This study reports a strong association between exposure to
commercially applied agricultural pesticides during a crucial period
in fetal development and the likelihood of fetal death due to congenital
defects. Associations were found for mothers living within a 9-square
mile area around the home. That is, fetal death is more likely
among mothers who are living within a 9-square mile area in which commercial
pesticide spraying takes place during pregnancy.
What did they do?
Bell et al.carried out a case-control study of the risk of
fetal death in relation to pesticide exposure. Their work built upon
two independent data sets:
- birth/fetal death/death certificates obtained
from the California State Vital Statistics Registry, including
mother's address, cause of death, data on the pregnancy and the
parents, etc.; and
- the California State Pesticide Use Report
(183-1984), which includes information on the application of all
restricted-use agricultural pesticides, including specific
chemicals used, amount applied, date and location for each
application. Bell et al. broke the pesticides examined in
their study into five categories: phosphates, carbamates,
pyrethroids, halogenated hydrocarbons, and endocrine disruptors.
Increases in risk were observed for all 5 categories.
Because both data sets included information on place and
time, Bell et al. break new ground in their work in several
- First, their measurement of exposure, while
imperfect (see below), does not depend upon memories of the
mothers and comes from a source independent of the assessment of
the birth outcome, a governmental database of pesticide
applications. Most epidemiological studies depend upon recall.
This makes them vulnerable to vagaries of memory and it also means
the asessment of exposure and outcome are not truly independent.
Bell et al. have avoided this problem.
- Second, they use a more refined approach to
the timing of exposure and vulnerability, focusing their attention
on the period of fetal development during which many of the most
basic aspects of fetal differention are taking place, weeks 3-8
(the period of "organogenisis"). Toxicological studies have clearly
established that the timing of exposure during fetal development
is crucial to determining the outcome. But the data are usually
not available in sufficient detail to allow the most vulnerable
period in human development, weeks 3-8, to be examined separately.
Studies are at best usually able to study the effects of during
different trimesters, and often cannot consider timing at all,
other than simply that exposure took place sometime during pregnancy.
This refinement by Bell et al. is very important because
it decreases the chance that irrelevant or less relevant exposures
(outside the period of maximum vulnerability) will dilute their
ability to detect real effects. They demonstrate the value of
this approach by showing that the calculated risk increases as
they focus their analysis: risk of fetal death is greater for
exposures in week 3-8 compared to week 1-20.
To calculate the effect of exposure on risk, Bell et
al. compared the cases of fetal and neonatal deaths due to
congenital defects that they found in the health records with
controls selected randomly from normal live births in the same
California counties in which the deaths occurred. The bottom line of
their analysis is that if pesticide exposure has no impact on risk,
then the mothers whose pregnancies resulted in fetal death should
not be any more likely to have been exposed to pesticides during
pregnancy than mothers whose babies were born normally. As noted
above, for each of the categories of pesticide studied, they found
that exposure increased the risk of fetal death.
The use of the California state Pesticide Use Report in this
study is an important breakthrough, but it has significant
limitations. It is good because, as noted above, it provides an
assessment of the likelihood of exposure that is independent of the
mother's recall and is specific to a time and place. The weakness is
that it is only a surrogate index of exposure. Clearly, if no
pesticide use took place within the mother's geographic area, then
no exposure to these agricultural chemicals should have been
possible. But the fact that pesticides were applied does not mean
that exposure to that mother actually took place. This limitation is
likely to have weakened Bell et al.'s ability to detect real