Prevalence of osteoporosis in high- and low-fluoride areas in North Dakota: The Bernstein Study.

In 1966, a study was published in the Journal of the American Medical Association (Bernstein et al, 1966) reporting that high levels of fluoride in the water reduced the incidence of osteoporosis. The study was conducted in two rural areas of North Dakota and was taken as conclusive evidence by some in the medical community that osteoporosis could be prevented, in part, by fluorides.

Included below is the abstract of this 1966 study followed by a discussion of major problems found with the study's methodology and findings.

According to Dr. B.E.C. Nordin, then head of the British Medical Research Council's metabolism unit, "the American (Bernstein) study was extremely inaccurate and did not use any objective quantitative method." (quoted in Medical News (London), September 26, 1969).


Bernstein DS, et al. (1966). Prevalence of osteoporosis in high- and low-fluoride areas in North Dakota. JAMA. 198: 499-504.

ABSTRACT: Roentgenograms of the lateral lumbar area of the spine were obtained from 1,015 subjects over age 45 who were residing in North Dakota. Three hundred lived in an area where the fluoride content of the water supply was high, 4 to 5.8 ppm, and 715 in an area where it was low, 0.15 to 0.3 ppm. Evidence of osteoporosis, reduced bone density, and collapsed vertebrae was substantially higher in the low-fluoride area, particularly in men. Limited information on milk and cheese consumption does not indicate that differences in calcium consumption are a significant factor. Fluoride consumption is important in the prevention of osteoporosis and may also play a significant role in preventing calcification of the aorta.


The Lancet, February 25, 1967


SIR, - In your annotation (Jan 28, p. 203) the preventive action of fluorides on osteoporosis is said to be confirmed by the recent findings of Bernstein et al. While it is quite possible that fluorides may have such an effect, the cited article has several deficiencies which in my opinion are severe enough to cast doubt on the validity of the results.

Of the factors known to influence the development and severity of involutional osteoporosis, age is of prime importance. In the paper under scrutiny, the material has been divided into age-groups without specifying the average ages within the groups. For statistical evaluation between the groups, this omission is particularly serious for the age group of 65 + years, because of possible wide age-variations upward, and especially because the high-fluoride group included only 38 male and 42 female subjects. A difference in the mean age of as little as 5 years between the high-fluoride and low-fluoride groups might be responsible for the entire statistical significance in differences of bone density and of percentage of compressed vertebrae between the two groups.

Secondly, it is most unusual and contrary to all previous experience that male populations should have a higher rate of severe osteoporosis than females. The extraordinary incidence of spinal compressions in males over 65 years old of about 50% indicates that the given study-population in general does not confirm to populations elsewhere as to the incidence of osteoporosis. Therefore, some other unknown (and unusual) factors may have been influencing the results. It is somewhat paradoxical that prevention of osteoporosis by administration of fluorides is suggested by a study in which men had more severe osteoporosis than women, and yet no difference was found in the prevalence of osteoporosis between the male subjects in the high and low fluoride areas.

These criticisms are not so much directed against Bernstein et al., who have made a good effort to solve a difficult problem, as against the too easy acceptance of early or preliminary results.

The Toronto Western Hospital,
Toronto, Canada


Other problems with Bernstein study

The fluoride content of the high and low fluoride counties were based on the concentrations of 2 to 3 towns from each county.

For the low fluoride county, Bernstein used the fluoride concentration of the towns of Grafton, Carrington, and New Rockford. The fluoride concentration range given for these towns was 0.15 to 0.30 ppm.

The town of Grafton was stated as having 0.3 ppm fluoride in its water. However, according to a 1959 Public Health Service publication (No. 655), Grafton is listed as having 0.9 ppm, while according to 1954 government records Grafton is listed as having 2.8 ppm, with 1938 records listing it as having 3.6 ppm.

In addition, the reference which the authors cite (North Dakota Geological Survey) for the water fluoride concentrations, showed that various wells in the low fluoride area had as much as 3 to 4 ppm.

Meanwhile, according to the same 1959 Public Health Service publication noted above, the concentrations of fluoride in the "high fluoride" area were listed as having less fluoride than the figures cited by Bernstein. According to the 1959 publication, the town of Hettinger had 2.2 ppm fluoride (versus the 4 to 4.8 ppm figure given by Bernstein), while the town of Mott had 1.2 ppm (versus the 5.5 to 5.8 figure given by Bernstein).

In addition to these discrepancies, it is noteworthy that Bernstein et al did not provide data on the mineral content (e.g. calcium and magnesium) of either area's water supplies, a factor which can influence both the toxicity of fluoride and the occurrence of osteoporosis.

According to Dr. B.E.C. Nordin, then prominent mineral metabolism researcher and head of the British Medical Research Council's metabolism unit, "the American (Bernstein) study was extremely inaccurate and did not use any objective quantitative method." (quoted in Medical News (London), September 26, 1969).

Nordin made these comments in the wake of a British study which found "no evidence that fluoride in non-toxic levels has any effect in preventing osteoporosis."


Return to Fluoride Bibliography