Meta-analysis of small for gestational age births and disinfection byproduct exposures

Some epidemiological studies show associations between disinfection byproducts (DBPs) and adverse developmental outcomes.

We undertook a meta-analysis of epidemiological studies on maternal exposure to trihalomethanes (THMs) and haloacetic acids (HAAs) and risk of small for gestational age (SGA) birth.

We identified forty-five publications including two reports and five theses via a 2020 literature search. Nineteen study populations from 16 publications met the inclusion criteria and were systematically evaluated. Effect measures were pooled using random effects meta-analytic methods along with cumulative, sub-group and meta-regression analyses to examine between-study heterogeneity and variation in risk across different DBP measures.

We detected a small increased risk for SGA with exposure to the sum of four (i.e., THM4) THM4 (odds ratio (OR) = 1.07; 95%CI: 1.03, 1.11), chloroform (OR = 1.05; 95%CI: 1.01, 1.08), bromodichloromethane (OR = 1.08; 95%CI: 1.05, 1.11) and the sum of the brominated THM4 (OR = 1.05; 95%CI: 1.02, 1.09). Larger ORs were detected for the sum of five haloacetic acids (i.e., HAA5) (OR = 1.12; 95%CI: 1.01, 1.25), dichloroacetic acid (OR = 1.25; 95%CI: 1.01, 1.41) and trichloroacetic acid (OR = 1.21; 95%CI: 1.07, 1.37). We detected larger SGA risks for several THM4 among the prospective cohort and case-control studies compared to retrospective cohorts and for the SGA3/5% (vs. SGA10%) studies. The THM4 meta-regression showed associations between SGA and the total quality score based on categorical or continuous measures. For example, an OR of 1.03 (95%CI: 1.01, 1.06) was detected for each 10-point increase in the study quality score based on our systematic review.

We detected a small increased risk of SGA based on 18 THM4 study populations that was comparable to a previous meta-analysis of eight THM4 study populations. We also found increased risks for other THM4 and HAA measures not previously examined; these results were robust after accounting for outliers, publication bias, type of SGA classification, different exposure windows, and other factors.