Comparison of Individual-level and Ecologic Studies of the Association of Drinking Water Contaminants on Adverse Birth Outcomes

Abstract

Listen

Translate article

Background Several epidemiological studies of the effects of drinking water contaminants have used ecologic designs, comparing groups defined by the community water system that they use. While ecologic studies are subject to bias, the degree of bias has rarely been assessed. Aims The aim of this analysis was to assess the bias from the use of ecologic study designs by comparing results from an individual-level study and an ecologic study of the association of arsenic in drinking water to adverse birth outcomes. Methods Maternal addresses at time of delivery were geocoded for over 651,000 live births (1989–2006) in the state of Utah, USA. Annual average arsenic levels were estimated for 476 Community Water Systems using water quality monitoring data. Arsenic levels were assigned to each birth by linking geocoded addresses to georeferenced water system service areas. Logistic regression was used to estimate the associations between individual arsenic levels and the occurrence of low birth weight (LBW), small for gestation age (SGA), pre-term and very preterm birth. For the ecologic study negative binomial regression was used to assess the relationship between annual average arsenic levels for each community water system and the proportion of births that were LBW, SGA, pre-term or very pre-term. Results There were significant associations between arsenic levels and the risk of LBW and SGA in both the individual-level and ecologic studies. Drinking water arsenic levels of 5 to <10 µg/L were associated with LBW in the individual-level study (aOR=1.18; 95% CI 1.06, 1.32) and in the ecologic study (IRR=1.09; 95% CI 1.00, 1.19). There were no significant associations with pre-term or very pre-term in either analysis. Conclusions Results from the ecologic study were comparable with results from an individual-level study of the effects of arsenic on LBW and SGA. This work was supported by the Centers for Disease Control and Prevention, National Environmental Public Health Tracking Program (contract number 200-2010-37353).