The Well Runs Deep: Examining the Latest Fluoride Research

Why Recent Studies Don't Support the Anti-Fluoridation Wave

The story of public health is one of hard-won victories. For over 75 years, community water fluoridation has stood as one of its greatest achievements - a simple, cost-effective intervention that has dramatically reduced tooth decay across all socioeconomic levels. But this cornerstone of public health now faces an unprecedented challenge…

We are witnessing a concerning trend across the United States: municipalities are removing fluoride from their drinking water based on misinterpretations of recent research. In Florida alone, multiple communities have voted to end water fluoridation following a vague court decision and the release of two widely misunderstood studies - a National Toxicology Program (NTP) report and a new meta-analysis published in JAMA Pediatrics. These decisions, driven by misapplied science, threaten one of public health's most successful interventions.

And, perhaps most notably, Utah is on the verge of becoming the first U.S. state to completely ban fluoride in public water systems. The unprecedented legislation, which has cleared its final hurdle in the state legislature and now awaits Governor Spencer Cox's signature, would strip cities and communities of their ability to decide whether to add the cavity-preventing mineral to local water supplies. Sponsors frame the legislation as "pro-informed consent and individual choice" rather than anti-fluoride.

The stakes extend far beyond dental health. As communities rush to make decisions based on misunderstood research, they risk creating a two-tiered system where access to fluoride's benefits becomes yet another marker of health inequality. To understand why these decisions are misguided - and potentially dangerous - we need to examine what the latest research actually says, and more importantly, what it doesn't. Let’s discuss…

Understanding the Meta-Analysis: A Fundamental Disconnect

The latest catalyst for this troubling movement is a meta-analysis examining fluoride exposure and children's IQ scores. While appearing comprehensive with 74 studies and over 20,000 children, a careful examination reveals critical limitations that make its findings largely irrelevant to community water fluoridation in the United States.

The fundamental disconnect lies in exposure levels. The meta-analysis primarily examines fluoride concentrations at or above 1.5 mg/L - more than double the U.S. recommended level for community water fluoridation (0.7 mg/L). In fact, only about 0.6% of the U.S. population is exposed to such high naturally occurring fluoride levels. Drawing conclusions about optimal water fluoridation from studies of substantially higher exposure levels represents a dangerous misapplication of evidence.

Garbage In, Garbage Out: A Data Science Perspective

A meta-analysis is only as good as the studies that comprise it. If those are flawed, so will the meta-analysis. Upon reviewing this particular meta-analysis, several critical methodological red flags that deserve careful examination. Let's break down the key issues that undermine its applicability to U.S. water fluoridation policy.

Study Quality and Composition Of the 74 studies included, an overwhelming 64 were cross-sectional studies, with only 10 being cohort studies. This heavy reliance on cross-sectional data fundamentally limits causal inference. More concerning, 70% of included studies (52/74) were rated as having a high risk of bias, yet were still included in the primary analyses. The geographic distribution is particularly problematic: 45 of 74 studies were conducted in China, with no U.S. studies included. This skewed distribution severely limits generalizability to U.S. populations.

Statistical Analysis: A close look at the analysis shows extreme heterogeneity (I² = 94%), or variation across study results. This level of variability cannot be attributed to chance alone and also makes makes pooled results highly questionable.

Perhaps most troubling is the publication bias correction through trim-and-fill analysis. A trim-and-fill analysis is a method used to check if there’s publication bias—which happens when studies with certain results (often positive ones) are more likely to be published than others. Their correction produced a striking asymmetry: 2 studies imputed or estimated missing studies that should exist (if there were no bias) on the right side versus 17 on the left. This pattern violates the fundamental assumption of trim-and-fill analysis that missing studies should be roughly symmetrical. More concerning, the correction actually strengthened the inverse association (SMD -0.63), contradicting the expected pattern where publication bias typically inflates effect sizes. This suggests the correction itself may be artificially exaggerating the relationship.

Dose-Response Relationship: Biological Implausibility The dose-response findings lack coherence with established toxicological principles. Consider the Yu et al. 2018 study, which found effects only between 1.60-2.50 mg/L but not at lower or higher levels. This pattern defies basic dose-response expectations where higher exposures should produce stronger effects. The lack of significant association below 1.5 mg/L - the level most relevant to U.S. policy - further undermines the study's relevance to community water fluoridation.

Individual Study Problems Several individual studies included in the meta-analysis have significant methodological flaws:

• Cui et al. 2020 found no significant fluoride-IQ association

• Trivedi et al. 2012 employed an incorrect statistical test, using paired-sample t-test for independent groups

• Khan et al. 2015 show the largest effect size but require independent validation

Let’s dig into some of these studies…

Critical Analysis of Key Fluoride-IQ Studies

1. Khan et al. (2015)

Despite reporting the largest effect size in the literature (a difference of nearly 15 IQ points between exposure groups), this cross-sectional study from Lucknow, India has significant flaws that undermine its dramatic findings. The study's design cannot establish causation, and the researchers failed to control for essential variables like socioeconomic status, parental education, and other environmental exposures. With only 429 children and measurements taken at a single time point, the study cannot account for historical exposure variations or demonstrate a clear causal relationship.

In simpler terms: While this study found the biggest difference in IQ scores between high and low fluoride groups, its poor design and lack of control for other important factors make its findings questionable.

2. Trivedi et al. (2012)

A fundamental statistical error invalidates this study's conclusions. The researchers inappropriately used a paired-sample t-test to analyze independent groups (children from high and low fluoride villages). This basic methodological error, combined with a small sample size of just 93 children and inadequate control of confounding variables, makes the reported IQ differences unreliable. The incorrect statistical approach raises serious questions about the overall quality of the study's analysis.

3. Yu et al. (2018)

The findings from this study challenge basic toxicological principles. The researchers reported a peculiar fluoride-IQ relationship that only appeared within a narrow exposure window (1.60-2.50 mg/L), with no effects at either lower or higher concentrations. This non-monotonic relationship contradicts fundamental dose-response principles and suggests either methodological flaws or the presence of uncontrolled confounding variables. The lack of a clear dose-response relationship seriously undermines the study's conclusions about fluoride's neurotoxic effects.

TL;DR? This study found that fluoride only affected IQ within a specific range, not at higher or lower levels. This unusual pattern doesn't make biological sense and suggests problems with the study's methods.

4. Cui et al. (2020)

Often cited but frequently misrepresented, this study actually found no significant association between fluoride exposure and IQ when properly analyzed. The research examined multiple variables including thyroid-stimulating hormone and dopamine levels, making it difficult to isolate any specific effect of fluoride exposure. The nonsignificant findings are sometimes overlooked in broader discussions of fluoride's cognitive effects.

When analyzed correctly, this study found no real connection between fluoride exposure and IQ scores, though this fact is often ignored when the study is cited.

5. Rocha-Amador et al. (2007)

This Mexican study stands out for several methodological strengths in examining both fluoride and arsenic exposure. The researchers measured both water and urinary concentrations of fluoride and arsenic, allowing better exposure assessment than many other studies. They found significant inverse associations between fluoride in urine and IQ scores (β values of -13.0, -15.6, and -16.9 for Performance, Verbal, and Full IQ respectively, all p<0.001), controlling for key confounders including blood lead levels, socioeconomic status, maternal education, height-for-age, and transferrin saturation. However, the study's primary limitation is the high correlation between fluoride and arsenic exposure (r=0.86), making it impossible to fully separate their effects. The study also highlighted exposure assessment challenges, noting that 53% of people in the highest-exposure area reported using bottled water for drinking but not cooking.

In plain terms: This study was well-designed and controlled for many important factors, but because fluoride and arsenic levels were so closely linked in the water supply, it's impossible to tell which substance might be affecting IQ scores.

6. Seraj et al. (2012)

This Iranian study had some methodological strengths, though some limitations remain. The researchers carefully controlled for iodine status - a key potential confounder - by measuring water iodine levels and confirming all households received iodine-enriched salt. They found modest but statistically significant IQ differences between groups (97.77±18.91 for normal fluoride vs 89.03±12.99 for medium and 88.58±16.01 for high fluoride areas), with a decrease in mean IQ scores as fluoride exposure increased. The study's relatively large sample size (293 children), careful documentation of residential history, and verification of exposure through dental fluorosis patterns strengthen its findings. However, several limitations warrant consideration: the lack of significant IQ differences between medium (3.1±0.9 ppm) and high (5.2±1.1 ppm) fluoride groups raises questions about dose-response relationships, individual-level socioeconomic data was limited, and like other cross-sectional studies, it cannot establish causation.

While this study was better designed than many others - with a good sample size and careful attention to important factors like iodine intake - it still can't prove that fluoride caused lower IQ scores. The finding that higher fluoride levels didn't lead to even lower IQ scores also raises some questions about the results.

7. Poureslami et al. (2011)

This Iranian study reported modest IQ differences between high and low fluoride groups (91.37 vs. 97.80), but these differences fall within the standard error range for most IQ tests. The small sample size (120 children) and failure to account for iodine status - particularly relevant given Iran's history of iodine deficiency - limit the study's reliability. The reported effect size is notably smaller than Khan et al., despite similar study conditions.

This study found small differences in IQ scores that might simply be due to chance, given the small number of children studied and failure to consider other important factors like iodine deficiency.

8. Eswar et al. (2011)

Though often cited as evidence for fluoride's neurotoxicity, this study from Karnataka, India actually found no statistically significant difference in IQ scores between high and low fluoride villages. The authors themselves acknowledged these limitations, though subsequent citations often omit this important detail. The null findings provide an important counterpoint to studies reporting large effects.

This study found no real difference in IQ scores between children from high and low fluoride areas, though it's often incorrectly cited as showing the opposite.

9. Sudhir et al. (2009)

While this study boasts a larger sample size (1000 children), its cross-sectional design and inadequate control of confounding variables limit its utility. The reported IQ differences might be explained by underlying socioeconomic and educational disparities between the study regions rather than fluoride exposure. The effect size was moderate compared to other studies in the literature.

Despite looking at many children, this study's poor design and failure to account for social and economic differences between communities make its findings questionable.

10. Wang et al. (1996)

This early Chinese study used outdated testing methods and reported relatively modest effects. Its relevance to contemporary water fluoridation practices is limited, given the significant differences in exposure levels and overall environmental conditions. The study's age and methodological limitations make it less reliable for current policy decisions.

This older study used outdated methods and looked at very different conditions than we see in modern water systems, making it less useful for current policy decisions.

Methodological Concerns Common Across Studies:

1. Reliance on cross-sectional designs that cannot establish causation

2. Insufficient control of confounding variables

3. Limited sample sizes in many studies

4. Inconsistent IQ testing methods

5. Poor measurement of total fluoride exposure from all sources

6. Inadequate consideration of other environmental neurotoxins

7. Limited accounting for socioeconomic factors

8. Questionable applicability to water fluoridation programs in developed countries

Effect Size in Context
The reported decrease of 1.63 IQ points per 1-mg/L increase in urinary fluoride requires careful interpretation. This small effect size falls within the standard measurement error for IQ tests. Moreover, studies rated as low risk of bias showed even weaker associations. When considering measurement error and the multiple confounding variables inadequately controlled for, the clinical significance of these small differences becomes highly questionable.

Missing Controls and Measurement Issues
A critical weakness of many included studies is their poor adjustment for confounding variables. Socioeconomic status, a well-known predictor of cognitive outcomes, was inadequately controlled across studies. Environmental co-exposures, particularly relevant in studies from regions with naturally high fluoride levels, were often not considered. This is especially problematic for the Chinese studies, where industrial pollution and other environmental contaminants could confound results.

The reliance on urinary fluoride as an exposure measure introduces additional uncertainty. Urinary fluoride represents short-term exposure and can vary significantly based on recent intake, hydration status, and other factors. It's not a reliable indicator of long-term fluoride exposure, which would be more relevant for cognitive development outcomes. The authors acknowledge that an individual's total fluoride exposure comes from multiple sources - water, food, and beverages - yet most studies failed to adequately account for this total exposure.

The Global Context: Addressing the "Europe Doesn't Fluoridate" Myth

unbiasedscipod

A post shared by @unbiasedscipod

A common argument against water fluoridation is that "97% of Europe is not fluoridated." However, this oversimplified claim misrepresents how European countries actually approach fluoride delivery:

Multiple Fluoride Delivery Systems

• Over 13 million Europeans receive fluoridated water

• Fluoridated salt reaches more than 70 million Europeans in countries including Germany, Switzerland, France, and Austria

• Many countries use school-based fluoride programs and milk fluoridation

• Some regions have naturally fluoridated water meeting optimal levels

European Infrastructure Context

• Many European countries choose alternative delivery systems due to logistical challenges with water systems

• Complex terrain and thousands of separate water sources make centralized water fluoridation difficult

• Cultural preferences (e.g., Italy's preference for bottled water) influence delivery methods

• These approaches complement rather than conflict with fluoridation programs

Universal Coverage Through Multiple Approaches European countries, especially in Scandinavia, ensure fluoride access through:

• Universal public dental coverage (e.g., Sweden provides free dental care through age 23)

• Regular fluoride treatments, including varnish applications and rinses

• Comprehensive school-based prevention programs

Public Health Implications

The timing of this meta-analysis is particularly concerning as communities across the United States face pressure to remove fluoride from their water supplies. In Florida alone, multiple municipalities have recently voted to end water fluoridation, citing this research and the NTP report as justification. These decisions overlook several critical factors:

Risk-Benefit Analysis

• Dental caries remain a significant public health concern

• Water fluoridation prevents at least 25% of tooth decay in children and adults

• Every $1 spent on fluoridation saves approximately $20 in dental procedures

• Poor dental health itself can impact cognitive development through chronic pain, inflammation, and nutritional deficiencies

Health Equity

Removing fluoride from water will impact historically underserved communities in multiple ways:

• Water fluoridation provides universal protection regardless of socioeconomic status

• Communities that have ceased fluoridation show increased social inequities in dental health

• The impact falls disproportionately on children from disadvantaged families

(We’ve written about this extensively and even generated data visualizations to help illustrate the health equity impact of water fluoridation.)

Conclusion

While ongoing research into fluoride safety is important, this meta-analysis does not justify changes to current water fluoridation practices. The methodological limitations, small effect sizes, and lack of significant findings at relevant concentrations provide no evidence that current U.S. fluoridation practices pose cognitive risks. The misuse of this research to argue against optimal water fluoridation represents a serious threat to public health that must be addressed through clear, evidence-based communication and policy decisions.

Stay Curious,

Unbiased Science & Nini and the Brain (Teching It Apart)

P.S. Want to support this kind of analysis? The best way is to subscribe to our Substack and share our work. While all our articles are always completely free to read, paid subscriptions help sustain our in-depth reporting on vaccine policy and public health. Thank you for considering it!