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


This was a cross-sectional study in 113 healthy white women, 20 to 88 years old, recruited as part of a longitudinal study. The rationale for including participants of wide age range was to examine the overall change in body composition with age and its association with different bone parameters. Similarly, we wanted to examine the changes of the effect of Ca on bone with age and hormonal status. However, because of the possibility of an age-specific association between the variables, we also divided participants into groups according to reproductive/menopausal status and examined the groups separately: group 1, premenopausal (age 20.0-45.9 years); group 2, perimenopausal (age 46.0-55.9 years); group 3, early postmenopausal (ages 56.0-65.9 years); and group 4, late postmenopausal (age > 66 years). The age of menarche, reproductive years, and years after menopause were assessed in the older population only (groups 3 and 4). Reproductive age was calculated by subtracting the age of menarche from the age of menopause. In the few subjects who had surgical removal of ovaries, the menopause was considered from the date of surgery.

Women could be included in the study if they had no current or history of metabolic bone disease and were not taking medications known to affect bone mass, such as steroids, insulin, and anticonvulsants. There were 6 women on estrogen, 1 on raloxifene (a portion of the data was recalculated without these 7 subjects, leading to the same overall conclusions and significance -- therefore, they were retained), 7 on oral contraceptives, and 16 on nondiuretic antihypertensive medications who were not excluded from this cross-sectional evaluation in order to achieve a more accurate representation of the health profile of today's women. All subjects were nonsmokers, although some did smoke in the past. Body weight and height were measured in indoor clothing without shoes. Thirteen subjects were overweight (body mass index [BMI] > 30 kg/m2). The study protocol was approved by the Institutional Review Board, and subjects signed informed consent forms.

Bone and body composition measurements were performed by dual x-ray absorptiometry technique with a Lunar DPX-MD densitometer (GE Medical Systems, Madison, Wisconsin) using specialized software for different skeletal regions: total body, lumbar spine, femur (neck, trochanter, Ward's triangle, shaft, and total), and forearm (ulna and radius at ultradistal and 1/3 distance from styloid). Densitometry analysis generates the bone mineral content (BMC) in g and the projected bone area in cm2. Areal BMD in g/cm2, which partially normalizes for bone and body size, is derived from the two and is most commonly used for bone mass interpretation and evaluation. Quality assurance for the densitometer was performed daily, and long-term stability and coefficients of variation were reported previously.[14]

The total body bone measurements enable the analyses of the whole body lean and fat tissue (kg) and percent body fat. We evaluated multiple skeletal sites in relation to independent variables to strengthen our conclusion in cases when a variable was associated with 3 or more skeletal sites as opposed to only 1 as well as to clarify whether a relationship existed between independent variables and predominantly trabecular (lumbar spine, femoral trochanter, ultradistal forearm) or predominantly cortical (femoral neck and shaft, proximal forearm and total body) bony tissue.

Two types of food frequency questionnaires were used to assess Ca intake, accommodating for variation in intake in younger and older cohorts. The conceptual basis of using a food frequency questionnaire is that it captures an individual's usual intake pattern over a long period of time. Both questionnaires were administered by providing subjects with detailed instructions and helping them fill out the food items and portion sizes. Ca intake in the younger cohort was assessed by a food frequency questionnaire, previously developed and used by one of the authors (JZI),[15,16] and in the older cohort by a shortened food frequency questionnaire, recommended by the National Osteoporosis Foundation.[17]

All subjects were asked to report any Ca-containing supplements, including multivitamins and minerals, other nutritional supplements, and antacids. An average daily Ca intake from supplements was determined on the basis of frequency and amount of consumption. Results are presented as daily intakes of dietary calcium (Ca food) and/or total calcium (Ca total) in mg/day. We analyzed separately the association between BMD and both Ca from food and total calcium.

Past physical activity and past and present walking were assessed only in the older population (> 56 years, groups 3 and 4) using an interview format with a modified version of the Allied Dunbar National Fitness Survey for older adults.[18] Past Physical Activity was assessed as percent of adult life, from age 18 years to the present, engaged in sport and recreational activities of an intensity of at least 4 kcal/min (such as low-impact aerobics) on a regular basis, at least once a week for 3 months of the year. Past Walking was assessed separately, as percent of adult life participating in regular walks, at least once a week for at least 2 miles. Present Walking was assessed through several modes: Walking Frequency, Walking Total Hours, and Brisk/Fast Walking. These modes considered walks of at least 1 mile of the following self-reported walking paces: (1) slow, (2) steady/average, ~ 2 miles/hour, (3) brisk, ~ 3 miles/hour, and (4) fast, ≥ 4 miles/hour. Walking Frequency assessed number of walks of at least 40 minutes or 2 miles at any pace; Walking Total Hours assessed total hours of walks of at least 1 mile at any pace; and Brisk/Fast Walking assessed walks of at least 1 mile or 20 minutes in a reported brisk/fast pace (3-4 miles/hour).

All data are presented as mean ± standard deviation, unless noted otherwise, and were analyzed using the statistical packages Data Desk (Odesta Corp, Northbrook, Illinois) and SAS (SAS Institute Inc, Cary, North Carolina). Assumptions for normality and homogeneity were examined and found satisfactory for the whole cohort and for each of the age groups. Simple association between bone mass, body composition, Ca intake, and physical activity were estimated by calculating Pearson's r. One-way ANOVA with Tukey's Studentized Range tests were used to assess group differences. To examine further whether potential predictor variables (Ca, components of body composition, and physical activity) are associated with bone mass independently or in combination, subgroup analyses were performed. In these analyses, Ca intake, lean body mass, total body fat, and past physical activity were categorized in the groups below and above median, and their combined effect on BMD was examined.

Single and multiple regression models were created to evaluate the influence of different potential predictors on BMD and/or BMC of measured skeletal sites that were treated as the dependent variables. The variables with skewed distribution were transformed to improve their symmetry. The Cp statistics were calculated to screen all possible regression models for the one with the highest predictive value and least bias for each bone variable. The Cp statistics account for the biases introduced in the estimates of the dependent variable by leaving the less predictive parameters out of the regression model. On the basis of Cp statistics selection, it is possible to develop a regression model with the highest predictive ability and the least bias.[19] Usually, the best model is one with the lowest Cp value that is closest to the number of independent variables chosen plus the intercept of the regression.

The independent variables presented to each Cp analysis were: height, weight, BMI, age, lean body mass (LBM), total body fat (TBF), and Ca intake for the whole cohort; age of menarche, years since menopause (YSM), number of reproductive years, measures of past activity, and walking modes for older subjects (groups 3 and 4) were also included in the analysis. In case of the set of strongly interrelated variables (eg, age, YSM, number of reproductive years, or height, weight, and BMI), only 1 variable from the set was entered in the model at a time. The variable with the highest level of significance from the set was retained in the final model. The accepted level of significance was P < .05.


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.