Do not WEED. This document
should be retained in the EPA
Region 5 Library Collection
y letuon.
FEBRUARY 1983
THE EPIDEMIOLOGY OF MOBILE HOME
FORMALDEHYDE VAPOR CONCENTRATION
AND RESIDENTS' HEALTH STATUS
By:
Henry A» Anderson, MoD,, Project Director
Kay Ao Dally, M.So, Industrial Hygienist
Lawrence P0 Hanrahan, M.S., Epidemiologist
Wisconsin Division of Health
Madison, Wisconsin
Ann D. Eckmann, B.S., Chemist
Wisconsin State Laboratory of Hygiene
Madison, Wisconsin
Marty S. Kanarek, Ph.D.
John Rankin, M.D.
University of Wisconsin Department of Preventive Medicine
Madison, Wisconsin
GRANT NO. CS806856-01 ^ $ Environmental Protection Agency
tegion 5, Library (PL-12J)
77 West Jackson Boulevard, 12tn rioor
Chicago, II 60604*3590
Project Officers:
Karl E. Bremer
Carolyn S. Hesse
Region V
U.S. Environmental Protection Agency
Chicago, Illinois
This study was conducted in cooperation with:
U.S. Environmental Protection Agency
Washington, D.C. 20460
-------
DISCLAIMER
This document has been reviewed and approved for publication
by the U.S. Environmental Protection Agency. Approval does
not signify that the contents necessarily reflect the views
and policies of the UoS« Environmental Protection Agency, nor
does mention of trade names or commercial products constitute
endorsement or recommendation for use.
-------
iii.
TABLE OF CONTENTS
Page
Executive Summary iv.
Assessment of Cooperative Agreement. .. 1
Objectives of Study 2
Development of Study Protocol 2
Summary of Implemented Epidemiologic Protocol 4
Results of Epidemiologic Survey 8
A. Quantification of factors influencing
formaldehyde concentration . 9
B. Health Status Assessment 10
C. Adult Survey Participants 15
Summary of Laboratory Activities ..... 16
Related Projects 20
-------
iv.
EXECUTIVE SUMMARY
Under the TOSCA Section 28 cooperative agreement with Wisconsn, three
epidemiologic sampling methodologies available to the Division of Health were
evaluated to select the most feasible method of enrolling mobile home
residents for studies of formaldehyde (HCHO) concentration measurements and
construction, environmental factors and health status. Utilizing property tax
records, an age (home) stratified, random sample cohort of 100 homes less than
3 years old and 37 homes of more than 3 years of age (mean 67 months) were
enrolled and followed prospectively at monthly intervals. A total of 976
monthly home visits were made (77% of homes had 6 or more monthly visits).
A total of 1874 air samples, 730 blanks and 573 impinger blanks were analyzed.
Formaldehyde air measurements ranged from below detection (0.1 ppm) to 2.84
ppm. The average monthly concentration from homes less than three years old
was 0.54 ppm and 0.19 ppm from homes older than three years. Regression
modeling found that the log of home age was the best environmental parameter
for predicting formaldehyde levels.
A total of 2,423 monthly self-administered symptom questionnaires were
completed by the 288 persons in the homes which remained under study for 6 or
more months. Upper respiratory symptoms were the most common (centering
around 40%) and the mucous membrane symptoms clustered about the 25%
frequency. Stepwise logistic regression analysis was performed. A positive
dependency of both burning eyes and watery eye symptom logits on the log
formaldehyde exposure parameter was demonstrated. A positive relationship
with respondents' age was also seen. Dry/sore throat, swollen glands and
diarrhea logits exhibited a lesser dependency on formaldehyde exposure.
Cough, phlegm and headache were related to cigarette smoking but not to
formaldehyde.
Fifty-three adults participated in clinical examinations. Evidence of ocular
or nasal irritation was more prevalent among those from households with the
higher formaldehyde levels (5 of 9 above 0.8 ppm) than those from the lower
level homes (10% among those less than 0.4 ppm). Pulmonary function
abnormalities were only associated with cigarette smoking.
Laboratory evaluation of analytic methodologies were performed and the NIOSH
P & CAM 125 method with two modifications utilized (sodium bisulfite as
absorbing reagent instead of distilled water and the NIOSH sodium sulfite
titration instead of the iodine method for standardization). Precision and
accuracy determinations for the method were performed. The coefficients of
variation for the sampling and analytical procedures was calculated at 32%
at 0.03 ppm to 5% at 0.4 ppm. Because of the high relative error at the low
concentrations (less than 0.1 ppm) 0.1 ppm was considered as the detection
limit for the method as used during the project.
-------
I. Assessment of Cooperative Agreement (TSCA Section 28) Overall Program
Objectives
Although Wisconsin's original toxic substances cooperative agreement
proposal included a broad range of timely topics of importance to both
the state and the EPA, the final agreement focused priority upon
assessing the potential health risks from exposure to indoor
contaminants, most specifically formaldehyde. The agreement assisted
Wisconsin to support expert professionals, and provided the necessary
technical equipment to scientifically evaluate and characterize the
health related questions being raised by state residents, legislators
and the medical community. The agreement also helped the state attract
other research scientists to join the core staff established by the
agreement. Without the assistance of this agreement the state would
have been seriously handicapped in responding appropriately to the
complexities of the formaldehyde issue. The results of the agreement
emphasize the importance of states being able to dedicate two or three
individuals to address specific problem areas of mutual importance to
states and the federal government. It is encouraging that a program of
modest size was able to have not only a local, but also a national
impact. The expertise gained by the program staff was sought and
utilized by more than 20 states faced with similar needs but who were
unable to mobilize a similar program. Preliminary results have also
been utilized by HUD and CPSC. Interstate cooperation and
collaboration eliminated some redundancy in state activities and
avoided the necessity of investing scarce dollars and time in
"reinventing the wheel."
Perhaps the best measure of the utility and success of the cooperative
agreement is the fact that all staff hired under the agreement have
been moved into corresponding state funded permanent positions created
as a result of the success of the project and the state Legislature's
recognition of the need for continuing such activities. In the current
state budgetary period, these were among the only "growth" positions
in the Division of Health. Thus the programmatic concepts in the
agreement will continue well beyond the expiration of the contract and
its funding. This would not have been possible without the "seed"
money and direction of the TSCA Section 28 cooperative agreement.
Throughout the agreement period EPA Region 5 has frequently called upon
the state's Section of Environmental and Chronic Disease Epidemiology
which was founded on the Section 28 agreement to assist in the
resolution of other issues brought to EPA's attention but also of
importance to Wisconsin. We expect this cooperative professional
interaction to continue.
-------
II. Objectives of Study
It was agreed that the specific project conducted under the cooperative
agreement would have the following objectives.
1. To evaluate the feasibility of different epidemiologic
methodologies; to select and implement the most
appropriate study design.
A. Quantify the concentration of formaldehyde in
the air of representative residential mobile homes
in Wisconsin.
B. Identify and investigate possible household
construction and environmental factors which might
influence the concentration of formaldehyde found in
mobile homes.
C. Assess the current and past health status of mobile
home residents and investigate possible associations
of health effects and formaldehyde concentrations.
2. Evaluate different laboratory methodologies for
collection and analysis of formaldehyde; establish
laboratory and field quality control procedures to assess
accuracy and reproducibility of study formaldehyde air
samples; establish inter-laboratory quality control
procedures.
III. Development of Study Protocol
During the first year of the agreement several different
epidemiologic methodologies were evaluated before finalizing the
main project protocol. Each design considered had different
scientific advantages but each had implementation drawbacks which
had to be evaluated. The final goal was to enroll a study
population identified in a random manner and which would have a
sufficient range of formaldehyde exposure levels to facilitate
investigation of associations of health status and exposure
levels.
The first methodology evaluated was a naturalistic probability
sample of all mobile home parks and home sites in the state
(sampling framework outlined in Table 1). A pilot sample was drawn
and investigated to test the feasibility of this methodology.
Prior to beginning, milestones were established for rejecting the
methodology.
-------
TABLE 1
MOBILE HOME INDOOR AIR QUALITY STUDY—FLOW CHART OF EVENTS
Suspend study if design
yields insufficient HCHO
exposure gradient, poor
response rate.
Using mobile home park registration
lists, select pilot sample(s) to obtain
100 homes for feasibility study.
Map randomly selected mobile home parks
for random selection of homes in park;
leaflet park, field staff solicits house-
hold cooperation in person.
Sequentially analyze incoming samples
to determine study efficacy.
Draw total sample of parks for full study,
Map parks, select 1,100 homes for full
study (yield 800 homes).
Leaflet homes, solicit household cooperation
in person.
Take indoor HCHO measurement (1 time sample)
health status assessed by follow-up question-
naire over the telephone.
Statistical analysis of data.
-------
While all mobile home parks and sites in the parks could be detailed
and sample sites selected, contact with owners and ultimate
participation was low (65 out of 208 pilot sites) making the effort
required to enroll the 800 desired homes beyond the time, manpower and
funding of the project. With only a park lot number and no resident
names, addresses or telephone numbers, the field staff had difficulty
making direct contact with many of the sample homeowners. This was
especially true for the many homes which turned out to be seasonal or
weekend vacation homes. The second reason for deciding this
methodology needed to be modified was that among the 65 homes sampled,
the observed formaldehyde exposure gradient was less than desired for
our health effects analysis. The majority of homes were more than
three years old and the formaldehyde concentration distribution
skewed toward the lower limit of detection. Although valuable
information was obtained from the pilot study, its design was not
selected as the best procedural method for obtaining a large
formaldehyde concentration gradient.
The second methodologic design utilized the mobile home retailers as
the source of identifying and enrolling study homes. The intent was to
interview and clinically characterize the householders before and after
they moved into their mobile home. The proposed study flow chart is
outlined in Table 2. Initially the Wisconsin Manufactured Housing
Association supported the study, but when participation was solicited
from the individual dealers, only a small number agreed to participate.
Working with this small group would not have provided sufficient homes.
Thus this study design was also rejected.
After determining that the initial two study designs were unsuitable
and could not be implemented in a manner consistent with the agreement
constraints of time, funding and personnel and still fulfill the
desired subject participation and scientific criteria established, a
modified protocol was developed incorporating the portions of each of
the first designs which were found feasible. The full-scale program
was initiated in March, 1980. A summary flow chart is provided in
Table 3.
IV. Summary of Implemented Epidemiologic Protocol
The final project design utilized a stratified random sampling
procedure and prospective observation with enrolled homes visited once
a month for six or more consecutive months; followed by a final sample
at the one year anniversary. The monthly visits utilized identical
staff, procedures and practices for each visit. Neither the field
staff nor the participant families were given the sample or
questionnaire results prior to the completion of all of the testing.
-------
TABLE 2
MOBILE HOME INDOOR AIR QUALITY STUDY—FLOW-CHART OF EVENTS
Retailer notification
of sale to Division of Health (DOH)
(All Buyers)
DOH call to purchaser
to enroll into Study
Non-Volunteers
dropped from
Study
Volunteers Sent
confirming letter
and consent forms
Survey Lab. calls
volunteer to give
extensive entry
questionnaire
Field Staff Environmental Study
of "Old" (Current) Home using a
short entry questionnaire;
schedule clinic visit;
Sign Human Subjects Release
Prior move
clinic visit
Bimonthly new home Environmental Visits;
monthly health questionnaire
Exit questionnaire by
Survey Lab
Study termination
clinic visit
Data base; statistical
Analyses
-------
TABLE 3
MOBILE HOME INDOOR AIR QUALITY STUDY—FLOW CHART OF EVENTS
Non-volunteers
dropped from study.
Non-volunteers
have self-admini-
stered health history
questionnaire given
by field staff.
Select late model year homes
(100 homes <3 years) and early
model year homes (37 homes >3 years)
from city, town, and county property
assessor offices, (public records)
Field staff call to enroll home
voluntarily into the study.
Volunteers sent confirming
letters and consent forms.
Field staff monthly visit to home
1. Monthly UCHO sample
2. Monthly home status form
3. Monthly health questionnaire
for all occupants
-1 time measurement outdoor ECHO for
each home.
-1 time measurement of indoor nitrogen
dioxide.
Voluntary clinic visit (1 time during
observation period) scheduled by field staff•
Volunteers clinically assessed
1. Physical exam
2. Pulmonary function assessment
3. Self administered health history questionnaire
Data base - statistical analysis
1. Monthly HCHO, home status, health
status analysis
2. Seasonal analysis of HCHO concentration
3. HCHO, home status, clinical exam data analysis
-------
A. Final Study Population Selection Methodology
Field offices and staff were located in three geographical
areas of the state (Madison, Green Bay and Eau Claire). Each
office contacted city and town assessors in their area for
lists of mobile homes. Staff were able to utilize the
statewide listing of mobile home parks developed during the
previous design assessments and concentrate on cities or towns
where mobile home parks were known. The lists generated were
ordered by age of the mobile home and proximity to the field
office. To minimize staff travel time, owners were called and
invitations offered beginning with those parks closest to the
offices. The goal was to enroll approximately 30 homes in
each area. To assure a sufficient formaldehyde exposure
gradient, only homes less than three years old were selected.
Selection and enrollment began in March, 1980 and continued
until July, 1980. The majority (78%) were enrolled and had
their initial visit by May 31, 1980.
B. Comparison Group for Symptom Assessment and Clinical Studies
The pilot study from the first design evaluated confirmed that
older mobile homes (greater than 3 years old) have
significantly lower levels of formaldehyde than new homes. It
was decided to use such homes as a comparison group. They
were most appropriate since most would share the same
socioeconomic and demographic characteristics as the study
population except for formaldehyde exposure. Additionally,
neither the participants nor the field staff would know which
group were controls and which the study population. Media
impact on attitudes would also be similar since all mobile
homes were being portrayed as "at risk," not just the new
homes. A target of approximately 10 comparison homes was
sought for each field office.
C. Symptom Assessment
Each month at the time of the sampling visit, each family
member was asked to complete a brief, self-administered health
history and symptom questionnaire covering the preceding
month. Compliance was excellent and a total of 2,423
questionnaires were completed during the study.
Midway through the prospective period, each individual was
asked to complete a much more detailed health questionnaire
requiring 30-40 minutes to complete. As might be expected,
compliance with that questionnaire was much more difficult and
only 170 (60%) questionnaires were returned.
D. Clinical Assessment
Midway through the observation period, each individual was
offered and encouraged to participate in clinical examinations
-------
conducted two days in each of the field office areas. At the
clinic a physical examination and pulmonary function tests
were performed. None of the clinic staff were aware of the
levels of formaldehyde in the homes or which examinees were
study subjects and which from the comparison group. A total
of 53 adults and 23 children were examined.
V. Results of Ep id end, o logic Survey
A total of 137 mobile homes were voluntarily enrolled into the
study over an eight week period. Formaldehyde samples were
collected in two rooms (usually the kitchen or living room and
bedroom) using personal sampling pumps (MSA Model 6 and Bendix
BDX 44). Air was drawn through midget impingers containing
15-20 mis of a one percent sodium bisulfite absorbing reagent.
Pumps were run at a flow rate of 0.7 L/minute for
approximately one hour. Gas appliances were shut off and
smoking was discouraged during the sampling period. Windows
were closed approximately 1/2 hour prior to sampling. Quality
control blanks were prepared to insure non-contamination of
the absorbing solution. Samples were analyzed at the
Wisconsin State Laboratory of Hygiene using a modified NIOSH
chromotropic acid procedure (NIOSH, 1977).
These data along with climatologic and home environment data
measured at the time of sampling were key punched to magnetic
tape by data entry staff of the Wisconsin Division of Health.
Statistical analyses were executed under version 79.5 of SAS
(Statistical Analysis System, Inc., 1979).
A total of 979 household visits were made during the course of
the study. Formaldehyde concentrations were found to range
from below the limit of detection (0.10 ppm) to 2.84 ppm. The
mean for this distribution was 0.46 ppm; the median was 0.39
ppm. The 75th percentile was 0.59 ppm, while the 90th
percentile was 0.85 ppm. The distribution exhibited a marked
positive skewness; taking the log transform of the readings
resulted in a symmetrical normal distribution. Here, the
geometric mean was 0.37 ppm.
The home age distribution exhibited a bimodal shape which
reflected the age stratified sampling design of the study.
Approximately 100 homes were less than 36 months old at
enrollment into the study (58 homes were less than 12 months
old). The remaining 37 homes had an average age of 67 months.
Homes less than three years old had an average monthly
formaldehyde reading of 0.54 ppm; homes over three years old
averaged 0.19 ppm each month.
-------
The uncorrected log formaldehyde prediction model exhibited the
greatest R-Square value (approximately 46%); the R-Square for
the corrected log formaldehyde model was 30%. The R-Square for
corrected and uncorrected arithmetic formaldehyde models was
18% and 33% respectively.
Preliminary investigations of additional environmental
parameters indicated that these predicative models may be
improved, raising the R-Square by as much as 20%. Significant
variables for additional consideration include the number of
windows in the home, the number of windows open prior to
sampling, wind speed, and households with occupants who smoked
tobacco, (all appear to have a negative relationship to
concentration). Humidity did demonstrate a positive
coefficient for predicting formaldehyde, however, it did not
reach statistical significance in this analysis. An improve-
ment for predicting concentration with home age would
undoubtedly be achieved with the quantitation of ventilation
rates and particle board emitting source load (i.e., board
area/room volume ratio).
Summary
The repeated measurement of mobile homes for formaldehyde
failed to exhibit a systematic correlation between the
sequential readings taken in each home. These data were thus
analyzed using least squares regression, treating each sample
as an independent observation. The log of the home age in
months was seen to be the single best predictor of both climate
corrected and uncorrected formaldehyde readings.
These findings are in accord with analyses previously performed
on complaint case series and random pilot mobile home data
(Dally et al., 1981). An improvement in these models'
predictive power is suggested by the inclusion of variables
that begin to quantify ventilation rates found in the home at
the time of sampling. It is further hypothesized that
accounting for source load will improve the models' efficacy.
B. Health Status Assessment
Health status assessment of mobile home residents was done to
check possible associations of health effects with indoor
formaldehyde concentrations. A monthly health status
questionnaire and a one time chronic health questionnaire and
clinical assessment were requested from all participants.
Subjects
A total of 339 persons were enrolled into the study. However,
early dropouts yielded a cohort of approximately 288 persons
which remained stable for six home visits. Ten percent of the
persons were under two years old; the median age was 24 years
10
-------
old. The 75th percentlle was 35 years of age, while only 10
percent of the group were over 60 years old. Forty-eight
percent were males, 52% were females. Roughly 37% were
cigarette smokers.
Methods - Questionnaire
A total of 2,423 monthly health questionnaires were completed
by the occupants. Symptoms queried included sensory and upper
respiratory tract irritation, respiratory problems, and
gastrointestinal disorders. The average monthly rate of
response for these conditions (based on 2,404 responses) is
depicted in Table 5.
TABLE 5
MONTHLY AVERAGE REPORT OF SYMPTOMS
Symptom Percent Responding
Burning eyes 25%
Watery eyes 20%
Cough 44%
Runny nose 35%
Sneezing 43%
Dry throat 24%
Phlegm 25%
Wheezing 12%
Breathlessness 10%
Difficulty breathing 11%
Chest pain 8%
Swollen glands 6%
Headache 29%
Dizziness 10%
Weakness 10%
Tiredness 24%
Difficulty sleeping 16%
Nosebleeds 4%
Nausea 7%
Vomiting 5%
Diarrhea 10%
Fever 8%
Rash 11%
Other 1%
N=2,404
11
-------
Higher rate of symptom report can be seen in the upper
respiratory symptoms (cough - 44%; sneezing - 43%; runny
nose - 35%) when compared to the remaining conditions. The
irritant symptoms of burning and watery eyes, dry/sore throat,
and phlegm center around a rate of 25%. Headache (29%) and
tiredness (24%) are two neuro-behavioral conditions which also
exhibit a relatively higher rate of report.
Stepwise logistic regression analysis (Cox, 1970) was utilized
in order to examine the effect formaldehyde exposure, and the
individual's age, sex, and smoking status would have on
predicting the log odds of symptom report [i.e., P(symptom
present)/P(symptom absent)]. Symptom logistic regression
models were separately constructed for each of the months
(April through December, 1980). Table 6 exhibits the results
of the analyses. A coefficient is depicted if its sign remains
unchanged and it achieves significance for at least three of
the nine months of observation.
12
-------
TABLE 6
SYMPTOM LOGISTIC REGRESSION MODELS
SIGNIFICANT REGRESSION COEFFICIENTS*
LOG FORMALDEHYDE AS EXPOSURE PARAMETER
Symptom Log Formaldehyde Age Sex** Smoking***
Burning eyes +(8) +(9)
Watery eyes +{7) +(9)
Cough +(7)
Runny nose
Sneezing
Dry throat +(3) +(4)
Phlegm +(9) +(9)
Wheezing +(3)
Breathlessness +(5)
Difficulty breathing +(5)
Chest pain +(6)
Swollen glands +(3)
Headache +(4) +(7)
Dizziness +(4) +(4)
Weakness +(8)
Tiredness +(7)
Difficulty sleeping +(5)
Nosebleed
Nausea
Vomiting
Diarrhea +(3)
Fever
Rash
* Sign indicates direction of association. Number in parantheses indicates
the number of months that the association remains significant (P<0.05)
** Sex: 0=male, 1-female
*** Smoking: 0=No, l=Yes
13
-------
As can be seen from the table, log formaldehyde concentration
was associated with the report of burning and watery eyes, and
dry sore throat. This is in accord with the known human
responses to formaldehyde exposure (National Academy of
Science, 1980; Schuck et. al., 1966; Kerfoot, 1975). The
report of swollen glands and diarrhea also exhibit a log
formaldehyde dependency. None of the other symptoms
demonstrated a similar consistent relationship to formaldehyde.
Cigarette smoking exhibited a consistent positive relationship
to the report of coughing, dry/sore throat, phlegm, and
headache; smokers would thus have a higher rate of report than
would non-smokers. Age was positively related to the report of
burning and watery eyes, respiratory condition (phlegm,
wheezing, breathlessness, difficulty breathing) chest pain, and
several neuro-behavioral conditions (dizziness, weakness,
tiredness, and difficulty sleeping). Sex predicted the report
of headache, and dizziness.
Summary
A positive dependency of both burning eyes and watery eye
symptom logits (log odds of report) on the log formaldehyde
exposure parameter was demonstrated. In addition, a
significant positive relationship with the respondent's age was
also seen. Dry/sore throat, swollen glands and diarrhea logits
exhibited a less robust dependency on formaldehyde exposure.
The report of cough, phlegm, and headache were strongly related
to cigarette smoking.
Increasing age was related to the log odds of symptom report in
many instances because of the differential rate of response in
younger persons. Continuing separate investigations of adult
and children groups will factor out this pervasive effect to
allow for an age-specific or stratified analysis.
Clinic and Chronic Health Assessments
After approximately six months, a more detailed comprehensive
medical history, reproductive history and review of symptoms
questionnaire was distributed and each study participant
encouraged to attend one of three clinics for physical
evaluation. The questionnaire asked each participant to
subjectively assess each of the symptoms they had reported
during the previous six months and record whether they felt the
symptom was related to conditions in their home and whether
they had experienced the same symptoms prior to moving into
their current residence. The clinical evaluation focused on
the eyes, nose, throat, respiratory system and skin looking for
objective signs of irritation. The examining physicians did
not know the levels of formaldehyde present in the homes of the
14
-------
clinic participants. They were asked to review the
participants current symptomatology and physical findings and
report whether current upper respiratory infection or acute
allergy reaction was present and whether the person was atopic.
Spirometry and single breath diffusing capacity were performed
on all participants over the age of 12. Clinics were conducted
at the end of September and during the month of October to
avoid the peak pollen allergy season and the winter flu season.
C. Adult Survey Participants
Prior to the distribution of the detailed questionnaires and
the clinics, twenty-three households dropped out of the study.
Explanations ranged from inconvenience of the monthly visits to
three homes destroyed by tornados. Of the remaining 114 homes,
62% have returned their detailed questionnaires. Thirty-two
households (28%) participated in the clinics. Utilizing the
information of the monthly questionnaires, no differences were
seen between the three groups, [1) non-respenders, 2) returned
questionnaires but not examined, 3) clinic participants] with
regard to monthly prevalence of symptoms, age, sex and smoking
habits. The mean formaldehyde level among the non-responder
homes was significantly lower than either of the other groups
(.37+ .2 ppm, .45+ .3 ppm, .49+ .2 ppm respectively). The
adults who participated in the clinical examinations were
significantly older than those who returned the questionnaires
but were unable to be examined (39+ 19 years versus 31+ 14
years), although they were not older than the total group of
not examined adults. All other parameters were similar between
the two groups which returned the questionnaires.
Of the adults who returned questionnaires, 72% reported having
experienced cough since the onset of the project and 58% had
experienced eye burning. However, when asked if they thought
the symptom was related to household conditions, only 11% of
those with cough felt it was probable or definitely related.
For eye burning, 71% felt it was related to the home. The
presence of cough was not associated with the level of
formaldehyde found in the home. However, individuals from
homes with mean formaldehyde over .8 ppm felt their cough was
related to the home environment significantly more frequently
than those from homes with levels below .8 ppm (31% versus 7%).
the occurrence of burning eyes was significantly associated
with the level of formaldehyde found in the home. Utilizing
the individual's subjective assessment strengthened this
association even more. Not only did the prevalence of burning
eyes increase significantly with increasing mean household
formaldehyde, but also the proportion of individuals who felt
their burning eyes were related to household conditions went
from 50% to 87%. The mean household formaldehyde for those who
felt their symptoms were related to their home was .59+ .3 ppm
15
-------
and was .4+ .2 ppm for those without symptoms or felt their
symptoms were unrelated to home conditions. These means were
statistically significantly different.
A total of 53 adults from 32 households completed the clinical
examination portion of the study. Four individuals were
determined by history and physical to have current "colds" and
thus their clinical signs of eye and nose irritation were not
attributable to formaldehyde exposure. They have been removed
from this portion of the report. Thirty-six percent of those
examined had a history of allergies. Signs of eye or nasal
irritation were not significantly more prevalent among those
with a history of allergy. Of the 49 adults, eighteen percent
had mild to moderately injected conjunctiva and 14% had inflamed
or congested nasal mucosa. Combined, 27% had evidence of mild
to moderate mucosal irritation. Two individuals had nasal
polyps (both predated moving into their mobile home). The
prevalence of irritation signs correlated with increasing mean
household formaldehyde (measurement closest to day of exam was
utilized). Ten percent of individuals from households with less
than .4 ppm formaldehyde had signs of irritation compared to 24%
of those greater than .4 ppm. Of the 9 individuals from
households with greater than .8 ppm, 5 (56%) had signs of
irritation. The mean formaldehyde for those with signs of
irritation was .7+ .3 ppm compared to .4+ .3 ppm for those
without signs of irritation. This difference was statistically
significant. Spirometry did not identify any associations other
than those already well recognized for cigarette smoking.
VI. Summary of Laboratory Activities
The Clinical Chemistry Section of the State Laboratory of
Hygiene (SLH) provided laboratory support during the project.
Their primary tasks included:
1. Evaluation of analytical techniques for formaldehyde.
a. Determine which method would be most appropriate.
b. Adapt the procedure for SLH use.
c. Assess the method's accuracy and precision.
2. Establishment of quality control procedures.
3. Analyze samples.
The method chosen by the SLH for formaldehyde analysis was the
NIOSH P+CAM 125 method or chromotropic acid method. The method
has been widely used and it had sufficient sensitivity and
reproducibility for use in residential sampling. The procedure
involves collection of air in two midget impingers connected in
16
-------
series. In each impinger there is an absorbing reagent which
forms a complex with formaldehyde as air is bubbled through the
impinger. The liquid is analyzed in the laboratory.
Chromotropic acid and sulfuric acid is added to an aliquot of
the specimen to form a purple chromagen. The absorbance, read
at 580 run on a spectrophotometer, is proportional to the amount
of formaldehyde in the solution.
Two major modifications were made in the NIOSH procedure. The
method specifies that distilled water should be used as the
absorbing reagent. However, with water there are potential
sample stability problems. Several studies indicated that
sodium bisulfite was more efficient at collecting formaldehyde
and formed more stable complex with formaldehyde than distilled
water. Since samples could not be analyzed immediately, a 1%
sodium bisulfite solution was used as the absorbing reagent.
The solutions were prepared in the field at two of the three
field staff stations by mixing preweighed portions of sodium
bisulfite and double distilled water. At the third location,
field staff used solutions prepared by SLH.
The second major modification was in the standardization
procedure. The NIOSH method recommends the use of a 37%
formalin solution diluted in distilled water or sodium
formaldehyde bisulfite in the standardization process. This is
titrated against an iodine solution to obtain the formaldehyde
concentration of the standard solution. However, the
reproducibility of this method was not adequate for the
determination of low levels of formaldehyde. The quality
control labs used by SLH had experienced similar problems with
this standardization procedure. The best laboratory agreement
occurred when the same stock solutions were used by the various
labs. As a result, the laboratory undertook an investigation of
various standardization procedures to determine which method
would be best. A method based on the titration of a stock
solution by sodium sulfite was used. The stock solution was
prepared by refluxing paraformaldehyde in water to form
methylene glycol. The stock solution was then diluted in a 1%
sodium bisulfite solution for the preparation of standard curves
since samples were collected in sodium bisulfite. This method
has since been published by NIOSH in 1981 as a replacement for
the iodine titration method for P+CAM 125 (NIOSH, 1981).
A total of 1,874 air samples were analyzed during the main
formaldehyde study.
Quality Control Procedures
Blanksi Two types of blanks were collected during the project.
Sodium bisulfite blanks were prepared daily in the field to
determine background concentrations of formaldehyde in the
collection liquid. If more than one batch of sodium bisulfite
was mixed and used per day, blanks were prepared from each.
17
-------
Each batch mixed was identified by a unique identifying number
and this was used to indicate which samples came from particular
batches of the absorbing reagents. A total of 730 blanks were
collected and analyzed.
Impinger blanks were also prepared to determine how adequately
impingers were being rinsed between uses. Under ideal
circumstances, impingers should be acid washed between each use
for thorough cleansing. However, due to the large numbers of
samples collected daily, the hazard of handling acid in the
field and the location of field staff away from the lab, this
could not take place. To ensure that carryover from sample to
sample was minimal, a strict rinsing procedure was used to wash
the impingers between each use. Before use, the impinger was
washed three times with sodium bisulfite and three times with
distilled water. Care was taken to fill the base entirely full
of rinse water and to thoroughly rinse the stem.
Impinger blanks were prepared in the field by following the
rinsing procedure and then adding 15 mis of liquid to the
Impinger, agitating it, and adding the contents to a sample
vial. The blanks were prepared at least once per day. A total
of 573 impinger blanks were collected and analyzed.
Sample Handling
Past experience at SLH indicated that certain types of bottle
cap liners contained substances that interfered with the
analysis. This has been confirmed by additional studies. To
prevent contamination, Nalgene polyproplyene bottles were used
to store samples. Several studies have indicated that sample
loss can occur when samples are stored at room temperature.
Samples could not be analyzed immediately, so special
precautions were taken to ensure that sample loss was minimized
during storage. The sodium bisulfite solutions were kept
refrigerated until use and stored in ice chests in the field.
All samples after collection were placed in ice chests and then
stored in a refrigerator. When it was necessary to ship
samples, specimens were placed in an insulated shipping
container with frozen foam packs with overnight delivery to the
SLH. Temperatures in the containers ranged from 9-12°C. upon
arrival.
Quality Control Procedures
During each analysis, samples from the previous run were
re-analyzed. These replicate samples included all sodium
bisulfite blanks exceeding 0.20 ng/4 ml aliquot, all impinger
blanks exceeding 0.5 ug/4 ml aliquot, all outdoor samples
exceeding 0.1 ppm, specimens collected within the same home
where the two room samples didn't agree within +0.10 ppm, and
random selections of high, medium, and low specimens.
18
-------
Assays from a pool solution was prepared and run in duplicate
during each run. Other chemists also participated in analyses
on several occasions. If deviations beyond the preestablished
limits in standard curves were detected, analyses were repeated
for all affected specimens.
During the project, samples were sent to other laboratories for
inter-laboratory quality control. Samples were sent to RalTech
Laboratory, Madison, Wisconsin three times during the main
study. Samples were analyzed within 24 hours. When the same
method of standardizing the stock standard was used, (sodium
sulfite tltration) excellent agreement was obtained between both
laboratories - both with split laboratory prepared solutions and
actual split air specimens. Agreement was not as good when
different standardization techniques were used.
Precision and Accuracy Determination
In order to determine the precision and accuracy of the method
used by the SLH, a study of pool solutions of known
concentrations was conducted. Concentration levels of 0.20,
0.30, 0.40 and 5.50 ug/4 ml aliquot were chosen for study. This
is equivalent to 0.01, 0.02, 0.03 and 0.40 ppm using our
sample procedures. Ten replicate assays of each concentration
were run daily for five days. Forty blank specimens were
prepared each day and one was paired with a pool specimen.
Specimens could be distinguished from blanks and each other at
each of the concentrations on the spectrophotometer. However,
the coefficients of variation (CV) at the three lowest
concentrations were quite high. The inter-run CV for the
analysis ranged from 32% at 0.03 ppm to 125% at 0.01 ppm. At
0.40 ppm the CV was only 1%. Interday variations of
spectrophotometer which reflect changes in sensitivity from day
to day were significant, however, when the ppm value is
calculated this difference shows up in the third or fourth
decimal place.
Combining the CV's for the sampling and analytical procedures a
total CV for the method was calculated. At 0.03 ppm the total
CV was 32% while at 0.40 ppm the total CV was 5%. At 0.03 ppm
the 95% confidence interval is 0.01 to 0.05 ppm while at 0.40
ppm it is 0.36 to 0.44 ppm. These represent the results from a
single impinger analysis. When two impingers are used, the
error will be slightly higher. At low concentrations (less than
0.10 ppm) values are subject to high relative error since the
concentration from each of the two impingers used in the sample
train are added together. Because of this, 0.10 is considered
the overall detection limit of the method as used during the
project.
19
-------
VII. Other Related Projects
Follow-up Study
Occupants of homes originally investigated in 1977-79 were
contacted in March-July, 1981 and asked to participate in a
follow-up study. In addition to measuring formaldehyde, CO,
respirable particles, and nitrogen dioxide were also measured to
give a better indication of total indoor air quality. There were
27 homes in the group and these included 17 mobile homes, 5
conventional homes with particle board products, and 5
conventional homes with U-F foam insulation. The original group
mean formaldehyde concentration was 0.58 ppm, while at the
follow-up visit these homes had a mean of 0.30 ppm. This
represented a significant decrease in formaldehyde concentration.
This would be expected since the homes had increased in age on
the average of 20.5 months. An interesting finding is that even
after this increase in age of the home, the mobile homes still
had a mean concentration of 0.37 ppm. As with our other
formaldehyde studies, good correlation was found between samples
collected in different rooms; and no affect in formaldehyde
concentration was found from the presence of smokers or gas
stoves.
Different sample collection techniques were utilized during this
study. Samples were collected side by side in distilled water
with impingers immersed in an ice bath, sodium bisulfite with
impingers immersed in an ice bath, and sodium bisulfite with
impingers not immersed in an ice bath (conventional method). No
significant difference in formaldehyde concentration was found
between the collection methods.
In addition, the samples collected in distilled water were also
analyzed using the pararosaniline method as modified by Lawrence
Berkeley Laboratory, Berkeley, California (Miksch et al., 1981).
The method was not available for evaluation until after the main
study had been completed. Since the pararosaniline method was
reported to be more sensitive to lower formaldehyde concentra-
tions, we evaluated it here to determine whether it could be used
for future projects. The correlation, however, between samples
analyzed by this method and by the chromotropic acid method was
not very good. It appeared that S02 emitted from the impingers
using sodium bisulfite may have been a serious interferent with
the pararosaniline method (Eckmann et al., 1982).
Nitrogen dioxide concentrations ranged from none detected to
156.6 yg/m^. Homes with gas stoves had a mean kitchen concentra-
tion of 64.9 ug/m^, while homes with electric stoves had a mean
kitchen concentration of 8.0 yg/nr>. These findings are similar
to those found in studies by other researchers. Respirable
particle concentrations ranged from 0.005 vg/m^ to 1.570 jjg/m^.
Homes with smokers had levels higher than homes without smokers.
Carbon monoxide concentrations ranged from none detected to about
10 ppm.
20
-------
Comparison of Passive Monitors with NIOSH Method
Three homes were visited in an effort to compare two dosimeters
with the impinger method of collecting formaldehyde. Since the
dosimeters were exposed for eight hours, eight sequential 1-hour
samples were collected in Implngers to cover the entire time
period the dosimeters were exposed. Three DuPont passive
monitors and two Envirotech dosimeters were placed in the kitchen
and in two bedrooms. In the kitchen and one bedroom two sample
trains with impingers were set up side by side, and in the third
bedroom one sample train was set up.
All the impinger samples were analyzed using the chromotropic
acid procedure; the DuPbnt badges were analyzed by the Indiana
Board of Health Lab in Indianapolis using the chromotropic acid
method and the Envirotech dosimeters were sent back to the
distribution for analysis.
The correlation between the passive monitors and the impinger
samples was not very good. One problem with the DuPont monitor
was probably insufficient sample time indicating that an eight
hour exposure of this badge at relatively low concentrations is
not adequate to detect the levels actually present in a home. It
was discovered later that the Envirotech badges had some design
flaws necessitating modification, so results from the dosimeters
used here are probably not valid.
Despite the problems with the passive monitors, the study was
useful in that many samples taken by one person over an eight
hour time period in different rooms could be compared. While
data analysis is not complete at this time, correlation between
the samples is good. Given that the home was sampled under
relatively constant temperature conditions, little variability in
sample results was seen.
01-176-70
21
-------
REFERENCES
Berge, A., Mellegaard, B., Hanetho, P. and Ornstad, E. B., (1980)
"Formaldehyde Release From Particle Board - Evaluation of a
Mathematical Model," Holz als Roh und Werkstoff, 3£, 251-255.
Cox, D. R., (1970), "The Analysis of Binary Data," Methuen, London.
Dally, K. A., Hanrahan, L. P., Woodbury, M. A., Kanarek, M., (1981),
"Formaldehyde Exposure in Nonoccupational Environments." Arch Environ.
Health, 3^(6):277-284.
Draper, N. R., and Smith, H. (1981) "Applied Regression Analysis," John
Wiley and Sons, Inc., New York, NY.
Eckmann, A. D., Dally, K. A., Hanrahan, L. P., Anderson, H. A., (1982),
"Comparison of the Chromotropic Acid and Modified Pararosaniline Methods
for the Determination of Formaldehyde in Air." Environ. Inter.,
JJ: 159-166.
Kerfoot, E. J., Mooney, T. F., (1975) "Formaldehyde and Paraformaldehyde
Study in Funeral Homes." Am. Ind. Hyg. Assoc. J., 36, 533-537.
National Academy of Sciences, National Research Council Committee
on Toxicology, (1980), "Formaldehyde - An Assessment of its Health
Effects."
NIOSH (1977), "NIOSH Manual of Analytical Methods, Second Edition,
Volume 1." DHEW (NIOSH) 77-157A, U.S. Department of Health, Education
and Welfare, Cincinnati, Ohio.
NIOSH (1981), "NIOSH Manual of Analytical Methods, Second Edition,
Volume 7." U.S. Department of Health, Education and Welfare, Cincinnati,
Ohio.
Statistical Analysis System, Inc. (1979), "SAS User's Guide 1979 Edition."
SAS Institute, Inc., Raleigh, North Carolina.
Schuck, E. A., Stevens, E. R., and Middleton, J. T., (1966), "Eye
Irritation Response at Low Concentrations of Irritants." Arch.
Environ. Health 31, 570-575.
22
-------
EPA-905/1-83-001
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE (date of preparatio i
February 15. 1983
4. TITLE AND SUBTITLE
The Epidemiology of Mobile Home Formaldehyde Vapor
Concentration and Residents' Health Status
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Henry A. Anderson, Kay A. Dally, Lawrence P. Hanrahan,
Ann D. Eckmann*, Marty S. Kanarek**, John Rankin**
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Wisconsin Division of Health, P.O. Box 309, Madison, WI
53701. *State Laboratory of Hygiene, 465 Henry Mall,
Madison, WI, 53706. **Department of Preventive Medicin|e
10. PROGRAM ELEMENT NO.
University of WI, Madison, WI, 53706.
11. CONTRACT/GRANT NO.
» CS806856-01-3
12. SPONSORING AGENCY NAME AND ADDRESS
US Environmental Protection Agency
Washington, DC 20460
13. TYPE OF REPORT AND PERIOD COVERED
Final; 1979 - 1983
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
A home age stratified random sample cohort of 100 homes less than three years old and
37 homes of more than three years of age (mean 67 months) were enrolled and followed
prospectively at monthly intervals. Formaldehyde air measurements ranged from below
detection (0.1 ppm) to 2.84 ppm. The average monthly concentration from homes less
than three years old was 0.54 ppm and 0.19 ppm from homes older than three years. Re-
gression modeling found that the log of home age was the best environmental parameter
for predicting formaldehyde levels. A total of 2,423 monthly self-administered symptom
questionnaires were completed by the 288 persons in the homes which remained under
study for six or more months. Stepwise logistic regression analysis was performed. A
positive dependency of both burning eyes and watery eye symptom logits on the log for-
maldehyde exposure parameter was demonstrated. A positive relationship with respon-
dents' age was also seen. Dry/sore throat, swollen glands, and diarrhea logits ex-
hibited a lesser dependency on formaldehyde exposure. Cough, phlegm, and headache
were related to cigarette smoking but not to formaldehyde. Fifty-three adults par-
ticipated in clinical examinations. Evidence of ocular or nasal irritation was more
prevalent among those from households with the higher formaldehyde levels (five of
nine above 0.8 ppm) than those from the lower level homes (10% among those less than
0.4 ppm).
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Formaldehyde
Indoor Air Pollution
Eye Irritation
Mobile Homes
8. DISTRIBUTION STATEMENT
No Limitations
19. SECURITY CLASS /ThiiRepprtj
Unclassified
•>'i NO.-OF PAGES
27
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4—77) PREVIOUS EDITION is OBSOLETE
------- |