A new study in JAMA Network Open (Malin, et al.) claims to have found an association between maternal urinary fluoride (MUF) levels and later deficiencies in children’s executive function. This study used urine “spot samples” to measure a pregnant woman’s overall exposure to fluoride.
Why Spot Samples Are Not Valid for this Study
Experts in assessing exposure to fluoride have determined that MUF can be a valid way to understand exposure on a community level, but not for an individual. Why aren’t spot samples a valid measure of an individual’s overall fluoride exposure? Fluoride levels vary throughout the day — and from day to day, too. If fluoride concentration in urine is measured at only one point in the day, it is just that – a measure of fluoride level at that moment in the day.
Most of our exposure to fluoride comes from food, beverages and dental products. About 50% is excreted from the body within 24 hours of consumption. Say the urine was collected less than 24 hours after someone consumed foods that are naturally high in fluoride. An MUF level would likely be higher than the level usually found in urine. Likewise, if the previous days’ diet was low in these foods and beverages, the level would probably be lower than usual. In either scenario, the information from a spot sample is limited. When medical tests of urine are used for an individual’s results, they show “if a person has been exposed recently to higher-than-normal levels of fluorides.”
Convenience Samples
Urine samples are routinely collected during pregnancy. Conveniently available data are common sources of information for studies like this one, hence the term “convenience sample”. It may seem intuitive that MUF is a sound way to measure a pregnant woman’s exposure to fluoride. It is not. To associate a subsequent outcome in children with an exposure during pregnancy, the all contributing factors must be properly measured. The scientific community is clear on this. A convenience sample cannot be used to infer statistical significance that applies to the general population. Any recommended changes to medical practice or policy should be made based upon statistically significant results.
Strong Evidence of No Association
In 2023, a study of children in Australia (Do, et al.) also explored the question of whether fluoride and children’s executive function were linked. It found no such association. The Australian study measured fluoride exposure using the status of water fluoridation rather than spot samples. The authors also analyzed a much larger data set, one that included 2,682 children. That’s >90% more children than studied by Malin et al, who relied on a sample of only 229 mother-child pairs. Those pairs were from one ethnic group and came from a single city. They were not, therefore, representative of the general population of children. That means that the study’s results describe only a single and small demographic. The findings of the Australian study are more generalizable to a large and diverse population like the U.S. This is important. When public policy recommendations result from research, the data must describe the public.
What comes next?
More research, properly designed, is needed on this important question in order to develop a robust body of evidence. Only then can clinical recommendations and public policy be evidence-based, consistent and sound. As always, we welcome new research to broaden our understanding on the safety of fluoride when the data are valid, sufficient and purposefully derived.