What’s in the water? Long-run effects of fluoridation on health and economic self-sufficiency.*

Over 70% of publicly-supplied drinking water in the United States is fluoridated and the CDC has named community water fluoridation as one of the 10 greatest public health achievements of the 20th century (Gooch, 2020). Despite strong evidence that exposure to low levels of fluoride is an effective way to strengthen teeth, many individuals, communities, and industrialized countries oppose water fluoridation out of concern for potential negative health risks. Existing research has not shown conclusive evidence of negative health effects from low levels of fluoride exposure. However, concerns about the safety of fluoride are supported by a body of research showing that early childhood exposure to high doses of fluoride can cause a wide variety of health problems including weakened bones and joints, and cognitive impairment.² The highest safe level of fluoride exposure is unclear. In this paper, I investigate the impact of early childhood exposure to community water fluoridation on long-run health and labor market outcomes.

The impact of fluoride on health varies based on both the amount and timing of fluoride exposure. This is true for both the positive impacts on dental health as well as the potential negative side effects on teeth, bones, and cognitive function. While it is well established that fluoride exposure makes teeth more resistant to decay, recent controversy has focused on the role of fluoride as a neurotoxin. The meta-study by Choi et al. (2012) concluded that early exposure to high fluoride levels results in decreased cognitive functions equivalent to nearly one-half of a standard deviation in IQ scores. While the majority of reviewed studies focus on subjects with fluoride levels well above recommended levels, some find negative cognitive effects at relative low levels as well. In a follow-up meta-study incorporating more recent evidence, Grandjean (2019) concluded that the levels currently recommended for water fluoridation likely exceed safe exposure levels.

Although cognitive effects may result in negative long-run impacts, recent work has suggested that improvements in dental health play a more prominent role. Glied and Neidell (2010) provides the best evidence in the U.S. context, leveraging variation in the timing of community water fluoridation programs to estimate long-run wage effects in the National Longitudinal Survey of Youth – 1979 (NLSY79). Due to the narrow group of birth cohorts in their sample (1957–1964), they are unable to make within-county comparisons; instead, their results rely on the assumption that unobservable county characteristics that affect labor market outcomes are uncorrelated with fluoridation status. They find positive but insignificant wage effects in the full sample—driven by a statistically significant 4% increase in wages among women.^3,4 A more recent paper leverages natural variation in fluoride levels between water treatment plants in Sweden and finds positive effects of fluoride on labor force participation and income (Aggeborn and Öhman, 2021). The authors also estimate effects on cognitive ability and health, finding no effect on either. Their findings rest on the assumption that variation in the geological characteristics, and associated fluoride levels, of local water sources are exogenous to cognitive ability and labor market outcomes. While the Swedish data provide significant precision and measurement advantages over the NSLY79, fluoride exposure is relatively low; over 90% of Swedish observations were exposed to fluoride levels less than those typically added in the United States (0.8–1.2 mg/L). The impact of fluoridating water to these higher levels is unclear.

In this paper, following the pre-analysis plan in Roberts (2021), I provide the first large sample and quasi-experimental evidence of the long-run health and labor market effects of community water fluoridation programs. This data includes both respondents to the longform 2000 decennial census as well as American Community Survey respondents from 2001 to 2016 and allows for vast increases in precision, requires much weaker identifying assumptions via inclusion of birth-county fixed effects, and includes a broader set of outcomes, birth cohorts, and communities relative to previous work.⁵ I use a stacked difference-in-differences strategy that compares outcomes of county-birth-cohorts with exposure to fluoridated water to those without any, while controlling for county and year of birth. This stacked design is unbiased even in the presence of heterogenous treatment effects (Cengiz et al., 2019). While the decision to add fluoride to water is made at the local level in the US, the amount of fluoride added typically follows Center for Disease Control (CDC) guidelines. This research evaluates the impact of exposure to CDC approved levels of fluoride. I find that exposure to fluoridated water from age zero to five is associated with a 1.9 percent of a standard deviation decrease in adult economic self-sufficiency, 1.2 percent of a standard deviation decrease in physical ability and health, a 0.4 percentage point increase in likelihood of being incarcerated, a 1.0 percentage point decrease in military service, and a 1.5 percentage point decrease in high school graduation. To put some of these results in context, I compare them to Bailey et al. (2023) who estimated the beneficial effects of early childhood access to food stamps using a similar set of outcomes. Taking the causality of point estimates from both studies at face value, this comparison suggests that early childhood fluoride exposure has the potential to erase approximately two-thirds of the large improvements in self-sufficiency caused by early childhood utilization of food stamps. The net effect of fluoride is negative; even at levels previously thought to be safe, the tooth-strengthening effect of fluoride provides less benefit than fluoride’s corresponding negative impact on other determinants of health and economic self-sufficiency. While it is difficult to disentangle all the mechanisms at play, the observed decrease in high school graduation rates is consistent with negative cognitive effects.⁶

A gradual re-evaluation of water fluoridation policies is already underway. In 2015 the U.S. Public Health Service slightly reduced recommended fluoride levels to 0.7 mg/L and acknowledged the need for more research into the risks of low-level fluoride exposure (DHHS, 2015). In 2019, the American Dental Association issued a statement that reaffirmed their support of water fluoridation while also welcoming additional research into the potential negative cognitive effects (ADA, 2019). Despite the acknowledged need for more research, fluoride is still added to a majority of public water supplies in the U.S. and regulations for regions with naturally high levels of fluoride allow water to carry up to 4.0 mg/L, nearly six times the recommended water fluoridation level. The results of this study demonstrate the need to accelerate our re-evaluation of water fluoridation policies. The observed negative impacts of fluoride, at CDC approved levels, combined with widespread access to the enamel-strengthening benefits of fluoride through toothpaste and dental treatments provides a strong argument for ending the practice of water fluoridation and lowering the maximum levels of fluoride allowed by safe drinking water standards. If water fluoridation practices continue, more research is needed to determine the optimal level of fluoride such that the marginal benefits to dental health are not overwhelmed by negative health costs.