Risk Factors for Osteoporosis: Prevalence, Change, and Association With Bone Density

Janet R. Guthrie, MSc, Dip Ed, PhD, Peter R. Ebeling, MB BS, MD, FRACP, Lorraine Dennerstein, AO, MB BS, PhD, FRANZP, DPM, John D. Wark, MB BS, PhD, FRACP

In This Article


Of the 438 women in the longitudinal Melbourne Women's Midlife Health Project, 330 (76%) participated in the BMD and risk factor profile study. Of these, 112 were premenopausal and 112 were early perimenopausal (mean age, 50.0 years [SD, 2.3; range, 46-56)] at the time of their first BMD measure; these 224 women form the cohort of this report.

Tables 1 and 2 show the mean values of the modifiable and nonmodifiable risk factors. Of the cohort, 52% had a total calcium intake of less than 800 mg/day, 46% had a caffeine intake of greater than 360 mg/day, 29% exercised less than 1.5 hours/week, 5% had a BMI of less than 20, 14% were current smokers, 23% were past smokers, 10% reported abnormal menstrual cycle length, and 7% had a history of at least 6 months of amenorrhea. A family history of fractures or "dowager's hump" was reported by 25%. At baseline, 23 (10%) women had BMD greater than 1 SD below the age-matched reference range. Medications that may have an adverse impact on BMD were documented; 0.7% of the women were taking adrenocorticoid steroids at the time of their initial BMD measurements. At baseline, geometric means (95% confidence interval) of serum estradiol and FSH were 238.9 (201.6-283.2) pmol/L and 12.4 (11.0-13.8) IU/L, respectively.

When the lifestyle risk factors of low calcium intake (< 800 mg/day), low physical activity (< 1.5 hours/week), and current smoking were examined, 78% of the cohort reported at least 1 of these risk factors, 13% reported 2 factors, and 1% reported all 3 risk factors. If we include reporting a family history of fractures as a risk factor, 83% have at least 1 risk factor.

At 4-year follow-up, 13 women had dropped out of the study, 12 had undergone a surgical menopause, 42 were on HRT; 100 had become late peri- or postmenopausal; and 57 were still pre- or early perimenopausal.

Hormones. Figure 1 shows the estradiol and FSH levels at baseline and at 4-year follow-up in women who experienced the menopausal transition and in those who remained pre- or early perimenopausal. There was an increase over time in FSH (P < .0005) regardless of change or no change in menopausal status. There was no significant change over time in estradiol levels in women who remained pre- or early perimenopausal, but there was a decrease in these levels in women who became late peri- or postmenopausal (P < .0005). Compared with the women who remained pre- or early perimenopausal, FSH levels were higher in women who became late peri- or postmenopausal both at baseline (P < .05) and at 4-year follow-up (P < .0005). Estradiol levels were lower at 4-year follow-up in women who became postmenopausal (P < .0005). SHBG decreased (P < .0005) during the menopausal transition, but there was no significant change in testosterone levels, and as a consequence mean FAI increased (P < .005).

Estradiol and FSH levels at baseline and at 4-year follow-up by menopausal transition group.

The changes that occurred in modifiable risk factors over the 4-year study period are shown in Table 3. There was a mean weight gain of 2 kg during the study period (P < .005). There was also an increase in mean suprailiac (P < .0005) and subscapular (P < .005) skin-fold measures and in waist/hip ratio (P < .05) over the 4-year follow-up period. There was a decrease in the number of women who had a calcium intake of less than 800 mg/day (52% at baseline and 40% at follow-up, P < .05). There was a reduction in caffeine intake (359 [SE, 16] mg caffeine/day at baseline and 322 [SE, 15] mg/day at follow-up, P = .08). There were no overall significant changes in the intake of alcohol during the study period. There was a nonsignificant increase in the percentage of women exercising more than once per week (61% at baseline and 64% at follow-up); 17% did increase their exercise to more than once per week, but 14% had decreased at follow-up. Four women (2%) gave up smoking during the study and 1 (0.5%) took up smoking. There was a very small increase in medication use, and at the end of the study period, 1% of the women in the cohort were taking adrenocorticoid steroids.

Linear regression analysis found that lumbar spine (LS)-BMD and femoral neck (FN)-BMD decreased with increasing age, but the relationship was only statistically significant for FN-BMD (P < .05). Table 1 shows that LS-BMD and FN-BMD were significantly associated with measures of body weight and fatness, and all of these associations were positive. In multiple regression analysis, weight was the strongest predictor of BMD (P < .005). There were no significant associations between any of the other lifestyle or gynecologic risk factors listed in Tables 1 and 2.

Of the 140 (63%) women who returned for a second BMD measurement, 57 (41%) had experienced the natural menopausal transition, and the date of their FMP was recorded. At the time of their 2nd BMD measurement, an average of 2.4 (SE, 0.2) years had passed since their FMP. The first BMD measurements were taken a mean of 1.72 (SE, 0.2) years before FMP. Time between BMD measurements was 4 (SD, 0.08) years. Over this period, the mean (SE) annual change in LS-BMD was -1.6% (0.1%) (range, -3.8% to +0.2%); the change in FN-BMD was -1.0% (0.1%) (range, -2.9% to +1.5%). The percentage change in FN-BMD became more negative with increasing time since FMP (P = .1). The percentage change in LS-BMD also became more negative with increasing time since FMP (P < .05); most loss occurred in the first 10-15 months after FMP (Figure 2). The percentage change in LS-BMD was more negative with decreasing time to FMP (P = .06) -- that is, the closer the first measurement was to FMP, the greater the bone loss.

In the time period studied, none of the lifestyle or gynecologic risk factors had a significant effect on bone changes. There was a significant association between bone loss at the FN and baseline FN-BMD (beta = -0.03 SE, 0.009; P < .005) such that the larger the initial bone mass, the greater the bone loss. Changes in LS-BMD were independent of baseline LS-BMD.

. BMD changes (%/year) measured longitudinally at the lumbar spine and femoral neck in relation to time since final menstrual period. The line drawn for the lumbar spine is a LOWESS curve.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: