A critique of the analysis would be that a negative association was measured, based on the data. But the data included observations with higher levels of fluoride than what the regulators were allowing or recommended. And so yes while the trend line decreases and yes at high enough doses fluoride does cause negative outcomes of IQ, at recommended and allowed levels the harm is insignificant or unmeasurable. The negative trend line shows a negative only from the weight of these observations with higher amounts.

https://ehp.niehs.nih.gov/doi/10.1289/ehp655

Image From the article with data points included: Try to ignore the gray shaded lines resulting from mathematical analysis and consider if the data points were cut off after 1.5 mg/L where would the resulting trendline be? Would it be less negative than the current one, would it be a flat horizontal line?

It’s hard to tell I wish scientific journals required the actual data to be released for publishing so I could just through this in and premature under that scenario.

This study does show a negative effect however, which means it’s reasonable to desire the minimum amount that meets the level required for the health benefits they have measured. And it’s true that studies like this one caused US regulators to lower the recommended dose from 1.2 to 0.7 mg/L however even if recommended the allowable amount is still higher. It would be a good question to any water supplier why their fluoride level is higher than 0.7 I checked my local supplier in Ohio averages 0.91.