Risk Factors for Nonvertebral Fracture in Obese Older Women

Abstract

Context:

A high prevalence of obesity has recently been reported in postmenopausal women with low trauma fracture, suggesting that higher bone mineral density (BMD) in obese individuals may not be protective against fracture.

Objective:

The aim of this study was to compare BMD and other risk factors for nonvertebral fracture in 1377 obese postmenopausal women.

Design:

Characteristics of obese women with and without incident nonvertebral fracture were investigated among the prospective cohort in the Study of Osteoporotic Fractures.

Setting:

The Study of Osteoporotic Fractures is a multicenter study of 9704 women (>99% Caucasian) aged 65 yr and over who were recruited between September 1986 and October 1988 from population-based listings at four U.S. clinical centers.

Main Outcome Measure:

The main outcome measure was nonvertebral fracture.

Results:

BMD T-scores in the spine, femoral neck, and total hip were significantly lower in obese women who experienced fractures than in obese women without fracture: mean differences, −0.56 [95% confidence interval (CI) = −0.73 to −0.39], −0.46 (95% CI = −0.57 to −0.36), and −0.51 (95% CI = −0.62 to −0.39), respectively (P < 0.0001 for all). A previous history of fracture [odds ratio = 1.69 (95% CI = 1.33–2.14); P < 0.0001] and femoral neck BMD [1.62 (95% CI = 1.42–1.85) per sd decrease in BMD; P < 0.0001] were independently associated with incident nonvertebral fracture.

Conclusions:

Obese postmenopausal women who sustain nonvertebral fractures have significantly lower BMD on average than obese women without fracture and are more likely to have a past history of fracture. Fractures in obese postmenopausal women thus exhibit some characteristics of fragility fractures.

Body mass index (BMI) is a major determinant of bone mineral density (BMD), and obesity is widely believed to protect against fracture (1–4). However, the higher BMD associated with obesity may reflect adaptation to increased mechanical demands on the skeleton and may not confer greater protection against fracture than lower BMD levels in leaner individuals, particularly in view of the greater trauma associated with falling in the obese population (5–8). Consistent with this hypothesis, we recently reported a high prevalence of obesity in postmenopausal women presenting to a fracture liaison service with low trauma fracture. Overall, 27.7% of those aged 75 yr or less were obese or morbidly obese, and over 80% had normal femoral neck BMD (T-score > −1), whereas osteoporosis was present in fewer than 2% (9).

If fractures in obese women reflect increased skeletal fragility, lower BMD values would be expected than in their obese counterparts without fracture. We have tested this hypothesis by comparing femoral neck BMD in obese postmenopausal women with and without incident nonvertebral fracture in the Study of Osteoporotic Fractures (SOF) cohort, a prospective population-based study of Caucasian postmenopausal women followed for a mean duration of 11.3 yr. We also investigated whether other risk factors for osteoporotic fractures, including previous fracture history, family history of fracture, lifestyle characteristics, neuromuscular function, physical activity, and fall frequency are associated with risk of nonvertebral fractures in obese postmenopausal women.

Subjects and Methods

Subjects

The SOF is a multicenter study of 9704 women (>99% Caucasian) aged 65 yr and over who were recruited between September 1986 and October 1988 from population-based listings at four U.S. clinical centers in Portland, OR; Minneapolis, MN; Baltimore, MD; and the Monongahela Valley near Pittsburgh, PA (10). Women were recruited irrespective of BMD and fracture history; those unable to walk without assistance and those with bilateral hip replacements were excluded. All participants provided informed consent, and the protocol was approved by the institutional review boards of the participating institutions.

From January 1989 to December 1990, all participants were invited to undergo a second evaluation. A total of 7978 of the 9339 women attending the second visit had a technically adequate measurement of BMD at the hip and spine. For most variables in this analysis, we used data from the second visit to provide baseline values because measurement of BMD at spine and hip was first made at this time. The exceptions were years of education and maternal history of fracture, information on which was obtained only at the first study visit.

Measurement and quality control procedures were rigorous (detailed elsewhere) and included a standardized protocol and clinic site training by the SOF Coordinating Center (11). Height was measured at the baseline visit with participants barefoot (or in thin socks) by a wall-mounted Harpenden stadiometer (Holtain Ltd., Dyved, UK) after a held full-inspiration in the Frankfort horizontal plane. Weight was measured in indoor clothing without shoes using a balance-beam scale. BMI was calculated by the formula weight in kilograms divided by the square of height in meters. Obesity was defined as a BMI of 30 kg/m² or higher. Waist and hip circumferences were measured to the nearest 0.1 cm with steel tape; waist was defined at the midpoint between the highest point of the iliac crest and lowest part of the costal margin in the mid-axillary line, and hip circumference was measured at the level of the great trochanter (12).

Lean mass was estimated by the bioelectrical impedance method using Valhalla 1990B Bio-Resistance body composition analyzers (Valhalla Scientific, San Diego, CA). To standardize hydration for the measurement, participants were instructed to maintain a normal fluid balance and to abstain from vigorous activity, alcohol, and caffeine for 12 h before the clinic visit. Fat mass was calculated by subtracting lean mass from total weight; percent body fat was fat mass expressed as a percentage of total weight.

Other measurements included tests of neuromuscular function: grip and quadriceps strength, time to chair stands, tandem walk, and walk speed. Grip strength was assessed using a grip dynamometer (Preston Grip dynamometer; Takei Kiki Kogyo, Tokyo, Japan), and quadriceps strength was measured with an isometric leg extension chair with load cell. Gait speed (meters per second) was measured as the average time to complete two trials on a standard 6-m course at a usual pace. Finally, the chair stand test evaluated how long it took to stand up from a chair five times, without using arms.

Other potential covariates assessed included age, education level, physical activity, smoking habits, coffee drinking habits, age at menopause, family history of fractures, personal history of fracture after 50 yr of age, and falls during the previous year. Social support was evaluated by the Lubben Social Network Scale (13). Women were also asked about medical conditions such as diabetes mellitus, current medications, and therapy during the previous year with estrogen, vitamin D-containing supplements, calcium supplements, glucocorticoids, sleeping aids, and anxiolytic drugs.

At the second visit (1988–1990), dual-energy x-ray absorptiometry (DXA) was first available, and BMD of the proximal femur (total hip and its subregions) was measured using Hologic QDR 1000 scanners (Hologic, Bedford, MA). The coefficient of variation was 1.2% for both sites (14, 15). Total hip and femoral neck T-scores were calculated using the 20- to 29-yr-old non-Hispanic White NHANES III population as reference. For lumbar spine T-score calculation, the reference mean and sd were provided by Hologic (16).

Participants were contacted every 4 months by postcard (with phone follow-up for nonresponders) after the baseline exam to ascertain incident hip and other nonspine fractures; more than 95% of these contacts were completed. Incident nonspine fractures were physician-adjudicated from radiology reports. Pathological fractures (including peri-prosthetic) and fractures secondary to trauma (e.g. motor vehicle accident) were excluded.

Statistical analysis

Outcome variables were nonvertebral fractures, hip fractures, and wrist fractures. The univariate analysis was performed using a t test for normally distributed continuous variables, a Mann-Whitney U test for other continuous variables, or Fisher's Exact test for dichotomous variables. All variables with a P value < 0.05 were included in a logistic regression model following forwards model selection. This model was first performed with any nonvertebral fracture as outcome and then repeated with outcomes of hip and wrist fractures. Differences were considered significant when the two-tailed P value was <0.05. The statistical analysis was performed using the SPSS statistics package (SPSS, Inc., Chicago, IL) for Windows version 16.0.

Results

BMI was available in 7978 women. Of these, 1480 had a BMI of 30 kg/m² or higher (obese), and information about incident nonvertebral fractures was available in 1377. Of these, 516 (37.5%) had at least one nonvertebral fracture during the follow-up period. Hip and wrist fractures occurred in 9.6 and 7.4%, respectively, and 83% of the fractures affected other sites [humerus (6.2%), ankle (4.0%), foot (3.7%), ribs (3.5%), pelvis (2.8%), and other (62.8%)]. The mean follow-up time was 11.3 (5.9) yr. The incidence of nonvertebral fractures in nonobese women [n = 5968; mean (sd) age was 71.6 (5.1) yr] was 44.3%.

At baseline, 9.5% of the obese women were on hormone replacement therapy. After 8 yr, the percentage was 13.1% and at 18 yr 1.8%. At yr 8, 2.9% of obese postmenopausal women were taking bisphosphonates, increasing to 13.0% at yr 18.

The prevalence of osteoporosis (femoral neck T-score ≤ −2.5) in obese women with incident fracture was 11.7%. Osteopenia (T-score −1 to −2.5) was present in 64.7% and normal BMD (T-score > −1) in 23.6%, compared with 4.6, 51.8, and 43.6% in obese women without fracture (P < 0.0001). The corresponding figures in nonobese women with incident fracture were 54.4, 40.3, and 5.3%. The mean femoral neck T-score in obese women with fracture was −1.52 [95% confidence interval (CI) = −1.59 to −1.44) and in nonobese women with fracture was −2.03 (95% CI = −2.06 to −2.0) (P < 0.0001).

Personal fracture history after age 50 yr and lower femoral neck BMD [mean 0.68 (0.11) vs. 0.73 (0.12) g/cm²; mean difference (95% CI) = −0.056 (−0.043 to −0.068)] were associated with nonvertebral fractures in the univariate analysis. There was no difference in age, weight, waist to hip ratio, years of education, social support network, tobacco use, menopausal age, parental fracture, physical activity, or whole-body fat measurement between obese women with and without nonvertebral fracture. Although there was a tendency for women with nonvertebral fractures to be nonestrogen users, to have worse neuromuscular function, and to report diabetes mellitus, use of sleeping medications, and falls in the past year, none of these associations were statistically significant. Family history of fracture was not a significant risk factor for fragility fracture in obese women. These data are shown in Table 1. Unadjusted odds ratios (OR) for fractures in obese and nonobese women according to previous fracture, family history, and BMD are shown in Table 2.

Table 1

Characteristics of the obese population according to the presence or absence of incident nonvertebral fracture

	Women with fracture (37.5%)	Women without fracture (62.5%)	P
Age (yr)	72.8 (4.6)	72.3 (4.4)	0.08
Weight (kg)	83.0 (8.8)	83.8 (9.1)	0.09
Waist to hip ratio	0.87 (0.07)	0.88 (0.07)	0.51
Years of education	12.1 (2.8)	12.1 (2.7)	0.89
Social support network (Lubben Social Network Scale)	3.2 (0.7)	3.2 (0.7)	0.62
Personal fracture history after age 50 yr	50.2	33.1	<0.0001
Maternal history of fracture after age 50 yr	27.4	27.6	0.51
Current cigarette smoking	4.5	5.7	0.38
Currently drinking coffee	66.4	65.0	0.66
Menopausal status
Years after menopause	23.1 (7.5)	22.3 (7.6)	0.08
Medical conditions
Diabetes (%)	14.1	10.8	0.07
Medications
Oral estrogens (%)	7.4	11.0	0.03
Vitamin D-containing supplements (%)	37.5	35.2	0.412
Calcium supplements (%)	33.6	29.9	0.17
Use of sleep medication (%)	15.9	12.1	0.05
Use of anxiety medication (%)	13.9	14.3	0.87
Use of glucocorticoids (%)	2.9	2.9	1.0
Physical activity
Walking for exercise (yes/no) (%)	34.0	41.0	0.48
Neuromuscular function and physical performance
Grip strength (kg)	19.0 (4.8)	19.7 (4.9)	0.02
Time chair stands (sec)	13.1 (5.0)	13.0 (4.8)	0.74
Quadriceps strength (kg)	28.4 (12.7)	30.4 (13.7)	0.01
Tandem walk (sec)	13.5 (6.6)	12.8 (6.5)	0.01
Walk speed (m/sec)	0.827 (0.21)	0.850 (0.20)	0.05
Whole-body fat measurement
Fat-free mass (kg)	44.6 (4.1)	45.1 (4.2)	0.04
Fat mass (%)	46.1 (3.1)	46.0 (3.2)	0.91
DXA
Femoral neck T-score	−1.52 (0.89)	−1.05 (0.96)	<0.0001
Total hip T-score	−1.12 (1.03)	−0.60 (1.03)	<0.0001
Lumbar spine T-score	−1.26 (1.48)	−0.70 (1.53)	<0.0001
Falls
Falls history in year before evaluation (%)	34.0	28.0	0.05
More than two falls in the year before evaluation (%)	14.6	11.3	0.08

Results are shown as means (sd) or percent. The variables present in this table refer to the second study evaluation with the exception of years of education and maternal history of fracture after age 50 yr, which were documented in the first study evaluation.

Table 2

Unadjusted OR of incident nonvertebral fracture according to previous fracture history, family history of fracture, and BMD in nonobese and obese women

	Nonobese women (BMI < 30 kg/m2)	Obese women (BMI ≥ 30 kg/m2)
Personal fracture history after age 50 yr	1.73 (1.55–1.91)	2.04 (1.63–2.55)
Maternal history of fracture after age 50 yr	1.29 (1.14–1.47)	0.95 (0.72–1.25)
Femoral neck BMD (per 1 sd decrease)	1.60 (1.47–1.75)	2.02 (1.77–2.32)
Osteopenia	1.74 (1.41–2.14)	1.74 (1.31–2.30)
Osteoporosis	3.03 (2.45–3.74)	3.66 (2.60–5.14)

Results are shown as OR (95% CI).

All obese women lost weight during the follow-up period. After 8 yr of follow-up, obese women with fractures had lost [median (interquartile range)] = −22.2% (−34.0 to −11.2) of their weight vs. −21.9% (−32.1 to −10.9) in women without a fracture (P = 0.6). At the end of the follow-up period, weight loss was −28% (−37.9 to −15.4) in women with an incident fracture vs. −23.2 (−35.6 to −14.4) in women without a fracture (P = 0.2). The OR of fractures for 10% weight loss in 8 yr follow-up was 0.78 (95% CI = 0.30–2.02).

When univariate analysis was repeated for hip fracture as an outcome, women with incident hip fracture were significantly older, had been menopausal for longer, were more likely to have a previous fracture history after 50 yr of age, and had lower BMD [femoral neck 0.63 (0.09) vs. 0.72 (0.11) g/cm²] (Table 3). Walk speed and fat mass were significantly lower in women with than without incident hip fracture. When wrist fractures were considered as the outcome, incident wrist fracture was associated with a significantly higher age, greater menopausal age, lower waist to hip ratio, higher prevalence of previous fracture after 50 yr and significantly lower femoral neck BMD [0.64 (0.10) vs. 0.72 (0.12) g/cm²] (Table 4). No other association between characteristics and incident hip or wrist fracture was observed in these secondary analyses.

Table 3

Characteristics of the obese population according to the presence or absence of incident hip fracture

	Women with fracture (9.6%)	Women without fracture (90.4%)	P
Age (yr)	74.0 (4.6)	72.3 (4.4)	<0.0001
Weight (kg)	81.7 (7.3)	83.7 (9.1)	0.01
Waist to hip ratio	0.88 (0.07)	0.87 (0.07)	0.10
Years of education	11.8 (3.0)	12.2 (2.7)	0.16
Social support network (Lubben Social Network Scale)	3.12 (0.73)	3.24 (0.71)	0.09
Personal fracture history after age 50 yr (%)	54.9	38.6	<0.0001
Maternal history of fracture after age 50 yr (%)	33.7	27.0	0.16
Current cigarette smoking (%)	3.6	5.9	0.43
Currently drinking coffee (%)	71.8	64.5	0.13
Menopausal status
Years after menopause	24.1 (7.3)	22.4 (7.7)	0.01
Medical conditions
Diabetes (%)	16.9	11.7	0.08
Medications
Oral estrogens (%)	6.4	9.8	0.23
Vitamin D-containing supplements (%)	34.4	35.7	0.71
Calcium supplements (%)	30.2	31.1	0.92
Use of sleep medication (%)	18.0	13.4	0.15
Use of anxiety medication (%)	9.3	14.8	0.08
Use of glucocorticoids (%)	3.7	3.2	0.80
Physical activity
Walking for exercise (yes/no) (%)	37.5	61.2	1.0
Neuromuscular function and physical performance
Grip strength (kg)	19.1 (4.8)	19.4 (4.8)	0.37
Time chair stands (sec)	12.9 (4.7)	13.0 (4.9)	0.81
Quadriceps strength (kg)	28.4 (13.4)	29.8 (13.4)	0.27
Tandem walk (sec)	14.0 (6.8)	13.0 (6.6)	0.08
Walk speed (m/sec)	0.803 (0.21)	0.847 (0.21)	0.02
Whole-body fat measurement
Fat-free mass (kg)	44.4 (3.85)	45.0 (4.22)	0.08
Fat mass (%)	45.5 (2.92)	46.1 (3.20)	0.02
DXA
Femoral neck T-score	−1.87 (0.75)	−1.16 (0.96)	<0.0001
Total hip T-score	−1.58 (0.83)	−0.73 (1.06)	<0.0001
Lumbar spine T-score	−1.50 (1.49)	−0.86 (1.54)	<0.0001
Falls
Falls history in year before evaluation (%)	27.9	31.0	0.50
More than two falls in the year before evaluation (%)	9.3	12.8	0.28

Results are shown as mean (sd) or percent. The variables present in this table refer to the second study evaluation with the exception of years of education and maternal history of fracture after age 50 yr, which were documented in the first study evaluation.

Table 4

Characteristics of the obese population according to the presence or absence of incident wrist fracture

	Women with fracture (7.4%)	Women without fracture (92.6%)	P
Age (yr)	73.4 (4.9)	72.4 (4.4)	0.03
Weight (kg)	83.3 (8.4)	83.6 (9.0)	0.70
Waist to hip ratio	0.86 (0.06)	0.88 (0.07)	0.002
Years of education	12.0 (2.6)	12.2 (2.7)	0.11
Social support network (Lubben Social Network Scale)	3.19 (0.71)	3.23 (0.71)	0.63
Personal fracture history after age 50 yr (%)	57.8	38.5	<0.0001
Maternal history of fracture after age 50 yr (%)	18.8	28.2	0.07
Current cigarette smoking (%)	5.5	5.3	0.83
Currently drinking coffee (%)	75.6	64.4	0.038
Menopausal status
Years after menopause	24.9 (8.4)	22.4 (7.5)	0.006
Medical conditions
Diabetes (%)	10.1	12.3	0.65
Medications
Oral estrogens (%)	6.4	9.8	0.31
Vitamin D-containing supplements (%)	33.9	35.9	0.76
Calcium supplements (%)	34.3	30.7	0.45
Use of sleep medication (%)	21.1	13.2	0.03
Use of anxiety medication (%)	20.2	13.8	0.09
Use of glucocorticoids (%)	1.8	3.3	0.58
Physical activity
Walking for exercise (yes/no) (%)	37.5	37.7	1.0
Neuromuscular function and physical performance
Grip strength (kg)	18.7 (4.5)	19.5 (4.9)	0.10
Time chair stands (sec)	13.4 (5.5)	12.9 (4.9)	0.69
Quadriceps strength (kg)	27.5 (12.1)	29.8 (13.5)	0.07
Tandem walk (sec)	14.2 (7.5)	13.0 (6.5)	0.10
Walk speed (m/sec)	0.79 (0.22)	0.85 (0.20)	0.01
Whole-body fat measurement
Fat-free mass (kg)	44.9 (4.0)	45.0 (4.2)	0.87
Fat mass (%)	45.8 (3.3)	46.1 (3.2)	0.39
DXA
Femoral neck T-score	−1.61 (0.86)	−1.20 (0.96)	<0.0001
Total hip T-score	−1.12 (1.01)	−0.78 (1.07)	0.002
Lumbar spine T-score	−1.29 (1.42)	−0.89 (1.54)	0.006
Falls
Falls history in year before evaluation (%)	31.2	30.8	0.92
More than two falls in the year before evaluation (%)	13.8	12.4	0.39

Results are shown as mean (sd) or percent. The variables present in this table refer to the second study evaluation, with the exception of years of education and maternal history of fracture after age 50 yr, which were documented in the first study evaluation.

Logistic regression confirmed that previous fracture history after age 50 yr and lower femoral neck BMD were independently associated with incident nonvertebral fractures in obese women. Obese women with a history of prevalent fracture were almost 1.7 times more likely to experience nonvertebral fracture [OR (95% CI) = 1.69 (1.33–2.14); P < 0.0001], and each sd decrease in BMD was associated with a 1.6-fold increase in the odds of nonvertebral fracture [1.62 (1.42–1.85); P < 0.0001]. The nonsignificant variables removed from the final model were oral estrogens, use of sleep medication, quadriceps strength, tandem walk, walk speed, fat-free mass, and falls history in the year before evaluation.

Only lower femoral neck BMD was an independent risk factor for hip fracture [OR = 2.61 (2.07–3.30) per 1 sd decrease in BMD; P < 0.0001]. Age, weight, personal fracture history after age 50 yr, years after menopause, walk speed, and fat mass were not associated with hip fracture.

Both a previous fracture history after age 50 yr and lower femoral neck BMD were associated with incident wrist fracture. In addition, number of years after menopause, waist to hip ratio, and the use of sleep medication were also associated with wrist fractures in this model (Table 5). These results were unchanged by introduction of an interaction term between fracture history and femoral neck T-score in the models (data not shown).

Table 5

Final logistic regression model for risk factors for incident wrist fracture in obese women after forward model selection with likelihood ratio testing at the 5% significance level

	OR (95% CI)	P
Personal fracture history after 50 yr old	1.65 (1.03–2.66)	0.038
Femoral neck BMD (per 1 sd decrease)	1.40 (1.07–1.84)	0.015
Years after menopause (per 1 yr increase)	1.04 (1.01–1.07)	0.014
Waist to hip ratio (per 0.1 increase)	0.58 (0.41–0.81)	0.002
Use of sleep medication in the year before evaluation	1.99 (1.11–3.57)	0.021

Variables included in the model are age, waist to hip ratio, personal fracture history after age 50 yr, currently drinking coffee, years after menopause, use of sleep medication, walk speed, and femoral neck BMD. Nonsignificant variables removed from the final model are age, currently drinking coffee, and walk speed.

Nonvertebral fractures other than hip and wrist fractures were also associated with lower BMD and previous fracture history in obese women. The regression model showed an OR (95% CI) of 5.05 (4.28–5.97) per sd decrease in femoral neck BMD and 1.33 (1.02–1.75) for previous fracture history.

Discussion

Our results suggest that nonvertebral fractures in obese postmenopausal women exhibit some characteristics of fragility fractures, being associated with significantly lower femoral neck BMD and a higher prevalence of previous fracture than in obese women without fracture. Other risk factors, including maternal history of fracture, smoking, alcohol consumption, glucocorticoid use, and diabetes, were similar in prevalence between obese women with and without fracture. Although there was a trend toward poorer neuromuscular function and higher frequency of falls in obese women with incident fracture, this failed to achieve statistical significance.

Of the obese women in the cohort, 37.5% sustained a nonvertebral fracture during the follow-up period compared with 44% of nonobese women. These data confirm our previously reported finding that fractures in obese postmenopausal women make a substantial contribution to the overall fracture burden in this population (9). Further evidence is provided by data from the Global Study of Osteoporosis in Women, a multicenter prospective observational study of 60,393 women aged 55 yr or more (17). In this cohort, the prevalence of fracture after age 45 yr was 23 and 24% in obese and nonobese women, respectively, fractures of the upper arm, ankle, and upper and lower leg being significantly more common and fractures of the wrist, pelvis, and hip significantly less common in obese women compared with nonobese women (18).

In this study, we have shown that BMD is significantly lower in obese women with fracture than in women of similar age and weight who have not sustained a fracture. Measurement of BMD in obese individuals is subject to errors in both accuracy and precision as a result of the increase in fat mass and changes in its distribution (19). Experiments using blocks of lard to simulate body fat indicate that the effect of obesity varies according to the type of DXA system used for measurement (pin-beam vs. fan beam), the distribution of body fat, the software versions used by the manufacturer, and the scan mode (20). Nevertheless, because body weight and composition were closely similar in obese women with and without fracture in this study, there is little reason to doubt the validity of the finding of a significantly lower BMD in those with fracture.

Higher BMD in obese individuals has been widely believed to protect against fracture, although the inverse association between BMI and fracture risk is mainly determined by the effect of low BMI on fracture risk (3, 4). The higher BMD associated with obesity therefore provides a skeleton that is appropriate for higher body weight because of the greater mechanical stresses likely to be experienced, whereas in a lean person, equivalent strength is provided by a lower BMD.

In particular, the higher serum PTH levels associated with obesity may result in a lower cortical volumetric BMD, as reported recently by Sukumar et al. (21, 22). In addition, greater trauma associated with falling in obese individuals as a result both of their higher body weight and poor protective responses may contribute to the development of fractures at higher BMD values (8, 23).

The finding of a significantly lower BMD on average in obese women with fracture compared with those without fracture, together with their significantly higher prevalence of previous fracture, provides important new evidence that fractures in the obese exhibit characteristics of fragility fractures. Whether and how risk factors for fracture in obese and nonobese women differ requires further investigation. In Global Study of Osteoporosis in Women, obese women with prevalent fragility fractures were significantly more likely to be cortisone users, to report early menopause, to report fair or poor general health, to use arms to assist standing from a sitting position, and to report at least two falls in the past year (18).

Evidence for the antifracture efficacy of pharmacological interventions in postmenopausal women at increased risk of fracture is well documented (24–26). However, this is based predominantly on studies in women with low BMD, the majority of whom were not obese and the efficacy of intervention in obese women with higher BMD has not been established. In a recent randomized controlled trial in which women were not selected on the basis of low BMD, a significant interaction between BMI and antifracture efficacy was shown in women treated with clodronate, fracture reduction being less in women with a BMI of at least 30 kg/m²; a similar, although not statistically significant interaction remained when femoral neck BMD was included in calculation of fracture risk (27). The management of fractures in obese postmenopausal women therefore requires further study, including investigation of the possibility that higher doses of bone-protective agents might be required than for nonobese women.

Our study has some limitations. SOF is a cohort of community-dwelling volunteers and not a population-based sample. However, characteristics of the SOF participants are similar to, or healthier than, those of the population-based NHANES III, and the prevalence of obesity in SOF (18.3%) was somewhat lower than that of White women in the general population (22.4%) (28). Second, we analyzed only nonvertebral fractures, and our results may not be applicable to vertebral fractures. In a recent study of 362 postmenopausal women, BMI was positively associated with the presence of vertebral fractures independent of age and other recognized risk factors, but these data require confirmation in prospective studies (29). Third, only White postmenopausal women aged 65 yr and older were included in this analysis, and it remains to be established whether our results can be generalized to men, younger women, and individuals of different races. Finally, for some risk factors, the study may not have had sufficient power to demonstrate significant differences between obese women with and without fracture.

We have shown that despite their higher average BMD, more than one third of the obese older women sustained a nonvertebral fracture during the follow-up period of 11.3 yr. Our results also demonstrate that, similar to the nonobese postmenopausal population, lower BMD and previous fractures are strong risk factors for fracture in obese older women. The pathogenesis of fractures in obese individuals has not been clearly defined, and additional studies are required to investigate the changes in bone density and composition associated with obesity. The significant incidence of fractures in obese women has important public health implications in view of the rapidly increasing incidence of obesity (30, 31) and emphasizes the need to develop effective strategies to reduce fracture risk in this population.

Abbreviations:

 
• BMD

  Bone mineral density

 
• BMI

  body mass index

 
• DXA

  dual-energy x-ray absorptiometry

 
• OR

  odds ratio

 
• SOF

  Study of Osteoporotic Fractures.

Acknowledgments

This work was supported by the National Health Service National Institute of Health Research and the Cambridge Biomedical Research Centre. M.O.P. was supported by CAPES foundation, Ministry of Education, Brazil (Grant 1204-08-4). The SOF is supported by National Institutes of Health funding. The National Institute on Aging provides support under the following grant numbers: AG05407, AR35582, AG05394, AR35584, AR35583, R01 AG005407, R01 AG027576–22, 2 R01 AG005394-22A1, and 2 R01 AG027574-22A1.

Disclosure Summary: The authors have no conflict of interest to disclose.