Critical Factors for Bone Health in Women Across the Age Span: How Important Is Muscle Mass?

Jasminka Ilich-Ernst, RD, MS, PhD, Rhonda A. Brownbill, MS, RD, Martha A. Ludemann, MS, RD, Rongwei Fu, PhD

In This Article


Association between bone, body composition, and Ca. Table 1 presents descriptive characteristics of all subjects and individual groups. The average age of menarche, number of years since menopause, and number of reproductive years were 13.2 ± 1.5, 20.5 ± 8.5, and 35.9 ± 4.8, respectively. Lean tissue and bone mass in all skeletal sites were significantly different across the age groups, with lower values in older women (groups 3 and 4) compared with younger ones (groups 1 and 2). TBF had the opposite trend, with significantly higher values in older women.

Figure 1 depicts the relationship of total body bone mineral content (TBBMC), LBM, and age. Both TBBMC and LBM decreased with increasing age within the population (regression equations: TBBMC(g) = -13.33[age] + 3238 and LBM(kg) = -0.14[age] + 47.35).

Relationship of total body bone mineral content (TBBMC) and lean body mass (LBM) as a function of age in all subjects (mean ± SE).

Ca intake from food, supplements, and total (food plus supplements)are presented in Table 2 . About 70% of subjects were taking supplemental Ca from various preparations. Although older women (groups 3 and 4) had significantly lower Ca intake from dietary sources, the total Ca intake was not different across the groups because of supplementation. The results from the correlational analysis for the entire population indicate age as a significant negative predictor for bone mass in all measured skeletal sites (total body, lumbar spine, hip, and forearm), with r ranging from -0.36 in the lumbar spine to -0.71 in the femoral neck, P < .05. However, the slope of the curve for most of the skeletal sites (except femoral neck and Ward's) had a sharper decline only after the age of about 45 years.

In older individuals (groups 3 and 4), age of menarche was significantly negatively related to total body bone mineral density (TBBMD) and forearm BMD (r = -0.25 and -0.29, respectively, P < .05). YSM was significantly negatively related to TBBMD, TBBMC, all sites of the hip (except trochanter), and all sites of the forearm BMD (r ranging from -0.31 to -0.43, P < .05). Number of reproductive years did not show significant association with any of the skeletal sites.

In all subjects, WT was a significant positive predictor of all measured skeletal sites, (r = 0.21-0.62, P < .05), with the strongest correlation with TBBMC. However, LBM showed even stronger association with all measured skeletal sites (r = 0.35-0.76, P < .05) and had the strongest relationship with TBBMC. TBF correlated significantly only with TBBMC (r = 0.37, P < .05). Because there was a positive association between TBF and LBM (r = 0.272, P = .0054), we examined whether they exerted a synergistic effect on BMD by categorizing subjects into groups below and above median of each and performing subgroup analyses. Table 3 presents LBM and TBF interaction on BMD of various skeletal sites. Higher LBM was associated with higher BMD in all skeletal sites, irrespective of TBF.

In the whole population, Ca from food was significantly positively associated with all regions of the hip (except shaft), with the best positive relationship with femoral neck and Ward's BMD, r = 0.32 and 0.33, respectively (P < .05). Somewhat surprisingly, there was a negative association between TBF and Ca intake, either from food or total (r = -0.335, P = .0009), but there was no relationship between LBM and Ca intake. To further investigate the above associations and possible combined effects on BMD, the subjects were categorized into groups below and above median for Ca intake and TBF. There was a trend toward higher Ca intake being associated with higher BMD of various skeletal sites in the lower TBF group (below median), but it did not reach statistical significance. Figure 2 depicts relationship between Ca intake, TBF, and BMD of Ward's triangle.

Relationship between Ca intake (from food), total body fat, and bone mineral density (BMD) of Ward's triangle in all subjects (mean ± SE). The effect of Ca was more evident in subjects with lower body fat, although not statistically significant. Values are adjusted for below and above median of Ca intake and total body fat.

The analysis of age groups revealed similar trends for body composition parameters as in the entire population, although the relationships were not as strong. Again, LBM had stronger positive effect on various skeletal sites than either TBF or WT. Figure 3 shows that Ca, either total or from food, was significantly associated with neck BMD in the youngest (group 1) and neck and spine BMD in the oldest cohorts (group 4).

Scatterplots of Ca intake (from food) and femoral neck bone mineral density (BMD) in group 1 (left) and spinal BMD in group 4 (right).

Multiple regression results. The results from multiple regression models selected on the basis of the Cp criterion for the whole population are presented in Table 4 . Each measured skeletal site was treated as a dependent variable, and the independent variables presented for each model were: age, WT, HT, BMI, LBM, TBF, Ca from food, and total Ca intake. Physical activity was not included in these models, as it was assessed only in the older cohort. The Cp statistics select the best predictors for each skeletal site; age, LBM, and/or WT contributed to the variance in every skeletal site. LBM was significantly positively associated with TBBMD, all regions of hip, and lumbar spine, whereas TBF was significantly positively associated only with TBBMD. Ca from food contributed to the variance in the model with trochanter BMD as the dependent variable, with borderline significance (P = .0557).

The older subjects (groups 3 and 4) were assessed with the Allied Dunbar National Fitness Survey[18] for their participation in past activity, from the age of 18 years on, as well as for present and past walking. Table 5 illustrates the hours and engagement in various modes of activity. Most subjects (83.1%) had engaged in past physical activity (sports and recreation) and past regular walking (77.8%). With regard to present walking, 29.7% of subjects engaged in walking of at least 2 miles once a week at a faster pace, whereas 62.5% engaged in walking of shorter length and slower pace.

In the single regression models, past physical activity, but not past walking, showed statistically significant positive association with BMD in several skeletal sites: total body, Ward's triangle, lumbar spine, and forearm (r = 0.26-0.37, P < .05) and borderline significance with the BMD of shaft, trochanter, and total hip (P = .06). The subjects were also stratified into groups based on greater than or less than median percent of time engaged in past activity (< or > 47%). Two sample t-tests were used to determine significant group differences. The greater percent of adult life engaged in regular past activity was associated with greater Ward's triangle and ultradistal radius and ulna BMD. The opposite relationship was seen between past activity and TBF (Figure 4). No association was found between any of the physical activity parameters and LBM.

Boxplots of total body fat in older subjects stratified based on lower or greater than median % time (calculated from age 18 years to present) engaged in past activity. Horizontal line across is median with 95% confidence intervals (shaded area). Whiskers are the lowest and highest connected data.

Because there was a positive association between past physical activity and Ca intake (r = 234, P = .072), we further investigated this association and possible combined effects on BMD by categorizing subjects into groups below and above median for Ca intake and percent of adult life engaged in regular activity (< or > 47%). Figure 5 presents the relationship between total Ca intake (food and supplements), past physical activity, and BMD of ultradistal forearm (ulna and radius) and Ward's triangle.

Relationship between total Ca intake (food and supplements), past physical activity, and BMD of ultradistal forearm (ulna and radius) and Ward's triangle in older subjects (mean ± SE). The lines with corresponding P values connect statistically significant sites. Note the dominant effect of physical activity, irrespective of Ca on forearm, and added effect of Ca on Ward's BMD. Values are adjusted for below and above median of Ca intake and past physical activity.

It seems that physical activity had a more dominant effect than Ca intake on BMD of both skeletal sites. BMD of ultradistal forearm was higher in a group with above median activity, irrespective of Ca intake. The BMD of Ward's triangle was also statistically higher in a higher activity group, although there was a notable added effect of Ca.

Walking pace showed significant positive correlation with BMD of both neck and Ward's of the hip (r = 0.26 and 0.32, respectively [P < .05]), as well as with the BMD of the ulna and radius at the ultradistal and 1/3 region, and total forearm, (r = 0.25-0.33 [P < .05]). Total hours of walking correlated significantly with radius and ulna ultradistal BMD (r = 0.32 and 0.29, respectively [P < .05]). When subjects were stratified into groups below and above median percent TBF, a lower percent TBF was associated with a greater level of activity for walking pace and intensity.

The multiple regression analyses were performed separately for the older cohort to include various modes of walking and past physical activity. All other independent variables possibly associated with bone mass were also entered, including age, age of menarche, reproductive years, YSM, HT, WT, LBM, TBF, Ca from food, and total Ca intake.

The regression results revealed that past activity contributed significantly to the variance for BMD of total body, spine, all regions of hip, and ultradistal regions of forearm in the models with age, HT, and LBM. The coefficients of determination, R2 (adjusted), for the models were above 30%, with P for past activity ranging from .0022 (in the model with Ward's BMD) to .0382 (in the model with femoral neck BMD). Total hours of walking contributed significantly to the variance of ultradistal forearm BMD. Other walking modes or past walking did not reach statistical significance in multiple regression models with any of the bone variables.


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.