Association Between Calcium Ingested from Drinking Water and Femoral Bone Density in Elderly Women: Evidence from the EPIDOS Cohort


  • EPIDOS Study Group

    Coordinators: G. Bréart, P. Dargent-Molina (Epidemiology); P.J. Meunier, A.M. Schott (Clinical); D. Hans (Bone densitometry and ultrasound quality control); P.D. Delmas (Biochemistry).

    Principal investigators: C. Baudoin, J.L. Sebert (Amiens); M.C. Chapuy, A.M. Schott (Lyon); F. Favier, C. Marcelli (Montpellier); E. Hausherr, C.J. Menkes, C. Cormier (Paris); H. Grandjean, C. Ribot (Toulouse).


Although the main source of dietary calcium is dairy products, the calcium contained in mineral water, which is as available as that of milk, could provide a valuable source of calcium. We analyzed the data from the EPIDOS multicenter study to evaluate the relationship between both dietary calcium and that supplied by drinking water and bone density measured at the femoral neck by dual-energy X-ray absorptiometry. The study included 4434 women over 75 years of age who had not received any treatment likely to interfere with calcium metabolism. A significant correlation was found between total calcium intake and bone density at the femoral neck (r = 0.10, p < 0.001). After adjustment for the main variables influencing bone density, an increase of 100 mg/day in calcium from drinking water was associated to a 0.5% increase in femoral bone density, while a similar increase in dietary calcium from other sources only led to a 0.2% increase; however, this difference was not significant. The consumption of calcium-rich mineral water may be of interest, especially in older women who consume little calcium from dairy products.


The decrease in bone density is a major risk factor for hip fractures in elderly women.(1-4) The higher prevalence of this type of fracture, partly due to the increase in age of the population, has prompted the search for ways of preventing bone loss in elderly women in an attempt to decrease the risk of fractures.

The possible relationship between bone mass and ingested calcium has given rise to controversy. A meta-analysis of all studies on this subject published prior to 1988 concluded that there was a positive influence of exogenous calcium on bone mass and that a high daily intake in postmenopausal women helped prevent loss of bone tissue.(5) It also appears that a high dietary intake of calcium is also associated with a reduced risk of hip fractures in elderly women.(6)

The main source of calcium in the diet is milk and dairy products. However, some mineral waters are rich in calcium (150–550 mg/l), and regular consumption can give rise to a significant intake of calcium. It has been shown that the calcium contained in these mineral waters is equally well absorbed and utilized as that from milk and calcium supplements.(7) In individuals who consume little milk or dairy products, consumption of calcium-rich water could thus provide a useful source of calcium.

In the course of a large multicenter study (EPIDOS) of risk factors for hip fractures in women over 75 years of age in France,(4) the influence of the calcium in the water ingested by these women was analyzed. A particular objective was to find out whether the amount of calcium absorbed from drinking water had any influence on bone density.



Between January 1992 and January 1994, 7575 women aged 75 years and over were recruited into the EPIDOS cohort from five towns in France: Amiens, Lyon, Montpellier, Paris, and Toulouse. The protocol was approved by the ethics committee of Lyon for all centers. The initial objective of the EPIDOS study was to determine the risk factors for hip fractures. All women residing in these towns aged 75 years and over were invited to participate. They were selected from electoral registers or health insurance databases and contacted by mail. The final population was chosen on a voluntary basis after checking for inclusion criteria. Those unable to walk unaided, with a history of hip fracture or with hip prosthesis, were excluded. All the women who were included had to understand and fill in a questionnaire.

Among the 7575 women in the EPIDOS study, we excluded for the present analysis women receiving treatment likely to interfere with calcium metabolism, i.e., those treated with cortisone for more than 3 consecutive months at any time or more than 3 months in any year, those treated with fluoride salts, or disodium etidronate, and those who had received calcium or vitamin D supplements within the previous 12 months. We also excluded the women who had received hormone replacement therapy. This gave a total of 4434 subjects.

Data collection

The study was based on the data collected on the initial examination in the EPIDOS survey, which included a questionnaire personally administered by a trained nurse or physician, and a clinical examination. Bone mineral density (BMD) was measured by dual-energy X-ray absorptiometry with a DPX Plus device (Lunar Corp., Madison, WI, U.S.A.), at the proximal femur (femoral neck, trochanter, and Ward's triangle). Grip strength was measured in each hand with a dynamometer, and the higher of the two values for each subject was used in that analysis.

The women were considered as taking regular exercise if they went out on a daily basis for shopping or carried out at least 2 h of housework per week.

The daily calcium intake was estimated from a self-administered food frequency questionnaire.(8) This questionnaire, adapted to French diet habits, inquired about the daily or weekly frequency intake of 20 foods or drinks. The importance of each portion was described as small, medium, or large. The dietary calcium intake was estimated from an analysis of normal intakes of calcium-rich foods. The calcium derived from drinking water was calculated from estimated daily water intakes and the calcium contents of the waters consumed. In the case of tap water, the calcium content was that supplied by the corresponding water utilities, which for 1993 were as follows: Amiens 110 mg/l, Lyon 80 mg/l, Montpellier 118 mg/l, Paris 95 mg/l, and Toulouse 43 mg/l. The fluoride intake was estimated in the same way from the fluoride contents of the water that the subjects reported drinking.

The social class of the women was evaluated by their income, and we considered three categories: <3000 French francs (FF)/month, 3000–6000 FF/month, and >6000 FF/month.

Statistical analysis

All quantitative results were expressed as means ± SD. SAS software was employed for the statistical analysis (SAS Institute, Cary, NC, U.S.A.).(9) The associations between BMD in each measurement site and the various variables were analyzed with conventional parametric tests (Pearson's correlation, t-test). Multiple linear regression analyses (Proc REG) were carried out with BMD as the dependent variable. We initially introduced in the models all parameters related to BMD in the univariate analyses (p < 0.05). The final models only included the significant variables for at least one measurement site, after adjustment using a stepwise backward procedure (p < 0.05).

We expressed the effect of each variable as a percentage by dividing the increase in bone density induced by one unit of the considered variable by the mean bone density of the sample.


Description of population

The main characteristics of the women are shown in Table 1. Ten percent had been treated for a thyroid disorder and 7% were diabetic. Fifty-nine percent of the women took regular physical exercise.

Table Table 1.. Main Characteristics of 4434 Women* from Five Cities of France Enrolled in the EPIDOS Study
original image

The women reported drinking 882 ± 482 ml of water/day. Forty-three percent only drank tap water, 36% only drank mineral water, while 21% drank both. The mean total consumption of calcium was 830 mg/day. On average, 15% of the total calcium intake was supplied by drinking water.

The mean BMD was 0.720 ± 0.111 mg/cm2 at the femoral neck, 0.589 ± 0.124 mg/cm2 at Ward's triangle, and 0.644 ± 0.117 mg/cm2 at the trochanteric site.

Univariate analysis

We found a significant relationship between total intake of calcium and femoral neck BMD (r = 0.10, p < 0.001), Ward's triangle BMD (r = 0.09, p < 0.001), and trochanteric BMD (r = 0.09, p < 0.001). The femoral neck BMD of 0.689 g/cm2 for women consuming < 400 mg/day of calcium rose to 0.758 g/cm2 in those consuming > 2000 mg/day (Fig. 1).

Figure FIG. 1..

Femoral neck BMD according to total calcium intake in the EPIDOS study.

The variables significantly linked to BMD in the univariate analysis are presented in Table 2. The variable most significantly related to BMD was body weight (r = 0.38). At the femoral neck, an increase in 5 kg was associated with a 0.021 g/cm2 (+2.9%) increase in BMD. An additional intake of 100 mg/day of dietary calcium was associated with a 0.3% increase in BMD, while a further intake of 100 mg/day of calcium from drinking water was associated with a 1.2% increase in BMD. Consumption of alcohol, tobacco, and the fluoride content of the drinking water were not found to be associated with bone density.

Table Table 2.. Simple Linear Regressions for Femoral Neck BMD Ward's Triiangle BMD, and Trochanteric BMD
original image

Similar results were found for Ward's triangle and the trochanter, with 0.4% and 0.3% increases in BMD, respectively, for an additional intake of 100 mg/day of dietary calcium, and a 1.4% increase in BMD in both sites for 100 mg/day of calcium from drinking water.

Multivariate analysis

In the multivariate analysis (Table 3), the association of weight with BMD remained the most important. The adjustment on weight reduced the effect of height which even became negative on trochanteric BMD. Grip strength, pregnancy, and thyroid treatment were no longer associated with BMD at any site, while diabetes exhibited a significant association with trochanteric BMD. Age adjustment reduced the effect of income on BMD, and only an income >6000 FF/month was associated with a moderate increase in Ward's triangle BMD. The change in bone density for 100 mg of calcium intake showed a nearly 2-fold decrease between the univariate and multivariate analyses (this difference was mainly due to the association between weight and calcium intake). In the final model for femoral neck BMD, we observed a 0.2% increase in bone density for an additional intake of 100 mg of dietary calcium and a 0.5% increase for 100 mg of calcium from drinking water.

Table Table 3.. Multiple Linear Regressions for Femoral Neck BMD Ward's Triiangle BMD, and Trochanteric BMD
original image


In this study, after allowance for the influence of dietary intake of calcium and other factors affecting bone density, we found a relationship between the amount of calcium ingested from drinking water and the femoral neck BMD in women over 75 years of age.

One of the main methodological difficulties in studies of this type designed to investigate the influence of ingested calcium is that of estimating calcium intake.(5, 10) In the present study, dietary calcium intake was ascertained from a food frequency questionnaire focused on calcium-rich foods. This method has the advantage of simplicity, an important consideration for use with the elderly. It has been validated and shown to correlate well with the calcium intake estimated from prospective weekly dietary inventories.(8, 11) The inaccuracy in the method could give an underestimate of the association between calcium intake and bone density rather than lead to bias. Nevertheless, in this cross-sectional study, we could not take into account the effect of duration of calcium intake on the increase in BMD.

In our multivariate model, the variables most strongly linked to bone density were weight and age of the women. We found comparable percentage changes in bone density for weight to those reported by Bauer et al. in a study of 9704 women over 65 years of age(12); the influence of age was slightly higher in their study, but their population was much younger (mean 71.1 years) than ours. The other variables significantly associated with bone density in our multivariate analysis were also found by Bauer et al. and are in line with other literature data: age at menopause,(13) physical activity,(14, 15) and mother with history of bone fractures.(16) The influence of socioeconomic status on BMD has been poorly studied. A Spanish study showed a positive influence of socioeconomic status on BMD.(17)

Our results are further support for a relation between calcium intake and femoral neck BMD. From a meta-analysis,(5) Cumming noted evidence for an influence of calcium intake on bone density,(14, 18 19) in particular with respect to cortical bone,(20, 21) although some studies failed to find any association between the two.(13, 22)

In the study of Bauer et al.,(12) an additional 400 mg of calcium in the daily intake led to a 0.7% increase in bone density, which is close to the value we noted for dietary calcium (0.2% for 100 mg). On separating out the influence of calcium from drinking water, bone density was still found to be significantly related to dietary calcium intake. However, the effect on bone density of 100 mg of calcium in drinking water appeared to be greater than that of 100 mg of calcium in the diet, although the difference was not statistically significant. The beneficial effect of calcium in water may come from the fact that it is well absorbed in this form(7, 23) or even better,(24) and that fractionation of ingestion throughout the day may also improve absorption.(25) In a study of bone density in women consuming mineral water rich in calcium, Cepollaro et al. noted a positive influence of total calcium intake without discriminating that provided by water from other dietary calcium.(26)

An American Consensus Development Panel recommends an optimal calcium intake of 1500 mg/day in postmenopausal women not receiving hormone replacement therapy.(27) The calcium intake currently recommended for the elderly in France is somewhat lower (1200 mg/day),(28) and it is of interest that in our study <15% of women had this high intake. Most women need therefore to be encouraged to increase their calcium intake. For those who do not consume milk and dairy products, a simple strategy without prescription of calcium supplementation therapy would be to drink calcium-rich water. From our results, a woman with a daily consumption of 1 l of water containing 400 mg of calcium would be estimated to have an equivalent bone density of a woman 7 years younger drinking a low calcium-content water. Nevertheless, this was obtained from a cross-sectional study and would need confirmation in a prospective study.


We thank Dr. M.J. Arnaud and Dr. A.M. Delabroise for encouraging this work. We greatly appreciate the wholehearted participation of all volunteering recruits who contributed to this epidemiological study. The EPIDOS Study was supported by INSERM-MSD-Chibret. This study received funding from the Water Institute Perrier Vittel.