The review estimated the prevalence of fluorosis (mottled teeth) and fluorosis of aesthetic concern at around 48% and 12.5% when the fluoride concentration was 1.0 part per million,9
although the quality of the studies was low. The evidence was of insufficient quality to allow confident statements about other potential harms (such as cancer and bone fracture). The amount and quality of the available data on side effects were insufficient to rule out all but the biggest effects.
Small relative increases in risk are difficult to estimate reliably by epidemiological studies, even though lifetime exposure of the whole population may have large population effects. For example, an ecological study from Taiwan found a high incidence of bladder cancer in women in areas where natural fluoride content in water is high. The authors attributed the finding to chance because multiple comparisons were made.11
Testing the hypothesis that drinking fluoridated water increases the risk of bladder cancer would need to take account of errors in estimating total fluoride exposures; potential lack of variation in exposure; the probable long latency between exposure and outcome; the presence of strong confounders such as smoking and occupational exposures; and changes in diagnostic classification of bladder tumours. Therefore, a modest association between fluoridation and bladder cancer would be difficult to detect, both in communities and in individuals. This is of concern because a modest (for example, 20%) increase in risk of bladder cancer would mean about 2000 extra new cases a year if the entire UK population was exposed.
The methodological challenges of detecting harms of long term exposure to fluoridation are further illustrated by a case-control study on hip fracture in England.12
It reported “no increase” in risk associated with average lifetime exposure of ≥0.9 part per million fluoride in drinking water. Although exemplary in many other aspects, the study had less than 70% power to identify an odds ratio of 1.5 associated with exposure. If the odds ratio was only 1.2—which would mean more than 10
000 excess hip fractures a year in England if the population was so exposed—the study would have a less than one in five chance of detecting it.
Thus, evidence on the potential benefits and harms of adding fluoride to water is relatively poor. This is reflected in the recommendations of the Medical Research Council (MRC)13
and the Scottish Intercollegiate Guideline14
on preventing and managing dental decay in preschool children (box 3). We know of no subsequent evidence that reduces the uncertainty.
Box 3 Key recommendations for future research on water fluoridation
- “Studies are needed to provide estimates of the effects of water fluoridation on children aged 3-15 years against a background of widespread use of fluoride toothpaste, and to extend knowledge about the effect of water fluoridation by . . . (socio-economic status), taking into account potentially important effect modifiers such as sugar consumption and toothpaste usage”13
- “A robust evaluation of the benefits of water fluoridation, as well as the potential risks of fluorosis . . . should be a health priority”14
There is no such thing as absolute certainty on safety. While the quality of evidence on potential long term harms of fluoridated water may be no worse than that for some common clinical interventions, patients can weigh potential benefits and risks before agreeing to treatments. In the case of fluoridation, people should be aware of the limitations of evidence about its potential harms and that it would be almost impossible to detect small but important risks (especially for chronic conditions) after introducing fluoridation.