Fluoride in drinking water, yea or nay?

The effects of fluoride on human health has long been a matter of controversy. In small quantities, fluoride has beneficial effects on oral health, especially in preventing dental caries in children, which is a major public health concern in the industrialized countries as well as others. Therefore, the Oral Health program of the WHO supports fluoride containing toothpastes and mouth-rinses, as well as fluoridation of drinking water where necessary. As indicated in a 2004 report, systematic reviews of available datasets concluded that there was no credible evidence of adverse health effects of water fluoridation, and WHO Water Quality Guidelines recommended a maximum fluoride level of 1.5mg/L.

Fluoride, well-absorbed in the GI tract (70-100% absorption depending on chemical nature of the fluoride salt), is readily incorporated into calcified tissues, such as bone and teeth, substituting for hydroxyls in crystals of the naturally occurring mineral, hydroxyapatite. Fluoride exchanges between body fluids and bone, both at the surface layer of bone (a short-term process) and in areas undergoing bone remodeling (a longer-term process). Bone contains approximately 99% of the body’s fluoride. Exposure to fluoride may be via drinking water (he largest source of exposure in the US), food (including beverages), dental products, pesticides, as well as fumes from burning coal.

However, as with most things in nature, an excess is bad. Prolonged and excessive exposure to fluoride causes a condition called fluorosis, which has detrimental effects on both teeth (enamel fluorosis) and bone (skeletal fluorosis). Enamel fluorosis compromises the protective function of the enamel (on dentin and pulp) by causing structural damage to the tooth. In skeletal fluorosis, chemical adsorption of fluoride onto growing bone crystals, which accumulates over time in case of prolonged exposure to high concentrations, may increase bone density and makes bone-forming cells (osteophytes) grow faster, resulting in painful and stiff joints, eventually leading to mobility loss. In addition, high exposures appear to elevate the risk of certain fractures, corroborated by animal studies which indicates that, although fluoride increases bone volume, there is less strength per unit volume. This is a particular public health problem in areas of the world where the populations have natural exposures to high levels of fluoride. There is a growing body of research that contends that prolonged exposure to high concentrations of fluoride may have severe adverse effects on human health, especially in growing children, giving rise to neurobehavioral changes including a lowering of intelligence and other cognitive functions.

Under the Safe Drinking Water Act (1974) of the United States, the Environmental Protection Agency (EPA) establishes the acceptable limits of substances (including those with possible adverse effect on human health; ‘contaminants’) in public drinking water. These limits include (a) the maximum contaminant level goal (MCLG; safe levels for primary use), (b) the maximum contaminant level (MCL; practically achievable levels, to be set as close as possible to MCLG), and (c) the secondary maximum contaminant level (SMCL; specialized guidelines for other uses).

For Fluoride, the EPA-established limits are: MCLG, MCL = 4mg/L; SMCL = 2mg/L – the total amount of fluoride allowed in drinking water. [NOTE: this is separate from the acceptable levels recommended for prevention of dental caries (0.7-1.2mg/L).]

With a view to ensuring continued protection of public health, the Safe Drinking Water Act requires periodic scientific assessment of the limits of contaminants. At EPA’s request, the US National Research Council (NRC) of the National Academies convened a Committee on Fluoride in Drinking Water in 2006, to evaluate independently the appropriateness and adequacy of the EPA established MCLG and SMCL levels for Fluoride, considering the total exposure from various sources.

The NRC committee, in its 2006 report on Fluoride in drinking water, reviewed fluoride research since early 1990s, gathering information on levels of exposure, pharmacokinetics (how the body handles the fluoride), adverse effects on various organ systems, and genotoxic and carcinogenic potential. Possible human health risks were evaluated by considering the collective evidence from in vitro assays, animal research, human studies, and mechanistic information.

The principal problem with this meta-analysis by the NRC committee was the paucity of good quality studies with consistent and conclusive outcomes, for all the effects they investigated, namely, (a) dental, (b) musculoskeletal, (c) reproductive and developmental, (d) neurotoxicity and neurobehavioral, (e) endocrine, (f) gastrointestinal, (g) renal, (h) hepatic, (i) immune System effects, as well as (j) genotoxicity and carcinogenicity. Amongst published studies, available evidence on many occasions did not bear out the universality of isolated observations.

• Based on existing studies, the committee couldn’t reach unanimous consensus whether enamel fluorosis amounted to a serious health hazard, at least in the US; they found very few case-control studies that used a fluorosis risk index, and the accuracy of diagnosis was also in question. In addition, although focused on children, the literature appeared to be inconsistent in determining a specific age of risk, or exact nature of the hazard. Studies on frequency of dental caries in US and elsewhere within populations exposed to 2-4mg/L or more of fluoride were few and inconclusive.
• Epidemiological and observational studies appear to conclude that lifetime fluoride exposure in drinking water at 4mg/L or greater higher may increase fracture rates in the population. However, the 4mg/L level could not be causally associated with increased frequency of bone fracture; the evidence was inconclusive at around 1mg/L. Animal models, with subtly different skeletal physiologies (except rabbits which closely approximate humans, bone-wise), have not been conclusive as to the exact nature of fluoride’s effect on bone strength. It was unclear if fixing MCLG at 4mg/L would help protect against mobility issues of skeletal fluorosis; at prolonged 4mg/L exposure, the bone fluoride levels may reach or exceed those seen in stage II or III skeletal fluorosis, but a causal effect couldn’t be established because levels at which skeletal fluorosis occurs vary widely, and stage III condition is rare in the US. Although suspected from some veterinary observations, fluoride exposure could not be causally implicated in development of osteoarthritis in humans.
• Despite occasional isolated reports, GI changes or symptoms could not be causally attributed to water fluoridation. In animal experiments, significant GI effects of fluoride, involving hyperacidity, vasculopathy, reduced GI perfusion and inflammation of the gastric epithelium, were observed at abnormally high fluoride concentrations (100-1000 times the levels attainable by drinking fluoridated water). Skeletal fluorosis is suspected to be associated with urolithiasis, but there is a general lack of North American data on relative incidences; studies in other parts of the world have often ignored confounding factors, which limits their utility. Human populations living in regions of endemic fluorosis, as well as people with acute accidental/procedural exposure high fluoride doses may have some risk of developing adverse renal effects in a dose-dependent manner, but observational and epidemiological studies have not been conclusive. Effects of fluoride on liver, which performs defluorination in vivo, have not been quantified either; long-term daily ingestion of fluoridated drinking water (4 mg/L) has not been shown to cause any significant derangement of the hepatic enzymes.
• Effect of fluoride on cellular immunity has not been established. Concentrations at which immune cells are exposed to fluoride may be different from what organs see in case of chronic exposure. Mechanistically, fluoride has the ability to affect immune cells in vitro, but whether consumption of fluoridated water at 4mg/L compromises immunity is unknown; there is no epidemiological data.
• Animal studies on putative reproductive and developmental effects of fluoride appear to indicate that adverse outcomes may occur only at very high concentrations (~250mg/L). Human studies of fluoride’s impact on reproductive hormones, fertility, and Down’s syndrome have been either inconclusive, or useless because of significant limitations in design and power.
• Fluoride exposure may affect normal endocrine function/response, but the individual effects are highly variable. Available studies of endocrine effects of fluoride suffer from several limitations in design and execution, including unaccounted-for confounding factors.
• Human and animal studies on the putative effect of fluoride in promoting cancers of the bone (specially, osteoma and osteosarcoma), as well as its genotoxicity, have been ambiguous, inconsistent, and generally negative. Human studies often have had serious deficits in design, recruitment, execution and/or reporting, thereby severely limiting their utilities in reaching a conclusion.

In addition to the above findings, the 2006 committee summarized its findings on putative neurobehavioral and neurotoxic effects of fluoride exposure thusly:

Animal and human studies of fluoride have been published reporting adverse cognitive and behavioral effects. A few epidemiologic studies of Chinese populations have reported IQ deficits in children exposed to fluoride at 2.5 to 4 mg/L in drinking water. Although the studies lacked sufficient detail for the committee to fully assess their quality and relevance to US populations, the consistency of the results appears significant enough to warrant additional research on the effects of fluoride on intelligence.

A few animal studies have reported alterations in the behavior of rodents after treatment with fluoride, but the committee did not find the changes to be substantial in magnitude. More compelling were studies on molecular, cellular, and anatomical changes in the nervous system found after fluoride exposure, suggesting that functional changes could occur. These changes might be subtle or seen only under certain physiological or environmental conditions. More research is needed to clarify the effect of fluoride on brain chemistry and function.

Drinking water in the US has its natural fluoride concentrations varying from 2-4mg/L or more. The 2006 committee’s review of 15-odd years of fluoride research literature did not yield a definitive case for accepting or rejecting the EPA-established MCLG of 4mg/L. However, the committee found that across adults with varying water intakes, high percentages of individuals were exposed to upto 4mg/L of fluoride, but more alarmingly, on a per-body-weight basis, infants and young children had approximately three to four times greater daily exposure than do adults. Therefore, the committee concluded that fluoride exposure at MCLG of 4mg/L may put children, especially those in the 6-8 years of life (when the enamel in the transitional or early maturation stage of development), at risk of developing severe enamel fluorosis, in addition to other health problems, and recommended that fluoride MCLG be lowered from 4mg/L.

Corroborating and extending the NRC Committee’s 2006 meta-analysis, Choi et al. (group of Phillipe Grandjean, professor of Environmental Health, Harvard School of Public Health) recently published a systematic review and meta-analysis (published online in Environmental Health Perspectives on July 20, 2012) of published studies further investigating the putative role of fluoride in causing neurobehavioral defects in children. Noting that vast tracts of China have fluoride-rich minerals on the surface leading to wide-scale contamination of the ground water that is used for drinking, and exposure in drinking water may reach 11.5mg/L, much higher compared to acceptable levels in the US, Choi et al. focused on that region of the world (along with two studies from Iran), and performed a meta analyses of 27 studies, with high and low/reference exposure groups, published over ~22 years (1980-2011).

Identifying studies of drinking water fluoride and neurodevelopmental outcomes in children from a a wide range of sources/databases (including PubMed/Medline, EMBASE, TOXNET, as well as the China National Knowledge Infrastructure (CNKI) database to identify studies published only in Chinese journals), they included high quality studies that contained high and reference fluoride exposure estimates, endpoints of IQ scores or other related cognitive function measures, and computed data on some mean outcome measure and associated measure of variance [95% confidence intervals/standard errors, and numbers of participants], and based the interpretations of statistical significance on α=0.05. They also undertook extensive statistical analysis to look for intra- and inter-study heterogeneities and to check for publication bias, and took note of confounding factors, such as co-exposures to arsenic, iodine, fluoride from coal burning and low quantities of lead.

In the included studies, general intelligence in the study populations was quantitatively measured using the Combined Raven’s Test – The Rural edition in China (CRT-RC; in 16 studies), Weschler Intelligence tests (3 studies), Binet IQ test (2 studies), Raven’s test (2 studies), Japan IQ test (2 studies), Chinese comparative intelligence test (1 study), and the mental work capacity index (1 study) – in order to estimate the possible effects of fluoride exposure on general intelligence. [NOTE: I shall leave the discussion of the validity of these tests to those who are qualified to do so.] This 2012 meta-analysis, as laid out in the article, found a 75-93% risk of low score in intelligence tests associated with high fluoride exposure, as well as an inverse association between high fluoride exposure and children’s intelligence, i.e. children who lived in areas with high fluoride exposure had lower IQ scores than those who lived in low exposure or control areas, although the authors couldn’t perform a formal dose-response analysis for lack of data.

However, looking at the distribution of the Standardized Mean Difference (SMD) statistic that Choi et al. reported (as a surrogate for the difference in IQ points between children from high-fluoride and low-fluoride areas), the average/overall SMD for all 27 studies was -0.56, i.e. the decrease of about a half IQ point, although the decreases in 26 studies were spread out from 1/10^th of a point to 0.95 points (laid out in the Figure 2). This seems to be a rather modest decrease in IQ of a supposedly vulnerable population, despite all the fluoride exposure.

Figure 2: Random-effect standardized weighted mean difference (SMD) estimates and 95% CIs of child’s intelligence score associated with high exposure to fluoride. Adapted from Choi et al., Environmental Health Perspectives, July 20, 2012.

The authors explain this odd observation saying:

“The estimated decrease in average IQ associated with fluoride exposure based on our analysis may seem small and may be within the measurement error of IQ testing. However, as research on other neurotoxicants has shown, a shift to the left of IQ distributions in a population will have substantial impacts, especially among those in the high and low ranges of the IQ distribution…”

To my mind, that impact – the possibility of fluoride-associated cognitive deficit – remains to be established conclusively. However, the most important caveat of this (and in fact, any) meta-analytical study, as the authors themselves point out, is that the analysis is only as good as the constituent studies, several of which – in this instance – had various limitations, not the least of which is that the actual exposure levels of the children are not known. Therefore, this study cannot be used to predict or set exposure limits. However, the strength of this study lies in the collated and analysed observational data which indicate that the effect of fluoride on human health is an area that is incompletely understood, and therefore, merits further and more detailed studies for elucidation – a point that was also made by the 2006 NRC Committee report, one that seems to have been largely ignored and/or forgotten.

What I would like to have is a few answers: (a) From any of the constituent studies, is there any evidence for lasting brain damage or irreversible cognitive impairment of children beyond 14 years? (b) Evaluation of ‘general intelligence’ is fine as a metric, but what kind of cognitive impairments are actually seen in the vulnerable population, exposed to fluorides, in China? And most importantly, (c) how are the Chinese data relevant to the US (different environmental conditions, different demographic)?

So the tl;dr version of my original question ‘yea or nay?’ is: I don’t know, although the currently available evidence, followed by the WHO and various national agencies, leans towards ‘yea’; however, as NRC said, the ‘nay’ needs to be checked out, too.

This brings me to how I got interested in this meta-analysis paper in the first place. Someone brought it to my attention that one Marge Dwyer had written a summary of Choi et al. in the HSPH News website; the summary is – perhaps because of necessity and space constraints – full of breathless assertions that prima facie sound rather alarmist, particularly in view of statements like:

Fluoride seems to fit in with lead, mercury, and other poisons that cause chemical brain drain,” Grandjean says.

One of those things is not like the other; Fluoride (a halogen salt) should not be lumped in the same class as lead and mercury (both metals). In the actual paper, Choi states, summarizing a 1982 observation by Grandjean, that “fluoride may be a developmental neurotoxicant that affects brain development…“, noting the concern that serum fluoride concentrations associated with high intakes from drinking-water may exceed 1mg/L, or 50µmol/L. Other known neurotoxicants, such as lead and methylmercury, can cause adverse effects (including neurodevelopmental damage) at low (>10nmol/L) blood levels, about a 1000-times lower than the above fluoride concentrations. However, it must be borne in mind that (a) serum fluoride concentrations may not accurately reflect tissue levels, and most importantly, (b) there is scant evidence that fluoride has any adverse effect in humans at levels below or at 1mg/L.

And ‘poisons’, as well as the gratuitous use of ‘toxic’ and ‘toxicity’ in the write-up by Dwyer, evoke a terrible imagery without offering any useful information. Besides, “Chemical brain drain”? Really?