An ecologic study  compared fracture rates in 216 counties with natural fluoride levels greater than 0.7 ppm with rates in 95 counties with naturally low fluoride (less than 0.4 PPM) in the drinking water (Dose Ecology Study). Hip fracture ratios used as a denominator the hip fracture rates in low fluoride areas. Medicare data obtained from the Health Care Financing Administration, DHHS, was used to obtain hospitalization rates for upper femur and lower spine fractures in men and women over age 65 during 1985 and 1986. The natural water fluoride levels were obtained from 1969 county estimates. No allowance was made for other sources of fluoride. The demographic characteristics were based on 1975 data.
Table 1 summarizes some of the key data from a preliminary analysis from the Dose Ecology Study. In general, with increasing dose of fluoride in the drinking water the hip fracture ratio also increased. When corrected for the expected lower incidence of hip fractures in blacks as compared with whites, however, there was no significant difference in the hip fracture ratio for populations at fluoride levels considered optimal for dental caries prevention (approximately 1 PPM). Hospitalization for spine fracture generally decreased with increasing fluoride levels. Because most spine fractures in the population aged over 65 (osteoporotic crush fractures of the vertebrae) do not result in hospitalization, the precise interpretation of these data is not clear.
|Table 1. Dose Ecology Study: standardized fracture ratios by low natural fluoride|
|Ratio of observed to expected||
Average fluoride levels in parts per million
|* – Probability that ratio = 1.000 is less than 0.01
a – Adjustment for black population in each county assumes fracture rates for blacks are one half those for whites.
Another ecologic study  considered the percentage of residents of specific counties who received fluoridated water (Exposure Ecologic Study). The 1985 Fluoridation Census data were used for the 438 counties with populations over 100,000, which represents about 70% of the US population. Most of these urban counties have a low natural fluoride concentration in the drinking water. The percentage of the population that received natural or adjusted fluoride (approximately 1 PPM) was estimated for each county. Medicare data for 1984-1987 were used to calculate the annual incidence of age-adjusted hip fractures for white males and females age 65 and older.
A comparison was made of the age-adjusted hip fracture rates obtained when the denominator (population at risk aged 65 and older in each county) was census data versus Medicare data. As the percentage of individuals exposed to fluoridated water increased within a county, the hip fracture rate generally rose for both sexes, but not in a smooth linear fashion when using census data as a denominator. When calculated with a Medicare denominator, the rates were somewhat random with no clear relationship.
Because the Medicare denominator more closely matches the source of the fracture data, the following results are based on that version of the calculations. The regression coefficients in the Exposure Ecology Study represent the increase in hip fractures per 1000 persons at risk for each 1% increase in amount of coverage with fluoride at a level of approximately 1 PPM For white females the value was 0.0016 (95% confidence interval -0.0013 to 0.0045, not significant) and for white males 0.0037 (95% confidence interval 0.0020 to 0.0054, significant). Adjustment for county latitude and longitude produced higher correlation values and significance for females (0.0084, 0.0060-0.0107) and males (0.0064, 0.0048-0.0080). The baseline data were adjusted on the basis of the first three digits of the Social Security Number (SSN) for each person with a hip fracture. These digits indicate where the SSN application was filed and is probably a better measure of long-term exposure to fluoride in that county. The values were significant for females (0.0068, 0.0024-0.0112) and males (0.0060, 0.0041-0.0078). Finally, based on these calculations and postulating that a fluoride-related effect was real, an estimate was made of the number of: (1) prevented fractures if there was no fluoride, from any source, in any US drinking water (5.6% total reduction for females and males), and (2) excess fractures if the county had 100% fluoridated water (5.3% total increase for females and males). Thus, these very large hypothetical swings in percentage of population exposed to fluoridated water are postulated to have a small relative impact on total hip fracture rates.
 Keller C. Fluorides in drinking water. Unpublished data 1991.
 May DS, Wilson MG. Hip fractures in relation to water fluoridation: an ecologic analysis. Unpublished data 1991.