer after layer of redundancy for safety's
sake.
If you depend on manual control of your mechanical support
system, it will probably work reasonably well, but a manual system
does not have the reliability of a system that processes many data
points per second. Such a system gives you real-time appreciation
of what the analyzers are showing in controlled environmental lab-
oratories.
The program concept of human studies began in 1969 — during
the period of Earth day, the end of the Vietnamese war, and the
general unrest on the nation's campuses. At that time, it was
151
-------
difficult for a federal agency, even a benign one such as the
National Air Pollution Control Administration, to communicate
effectively with university administrations. However, we were
well received at the University of North Carolina, and our reception
by the University is essential to any success that we may ever
attain.
Because the CLEANS/CLEVER program would take a long time to
get started, we decided to install a less sophisticated mechanism
that we could use right away. We built an acrylic chamber that
has been the source of much valuable data. In a few months we
were able to design and successfully conduct an experiment with
reasonably well defined sulfuric acid aerosols. The chamber has
become a valuable pilot laboratory for debugging experiments; it
will be moved into the bigger, more expensive facility.
We spent roughly two years preparing design specifications
for the research facility. The contract we awarded included
research, development, design, and construction. Research and
development was such a large fraction of the overall cost that we
decided that the construction of duplicate laboratories, in terms
of hardware costs, was not inordinately expensive. Consequently,
two laboratories, identical as far as physiologic data acquisition
is concerned, were installed side by side. The use of two lab-
oratories doubles the throughput of the entire facility at rela-
tively little additional cost.
Physiologic Data Acquisition System (PDAS)
? ^i°10gic data acquisition system provides us with a
' much of *>i°* concerns cardiopulmonary
*
revnfo™ m°S °e- *
receive information either directly from the subject, or from
" ^^f ^ t0 ^ ^^ ™e data measured by the
, such as gas concentrations and flow rates travels
=". nr
pecn ^ allOW the C0mputer to
siller tasks f^™^ ""^ *""** inte^ention. One of the
task is nof 1 * ? USe calibration of the instruments. This
treadmill duS^ " SimPle' fOr instance< « control of the
exercise stress te«trC1S?.StreSS ^^^ Developing software for
exercise stress testing (i.e., exercise electrocardiograph)
-
152
-------
If necessary, we can operate the facility without the computer.
In the event that the computer is inoperable for a predictably
short period, we can continue an expensive experiment by acquiring
our data offline until the computer is functional once again, it
would be inconvenient to insert offline data into the computerized
data, but I think I would prefer to do that rather than throw away
the whole experiment.
Our PDAS is not totally independent of the human operator.
The digital data base processed by the computer supplies the
operator with information he needs to decide whether to intervene
or to allow the system to operate independently. The software can
prompt the operator to perform certain tests, or it can perform
the tests automatically. During the course of an experiment,
direct visual communication with the subject inside the laboratory
is maintained by a closed-circuit television camera. In addition
to an electrocardiogram attached to the subject, there is an
oscilloscope readout of the subject's pulmonary functions. The
readout is displayed on two CRT screens as either digitized wave-
forms or as alphanumeric information.
The use of this equipment allows for human intervention in
our highly mechanized system. For example, when a simple spiro-
metry test is conducted, within a few milliseconds after the
forced expiratory vital capacity maneuver is completed, a digitized
waveform shows on the CRT screen. Identifying hatch marks indicate
where certain software decisions were made. If the operator is
not satisfied with the point where the pattern recognition algorithm
indicates the expiration occurred, he can capture the identifying
mark with the cursor and move it to where he thinks expiration
actually occurred. When the operator moves the cursor, the computer
will create a file that has both the original digitized data and
the derived data for the waveform as it was defined by the computer.
Next to that file another file is created containing the new
information calculated on the basis of where the new cursor mark
was placed.
The nerve center of the PDAS is what we call the functional
keyboard — a panel of 25 numbered pushbuttons (five rows of five
buttons). By manipulating the keyboard, we can call up a variety
of software programs. We can choose any part of a program by
simply pushing the right number on the keyboard. Whenever any
test is designed that we want to include in our system, we can
write the appropriate software for that test and insert it into
the entire algorithm.
When the keyboard panel is not lighted, the buttons cannot be
pushed. When the panel light flickers, the buttons are ready to
153
-------
be pushed, at which time the panel becomes solidly backlit. The
solid lighting indicates to the operator that he may proceed with
whatever routine he has chosen. If the operator wants information
on spirometry, the system does not compel him to go through the
entire program of spirometry in a predetermined sequence. If
desired, the operator may choose certain components of the program,
such as forced inspiratory volume or forced vital capacity. Thus,
the keyboard provides for human control of the PDAS; it gives us
the option of deciding when, and to what extent, to perform a
given test.
Another facet of this system is the alternative to either
throwing out or keeping data — in other words, a way of saving
data we suspect might be incorrect so that we can check it later.
When we want to save data of this type, we put a special flag on
it by hitting the "poor data" button on the keyboard. If desired,
we can enter free text in a file that might say, for example, that
the subject coughed halfway through his FVC and we are uncertain
whether or not the data has been changed.
The other portions of the software package are not yet written.
The system is really infinitely expandable and alterable because
it is totally software dependent. Expansion is limited only by
our imaginations, what we decide we want to do, how much core we
want to buy, and how many new programs we plan to write.
Environmental Control System
A class 100 clean room is an artificial, arbitrary engineer-
ing requirement, but it is one that would warm the cockles of most
investigators' hearts. We decided that a class 100 clean room
would be our objective for our environmental control system.
We can achieve, relatively easily, class 1,000 clean room
conditions. Such conditions may not be clean enough for assembl-
ing microcircuits, but they are clean compared to the clean air
that people breathe most of the time, and class 1,000 conditions
are a nice control for us.
The outside air is prefiltered. To properly control tempera-
ture, we must use both heating and cooling in series. To maintain
good control of relative humidity, we dry the air and then reinject
the desired amount of water vapor. The air comes in through the
top of the controlled environmental laboratory, where diffusers
provide reasonably laminar flow from ceiling to floor. The air is
exhausted out through the perforated floor of the laboratory,
through several banks of high-efficiency particulate filters. In
addition, the air is filtered through activated charcoal.
154
-------
We have several levels of safety control as far as our environ-
mental system is concerned. The delivery of pollutant gases into
the air flow system is through low pressure lines, so if any loss
of integrity of that system occurs inside the building, we are not
dealing with a high pressure system coming off the supply bottles.
There are, at various points in the system, critical orifices
through which you cannot push air any faster than a certain flow
determined by the size of the orifice. This puts a mechanical ceil-
ing on the level of pollutant that we can put into the laboratory.
The actual makeup of the laboratory air is determined by the pol-
lutants injected into the air flow just proximal to the axial vane
fan; therefore, pollutants are churned up and well distributed.
For any particular circumstance, the volume of air flow is a
function of the damper setting, the speed of the fan, and the pitch
of the impeller blades on the fan. When these three factors are
adjusted simultaneously, they will determine the air flow through
the laboratory.
Allow me to back up a moment to point out the difference in the
two laboratories that I mentioned earlier. They are identical with
respect to gaseous air pollutant capability — specifically, ozone,
sulfur dioxide, nitric acid, nitrogen dioxide, and carbon monoxide.
In one of the laboratories, we are installing the aerosol gener-
ating and monitoring capability. The air flow for one of the
chambers had to be a little bit different than the air flow for
the other in order to accommodate the aerosol capability.
The aerosol generators can put any water soluble substance
into the air stream. The number of particles and the size of the
particles is determined by the concentration of the solution
supplying the nebulizer, the solution flow through the nebulizer,
and the air flow through the nebulizer. These three parameters
determine the mass concentration in the chamber and the particle
size distribution. The composition is determined by whatever is
put into the solution. That is why, for instance, we do not bypass
the hepa filter if we get it down to clean conditions. Then, once
most of the unwanted particles are eliminated from the airstream,
we bypass those filters to filter out all of the aerosol we intro-
duce in the system.
The solenoids either opened or closed, and the length of time
spent in the open mode or in the closed mode, determines the concen-
tration of pollutant gas in the chamber. These are hydraulically-
operated, compressed air solenoid valves.
155
-------
A high-limit alarm allows the operator to intevene, if neces-
sary. Even before the alarm goes off, the operator can monitor the
actual analogue output of the sensors. Because he knows the
relationship between analogue outuput and pollutant concentration
in the laboratory, the operator is able to find out whether or not
the predetermined pollutant concentration is actually in that
laboratory.
If the computer does not maintain the predetermined concentra-
tion of pollutant, the operator is able to intervene by changing
potentiometer settings or by manually overriding the computer. In
the case of either computer malfunction or computer crash, we
think the operator can successfully maintain the predetermined
concentration in the controlled environmental laboratory so that
the experiment can continue. In the event that this is not pos-
sible, we have made provisions for a single button to be pushed by
the operator which will abort the experiment and purge the chamber.
The button turns on the fan and opens the fusable link dampers.
Within a few seconds the chamber is completely evacuated of that
particular pollutant concentration.
We are most concerned about data quality control. Health
Effects Research Laboratory personnel are responsible for and are
in charge of all experiments. However, because we do not have
adequate engineering support staff in our laboratory to conduct
these experiments, we have a rather substantial operating and
maintenance contract with Rockwell International. We keep a close
watch on the laboratory environments that are maintained by our
contract personnel. Conceptually and administratively we monitor
the data quality control for both the physiologic data acquisiton
system and the environmental control system in the same way.
From time to time, though, an independent audit is conducted
by a third party and gives EPA a report. From that and from our
observation of the operation and maintenance of the environmental
laboratory, we will produce a data quality report from time to
time.
I would like to touch on some of the main aspects of the data
quality control, and some of the philosophical points in data
quality control. Very simply, we constantly make mechanical
status checks. For instance, with regard to the air flows, we know
that in order to properly maintain the environment, air flows at
certain points must be within certain limits, and that is checked
constantly. For temperature and humidity control, we take correct-
ive action to make sure that dampers are accurately set and that
the integrity of the circuits is sound. Finally, all the analyze-
rs are frequently calibrated with NBS referable standards.
156
-------
As far as the pollutant system is concerned with respect to
data quality control, we measure gas concentrations, obviously,
but we also look carefully at the gas flow rates past certain
points. If they are not within certain bounds, we know that even
though gas concentrations might be what we think they should be,
there is something wrong with the system. Solenoid positions
obviously have to be maintained in a certain configuration in
order to produce the gas concentrations we want.
Data quality control of the physiologic data acquisition
system is somewhat more complex because it is more difficult to
make sure the computer is giving information that closely cor-
responds to the data that are actually coming from the subject.
One of the ways to assure the reliability of computer information
is to compare it with canned or known input. Canned input is
synthetic data; for example, an EGG simulator is used or spiro-
metry data are mechanically produced. Such data are the same day
after day. Anytime you wish to check whether the numeric data
flashed on the screen are the same as they were the day before,
you can run the canned, or known, data through the computer.
We use physical calibration just as anybody does in any
laboratory, but we are also striving to achieve a high quality of
electronic time of use calibration. This is accomplished by using
the computer, which can detect any deviation in electrical signals.
If one or five milliwatts is put through, a certain signal is
obtained from the computer. Split signal redundancy analysis is
yet another approach. This permits us to split a given electrical
signal and run it through both of our computers. If the computers
are working properly, the same answer is obtained from both of
them.
If we get different answers, we may not know which one is
right, but at least we know that a problem exists. Trend analysis
is simply the intelligent and knowledgeable human operator watch-
ing the computer output for suspicious changes in the data. This
is a very soft kind of data quality control, but the computer
output of good trend analysis information probably will tip us off
to something.
Algorithm error analysis is, of course, applicable to any
algorithm, regardless of whether the algorithm is operating a
chamber or whether it is processing data obtained from a piece of
physiologic equipment. This type of analysis provides us with
built-in range checks. We know, for example, that vital capacity
should not be 15 liters or 500 cc.'s. If a data bit is identified
as unbelievable by our definition in the algorithm, then the data
is flagged so that we can examine it.
157
-------
I would like to conclude with a quick overview of where we
might be going with this. We have completed the first 30 subjects
of ozone adaptation in clean air, and we hope the experiment will
reveal some answers about differences in lung performance as it
relates to the length of time a subject has been breathing ozone.
We are very concerned about nitrogen dioxide and other environ-
mental stresses, such as heat, in conjunction with various pollutant
stresses. By January 1978, we should have a good capability of
generating aerosols, and by July 1978 we should actually be able to
generate and monitor them with online computer control.
Finally, our staff is now preparing the comprehensive program
plan that will describe how this facility will be used for the
remainder of this decade. We know what research capabilities we
have and we know the priorities and needs of the Environmental
Protection Agency. We encourage your participation is this pre-
paration process. Your ideas, your thoughts, and your proposals
are a national resource. We invite you to join us as correspond-
ents or, on occasion, as visiting scientists.
158
-------
SPECIAL CONSIDERATIONS AND
APPROACHES IN ENVIRONMENTAL
CLINICAL RESEARCH
Moderator: Philip Bromberg, M.D.
-------
Introduction to Panel Discussion
Philip Bromberg, M.D., Moderator
School of Medicine
University of North Carolina
I should like to disregard the typical question of what is
response and what is disease. Dr. Bates, who unfortunately is not
here to defend himself, suggests that response is disease, and
that the burden of proof is on those who claim that response is
not disease.
I should like to proceed to the question: What does one
measure as the effects of the agents that we are examining? How
do we measure them? When do we measure them?
The first thing that is measured is gross toxicology, for
example, excess mortality during air pollution academics. After
that, the researcher looks at the causes of the symptomatic and
mortality effects that have been produced.
Perhaps the first approach is to look at lung function. As
researchers probe deeper into the functions of the lung, they
develop new concepts of how the lung works, and produce more and
more elegant and sophisticated tests of lung function. These
studies are applied to man, as well as to animals.
Then, newer indices of lung actions are developed. When we
look at these new indices, the way they are studied progresses
from gross (e.g., the effect of ozone in fairly large concentrations
on airways is epithelium slough) to refined, to extremely elegant.
161
-------
In more refined studies, the researcher can look not only at the
gross destruction of the epithelium, but also at the rate of turn-
over of epithelial cells, which is a more refined index of the
death rate of epithelial cells.
Finally, the researcher can begin to study even more elegant
indicators, such as how permeable the airway epithelium is to
materials deposited in the lumen of the airways. This permeability
overall may be affected by the so-called mucous layer, although it
is not clear that the mucous layer is really a continuous layer.
Because it is an epithelium, the airway lining is characterized by
"tight junctions" between the epithelial cells. In many tissues,
these are known to limit the permeation of both the small and
large molecules.
Some evidence indicates that noxious agents, gases, can
affect the intactness of the tight junctions of the respiratory
epithelium and its permeability characteristics. The latter
effects allow materials to be deposited in the airway, where they
penetrate the deeper layers of the epithelium and act upon smooth
muscle vasculature and mast cells.
What exactly are we to measure and how are we to measure it?
These are questions that are subject to continual development and
refinement. The highlight of this kind of approach is always
mechanisms. We are not just interested in gross effects, but also
in the mechanisms that produce these effects.
I think that to answer this question coherently, governmental
regulatory agencies, such as the EPA, which is responsible for
developing concrete air quality standards, must be concerned with
mechanisms. Even though this conference focuses on human studies,
we should not limit our attempts to define air pollutant levels to
such studies. Our studies must be accompanied by an attempt to
understand the mechanisms that produce the effects.
Another difficult topic is the independent variable. We have
already heard that there are many problems pertinent to the identi-
fication and measurement of the independent variable. There are
multiple agents. There is, of course, the ordinary dose effect
relationship. There is the problem of timing--when, after exposure,
should one look for effects? How long should exposures be? What
about tolerance? What kinds of interactions are present between
external noxious agents; interactions with other pharmacologic
agents; interactions with physiologic stimuli, such as exercise;
interactions with temperature, humidity, and particles? These
questions give rise to so many permutations and combinations that,
to perform really pertinent experiments, I think one must have a
particular objective in mind—not merely a broad hypothesis.
162
-------
For example, a researcher ought to examine the mechanisms of how
a particular noxious agent works, and consider what type of
experiment can be devised to obtain more information about it,
rather than attempt to answer a question such as "Is this gas
noxious?"
Another major problem is the subjects we use in our experiments.
Whom should we choose as our subjects? Should our subjects be
healthy, or should we be looking at subjects who have some type of
disease? Disease, as well as health, is difficult to characterize
in human beings.
Finally, what are the effects of informed consent on human
subjects? Does the subject's knowledge of what effects to expect
influence his symptoms, and consequently, the test's so-called
objective findings? Real problems can arise from this element of
the equation. Again, I plead that researchers should think
mechanistically rather than simply be satisfied with the observation
of one effect or another.
Now, the panel and the audience may or may not agree with
everything I have said. I would like each member of the panel to
have a chance to make a statement of his own. Then we can have
some discussion after each member of the panel has had an
opportunity to talk, and follow up with a general discussion.
163
-------
Statement
Robert Frank, M.D.
Department of Environmental Health
University of Washington
I want to respond to some of the things Dr. Bromberg mentioned.
My personal philosophy regarding clinical research can be stated as
follows: First, clinical research can be directed toward under-
standing phenomena (Dr. Bromberg referred to this); that is,
toward providing a basis for an observed response. Second, clinical
research can be applied to describing dose-response relationships
for a particular agent.
Both objectives are important, particularly the latter. Dose
response relations affect standard setting, which in turn can
significantly affect society.
I also think the investigator engaged in the latter type of
clinical research should be prepared to discuss the biological
significance of his results. At the very least he should attempt
to assign some weight to his results. If he fails to do this,
then other scientific investigators perhaps ought to take up the
task, because if they do not, the task will be taken up by non-
scientists*
As Dr. Bromberg indicated, it is entirely possible that the
particular functional parameter under examination — the so-called
dysfunction the researcher may have observed — may, by itself, be
readily tolerated or quickly reversible. Alternatively, it may be
insignificant, but have important secondary effects. Dr. Lippman
165
-------
has stated, for example, that the transient bronchoconstriction
associated with exposure to irritants may in turn lead to increased
deposition of inhaled particles, which obviously has implications
for biological dosage.
But, if the investigator suspects that a dysfunction has
significant secondary effects, then he should state his hypothesis
and test it. At the moment, I think there are a number of so-
called positive findings in clinical research that have received
more attention than they deserve.
There may be some reasons for this. No one likes to profess
that the product of his work, the effect he has observed, has only
questionable or minimal importance. To do so means monumental in-
attention to the researcher's study and a loss of funds.
A word about functional testing. We rely rather heavily on
measurements of ventilatory function and respiratory mechanics.
There is good reason for this reliance. In the past these measure-
ments have been extremely useful and revealing, but we must remain
resourceful and be prepared to move beyond them, to recognize when
testing our hypothesis may demand additional techniques.
For example, Lippman has shown with sulfuric acid aerosol
administered to donkeys that in the presence of little or no
changes in respiratory mechanics, there may be significant depres-
sion of mucociliary clearance.
I am somewhat surprised to notice that in assessing changes
in function that are due presumably to reflect bronchoconstriction,
many of us still depend exclusively on a ventilatory test that
requires a maximum inspiratory maneuver. Years ago, Nadel showed
that a maximum inspiration can abolish, at least temporarily,
bronchoconstriction. Also years ago, Arend Bouhuys showed that a
partial expiratory flow-volume effort, which avoids this danger of
abolishing reflex bronchoconstriction, can be more revealing than
a flow-volume effort begun following maximal inspiration. Perhaps
this is a test we ought to use more often.
Furthermore/ when we start experiments involving ultra-fine
or accumulation mode irritant aerosols, as for example sulfuric
acid, we ought to emphasize tests that assess small airway function.
After all, such particles are most likely to land in the periphery
of the lung. This suggestion, at least in my own experience, is
borne out by studies on guinea pigs exposed to submicrometric
aerosols. Their dynamic compliance can be affected more often than
their total flow resistance. It is not easy to assess small
airway function. The tests are often time-consuming and their
166
-------
interpretation may be open to debate. Such tests can become
distracting in studies in which the functional changes may be
occurring rapidly.
In my judgment, the measurement of closing volume has not
been too revealing. Also, the forced expiratory flow rate at 25%
of vital capacity has a rather high variance so that subtle changes
in small airway caliber may elude detection by this means*
Dynamic compliance, which may be useful for detecting un-
evenly distributed changes in small airways or parenchyma, is
another test that is difficult to accomplish. It requires an
esophageal catheter and a well trained subject. It is not always
easy to control what has to be controlled, that is end-expiratory
lung volume and tidal volume, when the subject breathes at dif-
ferent frequencies. Perhaps we should experiment more with the
helium oxygen techniques developed at McGill University. McFadden
found this technique to be quite revealing in his studies of
exercise-induced bronchoconstriction. One problem may be the time
required to perform it.
Presently, my belief is that we ought to focus more attention
on what I term the "hyper-reactive subject"; in other words, a
subject who is otherwise normal and with no known underlying chest
disease. For the sake of discussion, I will accept Amdur's defini-
tion of a hyper-reactive subject as someone whose functional
change during exposure to a pollutant is at least three times
greater than the average group response. Amdur reviewed the
literature and noted that a number of investigators working in the
field encountered one or two subjects in their group who suited
this definition. For the moment we will exclude allergy as the
basis for hyper-reactivity.
Several questions might be asked about this type of subject:
Is the hyper-reactivity typical of the subject? That is, if I
observe an exaggerated response on Monday to a pollutant, will it
be present on Tuesday, Wednesday, Thursday, and one month later?
Does it have some plausible or identifiable mechanism? For example,
is it a matter of internal dosage?
You may recall the observations of researchers at New York
University years ago suggesting that the pattern of particle
deposition in the periphery of the lung is determined in great
measure by morphometric factors. Just as the aerodynamic character-
istics of an aerosol influence its rate and site of deposition, so
may airway and alveolar dimensions and ventilatory pattern. All
of these biological factors can vary considerably within the
normal population. Such variations among lungs, and particularly
within a lung, may become very important once lung disease has
167
-------
been established. There are likely to be variations in the distri-
bution of ventilation within a diseased lung, leading to local
"hot spots." Morphological factors of this type are likely to
impose differences in dosage, among or within lungs, over long
periods of time.
I wonder how much differences in bronchoconstriction among
subjects inhaling drugs such as acetylcholine or mecholyl may be
due to differences in the deposition rate or dosage.
An alternative explanation is that the exaggerated response
may reflect an altered or increased responsiveness to the same
fixed dosage. In his recent work on dogs, Nadel found that follow-
ing ozone inhalation, there was a clear-cut increase in the broncho-
constriction in these animals produced by parasympathomimetic
drugs. This hyper-reactivity reached a peak about 24 hours after
the ozone inhalation and thereafter receded. In this instance,
neurophysiological changes in function produced by exposure to
ozone, apparently led to hyper-reactivity. He noted, too, that
infection may produce the same result.
Functional or inflammatory factors of this type are likely to
impose differences in dosage over limited periods of time. As I
see it, if we could identify and understand the basis for hyper-
reactivity, we might be able to predict who among the population
is at greatest risk. The rewards might be especially high for
occupational settings, wherein the workers at greatest risk might
be identified before any impairment of health occurred. I think
it is a worthwhile hypothesis toward which research efforts should
be directed.
To move to another subject, it is probably a truism that we
ought to combine pollutant exposure with other forms of stress
during clinical experiments. This is particularly true when we
administer realistic concentrations of pollutants to healthy human
subjects. The most obvious concurrent stress is exercise. Studies
including both these variables, that is, exposure plus exercise,
have been well exploited by Drs. Bates and Horvath. It is probably
true that the more vigorous the exercise, the more likely the
effect on the response to the air pollutant. Some of the reasons
for this phenomenon are known. The increased ventilation that
occurs during exercise causes a greater amount of the pollutant to
be introduced into the respiratory system per unit of time, and
the increase in instantaneous flow rates that occurs during exercise
is associated with a reduction in the scrubbing efficiency (fraction-
al uptake) of the upper airways for soluble gases and particles.
This is particularly true when the subject switches from nose to
mouth breathing during exercise.
168
-------
In addition, there is probably some alteration in the sites
of transfer of these gases and particles in the lower airways.
For example, when a subject ventilates at high flow rates while
inhaling a particle of sulfuric acid, as during exercise, the pH
of the particle at the instant of impaction in the airways is
probably lower than during quiet breathing. This difference in pH
may be the result of two phenomena. First, the particle may be
less hydrated during exercise when there is less time for it to
absorb water (before impaction) and therefore, less time for the
water to dilute the hydrogenion concentration. Second, there is
also less time for any chemical transformation or neutralization
of the particle by ammonia in the upper airways.
There are other forms of stress that have not been adequately
explored. Recently, Dr. Horvath has begun combining heat with
exposure to pollutants. We have not, to my knowledge, rigorously
studied the effects of a combining cold and a pollutant, especial-
ly when a soluble gas is combined with a droplet aerosol. Certain-
ly, one of the impressive features of epidemiologic studies is
that sudden changes in ambient temperature may overwhelm the
effects of pollution.
Finally, an approach that I think is potentially as rewarding
as any other is to study the effect of psychological stress on
response. We have not yet begun to design appropriate psycho-
logical stresses for this purpose, and I have no idea how to
proceed. It is possible that noise might be an effective co-
stressor.
fly last point is really designed to evoke a not-always-friendly
response. In clinical research, we are beginning a generation of
experiments in which multiple pollutants are administered simultane-
ously for several hours. The ostensible purpose of these experiments
is to see if the combined effects of the simultaneously-administered
pollutants are addictive, synergistic, or conceivably, antagonistic.
I have misgiving about a number of these studies as follows:
Unless there is a plausible chemical or physical basis for inter-
action among the pollutants that will cause them to form some new,
presumably more irritant species or end product, or unless there
is a plausible argument why two or more pollutants are likely to
affect the same biochemical or physiological function, or at least
affect separate functions that are interdependent, the results of
such studies are likely to be disappointing and ambiguous.
Experiments of this type are extremely complex in design.
They require repeated observations on the same individual to
establish whether synergism, et cetera, has occurred. I have been
impressed with the variance in the responses of subjects to the
169
-------
same pollutant on different occasions, as well as the variance in
the functional parameters that have been tested.
The researcher is likely to end up with the product of these
several errors/ which makes it extremely difficult to detect all
but the most dramatic effects. I could readily support an attempt
to study the biological consequences of interactions among sulfur
dioxide, ozone, and some droplet aerosol. The analytical chemist
tells us that there is a reasonable likelihood for a heterogeneous
reaction among these agents with the formation of an irritant
small particle.
But I would be less enthusiastic about mixing agents like
sulfur dioxide, nitrogen dioxide, and carbon monoxide. Unless I
am mistaken, these three gases do not interact and their sites of
effect are by and large different. I would seek the rationale for
such a mix and would not be satisfied to learn that it is "because
all three gases are found in polluted air." Our resources are too
limited for that sort of luxury.
170
-------
Discussion
DR. HAZUCHA: Dr. Frank mentioned hyper-reactive subjects.
Certainly, you have to find out how to identify and study these
subjects. But from my experience with testing 20 or 30 subjects/
we usually find one or two who do not react at all. To get a true
picture of the response, we have to identify not only hyper-reactive
subjects, but the hypo-reactive subjects as well*
DR. FRANK: I couldn't agree with you more. Again, I am
speaking as a member of a society concerned about protecting
people. It is true that you may learn as much about this phenome-
non of hyper-reactivity by pinpointing what it is about the subject
that protects him, or dampens the result, as you do by identifying
what it is that exaggerates the response.
DR. CAVENDA: You have not mentioned repeated tests (other
than the fact that the subject reacted on Monday, Tuesday, Wednesday,
Thursday) to determine if, over a period of time, a subject might
be tested for changes in response. For instance, a subject might
experience anxiety the first time he engages in a new activity
defined by the testing procedure. This may produce a response
that is the result of the effect of psychological stress, which
may diminish as the subject grows accustomed to performing the
activity on a continuous or chronic basis.
DR. FRANK: I did not emphasize it, but in speaking of hyper-
reactivity, I posed the question: "Do subjects respond the same
way when tested repeatedly?" I think you can define "repeated
exposure" as you choose.
I would agree that anxiety can influence response. We ought
to test this notion as an hypothesis. I think anxiety is a par-
ticularly important factor when you are testing a highly sensitive
population, such as asthmatics.
171
-------
DR. KNELSON: We specifically take into account the problem
that has just been raised. We do not test for intra-subject varia-
bility as Dr. Frank suggested, but at the Health Effects Research
Laboratory we try to control -it as a variable.
One of the most important parts of our experimental design is
the training period, the habituation period of all the subjects
prior to the onset of the experiment. This is the last hurdle that
the test candidate has to pass before he can become a subject for
experimentation. During this period, we find out if the subject is
capable of becoming a reasonable habituated, trained individual in
the environment we expect him to participate in for a few hours or
a few days.
We not only control for differences in response (we call it
"training effects") the way I have described, but we also test for
its persistence. Do the data on Monday differ from those on Tuesday,
Wednesday, Thursday, and Friday? No, we have found that they do
not.
The first experiment we ever conducted here was with carbon
monoxide. In that particular case, we designed a double-blind,
crossover study that specifically addressed the question of whether
the subject's performance was different on Friday than it was on
Monday. I think the idea of testing this as a hypothesis is very
interesting. I would suspect that we would have considerable amount
of variance.
DR. FRANK: I happen to think that many responses are char-
acterized by a considerable amount of variance. You can define
an experimental circumstance in such a way that you say, "I will
eliminate variance by dropping all subjects who do not satisfy
my definition of conformity."
But in doing that, you may exclude some significant proportion
of the population. I wonder if in some way — it might be very
difficult — we shouldn't begin looking at the excluded proportion
of the population as well.
DR. HIER: The variance within a subject could be so-called
stochastic variance, too; that is, the subject's response could
vary randomly from day to day. To account for random daily varia-
tion, we take repeated measurements and average out that kind of
variability. If you want to introduce a specific variable, for
example, anxiety, you introduce particular conditions that might
exacerbate anxiety in one setting or another, so that it becomes a
cross-condition effect.
172
-------
But I am concerned that variability from one time period to
the next is not being ascribed to strict experimental conditions —
that subjects are being subjected to more or less anxiety, to more
or less training effect. But even after you control for all these
variables, you still must contend with stochastic variability
The important thing is to take repeated measurements.
DR FRANK: I would like to respond to Dr. Hier. In the
typical experiment, one or perhaps two measurements of fuctional
parameter are taken, and then an animal or human subject is exposed
for a period of hours, followed by one or two measurements taken
during or at the end of the exposure. My impression is that this
procedure is extremely thin ice on which to characterize: one,
the baseline function; and two, the changes that may be imposed by
the stimulus you are applying. We very often stretch the data
beyond the weight it can really support.
DR. BENIGNUS: By "thin ice," do you mean the paucity of
measurements taken?
DR. FRANK: That is right, because the few measurements we
make do not truly characterize the response during the period
under study.
DR. BENIGNUS: Two measurements are not very much.
DR. OTTO: Just a quick follow-up on what you are saying. I
agree very much with the comments about how to deal with sta-
tistical variability, but there are problems with using a repeat-
ed measurement design. For instance, if you are dealing with a
chemical agent, you may be considerably limited by a repeated
measurement design because the chemical's cumulative effect may
thwart your attempts to obtain responses at the baseline level.
In short, responses may be modified by cumulative effects.
You may decide to wait three or six months to take another
baseline measurement, but by that time numerous factors far beyond
our control have changed—the temperature, the climate, the indivi-
duals' home lives, or the subjects may no longer be attending
school. These are particularly strong variables that we have had
to face when dealing with the student population.
DR. FRANK: These are highly complex experiments, and unless
you have a hypothesis that you can test plausibly, and unless you
pay real attention to this problem of describing the functional
parameter before and after you intercede, you are going to end up
with ambiguous data, which often happens in this field.
173
-------
DR. OTTO: The importance of the initial finding may be much
greater than later findings because the subject's tolerance to the
test substance may change. The strength of the response is going
to change and become much weaker.
So, I am not so much worried about the importance of the
initial changes that are observed. In fact, sometimes with carbon
monoxide/ we suspect that response changes may be an effect that
occurs very early at very low levels — that the organism is able
to adapt almost immediately, say, within 20 to 30 minutes; and, if
you do not catch the response as it occurs, you miss it forever.
It is a real problem.
174
-------
Statement
Bernard E. Statland, M.D., Ph.D.
Department of Pathology
University of North Carolina
I would like to discuss the general problem of intra-individual
variation of laboratory results. Intra-individual variation is
the variation in results that one would note over time in one
individual. This problem should be of great interest to all investi-
gators studying the effects of environmental stresses on humans.
So often investigators use mean changes in a group of individuals
rather than looking at a person as his own control. Since many of
these stresses of the environment will evoke a significant change
within an individual/ but not for a group as a whole, it is perti-
nent to understand the expected sources of intra-individual varia-
tion. As soon as we have an understanding of the sources and
magnitudes of the expected intra-individual variations, we will be
better equipped to handle the experimental problem.
The major sources of intra-individual variation of laboratory
results (and also most other variates which we measure in individ-
uals) can be grouped into three types:
1. Analytical variation
2. Variation due to preparation of the subject
3. Variation due to temporal considerations when obtaining
a specimen
Figure 1 schematically presents the source of analytical
variation. The analytical variation can be divided into two types:
175
-------
-J
o>
SOURCES OF ANALYTIC VARIATION
separation
decanting
storage and
processing
I
"SPECIMEN"
PRE-INSTRUMENTAL
PHASE
instrumental
analysis
743
result
"SAMPLE"
INSTRUMENTAL
PHASE
Figure 1
-------
the preinstrumental sources of analytical variation and the
instrumental sources of variation. The preinstrumental sources
of variation in results determined on a specimen of serum
would include the venipuncture procedure itself, the filling of
the tube(s) of blood, the transportation of the whole blood speci-
men, the centrifugation step, the variation occurring while the
serum sits on the clotted cells, the decanting of the serum
supernatant, the storage of the serum specimen, the freezing
and thawing of the specimen, and the processing of the specimen
prior to the specimen entering the instrumental station. The
instrumental sources of variation in the case of laboratory
results obtained on a serum sample include the dispensing of the
specimen, the uncertainty in the temperature during which the
analysis is performed, and the spectrophotometric uncertainty,
etc. Most approaches of quality control of the analytical variation
are truly only monitoring the instrumental phase and not the pre-
instrumental phase of the analytical procedure. Obviously, the
intra-individual variation in laboratory results obtained on an
individual will always include the analytical sources of variation.
The analytical sources of variation will be the sum of the within-
batch and the batch-to-batch variations.
The second major source of variation is that due to the prep-
aration of the subject. Subject preparation includes all those
factors done to the individual or by the individual independent
of the pathological process (or environmental stress) that one
is investigating. Examples of such factors are prior exercise,
previous diet, ethanol ingestion before specimen collection,
posture of the subject, and tourniquet application time.
The effect of exercise on serum enzyme values can be appreci-
ated from the information seen in Figure 2. In this figure, we are
depicting the percent change from baseline values in four subjects
at various hours after a one hour exercise stress. The four sub-
jects were males who underwent a 1 hour game of paddleball. At
1 hour, 5 hours, 11 hours, 19 hours, 29 hours, 43 hours, 53 hours,
and 67 hours after the stress, specimens were obtained from these
subjects. The activity values for the enzymes creatine kinase (CPK),
aspartate aminotransferase (AST), lactate dehydrogenase (LDH), and
alkaline phosphatase (AP) were determined on the serum specimens.
The values were compared with baseline values (those obtained prior
to the exercise stress). As noted in Figure 2, the mean creatine
kinase values were 120% above baseline at 11 hours after the stress.
The AST and LDH values also were elevated significantly. CPK, AST,
and LDH are all enzymes present in muscle. Alkaline phosphatase is
not present in muscle to any great degree. This enzyme showed very
little change after the exercise stress.
177
-------
LU
Q.
5 II 19 29 43 53
HOURS AFTER EXERCISE STRESS
Figure 2
67
+ 20
UJ
o
< +10
i
o
H
-z.
UJ
s °
UJ
Q_
-10
rr
|d
-
•7
V
b
-i
i
\
L.
• AS,'
D ALT
D LDH
D CK
C GGT
n AP
n
1
1 Jh,
n
.
1
i
15 hours 36 hours 60 hours 100 hours
HOURS AFTER ETHANOL INGESTION
Figure 3
178
-------
We evaluated the effects of ethanol ingestion on nine healthy
subjects. The subjects had baseline values determined for the
serum enzymes AST, ALT, LDH, creatine kinase (CK), gamma glutamyl
transferase (GGT), and AP. The ethanol stress consisted of 0.75
grams of ethanol per kg of body weight ingested on three separate
evenings. At 15 hours, 36 hours, 60 hours, and 100 hours after
the third evening of ethanol ingestion, specimens were obtained
and determined for the enzymes depicted in Figure 3. The most
impressive change was noted for serum GGT. The mean increase in
GGT was approximately 25% at 60 hours after ingestion.
We examined the effects of posture on a number of serum analytes
in 11 healthy subjects. These healthy subjects, all male aged 21
to 24 years, had blood obtained both in the supine position (lying
horizontally for one-half hour) as well as in the erect position
(standing for 15 minutes and sitting 1 minute before venipuncture).
As noted in Figure 4, the serum proteins, analytes bound to proteins,
and enzymes were significantly increased. The bars in Figure 4
represent the percent change going from the supine to the sitting
position. In the case when such a change was significant, the bars
are stippled.
The third major source of intra-individual variation is that
variation due to the effect of time of specimen collection. The
temporal considerations can be divided into the within-day (hour-
to-hour) and day-to-day changes. I will concentrate on the day-to-
day changes.
Figures 5, 6, 7, and 8 represent the concentration values for
serum thyroxine, serum cortisol, serum triglyceride, and serum
cholesterol in each of four healthy subjects. The subjects are
labeled number 3, number 5, number 7, and number 8 in these figures.
Blood was obtained at the same time of day after the subjects were
in the sitting position. The posture and tourniquet application
time were standardized. As noted in these figures, the intra-individ-
ual variation for thyroxine was greater than the inter-individual
variation. However, for the constitutents serum cortisol, serum
triglyceride, and serum cholesterol/ the intra-individual variation
was much less than the inter-individual variation.
The relationship of the intra-individual (within subject) to
the inter-individual (subject-to-subject) variation can be appreciated
for the values of alkaline phosphatase activity in the serum of four
healthy subjects. Figure 9 presents the values obtained on each of
6 days during a 10 day period for four healthy volunteers. It
is obvious that the subjects maintain their values for alkaline
phosphatase in a very narrow range during the course of the study.
179
-------
SODIUM
POTASSIUM
CALCIUM
CHLORIDE
PHOSPHATE
UREA
CREATININE
TOTAL PROTEIN
ALBUMIN
IRON
URIC ACID
TOTAL LIPIDS
CHOLESTEROL
AST
ALT
ALK. P'TASE
ACID P'TASE
0 +5 +10 +15
PERCENT CHANGE
Figure 4
180
-------
Q i-
8
o>
3
UJ
X
o
cc
>-
T.
h-
J_
J.
2
45678
DECEMBER 1975
10
50 r
40
Figure 5
o»
30
fe 20
O
o
10
45678
DECEMBER 1975
10 II
12
Figure 6
181
-------
Note that we are using arbitrary units rather than international
units for alkaline phosphatase.
Figure 10 depicts the ranges for alkaline phosphatase for each
of fourteen individuals studied over 10 days. Specimens were
obtained at 8:00 in the morning. They were analyzed in one batch
at the conclusion of the study in order to decrease the magnitude
of the analytical variation. In such a manner, we would only have
to deal with the within-batch variation and not the batch-to-batch
analytical variation.
Figure 11 shows analogous information for the concentration
values of the serum complement C4 values. Once again, we see that
the individual maintains a much narrower reference region than
does the group of individuals.
The conclusions drawn from our studies thus far on the intra-
individual variations or serum constituents are the following:
1. Under conditions in which the analytical variation and
the variation due to the preparation of the subject are
minimal, the intra-individual variation is often much
less than the inter-individual variation of mean values.
2. When the intra-individual variation is much less than
the inter-individual variation, it does make sense to
use a subject's baseline values as his own control.
When electing to use a person as his own control, there are
certain issues that must be taken into account. The first issue
is deciding on the appropriate biological model of time-series
that a value is expected to follow. Two major models have
been composed: the homeostatic model and the random walk model.
Figure 12 illustrates the homeostatic model. In this model it
is assumed that a subject has a mean value below and above which
all values should fall. The variations from the mean are taken
into account in order to compute a prediction interval within
which the next value should fall. In Figure 12, we are examining
four values of serum iron in a healthy subject. The fifth value
should be expected to be the mean of the four. The width of the
prediction intervals is dependent upon the standard deviation of
the previous four values.
In the random walk model (see Figure 13) the next value in a
series of values should most likely be the most previous value.
In Figure 13, we are looking at the same four serum iron values
seen in Figure 12. The prediction interval, however, is 8.2 to 21.1
for the 95% prediction interval as compared to 10.7 to 24.6 for the
182
-------
homeostatic model. It is quite obvious that one must choose a
model of biological time-series before being able to evaluate the
width and the location of the prediction interval.
When using a person's values as his own control, the analytical
variance plays a very important role in the width of the prediction
interval. The relationship of the analytical variance to the biolog-
ical variance is depicted in Figure 14. This figure represents the
relationship of the percent increase in intra-individual variation
due to analytical error to the ratio of the analytical variance over
the biological variance. When the ratio of the analytical variance
to the biological variance is 1.0, one would expect a 42% increase.
When the analytical variance to the biological variance is 4.0,
the expected increase is greater than 120%. It had been recommended
that the analytical variance to the biological variance ratio be
no greater than 0.25. In this latter case, the percentage of
increase would be approximately 12%. Such a guideline might be
very important in prospective clinical studies as well.
In conclusion, I have discussed the various sources of intra-
individual variation that we as clinical chemists must deal with.
However, the intra-individual variation has much wider applicability
than just to clinical chemistry. It should play a role in monitoring
a number of variates including blood pressure readings, respiratory
function tests, and other parameters that are followed over time.
Since more and more investigators are using a subject's previous
values to evaluate the effect of various stresses, the elements
going into the intra-individual variation should be noted.
The work I have presented here has been done in collaboration
with my colleague Dr. Per Winkel who is presently at the Finsen-
institutet in the Department of Clinical Chemistry in Copenhagen.
This work has been supported in the main by the Danish Research
Council, departmental funds from the University of Minnesota, and
departmental funds from the University of North Carolina, Department
of Hospital Laboratories.
183
-------
400
3OO
LD
Q
£ 200
UJ
52
01
100
45678
DECEMBER 1975
IO II
12
Figure 7
300
CHOLESTEROL
ro
O
O
184
4 5 6 7 8
DECEMBER 1975
Figure 8
-O 7
10
12
-------
_ _
5.0
3.0
2.0-lr
T_
2 345 6 7 89 10 I
DECEMBER 1975
00
Ul
Figure 9
-------
00
O)
RANGE OF ACTIVITY VALUES OF ALKALINE
PHOSPHATASE IN SERA OF FOURTEEN HEALTHY
SUBJECTS FOLLOWED OVER TEN DAYS
CO 4
3
2
1.0 2.0 3.0 4.0 5.0
ALKALINE PHOSPHATASE
(ARBITRARY UNITS)
6.O
0 60 120 180 240 300 360 420
COMPLEMENT C4 (mg/liter)
480
Figure 10
Figure 11
-------
PREDICTION OF SERUM IRON VALUE IN
HEALTHY SUBJECT USING HOMEOSTATIC MODEL
28
o3 24
1 20
z 16
o
cr
i12
en
LU
^ 8
4
26.8 -TT
24.6 |
- \ |
1
10.7^
••'.*
8.5^
[S3 95% PREDICTION INTERVAL
03399% PREDICTION INTERVAL
ii"1
|234)|i
Tv1
{•;•;
T*"
;!•
DAY NUMBER
Figure 12
187
-------
PREDICTION OF SERUM IRON VALUE IN
HEALTHY SUBJECT USING RANDOM WALK MODEL
28
w. 24
\
•o 20
O
cr
ID
en
L±J
(S)
16
23.1-
i1
1
rs o Ox.
95% PREDICTION INTERVAL
99% PREDICTION INTERVAL
I
I
234
DAY NUMBER.
Figure 13
188
-------
200
0
8
(ANALYTICAL VARIANCE )/
-------
Discussion
DR. BRQMBERG: Apparently, there are many sources of what in
the first approach we might consider as noise. However, part of
this noise apparently is not noise. It is a response to other vari-
ables that are not adequately controlled. So, in experiments,
the researcher tries to look at some of these variables* For
example, muscular exercise unfortunately has some prolonged effect
on certain enzymes.
Other sources of variability apparently remain completely
undiagnosed, and we call them noise. Evidently, there are two kinds
of noise, some of which we may understand and try to control in our
experiments, and others that are beyond us, which we have to live
with.
The worse the total noise, the more limited we are in looking
for subtle or small effects on signals. How do we deal with that
problem? How do people like Dr. Horvath, Dr. Frank, and Dr.
Battigelli—all members of the panel who are, in fact, very much
concerned with making measurements in man and picking up relatively
small effects—deal with this problem in a practical sense?
In addition, how small are the effects that we ought to look
for? Is there some limit we ought to impose on ourselves and not
attempt to look for effects beyond that limit? The experimental
situation is too complex for us to eliminate a certain residual
level of uncertainty. Perhaps we ought to look only for fairly
gross effects.
DR. PENGELLY: I don't know whether we were ahead of the
clinical chemists in recognizing this problem of unidentifiable
noise, but when we designed protocols for the ozone experiments
performed in the late 1960's, it was clear to us that we could not
191
-------
use a population standard for what we would expect. We did not
pick a population standard for FEV * and then compare a response
of an individual.
It was clear that we had to peform what we called sham experi-
ments, in which we allowed for whatever variable that was going to
occur. We changed only one variable in the experiment, the expo-
sure to ozone. Then, we repeated the experiment in exactly the
same way over exactly the same time period with exactly the same
individual and compared the two experiments.
I think this approach has long been the traditional approach
to exposures of human subjects to environmental contaminants of
this type. Moreover, I think Dr. Frank was very impressed with
the results of the controlled experiments conducted at the facility
here—how the individual variability is quite small and how well
individuals have been characterized. I think that the lesson of
the significance of intra-subject variability has been learned in
this area.
DR. BROMBERG: But you have not-controlled for the effect of
time in your procedure. You did an experiment on day one, and you
made some assumptions as to how long you had to wait to do a valid
experiment, a sham experiment, and then you conducted that sham
experiment several days of weeks later. You were still left with
some potential for variability.
DR. PENGELLY: I think the important point was to compare the
individual against himself.
DR. HORVATH: In one sense, I disagree with you, Dr. Pengelly.
In the first place, I believe that we all have an idea of how to
handle intra-subject variability, but I think Dr. Statland1s
observations indicate very clearly that this is all a figment of
our imaginations. It leads us into an uncomfortable situation of
evaluating the effects of anything on a human organism.
One of the points I made in my presentation was that the
investigator has to prepare the subject. Dr. Statland has empha-
sized this point. There are three variables that we have not
considered: first, the effects of changes in posture, which can
result in a 16-20% change in plasma volume; second, the change in
biorhythm, which we have ignored to some degree; and third, the
most important factor—what happens to an individual on day A and
how long does he have before he fully recovers? I repeat again the
*forced expired volume in one second
192
-------
statement I made, because I think most people have not looked at
all the literature in this field, and I am sorry to say that I did
not know about your work either.
It is certainly obvious that it takes many days to recover
from any kind of strenous activity. Recall Dr. Statland's data.
Out of 10 marathon runners, only two were able to complete a
short, 5 mile race 5 days after they had run a 2-1/2 hour marathon
race. Moreover, the two subjects who completed the short race
were unable to run it at the same speed they had run the 26-mile
marathon 5 days earlier. It took almost 7 days for some of the
blood parameters (hematocrit, for example) to return to normal. It
appears to me that some of the effects we are looking at are very
definitely hidden by our failure to know exactly what has happened
to the subject.
I am just as guilty as anybody else because I accepted the
same premises you did, Dr. Pengelly. But I am convinced now,
after what we have been doing the last 2 years in this area, that
we have made horrible mistakes* We have got to use an entirely
different approach in our testing.
DR. BROMBERG: Dr. Horvath, let me ask you a question.
Suppose I do an experiment where I compare vital capacity measured
on day two after exposure to gas "x," to the vital capacity measured
on day two following a double-blind exposure to scrubbed air. I
find the exposure to gas "x" in four out of four subjects has
apparently resulted in a reduction in the vital capacity of 5%,
plus or minus 1%. Every single subject went down, but one went
down 4%; another, 5%; another, 6%. Now, have I found anything?
DR. HORVATH: Of course you have found something, but what do
your findings mean? I think everyone has done about the same sort
of thing. However, the point is do your findings represent anything
in terms of what has happened to the subject? For example, one of
the things we did with ozone was to show very clearly that the
greater effect is observed immediately after the individual has
finished exercising. If you waited half an hour or 15 minutes,
the effect is less: that is, the decrement is smaller.
I feel very strongly that we have missed the boat with regard
to our approach. We have missed it because we still work on the
basis of old-fashioned statistics; namely, we use "T" test for
evaluating everything, and that is not a critical way to evaluate
the test results.
193
-------
I think we are in this same position right now. I just want
to say that Dr. Statland made a very nice presentation because he
brought out three important testing considerations: time, day-
to-day position of the subject, and the effects of the studies
done prior to the time the individual was used as a subject.
DR. BRQMBERG: You feel not that we have missed the boat, but
that we have over-interpreted the boat?
DR. HORVATH: Well, that is missing the boat.
DR. BATTIGELLI: I would like to mention that Theodore Hatch
said some 10 or 12 years ago when he stated that our ability
to measure and to identify factors of possible injury far exceeds
our ability to interpret the meaning of our measurements. He
stated that at times we may vastly exceed our ability to interpret
effectively what we can observe and quantify quite carefully.
But indeed, I think the reply to the question we are discussing
is what Dr. Statland and Dr. Horvath have presented to us, and what
Dr. Frank and Dr. Pengelly have stressed--the proper design of an
experiment. By proper design of experiments, I mean segregating all
the significant variables (whether irrelevant or important) that
may affect the detection or the observation of the effect. Once we
accomplish that, we can proceed to the next step, which is to
interpret change and/or effect.
But in recent times, and this is probably what Dr. Horvath
alluded to, all too often we have made interpretations based on
crude designs, mixing into that process various factors so that
the significance in the causation of the effect observed is lost
or equivocated.
DR. BROMBERG: What I think you are saying is that we ought to
design our experiments better so that expert statisticians can help
us obtain the absolute maximum from our experiments with the minimum
chance of mistake.
I would like to look at this issue a little bit differently.
I would like to postulate that we have found a 5% change in some
variable that may be of interest. But suppose we had a system of
electrical resistors arranged in series, with a total resistance
being the sum of all of the resistors. We tamper with one of these
resistors, causing its resistance to change 200%. Yet, when we look
at the overall resistance of the circuit we see only a small change.
It turns out to be a definite change, but a very small one. And
we wonder—what does this change mean? If we only focus on the
194
-------
pertinent part of the system, we might then see that we had indeed
uncovered a very, very big change.
If I am right, this illustrates the importance of focusing on
mechanisms; of trying to understand in detail what is happening,
rather than being satisfied with an overall, very small effect and
remaining uncertain. Does the response really mean anything? Is
it biologically meaningful? Is it clinically meaningful, or is it
only a trap? Is it due to poor design of the experiment?
DR. BATTIGELLI: Granted, but don't you find that Dr. Statland1s
approach is exactly the right way to identify a mechanism? In other
words, the statistician is not the only crucial person that can
provide us with an adequate design. It is the epidemiological
design that probably overrides or is as important as that factor
or design that satisfied only medical adequacy.
DR. HORVATH: I think it might be appropriate to have on the
record that we are going through our usual 10- or 20-year cycle
in research. I would like to point out that in 1966, the New York
Academy of Science published an extraordinary review of papers on
human variability. I think it would be to our advantage to read
those papers again. I think we have forgotten what we intend to
do and now we are back on that cycle of remembering again, which is
nice to know that it takes ten years to do.
DR. STACY: I would like to inject a note of heresy. Any
physiological scientist who has been around for very long knows
that there are many factors that comprise the results of any
single measurement. If a researcher knows that a factor is going
to influence his measurement, he controls that factor. Eventually,
though, you have controlled for all the factors that you know
about, and you are left with a stochastic residue with which you
must live.
But having controlled everything that you know about, you
proceed with these experiments and use what statistics you have to
get results and draw conclusions from the experiment. I think we
are worrying too much about the fact that there are a lot of
variables. Of course there are; we know that, so we design around
them as much as we can. And what we can not design around, we
treat exactly as we do in measuring average evoked responses. We
know that there is a lot of noise, thus we take enough measurements
to average out the noise.
DR. HORVATH: But my point is that we do control all the
variables we remember, but we do not remember all the variables we
should.
195
-------
DR. STACY: That is true.
DR. KNELSON: We might open up just one other area of discussion
briefly, and that is the way all of us are used to handling the
double-blind, randomized, crossover design where each individual is
his own control. It has become almost a cliche in experimental
design, but I think there is another aspect that we are missing,
which we are beginning to examine. When we study 30 subjects under
a set of environmental conditions (in this case, clean conditions),
and if we have properly followed all the criteria Dr. Otto talked
about in selecting these subjects, there is no reason to believe
that our test subjects are not as representative of the universe
that we are trying to estimate as the next set of 30 that we
select, using the same criteria.
We we study the next 30 subjects under the experimental
conditions, we do not have the compounding variable of a person
serving as his own control, but during that time he has experienced
a set of environmental conditions that may have changed his control.
So, using multivariant statistics where 30 subjects serve as a
characterization of the universe that we are trying to estimate
some parameter of, and then taking another 30 and using them in
only the experimental situation, you do not go back to the old "T"
test.
I think we get some statistical follow-up at times there. We
may also be getting a better representation of the population. In
clinical medicine, frequently the population of concern is a single
individual. That individual is the universe of concern. We are
concerned about 200 million citizens of the United States.
DR. HIER: I would like to expand on that, Dr. Knelson. The
idea of using an independent control group can be important in
that when you start, the subject who undergoes a "control" experi-
ment is permanently altered; consequently, you no longer have a
fair comparison if you use that subject as an experimental subject,
too.
You can counterbalance that by giving him an experimental
control, but that takes a long time. You can have your cake and
eat it too, in a way, by giving each subject a baseline run without
experimental intervention, giving one group of subjects two baseline
runs and the other group of subjects a baseline and experimental
run. The baseline run that is given to one group of subjects, which
is the next experimental line, serves as a control for individual
differences.
196
-------
Then you can subtract or use an analysis of covariance for
each individual difference, for each individual's characteristics.
That way you do not have to worry about the permanent alterations
that occur every time you do something to a person.
DR. STATLAND: You are right. You are about ten years ahead
of clinical chemistry. But the reason why you are is that the
concept of preventive medicine has only come about recently within
clinical medicine. Basically, we have been very therapeutic in
nature. We have looked at the disease after the cause. I think
medicine is changing in general, not just clinical chemistry.
The second reason we clinical chemists are now looking at the
same issues as you is because our analytical procedures are now so
much more precise and so much more accurate. We can do a glucose
test today much more accurately than we could in the past. There-
fore, the question is: With this extra analytical precision and
accuracy, what are the other issues that we want to deal with, that
we want to consider and control?
Just a few comments about points that have been mentioned
before. One of the concerns when you compare two groups of 30
individuals is trying to define analytical variations. And, of
course, that occurs within the subject as well.
As you become more proficient in conducting experiments, you
also may change certain things, for example, the number of baselines
to have on the subject or the type of biological time series model
that one must adopt.
197
-------
Statement
L David Pengelly, Ph.D., P.Eng.
McMaster University
Hamilton, Ontario
I am approaching this topic, "Special Considerations and
Approaches," wearing three hats. My first hat is an electrical
engineer. As an instrumentation engineer, which I was for many
years with David Bates, I am concerned about the precision of all
environmental and biological measurements. We had quite a lot of
good discussion about this, and I think we could perhaps still have
a little more.
My second perspective is as a physiologist. I am concerned
about the need for a broad testing approach, and in fact, we have
already touched on some of the effects of biological variables. One
variable that has not been mentioned, which I think is very impor-
tant, is the age of the subjects.
Facial origin, for example, may be a very important factor in
test responses. The hormonal effects, such as the estrus cycle in
females, is obviously a factor that should be taken into account.
These variables have already been referred to. The question of
whether you are measuring the right thing should also be included in
the broad approach. One always has the suspicion that he may have
overlooked a sensitive factor that should have been measured.
Last of all, right up there on top, is my third hat—as an air
pollution health effects scientist. I feel strongly that early stud-
ies on the synergistic effects of pollutant gases and so-called inert
199
-------
aerosols will be necessary to enhance the credibility of the chamber
studies that we are conducting.
Perhaps I might take issue with Dr. Frank on this point. Ob-
viously, one should not begin this type of study with a "Let's see
what happens" attitude. I think there is ample evidence that the
aerosol particle is a very special vehicle for conveying pollutants
to nasty places.
I would now like to elaborate on these points. The first issue
concerns the precision of measurements. In the early days, many of
us used simple homemade chambers made of plastic and a few bits of
angle iron. In Figure 1 note the high-volume sampler we used to
measure suspended particulates, and that the subject is sitting at a
desk. That is the sort of elementary approach we used.
Figure 2 shows our pulmonary data acquisition system. It is a
very simple acquisition system, but I think it is aesthetically more
pleasing than the ones that you have here. It is capable of a
certain amount of independent programming. I included these figures
for historical interest, and to show you that we use a very simple,
inexpensive approach.
We cannot compete with the EPA in terms of throughput or in the
detail of the meticulous studies you are capable of here at the
Health Effects Research Laboratory. Consequently, I think that
places a special responsibility on you for giving us the data on
the studies you have performed. The tools that you are using—
computer data acquisition on pulmonary function performed—are by
far sharper tools than the ones we have been able to use in the
past.
The fact that analytical variability has now been reduced af-
fords us the opportunity to produce unique information on the vari-
ability of baseline levels on which many pulmonary functions are
based. The problem of variability and precision of measurements
is one that you are in a very good position to attack and document.
I suggest that it would be wise to document and analyze these
baseline data formally for the rest of us so that we can have the
advantage of your experience, in addition, I think you should very
carefully establish the type of measurement protocol that you plan
to follow. If you use a protocol that is not generally accepted, we
need explanations as to why you think it is better.
The second point I should like to examine concerns the choice
of subjects. There are a number of factors that play an important
role in subject selection. When choosing subjects, one must be con-
cerned about the sources of known variability. The age of subjects
200
-------
Figure 1
Figure 2
201
-------
is one example. I am not completely convinced that it is sufficient
to examine the response of young, normal, white, male subjects
which, if one examines the literature today, represent by far the
greatest number of human exposures to pollutant gases. I think it
is very important that researchers be prepared to study a wide age
spectrum—probably as wide as from seven to seventy years.
Other factors of variability include the difference between
smokers and non-smokers; the effects of cyclic, natural, and
therapeutic hormones; and, perhaps even as has been suggested, the
influence of diet. The studies done by Jack Hackney suggest that
factors such as place of residence, and polluted or non-polluted
environment, may very well play a role in variability.* For
example, using gas for home cooking might be a very important
determinant as to how a person responds to oxides of nitrogen.
As a respiratory physiologist, I also suggest that the kind of
pulmonary function test chosen is important. I think that the
current use of the single-breath foreign gas dilution, the so-
called "closing volume test," and the flow-volume curves done with
air and helium-oxygen mixtures, could be useful adjuncts to a
protocol.
I am sure you have agonized over what sort of test would be
the most useful and sensitive. I think the evaluation of changes
in non-specific bronchial reactivity is also likely to be useful,
but again, there is a limit on what you can inflict on test sub-
jects.
My last point is in support of what Dr. Bromberg has suggested.
I, too, would like to make a plea for the study of mechanisms. I
suggest that resources such as people who are able to interpret
mechanisms would be a very useful adjunct to the operation of the
Health Effects Research Laboratory. I think it is not enough to
say that a given level of a given substance does not produce a
specific effect of a certain class. If we are to gain insight into
the lung's defenses, into other disease entities, if we are to get
more output from these kinds of studies than just that output direct-
ed towards standards settings, then we have to be in a position to
shed light on the mechanisms.
*Hackney, J.D., W.S. Linn, S.K. Karusa, R.D. Buckley, D.C. Law,
D.V. Bates, M. Hazucha, L.D. Pengelly, and F. Silverman: Effects
of ozone exposure in Canadians vs. southern Californians: Evidence
for Adaptation. Arch. Env. Health 32 (3): 110-116, 1977.
202
-------
Discussion
DR. OTTO: Obviously, the lungs are one of the primary sites
through which air pollutants are ingested, but it seems to me that
we have inordinately emphasized pulmonary function. I think the
panel should discuss other organs that might be targets of air
pollution effects.
DR. BROMBERG: It is true that carbon monoxide is the noxious
gas that is most often considered to affect other systems, and the
lung is the portal of entry for carbon monoxide. Work with other
gases has focused primarily on pulmonary effects.
On the other hand, for ozone there is evidence that red cells
are affected. In a more hypothetical vein, immunocytes present in
the bronchus, which is associated with lymphoid tissue just under-
neath the epithelium, may be affected. These immunocytes are local,
but they do communicate with the body's overall immune system.
Other gases may cause changes in capillary endothelial cell
function, which would result in more widespread changes in the
circulation. These effects are a bit more remote and probably are
not as attractive as topics for study as the pulmonary effects.
The clinical side of the coin is that people die of lung
disease or suffer from chronic pulmonary symptoms: consequently,
lung function has received more attention than other, less directly
affected sites in the body. I would like to know what other people
think.
DR. KNELSON: We should make it explicit in this discussion
that we are addressing only a few facets of a large problem. Much
of what we have talked about touches only the periphery of the air
pollution testing program currently underway at the Health Effects
Research Laboratory. In our laboratory, we invest as many resources
203
-------
in the study of in vitro cellular toxicity testing as we do in
pulmonary function testing. For several years, we have had a
relatively ambitious psychophysiology neurobehavioral program. We
collect venous blood cells from exposed subjects and study the
immune competence in these cells in vitro. We also conduct rather
aggressive metabolic tests on these subjects.
DR. HAZUCHA: I would like to comment on Dr. Otto's remark.
It is true that we are concerned with pulmonary function because
it is a primary target, and I have no doubt that different pol-
lutants will affect the systems within the body. But if we
compromise in measuring pulmonary functions, we will have to be
prepared to compromise in studying other systems.
When I measure vital capacity, I do not care if the subject
drinks a glass of water or eats a sausage in the morning. If,
however, I want to measure some level of enzymes, I certainly will
be concerned with what the subject has eaten. The more sophisticated
test I use, the more I have to be concerned with factors such as
the subject's posture or the time of day, or any of those factors
mentioned by Dr. Statland earlier.
DR. KNELSON: I would like to amplify what Dr. Hazucha is
saying. We have already designed experiments that are impossible
to conduct. The interests of the psychophysiologist, the interests
of the clinical chemist, the interests of the cardiologist and the
pulmonary physiologist are converging on a group of subjects that
allows us to write a protocol we can never implement. As a result,
we sit back and say, "Well, which variables can we afford to
measure at this stage; which ones are we going to have to put off
for the next experiment?"
DR. BENIGNUS: I have heard several of you comment that theo-
retical work, backup work on the mechanisms of the effects we
observe, is important and necessary. Because it contributes to
the body of knowledge about physiology, such work is not only
necessary from the academic point of view, but also from the point
of view of credibility.
I do not, as a scientist, have much faith in results that show
only a decrement in some kind of performance—for example, the dose
response curve. As a scientist, I would have much more faith in
results that explain to me the mechanisms that produce the effects.
Work of this type is not just a matter of contributing to the
academic pool; it is a matter of gaining credibility in the
scientific community.
204
-------
DR. HAAK: To amplify that, Dr. Pengelly mentioned that perhaps
our activities are directed toward setting standards. I believe
that when our recommendations for air quality standards, which may
be based on the 5% decrement in pulmonary function that Dr. Bromberg
mentioned, are tested in court, the judges will require some
mechanistic formula to interpret how many millions of dollars our
recommendations will cost. Scientists from industry will argue
that a 5% decrement does not warrant the expense of millions of
dollars. Other scientists will be asked to explain what impact a
5% decrement will have on public health.
As long as the mechanism is not clear, the court and the judges
will be unwilling to make a multi-million dollar decision. The
issue will be a stalemate. The scientists will be sent back to the
laboratory to do more research to support their recommendation of a
particular standard. And that research will have to be based on
the expected mechanism of the pollutant at issue.
DR. OTTO: Allow me to cite one example that puts this issue
into perspective. We have not said very much about our neuro-
behavioral research. In fact, the results of our studies, which
in some cases reveal positive findings, have had virtually no
impact on standard setting in this country, even though we have
been collecting data for five years.
One of the reasons that we have not had any impact is that
when we measure brain waves, for example, and we notice an increase
in a given component, we do not know what the underlying mechanism
is for the change in that component. Therefore, our friends in
Washington, who have to evaluate this data, say, "So what? You
showed a change in the brain wave function; I can pick up my cup
of tea and joggle it around and show changes there, too, but
what is the functional significance? Until you can tell me the
mechanism and what its functional significance is, I really do not
know how to evaluate your data." The mechanisms must be described
before we can really interpret the effects that we observe.
DR. BROMBERG: I do not know whether I know David Bates' point
of view well enough to speak for him, but my impression is that he
would say, "If you have demonstrated a reproducible effect, that
effect speaks for itself. The burden is on those who would deny
the significance of that effect to prove that it has no significance.
To those individuals, I simply say, 'Well, I saw something happen,
and if you cannot disprove that it happened, then I have proved my
case.'"
What about that point of view? Do you deal with this at all,
Dr. Frank, in thinking about what it is you want to measure?
205
-------
DR. FRANK: Yes, I do. I have stated that dose response
relationships have considerable impact in this field. Let's face
it—they do. Ultimately one of the issues that we should address—
and this is kind of circular—is that basic research, dose response
relationships, animal toxicology and epidemiology all feed into
one another. To the extent that any one of these elements is
isolated, you can forget about it, if it cannot be drawn into the
entire picture. If animal toxicology and clinical research do not
inform and influence the design of epidemiologic studies, then we
are spinning wheels. I think we all agree on that.
My bias is toward studies of mechanisms. This is what fasci-
nates me. For example, we are in the process of trying to find a
basis for what may be the vulnerability of certain individuals in
the population to sulfuric acid. There is a good analytical
chemical basis for this notion, and I think it could be tested and
that such a test would not only be informative, but also great
fun.
As scientists we ought to anticipate being confronted with
the problem of how to relate our observations to some remote
entity, such as lung disease, or even "public health." We should
be prepared to say, "I have made an observation and I do not
understand it, but I think it has the following implications."
Then we should begin to test the hunch.
We should not say, "Here's my observation; it is an effect."
Implicit in that statement is the notion that all effects are bad.
I do not think they are. I have seen some data on dose response
relationships to ozone obtained from an extremely good laboratory.
The data were presented as evidence that there was an adverse
effect, beginning at about 0.3 ppm of ozone. My reaction to the
data was that it was reassuring that there was virtually no adverse
effect in the subject when he was breathing at rest. In my judg-
ment, the deviation from zero was almost trivial; it was within
the variance of the measurement in the population.
DR. BRQMBERG: It is hard to know what to do with small
changes. One man's artifact is another man's Nobel Prize. A small
change may be a very significant piece of information if we only
knew how to isolate it, to focus on it, and to amplify it and
understand it.
DR. FRANK: I couldn't agree more.
DR. BROMBERG: On the other hand, a small change may really
be nothing.
206
-------
DR. FRANK: Again, I couldn't agree more. What is an ap-
parent certitude today may change radically in its implications
tomorrow. We do not have final truth on anything, and there is
always the danger of over-interpreting data. But we do end up
with a lot of curiosities, and it seems to me that they get in our
way. The curiosities exist, but we avoid exercising judgment
about them.
It would be ideal if we could eliminate all environmental
stress, but we are not about to do that. What we have to try to
do is exercise a certain amount of judgment about how much is
. tolerable.
DR. HORVATH: I would like to make one facetious comment and
one serious comment. The facetious comment has to do with the
"beneficial effects" of ozone exposure. My reaction to that is as
follows: one of the things that we discovered was that the maximum
aerobic capacity—in other words, the capacity of a man to work at
a sustained level—seems to be reduced because of ozone exposure.
I say that may be beneficial because it may prevent all of us from
becoming over-worked; so, if we are exposed to ozone, we may
actually be benefitted by it.
My serious comment concerns Dr. Pengelly's important sug-
gestion; namely, that we are ignoring the effects of age. The
only studies I know of that contain any data on age were the
studies we did with carbon monoxide. We looked at young Caucasian
males, middle-aged Caucasian males, and we looked at smokers and
non-smokers, depending on whether they had smoked at least one
pack of cigarettes a day for a couple of years or for 16 to 17
years. We noticed differences among these groups.
But that is the only study on age groups that I know of. We
have ignored the effects of age for all the other pollutants, and
I feel strongly that the age factor must be taken into account
because we do know there are obvious changes that occur in indivi-
duals as a consequence of age and as a consequence of sex.
Age and sex have to be involved in any consideration we give
to what happens to the population. After all, 10.9% of the
population in the United States is over the age of 60. That is a
huge percentage of our population. The point that Dr. Pengelly
did not bring up, however, which annoyed me considerably, especial-
ly since in my laboratory we have some very strong people who are
of the opposite sex, was the influence of sex hormones. He
hinted a little bit, but I think we have to consider the striking
differences between males and females. We are completely ignoring
53% of our population. Every time I ask someone if they support
207
-------
studies on females, they look at me and say, "Well, females are
females." They are, indeed, and from the standpoint of being good
subjects, they are much more attractive than males.
We are also ignoring the elderly. When you talk about pol-
lutant effects on humans, what groups are concerned? Who dies?
It is not the young Caucasian male and female. It is the old
Caucasian male and female, and we have paid no attention to them.
This is one of the real difficulties that I find with all of the
experimental work that is being conducted on air pollutants. We
will have wasted a great deal of money if someone does not start
to support studies involving females and the elderly. The cost
benefit problem is of no importance if you relate it only to young
Caucasian males.
The factor of race is another problem that we have ignored.
Studies suggest that the responses of other races to environmental
stresses are different from the responses of Caucasians. We must
seriously consider the factors of age, sex, and race. We must
move quickly to incorporate people from these groups into our
studies, even though it is not easy to recruit subjects from the
population at large.
fly feeling is that no human organism can survive if he isn't
under some environmental stress, and I do not think we can remove
stresses from the organism. I am not sure that we can benefit by
studying ozone or dioxides as stresses, but I do think that we
have to have a certain degree of stress.
Before we decide whether or not these minor changes that we
have been recording are bad for us, I think we must ask ourselves,
"How much do we need of this stress in order to survive?"
208
-------
Statement
Mario C. Battigelli, M.D.
School of Medicine
University of North Carolina
I shall briefly attempt to offer some encouragement to consider
human experimentation as a legally/ socially, academically, and
medically acceptable endeavor.
Experiment is such a common event in life--certainly it is
common in many professional habits and commitments—that maybe
something is lost by using the term. Some of you may remember
the lovely poem by Emily Dickinson that ends something like this:
"Experiment to me is everyone I meet."
Some of you may remember the well-known British cardiologist
who stated that any time a physician embarks on a treatment of a
patient, he is conducting a clinical experiment. Justice Holmes
remarked that "All life is an experiment; the best test of truth
is the power of the thought to get itself accepted in the competi-
tion of the market."
"A test, a trial, a tentative procedure" is the dictionary
definition of experiment. The thought process that we have accept-
ted in embarking on experimental work will be exemplified by
discussing our experience in byssinosis.* I shall from the outset
*Byssinosis is a form of pneumoconiosis due to the inhalation of
cotton dust.
209
-------
define that we are not testing the entire pathogenesis of bys-
sinosis, but only a very small aspect of it--the acute ventilatory
response to calibrated clouds of cotton dust.
Whenever we deal with cotton dust, we face a complex, hetero-
geneous mixture of particles that may be derived from contaminat-
ing dirt or from various parts of the cotton plant—the fibers,
the shell or pericarp, the bracts (small leaf-life appendages at
the base), the stem, and the leaves.
Exposure to cotton dust is associated with the development of
a syndrome we call byssinosis. This syndrome is an airways response,
a sort of asthma, which is associated with well defined symptoms
that have a characteristic cyclic expression, at least at the be-
ginning of the natural history of the disease.
The subjective syndrome is part of a phenomenon that may be
objectively identified as a respiratory decrement, a respiratory
loss following exposure to dust, well described by the spirogram.
We also recognize that not all of the individuals exposed to
cotton dust come down with a significant response, and that the
proportion of those affected increases in proportion to the in-
tensity of the exposure. It has also been ascertained that indi-
viduals not used to the occupational exposure may, on exposure to
sustained concentration of dust, exhibit the symptoms, as well as
the signs of the cotton dust effect.
In any area of knowledge, there are certain aspects that are
out of focus, incomplete, erased, or totally missing. One of the
aspects that we thought interesting was to verify how specific was
the acute ventilatory response related to other conditions that we
distinguish from byssinosis. What is the dose-relationship of the
specific response, and the specific agents causing these responses
in patients, volunteering as subjects, or in normal healthy indivi-
duals who are free of disease and unexposed.
The primary justification for collecting information on dose-
response relationships is that such information is helpful in
formulating appropriate medical management procedures, including
removal from exposure, reassignment, relocation, retirement.
Hence, experimental exposure may assist in defining guidelines for
this type of decision. There is more to it.
We considered several questions as clinically significant to
our study. First, is there a minimal concentration and/or duration
of exposure that is effective in identifying the specific, segre-
gated ventilatory response, whatever the relationship of this may
210
-------
be to the whole syndrome? Second, are there any specific character-
istics of reactivity (such as symptoms or objective ventilatory
parameters) that can be defined and perhaps related to the patho-
logical background of each subject? Third, can these definitions
of reactivity characteristics be used to formulate pathogenetic
considerations?
The fourth question is, are there specific components or
characteristics of the exposure, the nature of the dust, that can
be identified as most injurious and therefore necessary to the
understanding of the natural history and the pathogenesis of the
disease?
In designing and executing the experiments, considerations of
safety are quite straightforward. Our continuous monitoring pro-
gram is maintained by assaying dust level continuously and by
having a physician exposed to the same conditions as the subjects,
in the exposure room.
The nature of the response, at this concentration, is limited
and invariably reversible. We must recognize that the true experi-
mental value to these activities is simply that we duplicate what
happens in the real world within total uniformity and continuous
control.
In doing so, we find justification in trying to remove the bad
connotation from the word "experiment." This term conjures visions
of vivisection, which, of course, excites the minds of those who
tend to see greater risks than actually are warranted in these
circumstances.
I would like to pass on this word of encouragement, quoting
from no one less than Jacob Bronowski, who said, "We live surrounded
by the apparatus of science, the diesel engine, the experiment,
the bottle of aspirins, and the survey of opinion. We are hardly
conscious of that, but behind that, we are becoming conscious of a
new importance in science. We are coming to understand that science
is not a haphazard collection of manufacturing techniques, carried
out by laboratory dwellers with acid-yellow fingers and steel-rimmed
spectacles and no home life.
Science, we are growing aware, is a method and a force of its
own, which has its own means and style, and its own sense of
excitement."2
One may wish not to call an experiment an experiment; let us
call it a justified verification of an hypothesis, but the fact is
that that type of.endeavor is here to stay. We are bound to conduct
211
-------
experiments, and perhaps a safe beginning is with a restatement,
according to what Socrates suggested, "The beginning of wisdom
is the definition of your own terms."3 I suggest this for your
own activities.
REFERENCES
1. Holmes, Jr., O.W.: Abrams v. United States, 250 U.S., 616.
2. Bronowski, J: Common Sense and Science, Heinemann (ed.), 1951,
3. Vitae Philosophorum. Cited by Diogenes Laertius (Long, H.S.
ed.). Oxford Univ. Press, 1964.
212
-------
Discussion
DR. BROMBERG: Yesterday Dr. Bates said that he thinks certain
experiments are best conducted in the environment where the problem
actually exists. If, for example, you want to study the long-term
effects of sulfur dioxide, perhaps the best place to do so is in
the work setting of lead smelters, and not attempt to study the
long-term exposures in a chamber.
Perhaps the smaller, better controlled studies that we do in
the laboratory should be viewed in this context. As long as we do
not exceed the exposure levels that are occurring in, and are accept-
ed by, our industrial society, perhaps we should consider the real
environment as a standard of reference for what we should permit
ourselves to do or not to do.
DR. CRIDER: If you use the workers in a particular industry
as your subjects, you may be omitting a large percentage of people
who may be susceptible to a particular agent. Individuals who
can survive potentially harmful environments tend to continue to
work in them; those who cannot, do not. I think we would have to
be careful in conducting the type of studies that you suggest, and
in selecting subjects for them.
DR. BROMBERG: I am not saying that one has to go into the field
and find individuals and study their responses to controlled lab-
oratory exposures. I am just saying that our society has indicated,
for the moment, its willingness to tolerate certain levels of
human exposure. By using society's willingness to tolerate exposure
as a standard of reference, perhaps the much more careful approach
we use in the laboratory is really quite defensible. There may be
another perspective from which to view our experiments other than
the almost frightening portrayal suggested by moralists, chairmen
of human use committees, and doctors of jurisprudence.
213
-------
Dr. Battigelli did not really discuss in detail any of the
work that he has done, but many of his subjects are not people who
are occupationally exposed to cotton dust, or who have any history
of byssinosis. They are people who have well-defined, chronic ob-
structive lung diseases of more common varieties, but who are
nonetheless willing to be exposed to cotton dust. They are ex-
posed along with the experimenter himself, who sits in the chamber
with the subjects and is as close as the touch of a hand.
Very interesting information has been obtained from Dr.
Battigelli's studies, and I think in our studies of pollutant
gases, for example, that we ought to pay more attention to the
possibility of using that difficult-to-control, more worrisome
kind of subject, the individual with well-established chronic
disease.
DR. BATTIGELLI: Because you conduct experiments within a
controlled laboratory setting, you can easily defend your activi-
ties by demonstrating that the subjects you expose are immensely
safer than other subjects who are exposed to the same variable in
a less controlled situation.
With regard to the variety and characteristics of exposure
subjects, one should include the ill and impaired. There is no
question that the impaired individual is quite often, if not
always, a very sensitive responder. I am sure that your protocols
have included exposure of individuals with significant disorders
because such people are, perhaps, more sensitive to the various
factors you wish to study.
DR. PENGELLY: Earlier, someone asked about the meaning of a
5% decrement in vital capacity. Vital capacity is a measurement
of pulmonary function, and it is well known that it decreases with
age. After growth stops, vital capacity drops approximately 0.5%
per year. If one's vital capacity declines 5% over a two-hour
period, that is comparable to ten years of aging. Even though the
vital capacity may return to its original level, I think this
suggests that age is an important factor to consider when we
decide what we should measure and how people respond.
It is well known, for example, that cigarette smoking produces
a vital capacity decrement that is greater than the normal decrement
associated with age. If exposure to some toxic material causes a
decrement in excess of the known decrement, I think that is a suf-
ficient argument to say that although this person may still have,
functionally, enough vital capacity to get on with life, the fact
that he will deteriorate at a more rapid rate than he would other-
wise is a significant point*
214
-------
Knowing the effects of cigarette smoking on pulmonary function,
knowing that the cigarette causes 60% of all cancer mortality, how
can we justify the question of sensitivity to air pollutants in
the face of such an enormous known stress to the lungs?
DR. FRANK: I would like to respond to this, not with respect
to cigarette smoking, but with respect to vital capacity. We
could be dealing with substances X, Y, and Z—call them pollutants
for the moment—and we could administer each of them to a care-
fully controlled group in which we know the subjects' previous
diets and all the factors that may influence their functional
response.
To measure vital capacity, we expose the subjects to pol-
lutants X, Y, and Z for 10 minutes. Each person shows a 5% re-
duction. At this stage, there is no way of weighing the im-
portance of the changes in vital capacity. We are not yet in a
position to judge the relative significance of the reductions in
vital capacity. So, we expand the experiment; we may also use
information obtained from the subjects while seeking answers to
other experimental questions. Which of these pollutants is the
most important? Are they all equally important, or are none of
them important? We may find that if the subject is exposed to
agent A for ten minutes on each of three days there will be a
constinuous drop in vital capacity; but for agents B and C the
subject will revert to control with repeated exposure. Perhaps
now we are developing a reasonable hypothesis to suggest that one
of these agents is more significant than another.
At the same time, some other researcher develops some epidemio-
logical data that either reinforces or weakens the hypothesis.
What I am saying is that it is incumbent on us, as scientists/ to
continue to test the implications for health, if you will, of the
observations we make. We cannot just say that, because an effect
has been noted, we have to accept the effect as significant for
health.
DR. BROMBERG: I think Dr. Pengelly uses a very convenient
calibration scale. He uses those very shallow slopes of function
versus age, and extrapolates the data he gets in terms of life
expectancy, and says, "Look! This curve crosses the time axis at
100 years, and now you have put this subject, by your little
exposure, onto a different curve, which if extrapolated, gives him
a life expectancy of only 70 years. You have reduced his life
expectancy by 30 years." Is that what you intended to say. Dr.
Pengelly?
215
-------
DR. PENGELLY: No, of course not. What I said was that one
has to examine the entire context of the situation, which rein-
forces what Bob Frank just said. The one little bit of infor-
mation we get in the laboratory has to be fitted into the other
pieces of known information—perhaps in part from animal experiment-
ation and in part from epidemiological studies. The point of view
one must have when looking at the response in the lab should be
conditioned by the other information at hand.
I certainly did not want to imply that a 10-minute exposure
to ozone or some other agent, was going to have some effect on the
subject. But, in order to understand what is taking place, one's
observations must be made within the context of all the other
observations that relate to that variable.
DR. BROMBERG: I think most people would agree with that
approach. That does not provide an easy answer to the question of
what to do with that 5% decrement. When you were talking, I
thought you were saying that the 5% decrement in vital capacity
over two hours is the same decrement you would observe after 10
years of aging, and that you can turn around and say to the inform-
ed layman, "For you to understand what that decrease in vital
capacity means, let me tell you that it is equivalent to 10 years
of aging." That is very dramatic, but is it'right?
DR. PENGELLY: No, although it does give the layman, in a
sense, a framework in which to judge things, but it is not to be
interpreted as a value judgment for what happens. For example, if
one routinely measured one's weight and noted a weight change of a
few pounds over several weeks, such a change is probably not
significant. But a consistent change of a few pounds over several
days is well known to be significant.
One must keep the scales of age in balance with other infor-
mation. There are a lot of implications that I have not mentioned;
many hidden messages in saying that a change of weight over a few
days is important. Many unsaid things come into play because of
one's other experience. I think it is important to keep the
context in balance.
DR. BATTIGELLI: To extend again what I believe Dr. Frank
implied, perhaps a short-term experiment may offer us guidelines
and help our progress on long-range experiments. Probably most
chronic experiments can be studied better from an epidemiological
approach because the study of so-called natural experiments cannot
be duplicated in the laboratory by any stretch of the scientific
imagination. The interplay between the short-term, bench-type
laboratory test and the epidemiological investigation may provide
216
-------
mutual support for both types of experiments and may sharpen
attention on important target areas*
DR. BROMBERG: So the lesson, then, for a specially-designed
facility like the EPA Human Research Facility here on the campus
of the University of North Carolina, is that isolation should be
guarded against; the design of the experiment should take into
account a wide range of information; and interpretations should be
based on an attempt to integrate observations into a coherent
picture. The development of this picture is based on animal
experiments, on short-term human experiments, on medium-term human
experiments, and on epidemiological studies.
I am afraid, though that there will be great difficulties in
interpreting the state of the art for a specific regulation that
Congress might want to consider. But we must do our best with
that problem, and not cave in as scientists to that demand.
DR. HAZUCHA: There are some other implications of acute
experiments. It is well known that only ideal systems, when they
are stressed, will return to pre-stress conditions. A living
organism, to which stress is applied, positively will not revert
to pre-stress conditions.
In an acute experiment, the difference between pre-stress and
stress conditions may be miniscule. With regard to any of these
tests that show a 0.10% decrement, perhaps you cannot state cate-
gorically that, over a 70-year life span, there will be a reduction
of 7% or seven years of life. But when a subject is exposed many
times for many years, the resulting loss of vital capacity will
certainly impair the quality of life.
DR. BROMBERG: That all depends on our accepting your origin-
al premise, which was that if every event has a permanent effect
on the subject, the subject never returns to the pre-stress con-
dition. I am not quite prepared to believe that. For example, we
are losing 1% of our red cells every single day. They are dis-
appearing, being eaten up by our reticuloendothelial system. Yet,
our red cell mass, over long periods of time, remains quite constant;
so it is possible for steady-state mechanisms to maintain constant
levels. I think it is possible to undergo some stress and to
completely recover from it, and to do so again and again.
I am not sure that either one of us can prove our points, but
I think that the assumption you are making, which if we do accept
it, would lead us to say, "Well, yes, we have to admit then, that
if the same stress is repeated sufficiently often, there will be
irreversible, gross, and progressive changes." I am not quite
prepared to believe that.
217
-------
DR. HAZUCHA. It is true that red blood cells are renewed,
but the hemopoietic system is stressed, and the hemopoietic system
as such will not return to the exact pre-stress condition.
DR. BROMBERG: I am not sure about that.
DR. HAZUCHA: Well, as you said, it is difficult to prove,
almost impossible.
DR. BROMBERG: I think you are addressing yourself, in a
sense, to your own point of view about the significance of small
changes. We have been talking around it, and we have arrived at
different ways of looking at the significance of small changes. I
think you have developed a way of looking at this which is satis-
factory to you, but it still remains a touchy problem. When con-
fronted with a specific circumstance, I suspect various people in
this room would come up with a different interpretation of what
small changes mean, even after all this discussion.
DR. PENGELLY: I should like to return to Dr. Frank's comments
about the importance of observed effects. I am reminded of a
discussion I had with a person who supported the development of an
airport quite close to a major city. He said that, as he under-
stood it, people living close to airports adapt to airport noise
and eventually become accustomed to it. My response to him was
yes, there is an adaptive mechanism—it is called deafness, but I
do not think deafness is an appropriate adaptation.
Recent studies suggest that people who are not constantly ex-
posed to ozone tend to react more to the gas than people who are
regularly exposed to it. The implication of this argument is that
individuals exposed to ozone are better off because they do not
respond. I argue that the method of adaptation must be con-
sidered. Deafness is an adaptation to airport noise, but deafness
certainly is not desirable. We must take the factor of adaptation
into account; we must make sure it is appropriate.
DR. FRANK: At the risk of being boring or repetitious, I
think the new legislation in TOSKA attempts to weigh a risk
because weighing a risk is important in deciding what to do about
it. I think it is reasonable, too, to say that our judgment of
the importance of a risk may be subject to change. We deal always
with uncertainty, and what may appear to be, from my point of view,
flimsy evidence today, may be quite impressive, overwhelming
evidence, if not proof, a month from now.
Consequently, we have to be prepared to change our judgment.
But I think, too, that as a society, we are compelled to make
218
-------
judgments because there is no such thing as a benign environment.
We have to weigh what it is that we recognize as being most import-
ant/ and how much of an effort we should make to mitigate harmful
stress.
I do not see anything wrong with making judgments. I do not
think we should just make observations and put them forth so that
finally they become curios. I think if we have any reason to
suspect that an observation has an implication, then we, as scient-
ists should pursue our observation and devise experiments designed
to test its importance for health.
Another thing that really impresses me is that the biological
system is marvelously well constructed to adjust to change. I
recognize there is a "price" for all adjustments, but we do manage
to adjust.
A hypothesis about to be tested in animals suggests that
total dosage may not be important for many of the pollutants we
are concerned about. (I am not referring to mercury or lead.
They are stored in the body and obviously have cumulative effects.)
What is important, however, is the variation in dosage. If an
animal is exposed to sulfur dioxide for "X" number of years, the
exposure will have no effect if the dosage is consistently low.
On the other hand, there will be measurable effects resulting from
the same total dosage over the same period of time if the dosage
of sulfur dioxide is varied each time the gas is administered* If
this hypothesis proves true, it may reveal impressive epidemio-
logical evidence that we can apply to our studies of morbidity and
mortality rates.
If humans are exposed to sulfur dioxide more effects will be
noticed with a dosage of 10 ppm for one minute than with 1 ppm for
10 minutes. And I suspect you could develop an argument for
epidemiologic data to reinforce that as well.
DR. KNELSON: Our two ozone studies demonstrated the importance
of variation in dosage very nicely. In both studies, using 0.4 ppm
for four hours (which could be expressed as 1.6 ppm of total dose)
the effect on lung mechanics was considerably less dramatic than the
effect of 0.6 ppm given for one hour, which is less than half of
the total dose in the 4 hour exposure to 0.4 ppm ozone.
Our laboratory has also conducted dose-rate experiments, using
death in animals as an end point, and we have obtained elegant data
showing that ozone does exactly what you have described. These
experiments show a continuum of dose-rate response, which is the
rate a given dose is administered. We have quantified the dose-
rate response, and we have found that short exposures to higher
219
-------
levels of ozone are much more detrimental than long exposures.
That does not surprise you, does it, Dr. Bromberg?
DR. BROMBERG: No, but when you are dealing with a potentially
noxious agent and with rate-limited disposal mechanisms, and if
you increase the rate of delivery to some critical point, you will
eventually see a very marked effect. A case in point is the agent
potassium chloride. If you make a mistake by injecting 40 milli-
equivalents of potassium chloride into a patient's vein, in a
few minutes the patient dies from cardiac arrest because the
circulatory mechanisms that dispose of noxious agents are too slow
to cope with that much potassium chloride.
It is not a surprise that if we push hard enough with a concen-
trated enough dose in a short period of time, we can overcome the
effectiveness of the body's disposal mechanisms.
DR. KNELSON: But that is not a consistent tenet in toxicology.
You must pay attention to what you are examining. We both know the
dose-rate response is significant for toxic substances that the
body's repair mechanisms cannot handle.
On the other hand, there are toxic substances that elicit a
compensatory response. If you administer a low dose of these sub-
stances for a long period of time, the compensatory response will
not be invoked, and the overall toxic effect will be greater than
if you give the same dose over a short period of time and trigger
the compensatory mechanisms.
DR. HAAK: I would just like to pose a final question to this
group. If we do not completely understand the mechanism of action,
how are we going to get enough concurrence among scientists, our-
selves for example, to make a scientifically sound argument before
a court to justify the large costs incurred by setting a standard
for a given pollutant?
The safe limits for ozone, for example, have been exceeded
nearly everywhere. It is going to be a great expense to apply air
quality standards to ozone. How are standards for ozone to be
defended? Even if we could meet in court, EPA on one side and
industry on the other, society will be left without a conclusion.
DR. BROMBERG: We have heard that there is no easy formula.
It is clear that various people at this meeting have developed a
personal approach to this difficult interface between science and
society. They do not pretend that they have the right answer, but
at least they have the satisfaction of knowing that they are
involved in the issue and have some kind of approach to dealing
220
-------
with it. We do not know the answers to most of the questions that
are being posed, and I do not think we are going to know the
answers in the very near future. All of us must live with that as
best we can.
221
-------
Appendix
Program Participants
George Armstrong, M.D.
Director, Health Effects
Division
Office of Health and Ecological
Effects
U.S. Environmental Protection
Agency
Office of Research and
Development
401 M Street, S.W.
Washington, D.C. 20460
David Bates, M.D.
University of British Columbia
Vancouver, British Columbia,
Canada
Mario C. Battigelli, M.D.
Professor of Medicine
Division of Pulmonary Medicine
School of Medicine
University of North Carolina
Chapel Hill, North Carolina 27514
Edward Bishop, M.D.
Professor of Obstetrics and
Gynecology
Chairman, Committee on the
Protection of Rights of
Human Subjects
School of Medicine
University of North Carolina
North Carolina Memorial Hospital
Chapel Hill, North Carolina 27514
Philip Bromberg, M.D.
School of Medicine
University of North Carolina
724 Clinical Sciences Building
(229H)
Chapel Hill, North Carolina 27514
Charles E. Daye, J.D.
Associate Professor
University of North Carolina
School of Law
Chapel Hill, North Carolina 27514
Robert Frank, M.D.
Department of Environmental Health
University of Washington, SC-34
Seattle, Washington 98195
Milan Hazucha, M.D.
Research Scientist
Clinical Studies Division
Health Effects Research Laboratory
U.S. Environmental Protection
Agency
Environmental Research Center,
MD-73
Research Triangle Park,
North Carolina 27711
Steven Horvath, Ph.D.
Director and Professor
Institute of Environmental Stress
University of California
Santa Barbara, California 93106
223
-------
John H. Knelson, M.D.
Director
Health Effects Research Laboratory
(MD-73)
U.S. Environmental Protection
Agency
Research Triangle Park,
North Carolina 27711
G. Guy Knickerbocker, Ph.D.
Emergency Care Research Institute
5200 Butler Pike
Plymouth Meeting, Pennsylvania
19462
Morton Lippman, Ph.D.
Department of Environmental
Medicine
New York University Medical
Center
550 First Avenue
New York, New York 10016
Michael V. Mclntyre, J.D.
P.O. Box 801
Santa Monica, California
90406
David Otto, Ph.D.
Research Psychologist
Clinical Studies Division
Health Effects Research
Laboratory {MD-73)
U.S. Environmental Protection
Agency
Research Triangle Park,
North Carolina 27711
L. David Pengelly, Ph.D.
Associate Professor
Department of Medicine
McMaster University
1200 Main Street West
Hamilton, Ontario, Canada
Russell Pimmel, Ph.D.
Associate Professor
Department of Medicine
University of North Carolina
Chapel Hill, North Carolina 27514
Carl Shy, M.D.
Professor of Epidemiology
Department of Epidemiology
University of North Carolina
Chapel Hill, North Carolina 27514
Harmon L. Smith, Ph.D.
Divinity School
Duke University
Durham, North Carolina 27706
Ralph Stacy, Ph.D.
Research Scientist
Clinical Studies Division
Health Effects Research Laboratory
(MD-73)
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina
27711
Frank Starmer, Ph.D.
Associate Professor of Computer
Science and Associate Professor
of Medicine
P.O. Box 3181
Duke University Medical Center
Durham, North Carolina 27710
Bernard E. Statland, M.D., Ph.D.
Associate Professor
Department of Pathology
University of North Carolina
North Carolina Memorial Hospital
Chapel Hill, North Carolina 27514
Thomas Wagner, Ph.D.
Acting Associate Director
Clinical Studies Division
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Environmental Research Center, MD-73
Research Triangle Park,
North Carolina 27711
224
-------
TECHNICAL REPORT DATA
(Please read Inunctions on the reverse before completing/
1. REPOHT NO.
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
METHODOLOGIES AND PROTOCOLS IN CLINICAL RESEARCH:
Evaluating Environmental Effects on Man -
Proceedings of a Symposium ...
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
fl. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Clinical Studies Division
Health Effects Research Laboratory
Office of Research and Development
Research Triangle Park, N.C. 27711
10. PROGRAM ELEMENT NO.
1AA601
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Health Effects Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
-Research Triangle P^r-Tc. w.r. 27711
13. TYPE OF REPORT AND PERIOD COVERED
RTP,NC
14. SPONSORING AGENCY CODE
EPA 600/11
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The report is a proceedings of a symposium convened at Chapel Hill, North
Carolina, October 26-28, 1977. Five major topics pertaining to the use of_humans
as experimental subjects are addressed in this volume: philosophy of clinical
research, environmental and physical safety considerations in human exposure
facilities, the EPA human studies programs, and special considerations and approaches
in environmental clinical research. The first four parts consist of twelve formal
papers covering issues such as ethical and legal considerations surrounding the
use of human subjects, environmental controls and safeguards, and electrical
surveillance and integrity. Following each paper is a summary of the discussion
that took place after the paper was presented to the symposium participants. Part
five is a panel discussion composed of four brief presentations and an exchange of
comments among panel participants.
The purpose of these proceedings is to help identify, through open discussion,
the problems connected with using human subjects in clinical research.
7.
KEY WORDS AND DOCUMENT ANALYSIS
'SCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Gioup
laboratories
Laboratory tests
humans
06 F, L
3. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
20, SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
------- |