Public Health and Ecological Interconnectivity:
A Conditional Probability Approach Associating
Degradation of Streams and Infant Mortality
John F. Paul, Michael E. McDonald, Steven F. Hedtke
U.S. EPA, Office of Research and Development,
National Health and Environmental Effects Research Laboratory, Research Triangle Park, NC 27711
Public Health Goal
Major improvements in
public health by reductions
in pervasive diseases
Pump linked to cholera outbreak
Problem was viewed as health
effect (mortality from cholera
in population)
Tracked back to water pump as
likely source of exposure
Current Approach
linical exposure-based studies on
individuals determine causation,
but there are thousands of
exposure agents.
This implies possible large
expenditure of funds on studies
with no adverse health effects
Possible Improvement
Use Snow's large-scale approach
as a screen and then follow with
appropriately designed individual-
based studies
Death Records
Demographic barometer of
community health since
19lh century
Mortality and morbidity data still
collected and publicly available in
CDC national data bases
Suggested Approach
Explore associations between
infant mortality and environmental
conditions (look for patterns in
associations) for more effective
public health intervention
Infant Mortality
Major health status indicator of
populations and available at the
county level of aggregation
(Environmental Condition)
Health of bottom-dwelling
communities in streams is
| integrative of recent water quality
degradation and reflective of
environmental condition
(intent of This Research!
Explore possible relationship
| between environmental condition
and infant mortality
Indicator of Public Health:
probability of infant mortality rate
(IMR) greater than 8.2/1000
(national norm for 1989-98)
Indicator of E-Condition:
percent of stream miles degraded
(low score for benthic index)
EJBD
ARCHIVE
EPA
601-
F-
06-
001
MD counties with > 60%
degradation
For Maryland counties, is there a
relationship between degraded
stream condition and infant
mortality?
Formulate and test as null
hypothesis
Conditional Probabilit
Probability of something occurring
when something else has occurred
Example: Probability of IMR in a
county > 8.2 per 1000 if at least
half of streams in county are
degraded, P ( Y | X > 50%)
Empirical conditional
probability curve
There is a real
association
(but it is not cause-effect)
Make the association quantitative
by trying to disprove null
hypothesis
Null Hypothesis
P(Y|X>XC)=P(Y)
or
conditional probability equals
unconditional probability
Result of bootstrap
resampling
Less than 2 % chance that
P ( Y | X > 78%) = P ( Y )
could have occurred randomly
Null hypothesis disproven!
Results Expressed as Relative Risk
Similar Shape for PA and WVJ *• • •
V
*V** if | a M_
^\»HHtJr | f a-
What would you conclude
from these results? What
are the next steps?
Please share your thoughts
with us on post-it notes.
Our Conclusions and Next
Steps are on handout.
Conclusions
A quantitative relationship exists
between environmental condition
and public health outcome.
If we protect the environment,
can we also protect human
health?
If we understood the link (eco-
human web), then management
action could protect both
Confirm with additional data
sets and in other areas
ENVIRONMENTAL
PROTECTION AGENCY
John F. Paul, Ph.D.
Rr-w*rih (-IT. iriionifntjl S.K"
Njtioiul rtrtlth Jnd Environ
R«*arch Ubor-turv ORD
epascienceforum
Your Health • Your Environment • Your Future
------- |