This type of study design can indicate whether a forest fire is burning but not whether there are smoldering embers in the forest causing other types of damage. In such a large population even a small effect could theoretically be identified but exposure assessment and misclassification confound the results. All things considered the study does not offer much evidence one way (effect) or the other (no effect). The assumptions made by the researchers going into the study define the resulting conclusion.
Interesting how “negative” studies on fluoride such as this are published while “negative” studies showing little or no benefits to communal fluoridation are not published. This is the classic publication bias whether intended or unintended.
Peggy Näsman, Fredrik Granath, Jan Ekstrand, Anders Ekbom, Gunilla Sandborgh-Englund, C. Michael Fored. Natural fluoride in drinking water and myocardial infarction: A cohort study in Sweden. Science of The Total Environment, Volume 562, 15 August 2016, Pages 305-311.
Large geographical variation in the coronary heart disease (CHD) incidence is seen worldwide and only a part of this difference is attributed to the classic risk factors. Several environmental factors, such as trace elements in the drinking water have been implicated in the pathogenesis of CHD. The objective was to assess the association between drinking water fluoride exposure and myocardial infarction in Sweden using nationwide registers. This large cohort consisted of 455,619 individuals, born in Sweden between January 1, 1900 and December 31, 1919, alive and living in their municipality of birth at the time of start of follow-up. Estimated individual drinking water fluoride exposure was stratified into four categories: very low (< 0.3 mg/l), low (0.3–<0.7 mg/l), medium (0.7–<1.5 mg/l) and high (≥ 1.5 mg/l). In Cox regression analyses, compared to the very low fluoride group, the adjusted Hazard Ratio for the low fluoride group was 0.99 (95% confidence interval, 0.98–1.00), for the medium fluoride group 1.01 (95% confidence interval, 0.99–1.03) and 0.98 (95% confidence interval, 0.96–1.01) for the highest fluoride group. Adding water hardness to the model did not change the results. We conclude that the investigated levels of natural drinking water fluoride content does not appear to be associated with myocardial infarction, nor related to the geographic myocardial infarction risk variation in Sweden. Potential misclassification of exposure and unmeasured confounding may have influenced the results.