Epistemic Status: back-of-the-envelope. I am about to present a bad cost-benefit analysis of fluoridation. However, something is better than nothing, and I have not seen anyone else do this.
Summary: Fluoridation causes more monetary harm than good if it causes an average IQ point loss of more than 0.2. According to a recent meta-analysis, the effect size from high-quality studies is 1 IQ point loss per 1 ppm fluoride. Fluoridation adds about .5 ppm. If this effect size is accurate and fluoride has a linear population-level dose-response function, then fluoridation would lower IQ by 0.5 points. If you place more than 40% credence on this, and/or if you think that the non-monetary harms of IQ loss are larger than for caries, then you should oppose fluoridation.
Background: I used to think that fluoridation was mostly harmless, and that fears were overblown. But I have changed my mind based on new evidence appearing in the scientific literature. I agree that this new evidence has problems, but it shows that there is real uncertainty and risk.
If we lived in a world without fluoridation, then under the precautionary principle, we would not be starting it now. Everyone agrees that high amounts of fluoride (>2 ppm) are harmful, nobody knows what the safe threshold actually is, and we do not have enough evidence to rule out a linear-no-threshold dose-response function.
But I prefer to do public policy based on cost-benefit analysis rather than the precautionary principle. The question then becomes how to weigh the possible IQ loss against the known benefits of fluoridation.
Breakeven Analysis: When the situation is uncertain, it helps to conduct a breakeven analysis. So I calculate how much IQ loss would cancel the known benefits of fluoridation.
The CDC says that fluoridation has an economic value of $32 per person per year, and other sources agree. So over your lifetime, you will get about $2500 of money benefit from fluoridation. The monetary value of one IQ point is about $12,000, which is about 5x the benefit of fluoridation. So if fluoridation costs the population 1/5th of an IQ point, it does more economic harm than good.
In reality, money is a small part of the equation. Cavities cause pain and suffering, and the DALY loss for lower intellectual function is significant. But if the ratio of non-monetary to monetary harms are the same for both conditions, then the breakeven will be the same.
Personally, I think that the non-monetary harm from IQ loss is significantly larger than the non-monetary harm from caries, such that the total harm from losing 0.1 IQ point is roughly equal to the total benefits of fluoridation reducing cavities.
Risk of Harm by Fluoride Dose: On an individual level, the dose-response function is rarely linear. There is usually a non-zero point where harm is minimized, and the marginal harm per unit usually increases as the dose (or the background exposure before the dose) increases.
However, we are not dosing an individual. We are dosing a population. Most people have significant background exposure to fluoride from other sources. If a significant number of people already have a baseline exposure to fluoride greater than the harm-minimizing point, then any amount of fluoride added to the water will harm them. We know that at least single-digit millions of people are above this level, and it is possible that many more are.
The population-level dose-response function is based on how many people are already beyond the harm-minimizing point, and how far they are beyond it. At low levels, it only harms vulnerable sub-populations, but eventually it harms everyone.
We do not know, based on the data we have, what the population-level dose-response function is between IQ and fluoride in drinking water between 0.2 and 0.7 ppm in a developed country. But here's what we do know from the meta-analysis (Table 2):
1) There is strong evidence that, above 1.5 ppm, 1 additional ppm fluoride in drinking water will lower population-average IQ by about a point.
2) There is weaker evidence that the dose-response is roughly similar below 1.5 ppm, and not enough evidence to rule out a linear effect model.
Conclusion: Personally, I would assume that the dose-response at lower levels is about half what the paper estimates, meaning that fluoridation has caused an IQ loss of about 0.25 on average. I would further assume that most of the loss is concentrated in vulnerable subpopulations, with a median IQ loss of 0.
So I believe that fluoridation is probably beneficial for most people, and that it probably causes total harm to the population slightly larger than its benefits. Getting rid of it would slightly increase social welfare, and also decrease inequality.
No comments:
Post a Comment