Community water fluoridation (CWF) has been a cornerstone of public health policy to prevent dental caries for nearly seven decades. Its effectiveness in reducing tooth decay is well established, though its marginal benefits have reduced over time due to the widespread use of other fluoride-based preventive measures like dental sealants and toothpaste. In recent years, concerns have emerged about potential unintended adverse effects of fluoride exposure, specifically during prenatal and early life periods. Much of this debate has focused on neurodevelopmental outcomes, including intelligence quotient (IQ), but the existing evidence largely derives from observational studies that are vulnerable to confounding and limited in their ability to support causal inference.
The aim of this study was to assess the link between prenatal exposure to CWF and adverse birth outcomes, with a primary focus on birth weight as an indicator of infant health and later-life outcomes. Secondary outcomes included low birth weight, gestational length, and prematurity. The study leveraged the staggered implementation of CWF across the United States to strengthen causal inference compared with prior cross-sectional analyses.
This cohort study used US population-level data spanning January 1968 to December 2988. Information on water fluoridation was obtained from the Centers for Disease Control and Prevention’s (CDC) 1992 Water Fluoridation Census, which documents the timing, type, and population coverage of fluoridated public water systems. The month and year of fluoridation initiation were detected for each country using established methods. Country-level fluoridation exposure was estimated by calculating the proportion of the population served by fluoridated water systems, which assumes coverage remained constant over time, and constructing a country-month panel of CWF exposure.
Birth outcome data were obtained from the National Vital Statistics System Natality Detail Files, which include birth certificate microdata from all U.S. states and the District of Columbia: the analytic sample excluded multiple births, records with missing birth weight, and births lacking county identifiers. Birth outcomes were aggregated to the county–month–year level and merged with the fluoridation exposure data. The primary outcome was the mean county-level birth weight. Secondary outcomes involved the proportion of low-birth-weight births (<2500 g), gestational length (in weeks), and prematurity (<37 weeks of gestation). Covariates involved maternal demographic characteristics like age, education, marital status, parity, infant sex, and maternal care.
The primary analytical method was a staggered-entry difference-in-differences (DID) event study design, which compared within-county changes in birth outcomes before and after CWF adoption to changes in counties that never adopted fluoridation or had not yet adopted it. This design allowed assessment of pre-treatment trends to assess the parallel trends assumption and estimation of dynamic post-treatment effects. Traditional static DID models were estimated, defining three pretreatment periods as 1 year before CWF adoption and posttreatment periods as 9 to 21 months after adoption, corresponding to full prenatal exposure. Sensitivity analyses involved models with state-by-year fixed effects. Statistical significance was assessed by using two-sided tests with a threshold of P < 0.05.
The final analytic sample comprised 170,604 county-month observations from 677 counties, representing 11,479,922 singleton births. Of these counties, 408 (60.3%) adopted CWF during the study period, and 269 (39.7%) never adopted it. Mean (SD) birth weight was 3337.4 (172.8) g, and mean (SD) gestational age was 39.5 (0.8) weeks. Birth outcomes and maternal characteristics were similar between treated and never-treated countries.
Event study analyses showed no evidence of differential pretreatment trends in birth weight, which supports the internal validity of the DID design. Estimated effects on birth weight were small and statistically insignificant in all posttreatment periods after CWF adoption, which range from −8.44 g (95% confidence interval [CI]: −20.41 to 3.53 g) to 7.20 g (95% CI: −5.45 to 19.85 g). The overall DID estimate was −0.53 g (95% CI: −4.75 to 3.70; P = 0.81). CI in posttreatment estimates ranged from −21.2 g to 20.3 g, which indicates that even the most extreme plausible effects were below 1% of mean birth weight. Secondary outcomes similarly showed no association with CWF: low birth weight (DID estimate: 0.15, 95% CI: −2.26 to 2.57, P = 0.91), gestational length (−0.01 weeks, 95% CI: −0.04 to 0.02, P = 0.34), and prematurity (0.88, 95% CI: −3.45 to 5.41, P = 0.70). Results were robust to multiple sensitivity analyses, including high-coverage counties and models with state-specific time trends.
In conclusion, this large population-based cohort study found no evidence of an association between CWF and adverse birth outcomes like weight and gestational age. The estimated effects were small and statistically nonsignificant, supporting the safety of CWF for prenatal and infant health. By employing a rigorous event-study difference-in-differences design, this study minimizes confounding and underscores the importance of cautious interpretation of earlier research on fluoride exposure and health outcomes.
Reference: Krebs B, Simon L, Schwandt H, Burn S, Neidell M. Community Water Fluoridation and Birth Outcomes. JAMA Netw Open. 2026;9(1):e2554686. doi:10.1001/jamanetworkopen.2025.54686
