Relationship Between Fluoride Concentration in Drinking Water and Mortality Rate from Uterine Cancer in Okinawa Prefecture

The Okinawa Islands located in the southern-most part of Japan were under U.S.administration from 1945 to 1972. During that time, fluoride was added to the drinking water supplies in most regions. The relationship between fluoride concentration in drinking water and uterine cancer mortality rate was studied in 20 municipalities of Okinawa and the data were analyzed using correlation and multivariate statistics. The main findings were as follows.

(Figure 1).Okinawa is subtropical, oceanic and insular, and therefore its natural evironment differs considerably from mainland Japan.Within Okinawa prefecture itself, environ-mental conditions such as weather, dietary habits and way of life do not differ considerably.However, other factors such as geological features, soil, population density, degree of development, area ratio of military bases show regional differences.Regional characteristics are also seen in the quality of drinking water supplied to the municipalities.
Administrative power of Okinawa was returned in 1972 from the U.S. to Japan.However, the water treatment methods developed during the U.S. administration which included the addition of sodium fluoride and sodium metaphosphate to tap water were continued.Thus, studying the ecologic epidemiological influence of the drinking water quality on the health of Okinawan people is considered highly relevant.
The present author has previously reported a correlation between the concentration of 15 components in drinking water and the mortality rate from various diseases (14 in female, 12 in male, 12 in both sexes) in Okinawa prefecture and a statistically signifiant con-elation been observed between fluoride and Received August 13, 1996; accepted November 12, 1996.Department of Preventive Medicine, School of Medicine, University of the Ryukyus, Okinawa, Japan.Relationship between Fluoride and Uterine cancer.Address for correspondence : Eiichi Tohyama, Department of Preventive Medicine, School of Medicine, University of the Ryukyus, 171 Uechi Okinawa City, 904 Japan.uterine cancer as well as between ABS and ischemic heart disease, NO3 levels and liver cancer 18).The present paper explores epidemiologically the association between fluoride level and uterine cancer mortality using correlation and multivariate analyses.Uterine cancer here includes both cancer of the corpus and the cervix.
Although fluoride treatment of community water has been surrounded by controversy regarding the possibility of adverse health effect, the carcinogenicity of fluoride remains unclear 19).

Municipalities Investigated
Twenty of the 53 municipalities in Okinawa prefecture, Naha City, Okinawa City, Ginowan City, Urasoe City, Itoman City, Gushikawa City, Ishikawa City, Nago City, Kin Town, Kadena Town, Chatan Town, Nishihara Town, Yonabaru Town, Sashiki Town, Haebaru Town, Kochinda Town, Motobu Town, Katsuren Town (the above being in the Okinawa Island proper), Hirara City (Miyako Island), Ishigaki City (Ishigaki Island) were investigated.Thirty-three municipal villages were excluded from the analysis, because their population scales (mean population: 6,882 in 1985) 20) were very small and their socioeconomic-medicare levels (mean income gap: 34.6The population according to age of municipalities over a 5year period was computed by averaging the numbers for 13 years that correspond to the number of deaths using the Okinawa prefecture's Statistical Yearbook21).The 5-year mortality rate between 1973 to 1992 was determined in order to study the time trend using the same method described above.S M R24) by cause of death for each municipality in Okinawa was used for consultation.

Socioeconomic and Uterine Variables
The population density, the primary and secondary industrypopulation ratios and the income gap in 1985 (with the national average of Japan considered as 100%) for the 20 municipalities of Okinawa prefecture were obtained from the Minryoku 20).
The tap water diffusion rate in 1975 and percapita water consumption per day were obtained from the Outline of Tap Water in Okinawa Prefecture21), and stillbirth, marriage and divorce rates were found in the Okinawa prefecture statistical yearbook Analysis PC-SAS (Statistical Analysis System)26,27) was used for basic statistical analysis.Personal Computer Statistical Analysis Handbook II, Multivariate Analysis28) was used for multiple regression analysis using a backward elimination procedure in order to screen variables for inclusion in the final model.

Drinking water quality
The concentrations of 15 components in drinking water from 1968 to 1980 in the 20 municipalities were averaged in order to investigate the overall water quality in Okinawa prefecture.The results are shown in Table 1.As shown in Table 1, dissolved solids range from 106.17mg/l to 485.49mg/l, and the range is wide.Also, fluorides are scattered with a minimum 0.02mg/l and maximum 0.37 mg/l.These concentrations can be divided into two groups, namely one greater than 0.19mg/l and the other less than 0.03mg/l.

Relationship Between Fluoride Concentration and Uterine Cancer Mortality Rate
Pearson's correlation analysis was used to evaluate the relationship between fluoride concentration in drinking water and uterine cancer mortality rate in the 20 municipalities of Okinawa prefecture.The results showed a correlation coefficient of r=0.626 (p<0.005).A scatter graph of these data is shown in Figure 2.

Time Trend
The time trends of water fluoridation were studied by analyzing the changes in fluoride concentration over time in several municipalities.The results are shown in Figure 3 and 4. The 5-year uterine cancer mortality rates in either fluoridated or non-fluoridated municipalities were also compared over 4 consecutive time periods (1973-'77, 1978-'82, 1983-'87, 1988-'92).The results are shown in Figure 5 As shown in Figure 3, the fluoride level declined suddenly in 1972 when sodium fluoride addition was discontinued.The fluoride level in Figure 4 is very low and does not change substantially.When the two curves in Figure 5 are compared, the slope of curve A-fluoridated (15 municipalities) appears to be sharper than that of curve B-non-fluoridated (5 municipalities).The mortality shown by curve A is higher for every time period compared to that of curve B. However, in the last period (1988-92), the mortality shown by curve A approaches that of curve B.

DISCUSSION
Surface water is the primary source of drinking water in Okinawa prefecture.Compared to the average qualiti of river water in the rest of the world, Ca, HCO3 29) levels are higher outside of Japan because of the high lime stone content, whereas SiO230) is higher in Japan due to the influence of volcanic geology and soil, and Cl, Na, SO431) levels are higher in Okinawa due to direct influence of sea water on the small islands.The mean concentration of dissolved solids is 119.8 mg/l in the rest of the world, 86.4 mg/l in Japan, and 253.4 mg/l in Okinawa.When Okinawa is divided roughly into two regions, the non-calcareous area and the calcareous area, the dissolved solid content is markedly higher in the latter region, 592.2 mg/l compared to 111.2 mg/l 32).The drinking water quality in the 20 municipalities in Okinawa prefecture reflects the same characteristics as those of the water sources.Judging together from Table 1, Figure 1, Figure 3 and Figure 4, fluoride concentrations less than 0.03mg/l are attributed to natural origins and do not seem to be subject to the infuences of other water contents and the local calcareous or non-calcareous geology.However, concentrations greater than 0.19mg/l occur as a result of fluoridation of municipal water supplies.Fluoride supplementation was added to tap water in municipalities such    as Naha City, Ishikawa City, Gushikawa City, Ginowan City, Urasoe City, Itoman City, Okinawa City, Kin Town, Kadena Town, Chatan Town, Nishihara Town, Sashiki Town, Yonabaru Town and Haebaru Town during the 27-year U. S. administration of Okinawa prefecture from 1945 to 1972, whereas municipalities such as Nago City, Motobu Town, Kochinda Town, Hirara City and Ishigaki City received no fluoride supplementation.The mean fluoride concentration over the 13 years in the municipalities supplemented with fluoride was between 0.19 and 0.37 mg/l, whereas that in municipalities which recieved no fluoridation was between 0.02 and 0.03 mg/l.As can be seen in Figure 2, the uterine cancer mortality rate shows a significantly positive correlation with the fluoride concentration in drinking water (r=0.626,p<0.005).Using the regression line in Figure 2, the uterine cancer mortality rate in the municipalities supplemented with fluoride at a concentration of 0.30 mg/1 (mean over 13 years) is 69.5% higher than that in the municipalities at zero fluoridation.
Weak associations in epidemiological studies are always sensitive to the influence of confounding variables.In Table 2, the tap water diffusion rate, percapita water consumption, population density, population in primary and secondary industies and stillbirth rate, marriage rate, divorce rate are considered to be confounding variables.The first 5 variables are considered to be socioeconomic factors and the latter 3 uterine factors.However, the carcinogenicity of fluoride in animals and humans remains unclear despite numerous biomedical and epidemiological studies7.[42][43][44][45][46].The present study was able to demonstrate a possible association between water fluoridation and uterine cancer mortality.Further research, therefore, will be necessary to evaluate this association.These studies should emcompass other individual predictors of cancer risk (or confounding factors), such as smoking,dietary patterns and sexual activity.

(
1980) 21).The differences of two above mentioned values between 1975 and 1985 for the individual municipalities are well below 10% the sums of the same two values.

Figure 3 .
Figure 3. Secular changes in fluoride concentration of drinking water for 3 typical fluoridated municipalities.

Figure 4 .
Figure 4. Secular changes in fluoride concentration of drinking water in 2 typical non-fluoridated municipalities.

Table 1
, tap water diffusion rate (1975), percapita water consumption per day (1975), population density (1985), primary and secondary industry population ratios (1985), income gap (1985), stillbirth rate (1980), marriage rate (1980) and divorce rate (1980) shown in Table 2 as the explanatory variables, multiple correlation analysis by the variable selection procedure was carried out.The results are listed in Table 3.The explanatory variables selected in the final model were fluoride concentration, income gap, stillbirth rate, tap water diffusion rate and divorce rate.The highest significance was that of fluoride concentration.

Table 2 .
Independent variables used in multiple linear regression analysis of uterine cancer rates of 20 municipalities in Okinawa, Japan.

Table 3 .
Results of multiple regression analysis of uterine cancer mortality versus variables listed in Table2for 20 municipalities in Okinawa, Japan.