Katherine Fu
EPI 225
Association Between Maternal Fluoride Exposure During Pregnancy and IQ Scores in
Offspring in Canada (Green et al., 2019)
This was a prospective cohort study, where maternal fluoride levels were the exposure
and their child’s IQ was the outcome variable. This is an observational study, which is
appropriate because it is not ethical to randomly assign mothers to higher fluoride levels.
Usually, cohort studies do not begin with individuals with IQ data (outcome), instead of the
individuals who are eligible. However, the data about IQ was only collected in 601 individuals.
From those with data, there was a relatively high proportion of individuals eligible in both
groups. Additionally, Table 1 shows that the groups were relatively balanced on a variety of
demographic variables. In terms of the exposure, maternal urinary fluoride concentration was
collected in each trimester and then averaged. This is an objective measure, and the validity is
supported by the community water fluoridation status data. However, information on the
consumption of tap water was based on a self-report questionnaire and not validated.
Therefore, while the exposure information was non-differential because both groups were
measured the same way, there is a high risk of bias. In terms of the outcome, IQ is objective and
the same in both groups, so there is a low risk of measurement error. In terms of confounders,
the researchers listed many factors potentially associated with fluoride metabolism and IQ in the
covariates paragraph. Although it was an observational study, the potential threat to internal
validity due to confounding is low because they included many important potential confounders.
Given this, I think that they main threat to internal validity is from the measurement of the
exposure, which would blunt the signal towards the null.
In terms of the results, linear regression was appropriate for the study design. The
researchers found that higher fluoride exposure was associated with lower IQ scores. Higher
maternal urinary fluoride was associated with lower IQ scores among boys. While the results
were statistically significant, there is less support for clinical significance, such that the
associated change in IQ may not be enough to warrant a public health response. Additionally,
the sex difference was not seen for fluorine intake as the exposure. Therefore, while it is
relevant to explore environmental impacts on maternal and fetal health, I would exercise caution
when interpreting the actual clinical impact of the results of this study.