SEARCH

SEARCH BY CITATION

Keywords:

  • Hip Fracture;
  • Physical Activity;
  • BMI;
  • PROSPECTIVE STUDY;
  • POSTMENOPAUSAL WOMEN

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Hip fracture risk is known to increase with physical inactivity and decrease with obesity, but there is little information on their combined effects. We report on the separate and combined effects of body mass index (BMI) and physical activity on hospital admissions for hip fracture among postmenopausal women in a large prospective UK study. Baseline information on body size, physical activity, and other relevant factors was collected in 1996–2001, and participants were followed for incident hip fractures by record linkage to National Health Service (NHS) hospital admission data. Cox regression was used to calculate adjusted relative risks of hip fracture. Among 925,345 postmenopausal women followed for an average of 6.2 years, 2582 were admitted to hospital with an incident hip fracture. Hip fracture risk increased with decreasing BMI: Compared with obese women (BMI of 30+ kg/m2), relative risks were 1.71 [95% confidence interval (CI) 1.47–1.97)] for BMI of 25.0 to 29.9 kg/m2 and 2.55 (95% CI 2.22–2.94) for BMI of 20.0 to 24.9 kg/m2. The increase in fracture risk per unit decrease in BMI was significantly greater among lean women than among overweight women (p < .001). For women in every category of BMI, physical inactivity was associated with an increased risk of hip fracture. There was no significant interaction between the relative effects of BMI and physical activity. For women who reported that they took any exercise versus no exercise, the adjusted relative risk of hip fracture was 0.68 (95% CI 0.62–0.75), with similar results for strenuous exercise. In this large cohort of postmenopausal women, BMI and physical activity had independent effects on hip fracture risk. © 2011 American Society for Bone and Mineral Research.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Hip fractures result in a significant public health burden in the United Kingdom.1 Previous studies have reported a decreased risk of hip fracture among physically active women.2–9 Regular physical activity also has been associated with a lower risk of osteoporosis, obesity,10 and falls.11, 12 The protective effect of physical activity on hip fracture has been attributed to increased forces on the bones that may slow the rate of decline in bone mineral density (BMD) related to aging13, 14 and/or prevent falls by improving balance, coordination, and muscular strength.12 Obesity protects against hip fractures2, 3, 15 but increases the risk of having other conditions.16 This protective effect has been attributed to higher levels of BMD resulting from increased chronic strain on the bones, an increased production of estrogens from larger stores of adipose tissue, and/or a reduced impact from falls because of a greater cushioning by subcutaneous adipose tissue.17–19 Since body mass index (BMI) may be part of the causal pathway for the effect of physical activity on hip fracture risk, and because active people tend to have a lower body weight than inactive people,9, 20 the independent effects of BMI and physical activity on hip fracture risk need clarification.

Women have a higher incidence of hip fracture than men, and postmenopausal women have a higher risk than premenopausal women of the same age.21 The aim of this study was to describe the separate and combined effects of BMI and physical activity on the incidence of hip fracture in postmenopausal women.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Participants and data

The Million Women Study is a prospective cohort study designed to examine the effects of hormone-replacement therapy (HRT) usage and other major risk factors for a variety of common diseases in women, as described elsewhere.22 Briefly, in 1996–2001, 1.3 million women who were invited for breast screening by the National Health Service (NHS) completed a self-administered questionnaire on anthropometry, physical activity, and other factors, including the use of HRT, and gave written informed consent for follow-up. Repeat questionnaires were sent out at approximately 3-yearly intervals (see www.millionwomenstudy.org).

Each study participant was registered with the NHS with a unique NHS identification number. The NHS number was used, in conjunction with other personal details, to link to information on NHS hospital admissions, cancer registrations, deaths, and emigrations. The details of hospital admissions of the study participants were available for women in Scotland from January 1, 1981 through December 31, 2003,23 and for women in England from April 1, 1997 through March 31, 2005.24 Date of admission and principal reason for admission are coded by those providing the data for each hospital episode, with up to a maximum of 13 further clinical diagnoses. All diagnoses are coded according to the World Health Organization's International Classification of Diseases, 9th and 10th revisions (ICD-9 and ICD-10).25 In addition, up to 12 operations and procedures are coded using the Office of Population Censuses and Surveys Classification of Surgical Operations and Procedures, 3rd and 4th revisions (OPCS-3 and OPCS-4).26 The coding of all diseases and operations in our study was done by those providing the hospital data. For the analyses, incident hip fracture was defined as the first hospital admission with a fractured neck of the femur as the primary diagnosis (ICD-10 codes S72.0–S72.2) recorded after recruitment into the Million Women Study. The Oxford and Anglia Multi-Centre Research Ethics Committee provided ethical approval for the study.

Body mass index (BMI), physical activity, and other factors

Height was self-reported at recruitment in feet and inches and weight in stones and pounds and converted to centimeters and kilograms, respectively. BMI was calculated as weight in kilograms divided by the square of height in meters. Subsequently, height and weight were measured in a randomly selected validation subsample of 2772 study participants. Women were classified by categories of BMI based on baseline self-reported data (<20, 20.0 to 22.4, 22.5 to 24.9, 25.0 to 27.4, 27.5 to 29.9, and ≥ 30.0 kg/m2).

Each questionnaire sent to Million Women Study participants asked about physical activity. At recruitment, the question asked was: “How often do you do any strenuous exercise? (that is, enough to cause sweating or a fast heart beat),” and another question was added after the first 9% of women had been recruited: “How often do you do any exercise?” Study participants could answer rarely/never, less than once a week, once a week, 2 to 3 times a week, 4 to 6 times a week, or every day for each question. Where the activity levels of women are described here, inactive means those reporting rarely/never participating in the given activity, whereas active refers to those who reported participating in the respective activity more often.

Approximately 3 years after recruitment, women were asked different questions on physical activity (see www.millionwomenstudy.org). The questions asked then were based on session-based measures of physical activity used in the International Physical Activity Questionnaire27 and the Active Australia Survey.28 Responses to these questions on the number of hours spent doing housework, gardening, walking, cycling, and doing any activity that caused sweating or a fast heartbeat (averaging summer and winter values) were used to estimate energy expenditure, assigning metabolic equivalents (METs) to each activity using the compendium of physical activities by Ainsworth and colleagues.29 All METs of physical activity were calculated as excess activity above baseline MET values. Women also were asked: “In the last year, how many falls have you had?” Responses to these questionnaires and estimated METs were compared across categories of physical activity reported at recruitment.

Statistical analysis

The statistical package Stata, Version 10.1,30 was used for all analyses. Analyses were restricted to postmenopausal women, defined as those women who reported having undergone natural menopause or bilateral oophorectomy or women 53 years of age or older at baseline whose menopausal status at recruitment was unknown. Women were excluded from all analyses if they had missing data on BMI or strenuous physical activity or had been diagnosed with cancer before recruitment or if they reported prior treatment for osteoporosis or stroke.

Person-years were calculated from the date of recruitment to whichever came first, the date of first hospital admission for hip fracture, the date of emigration, the date of death, or the end of follow-up. The last date of follow-up was March 31, 2005, for participants in England and December 31, 2003, for participants in Scotland. Record linkage was incomplete for hospital records of later dates. For women recruited in England prior to the availability of hospital records (n = 61,228), person-years were calculated from the first date of available hospital records, April 1, 1997.

Cox proportional hazards models were used to estimate the relative risk (RR) and 95% confidence interval (CI) for incident hip fractures according to BMI and frequency of physical activity. Attained age was used as the underlying time variable. Heterogeneity was assessed across categories of these exposures using likelihood-ratio tests. All analyses were stratified by and adjusted for variables recorded at recruitment. Stratification was by geographic region (10 regions), and adjustments were for smoking status (ie, current, prior, or never), alcohol consumption (ie, 0, 1 to 2, 3 to 6, 7 to 14, or ≥ 15 drinks per week), socioeconomic status (quintiles), parity (ie, nulliparous or parous), HRT use (ie, never, prior, or current), history of heart disease/thrombosis (yes, no), osteo/rheumatoid arthritis (yes, no), thyroid disease (yes, no), and diabetes (yes, no). Some analyses additionally were adjusted for height (ie, <155, 155.0 to 159.9, 160.0 to 164.9, 165.0 to 169.9, or ≥ 170 cm), BMI (ie, <20, 20.0 to 22.4, 22.5 to 24.9, 25.0 to 27.4, 27.5 to 29.9, or ≥ 30.0 kg/m2), and strenuous physical activity [ie, rarely/never (inactive), at most once per week, or more than once per week] as appropriate. Crude incidence ratios also were calculated per 100 over 5 years. Missing values for adjustment variables (generally less than 2% for each variable) were assigned to a separate category for that variable. We assessed our analyses for the proportional-hazards assumption and established that this assumption was not violated by our models.

A likelihood-ratio test was used to determine if there was an interaction between BMI and physical activity and risk of hip fracture, as well as between BMI and HRT use and physical activity and HRT use with respect to hip fracture risk. When more than two categories were used in risk comparisons, results are presented in the tables and figures as relative risks (RRs) with their corresponding floated confidence interval (FCI)31 so that valid comparisons can be made between any two groups. When only two groups were compared, and when reporting RRs in the text, conventional confidence intervals were used.

Measurement errors in BMI cause regression dilution in the effects of BMI.32 Since in some studies heavier individuals underreport their weight to a greater degree than leaner individuals,33, 34 height and weight were measured in the validation subsample. BMI based on self-reported and measured data were correlated (Fig. 1; correlation coefficient = 0.85). The line of best fit between measured and self-reported values is almost parallel to the line of equality. This suggests that substantial regression dilution is unlikely; nevertheless, to account for any residual confounding owing to measurement error, we plotted relative risks against the mean measured BMI value within each category of self-reported BMI and used these in analyses of trends in risk per unit change in BMI. By plotting category-specific relative risks against the mean measured BMI values in each category, results based on self-reported data can be interpreted on a more objective scale.

thumbnail image

Figure 1. Scatter plot of BMI calculated from measured height and weight versus BMI calculated from self-reported height and weight.

Download figure to PowerPoint

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

In total, 925,345 women were eligible for inclusion in these analyses, with an average follow-up of 6.2 years per woman (5.7 million person-years in total). During this follow-up period, 2582 women had a first hospital admission with a primary diagnosis of hip fracture. Of these fractures, 2091 were classified as a fracture of the neck of the femur, 409 as a pertrochanteric fracture, and 82 as a subtrochanteric fracture. Hip fracture incidence rates for women 50 to 54, 55 to 64, and 65 to 74 years of age were 0.8, 1.7, and 4.4, per 1000 over 5 years, respectively. The cumulative risk of being admitted to hospital with a first hip fracture from age 50 to age 74 was 13 per 1000 women.

Table 1 shows various baseline characteristics for women admitted to hospital with hip fractures and for the population at risk. Compared with the total study population, women admitted with an incident hip fracture were, on average, older, taller, weighed less, had a lower BMI, and were more likely to smoke and not to use HRT.

Table 1. Baseline Characteristics of the Study Population and Those With Incident Hip Fracture for Postmenopausal Women in the Million Women Studya
Characteristics at recruitmentWomen with incident hip fracture (n = 2582)Population at risk (n = 925,345)
  • a

    Women with missing values for a variable were excluded when calculating the means or percentages for that given variable.

Mean age, years (SD)60.3 (5.4)57.4 (4.4)
Mean height, cm (SD)163.6 (7.1)162.0 (6.6)
Mean weight, kg (SD)65.6 (11.6)68.9 (12.5)
Mean BMI, kg/m2 (SD)24.5 (4.3)26.3 (4.6)
Mean alcohol, g/d (SD)5.7 (7.9)6.1 (7.5)
Mean number of children (SD)2.2 (1.4)2.2 (1.3)
Current smokers (%)28.319.4
Socioeconomic status: Lowest fifth (%)23.119.1
Never users of HRT (%)63.549.0
No physical activity (%)28.621.2
No strenuous activity (%)61.048.8

Table 2 shows, for categories of physical activity and BMI at recruitment, the average number of hours spent walking each week, energy expenditure (METs), and frequency of falls. As expected, increasing levels of physical activity at baseline were associated with decreasing BMI, a greater number of hours spent walking, greater levels of energy expenditure, and fewer falls in the previous year. Conversely, increasing BMI at baseline was associated with less physical activity (including walking and energy expenditure) and increasing proportions reporting a fall in the previous year. Among women experiencing a hip fracture in this study, 36% reported falling at least once in the year prior to the second resurvey compared with 28% of women without an incident hip fracture.

Table 2. Association Between Physical Activity and BMI Based on Information Reported at Recruitment With Subsequent Measured BMI and Subsequent Reporting of Factors Associated with Physical Activity or Fracture Risk
Recorded at baselineMean measured BMI [kg/m2 (SD)]aMean hours spent walking per week (SD)bMetabolic equivalents (METs)Falls in the previous year (%)c
Strenuous activity (SD)bAny activity (SD)b
  • a

    Among 2772 randomly selected study participants with measured height and weight, restricted to postmenopausal women without prior treatment for osteoporosis or stroke.

  • b

    Among 434,814 study participants who reported information on any physical activity at subsequent survey, restricted to postmenopausal women without prior treatment for osteoporosis or stroke.

  • c

    Among 12,526 randomly selected study participants; the percent that reported one or more falls in the previous year, restricted to postmenopausal women without prior treatment for osteoporosis or stroke.

BMI categories (kg/m2)
 <20.020.4 (2.2)5.1 (5.6)9.6 (24.8)63.0 (43.5)22.9%
 20.0–22.422.6 (2.0)5.0 (5.3)10.4 (23.1)62.4 (40.7)24.3%
 22.5–24.925.2 (2.2)4.8 (5.2)9.8 (22.6)61.0 (40.4)25.1%
 25.0–27.427.7 (2.5)4.6 (5.1)9.1 (22.6)59.9 (40.8)28.1%
 27.5–29.930.4 (2.7)4.4 (5.2)8.9 (23.2)59.1 (41.3)33.1%
 30+34.5 (4.6)4.0 (5.1)8.7 (24.7)56.7 (42.8)34.3%
Strenuous physical activity
 Rarely/never (inactive)26.2 (4.3)3.9 (5.0)5.3 (20.3)52.8 (38.2)29.3%
 Up to once per week25.2 (3.9)4.7 (4.9)9.2 (19.6)60.3 (37.9)27.1%
 More than once per week24.6 (3.5)5.7 (5.8)18.0 (30.1)74.1 (47.5)27.8%
Any physical activity
 Rarely/never (inactive)26.7 (4.5)3.2 (5.0)5.4 (23.5)50.6 (40.6)30.5%
 Up to once per week26.2 (4.5)3.6 (4.4)7.2 (19.1)53.3 (36.6)27.9%
 2 to 3 times per week25.3 (3.8)4.4 (4.5)10.9 (19.9)60.4 (37.0)27.2%
 More than 3 times per week24.6 (3.5)6.1 (5.9)12.1 (27.3)69.7 (45.4)27.9%

The relative risk of hip fracture was higher the thinner the women were (Table 3). Compared with obese women, for women of healthy weight (BMI of 20.0 to 24.9 kg/m2), the RR of hip fracture was 2.55 (95% CI 2.22–2.94), and for overweight women (BMI of 25.0 to 29.9 kg/m2), the RR for hip fracture risk was 1.71 (95% CI 1.47–1.97). From Fig. 2 it is evident that this trend is especially steep in the leanest women; therefore, we further examined trends of the RR of hip fracture separately for women with BMIs above and below 25.0 kg/m2. The decrease in risk of hip fracture per 5-unit increase in BMI among normal and underweight women was 57% (RR per 5 kg/m2 = 0.43, 95% CI 0.38–0.49), significantly greater than the 37% decrease in risk per 5-unit increase in BMI in overweight and obese women (RR per 5 kg/m2 = 0.63, 95% CI 0.56−0.71) (p < .001 for heterogeneity).

Table 3. Relative Risk of Hip Fracture in Postmenopausal Women According to BMI and Frequency of Physical Activity
 Incident cases of hip fracture/population at riskIncidence rate per 100 over 5 yearsRelative risk for hip fracture
Minimal adjustmentaFully adjustedb (95% FCIc)
  • a

    Stratified by study region and adjusted for age and socioeconomic status.

  • b

    Stratified by study region and adjusted for age and socioeconomic status and additionally adjusted, where appropriate, for smoking, alcohol consumption, parity, use of HRT, height, heart disease/thrombosis, diabetes mellitus, thyroid disease, rheumatoid arthritis/osteoarthritis, and strenuous activity or BMI.

  • c

    FCI = floating confidence interval for RR.

  • d

    Information not collected on 9% of the study population (see “Materials and Methods”).

BMI (kg/m2) (mean measured)
 < 20.0 (20.4)262/33,3000.645.485.29 (4.68−5.98)
 20.0−22.4 (22.6)591/140,4010.343.003.24 (2.99−3.52)
 22.5−24.9 (25.2)703/249,3920.231.972.20 (2.04−2.37)
 25.0−27.4 (27.7)501/204,4390.201.661.84 (1.69−2.01)
 27.5−29.9 (30.4)283/132,7100.171.421.52 (1.36−1.71)
 30 + (34.5)242/165,1030.121.001.00 (0.88−1.14)
p Value (heterogeneity)   < .001
Strenuous activity
 Rarely/never (inactive)1574/451,5920.281.001.00 (0.95−1.06)
 Up to once per week594/279,0670.170.680.70 (0.65−0.76)
 More than once per week414/194,6860.170.670.64 (0.58−0.71)
p Value (heterogeneity)   < .001
Any activityd
 Rarely/never (inactive)639/177,4430.301.001.00 (0.92−1.09)
 Up to once per week449/193,9190.190.700.76 (0.69−0.84)
 2 to 3 times per week413/199,2310.170.590.60 (0.55−0.67)
 More than 3 times per week733/267,1700.230.750.68 (0.63−0.73)
p Value (heterogeneity)   < .001
thumbnail image

Figure 2. Hip fracture risk in postmenopausal women, by BMI. Relative risks, adjusted for age, region, socioeconomic status, smoking, alcohol consumption, parity, use of HRT, height, heart disease/thrombosis, diabetes mellitus, thyroid disease, rheumatoid arthritis/osteoarthritis, and strenuous activity. Relative risks are plotted against the mean measured BMI in each category (< 18.5, 18.5 to 19.9, 20.0 to 22.4, 22.5 to 24.9, 25.0 to 27.4, 27.5 to 29.9, and ≥ 30.0 kg/m2).

Download figure to PowerPoint

Participating in strenuous or any physical activity was associated with reduced risk of hip fracture (Table 3). The most substantial decrease in hip fracture risk was between women who reported rarely/never being active compared with those participating in activity up to once per week. For strenuous exercise versus no strenuous exercise, there was a 32% reduction in hip fracture risk (RR = 0.68, 95% CI 0.62–0.73), and for any exercise versus no exercise, there was a 32% reduction in hip fracture risk (RR = 0.68, 95% CI 0.62–0.75).

The separate effects of BMI and physical activity with hip fracture risk are shown in Figs. 3 and 4. The greatest relative risk was found in inactive women with a BMI of less than 20 kg/m2. Although a progressive decrease in risk of hip fracture was noted with each 5 kg/m2 increment in BMI, active women, irrespective of activity level, were consistently at a lower risk for hip fracture than inactive women within each BMI category. Differences in risk with increasing levels of physical activity were not as clear, and there was no statistically significant interaction between BMI and either strenuous physical activity (p = .92) or any physical activity (p = .87).

thumbnail image

Figure 3. Relative risk of hip fracture in postmenopausal women by frequency of strenuous physical activity and BMI. Relative risks, adjusted for age, region, socioeconomic status, smoking, alcohol consumption, parity, use of HRT, height, heart disease/thrombosis, diabetes mellitus, thyroid disease, and rheumatoid arthritis/osteoarthritis. Relative risks are plotted against the mean measured BMI in each category.

Download figure to PowerPoint

thumbnail image

Figure 4. Relative risk of hip fracture in postmenopausal women by frequency of any physical activity and BMI. Relative risks, adjusted for age, region, socioeconomic status, smoking, alcohol consumption, parity, use of HRT, height, heart disease/thrombosis, diabetes mellitus, thyroid disease, and rheumatoid arthritis/osteoarthritis. Relative risks are plotted against the mean measured BMI in each category.

Download figure to PowerPoint

Several sensitivity analyses were done (Table 4). When analyses were restricted to women without missing data for any of the adjustment variables, results showed similar risk relationships, as did results from a sensitivity analysis that excluded the first 2 years of follow-up. There appeared to be small differences in the pattern of risk of hip fracture according to BMI between current users of HRT and those not currently using HRT, but these differences were not significant (p = .15 for interaction). There also were no significant interactions between HRT use and strenuous activity (p = .16) or between HRT use and any activity (p = .30). Although a sensitivity analysis restricted to nondiabetics gave similar risk relationships as the full sample (Table 4), all analyses were adjusted for diabetic status because diabetics are at an increased risk of hip fracture.35 There were 155 hip fractures among diabetic women, too few for separate sensitivity analyses subdivided by BMI and physical activity. Finally, we considered the differences in risk relationships between two fracture subtypes. The inverse association between BMI and risk of fracture was stronger for fractures that occurred at the neck of the femur than it was for pertrochanteric fractures. There were too few subtrochanteric fractures to repeat a similar sensitivity analysis among this group.

Table 4. Sensitivity Analysis of Relative Risk of Hip Fracture in Postmenopausal Women According to BMI and Frequency of Physical Activity
 Relative risk for hip fracturea (95% FCIb)
Excluding the first 2 years of follow-upExcluding women with missing values for adjustment variablesRestricted to current users of HRTRestricted to women not currently using HRTRestricted to women without diabetesRestricted to cases of fractured neck of femurRestricted to cases of pertrochanteric fracture
  • a

    Stratified by study region and adjusted for age and socioeconomic status and additionally adjusted, where appropriate, for smoking, alcohol consumption, parity, use of HRT, height, heart disease/thrombosis, diabetes mellitus, thyroid disease, rheumatoid arthritis/osteoarthritis, and strenuous activity or BMI.

  • b

    FCI = floating confidence interval for RR.

  • c

    Information not collected on 9% of the study population (see “Materials and Methods”).

BMI (kg/m2) (mean measured)
 < 20.0 (20.4)5.12 (4.45−5.89)5.54 (4.88−6.29)3.58 (2.67−4.79)5.83 (5.08−6.67)5.14 (4.54−5.82)6.11 (5.34−6.99)3.39 (2.43−4.72)
 20.0−22.4 (22.6)3.17 (2.89−3.48)3.22 (2.96−3.52)2.32 (1.92−2.80)3.44 (3.13−3.77)3.08 (2.84−3.35)3.47 (3.16−3.80)2.98 (2.46−3.60)
 22.5−24.9 (25.2)2.26 (2.08−2.45)2.18 (2.02−2.36)1.77 (1.50−2.09)2.27 (2.09−2.47)2.06 (1.91−2.23)2.43 (2.24−2.64)1.66 (1.38−2.01)
 25.0−27.4 (27.7)1.84 (1.67−2.03)1.86 (1.70−2.04)1.47 (1.20−1.79)1.90 (1.72−2.10)1.72 (1.57−1.88)2.13 (1.94−2.34)1.03 (0.79−1.33)
 27.5−29.9 (30.4)1.57 (1.38−1.79)1.54 (1.36−1.74)1.42 (1.10−1.85)1.52 (1.33−1.74)1.45 (1.29−1.64)1.68 (1.47−1.91)1.03 (0.75−1.40)
 30+ (34.5)1.00 (0.86−1.16)1.00 (0.87−1.15)1.00 (0.74−1.36)1.00 (0.87−1.16)1.00 (0.87−1.15)1.00 (0.86−1.16)1.00 (0.75−1.33)
p Value (heterogeneity)< .001< .001< .001< .001< .001< .001< .001
Strenuous activity
 Rarely/never (inactive)1.00 (0.94−1.06)1.00 (0.95−1.06)1.00 (0.88−1.13)1.00 (0.94−1.06)1.00 (0.95−1.06)1.00 (0.94−1.06)1.00 (0.88−1.14)
 Up to once per week0.71 (0.65−0.78)0.71 (0.65−0.77)0.77 (0.65−0.91)0.69 (0.63−0.75)0.70 (0.65−0.76)0.70 (0.64−0.77)0.66 (0.53−0.81)
 More than once per week0.66 (0.59−0.74)0.64 (0.58−0.71)0.56 (0.45−0.71)0.66 (0.60−0.74)0.65 (0.59−0.72)0.66 (0.60−0.74)0.56 (0.43−0.72)
p Value (heterogeneity)< .001< .001< .001< .001< .001< .001< .001
Any activityc
 Rarely/never (inactive)1.00 (0.91−1.10)1.00 (0.92−1.09)1.00 (0.83−1.21)1.00 (0.91−1.10)1.00 (0.92−1.09)1.00 (0.91−1.09)1.00 (0.81−1.23)
 Up to once per week0.81 (0.73−0.89)0.75 (0.68−0.83)0.65 (0.53−0.81)0.79 (0.71−0.88)0.77 (0.70−0.84)0.76 (0.69−0.84)0.74 (0.58−0.95)
 2 to 3 times per week0.61 (0.55−0.68)0.60 (0.54−0.67)0.63 (0.51−0.77)0.59 (0.53−0.66)0.62 (0.56−0.69)0.60 (0.54−0.67)0.63 (0.49−0.80)
 More than 3 times per week0.73 (0.67−0.79)0.67 (0.62−0.72)0.72 (0.61−0.85)0.66 (0.61−0.72)0.68 (0.63−0.74)0.67 (0.62−0.73)0.70 (0.59−0.85)
p Value (heterogeneity)< .001< .001.005< .001< .001< .001.026

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

In this study of almost a million postmenopausal women followed for an average of 6.2 years each, although BMI and physical activity were inversely related, each had an independent effect on the risk of hip fracture. Hip fracture risk decreased with increasing BMI and was greater in inactive than active women. Women who reported participating in any or in strenuous exercise had a 30% reduction in fracture risk compared with women who exercised rarely or never. In addition to this, decreasing BMI was associated with a greater adverse effect in lean and normal-weight women than in overweight and obese women. Physically active women had lower risks of hip fracture than inactive women at every level of BMI.

Women's age has a profound effect on the incidence of hip fracture, with rates approximately seven times higher among women aged 70 to 74 years than among those aged 50 to 54 years.21 The cumulative risk for first hospital admission for hip fracture between the ages of 55 and 74 years was 12 per 1000. A study from Nottingham, United Kingdom, reported cumulative risks between 55 and 74 years of 29 per 1000 (using census data to calculate rates). It is unclear why these rates differ.

It is well known that overweight and obese women are at a reduced risk of hip fracture,2, 3, 15 and our results are consistent with those from previous reports. A meta-analysis of 12 prospective population-based cohorts with 1141 incident hip fractures investigated whether the impact of BMI on hip fracture risk varies according to the absolute level of BMI.15 Our findings based on 2582 hip fractures cases, which were adjusted for measurement errors, are similar. Women were categorized on the basis of self-reported data, but there is an excellent correlation between BMI calculated from self-reported and objectively measured data (Fig. 1; correlation coefficient = 0.85), and by plotting our RR estimates against mean measured BMIs (rather than self-reported values) for each BMI category, we effectively correct for differences between self-reported and measured values.32

Three possible mechanisms through which the protective effect of higher BMI is mediated have been suggested. First, an increased strain on the bones imposed by higher body mass may lead to increased BMD and improved structural integrity of the bones.17 A second protective mechanism is that the greater amount of adipose tissue available enhances the ability to produce endogenous estrogens, which preserve BMD.18 Third, women who carry more fat mass may benefit from cushioning of their hip by gluteofemoral adipose tissue, which reduces impact forces when they fall and hence their chance of fracture.19

A protective association of physical activity, often measured as leisure time activity in older studies, with hip fracture has been reported previously.2–9 One previous study has considered the potential competing effect of BMI with respect to the effects of walking and other leisure time activities on hip fracture risk. Total physical activity, made up of both leisure time and incidental physical activities, is increasingly recognized as a key determinant of disease risk.36, 37 For many people, particularly women, domestic activities contribute substantially to overall physical activity levels,38 and the contribution from leisure activities is of lesser importance.36 In the Million Women Study, energy expenditure (METs) from all activities, including both domestic activities and purposeful physical activity, is strongly associated with physical activity reported at baseline (Table 2). The measures of physical activity used in this article were based on validated session-based physical activity measures used in the International Physical Activity Questionnaire27 and the Active Australia Survey,28, 39 but we did not conduct our own separate validation study.

Most bone mineral deposition occurs during the first two to three decades of life,40 but the rate at which this is lost over the lifetime can be slowed by weight-bearing physical activity.13, 14 Physical activity also may decrease the risk of falls by improving balance, coordination, and muscular strength.11, 12 Our results corroborated this, showing that more active women were less likely to experience one or more falls subsequently.

The beneficial effects of physical activity on BMD, strength, balance, and coordination are likely to result from the variety of activities constituting general physical activity. Our results show a direct relationship between hip fracture incidence and strenuous and other activity. Physically active women generally are less likely to be frail and to have major comorbidities, key risk factors for hip fracture.3 However, the benefits of exercise on hip fracture risk also may be tempered by an increased risk of injury during leisure time activities because a higher number of leisure time and sports-related injuries have been reported among physically active adults when compared with inactive adults.41

That a higher body mass and physical activity each have an independent effect on hip fracture risk may seem contradictory because these factors are inversely associated in this study as in other populations.20 We found no significant interaction between BMI and physical activity in relation to the risk of hip fracture, although it is theoretically possible that measurement error may have influenced these results through loss of power and residual confounding. The plotted RRs (Figs 3 and 4), which allow for measurement error in self-reported BMI within categories of physical activity, provide no suggestion of an interaction. We were unable to conduct a similar adjustment for measurement error in physical activity, which may have resulted in some residual confounding such that we cannot exclude interactions at the more extreme ranges of physical activity and BMI.

This is the largest prospective study of hip fracture in relation to BMI and physical activity to date, with 2589 incident cases, and it allows detailed evaluation of the separate and combined relationships of BMI and physical activity to hip fracture risk. The only other study to consider the combined effects of BMI and physical activity on hip fracture was done in the United States, where levels of physical activity were substantially lower than in our study. For example, in that study, women spent a median time of 1.25 hours walking per week,9 whereas in this study, the median time spent walking was 3 hours per week. We used objectively recorded incident hip fractures and had virtually complete follow-up. All women were registered with the NHS. Use of private hospitals is rare among them,42 and for the small numbers that may have gone to a private hospital with a hip fracture, many of the private admissions are also captured by the NHS hospital admissions data. Almost all hip fractures result in hospital admission, and the NHS hospital admission data have been shown to record the reason for admission reliably.43

The main limitation of this study is the lack of a measure of BMD, which may modify protective effects of increasing BMI.15 The limited information on falls suggests a lower frequency with increasing physical activity but a higher frequency with increasing BMI. Therefore, part of the protection with increasing physical activity could be due to the lower number of falls. For BMI, however, the reverse may be true because obese women fall more often than thinner women but are at a reduced risk of hip fracture. About 99% of the women in this study were white, and hence these findings may not be relevant in other populations, especially those with different levels of obesity and physical activity. While we did not exclude women with prior hip fracture from our main analyses, in Scotland (with hospital admission data going back to 1981), only 9 of 250 women (3.6%) with a hip fracture included in these analyses had experienced a prior hip fracture. Such a small proportion would be insufficient to influence our findings.

In conclusion, both high BMI and physical activity reduce the incidence of hip fracture. In this large observational study, BMI and physical activity appear to work independently of each other to affect the risk of hip fracture in postmenopausal women.

Disclosures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

All the authors state that they have no conflicts of interest.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We thank all the women who participated in the Million Women Study, collaborators from the NHS Breast Screening Centres, members of the study coordinating centre, and the study steering committee (listed below). We also thank the general practitioners for providing height and weight measurements and the Information Centre for Health and Social Care and ISD (Information Services Division) Scotland for providing hospital records.

This work was supported by public funds from Cancer Research UK and the UK Medical Research Council. The funders did not influence the conduct of the study or the preparation of this article.

Steering committee: Joan Austoker, Emily Banks, Valerie Beral, Judith Church, Ruth English, Jane Green, Julietta Patnick, Richard Peto, Gillian Reeves, Martin Vessey, and Matthew Wallis.

NHS Breast Screening Centres collaborating in the Million Women Study (in alphabetical order): Avon, Aylesbury, Barnsley, Basingstoke, Bedfordshire & Hertfordshire, Cambridge & Huntingdon, Chelmsford & Colchester, Chester, Cornwall, Crewe, Cumbria, Doncaster, Dorset, East Berkshire, East Cheshire, East Devon, East of Scotland, East Suffolk, East Sussex, Gateshead, Gloucestershire, Great Yarmouth, Hereford & Worcester, Kent (Canterbury, Rochester, Maidstone), Kings Lynn, Leicestershire, Liverpool, Manchester, Milton Keynes, Newcastle, North Birmingham, North East Scotland, North Lancashire, North Middlesex, North Nottingham, North of Scotland, North Tees, North Yorkshire, Nottingham, Oxford, Portsmouth, Rotherham, Sheffield, Shropshire, Somerset, South Birmingham, South East Scotland, South East Staffordshire, South Derbyshire, South Essex, South Lancashire, South West Scotland, Surrey, Warrington Halton St Helens & Knowsley, Warwickshire Solihull & Coventry, West Berkshire, West Devon, West London, West Suffolk, West Sussex, Wiltshire, Winchester, Wirral and Wycombe.

Million Women Study Coordinating Centre: Simon Abbott, Miranda Armstrong, Krys Baker, Angela Balkwill, Vicky Benson, Valerie Beral, Judith Black, Anna Brown, Diana Bull, Benjamin Cairns, James Chivenga, Barbara Crossley, Gabriella Czanner, Dave Ewart, Sarah Ewart, Lee Fletcher, Toral Gathani, Laura Gerrard, Adrian Goodill, Jane Green, Isobel Green, Joy Hooley, Sau Wan Kan, Carol Keene, Oksana Kirichek, Nicky Langston, Maria-Jose Luque, Lynn Pank, Kirstin Pirie, Gillian Reeves, Andrew Roddam, Emma Sherman, Moya Simmonds, Elizabeth Spencer, Helena Strange, Sian Sweetland, Alison Timadjer, Sarah Tipper, Joanna Watson, Stephen Williams, Lucy Wright.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  • 1
    Lawrence TM, White CT, Wenn R, Moran CG. The current hospital costs of treating hip fractures. Injury, Int J Care Injured. 2005; 36: 8891.
  • 2
    Johnell O, Gullberg B, Kanis JA, et al. Risk factors for hip fracture in European women: the MEDOS Study. Mediterranean Osteoporosis Study. J Bone Miner Res. 1995; 10: 18021815.
  • 3
    Robbins J, Aragaki AK, Kooperberg C, et al. Factors associated with 5-year risk of hip fracture in postmenopausal women. JAMA. 2007; 298: 23892398.
  • 4
    Cummings SR, Nevitt MC, Browner WS, et al. Risk factors for hip fracture in white women. N Engl J Med. 1995; 332: 767774.
  • 5
    Paganini-Hill A, Chao C, Ross RK, Henderson BE. Exercise and other factors in the prevention of hip fracture: the Leisure World Study. Epidemiology. 1991; 2: 1625.
  • 6
    Høidrup S, Sørensen TIA, Strøger U, Lauritzen JB, Schroll M, Grønbæk M. Leisure-time physical activity levels and changes in relation to risk of hip fracture in men and women. Am J Epidemiol. 2001; 152: 6068.
  • 7
    Gregg EW, Cauley JA, Seeley DG, Ensrud KE, Bauer DC, for the Study of Osteoporotic Fractures Research Group. Physical activity and osteoporotic fracture risk in older women. Study of Osteoporotic Fractures Research Group. Ann Intern Med. 1998; 129: 8188.
  • 8
    Moayyeri A. The association between physical activity and osteoporotic fractures: a review of the evidence and implications for future research. Ann Epidemiol. 2008; 18: 827835.
  • 9
    Feskanich D, Willett W, Colditz G. Walking and leisure-time activity and risk of hip fracture in postmenopausal women. JAMA. 2002; 288: 23002306.
  • 10
    US Department of Health and Human Services. Physical activity and health: a report of the surgeon general. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion; 1996.
  • 11
    Province MA, Hadley EC, Hornbrook MC, et al. The effects of exercise on falls in elderly patients. JAMA. 1995; 273: 13411347.
  • 12
    Nelson ME, Fiatorone MA, Morganti CM, Trice I, Greenberg RA, Evans WJ. Effects of high-intensity strength training on multiple risk factors for osteoporotic fractures: a randomized controlled trial. JAMA. 1994; 272: 19091914.
  • 13
    Wolff I, van Croonenborg JJ, Kemper HCG, Kostense PJ, Twisk JWR. The effect of exercise training programs on bone mass: A meta-analysis of published controlled trials in pre- and postmenopausal women. Osteoporos Int. 1999; 9: 112.
  • 14
    Lanyon LE. Functional strain as a determinant for bone remodelling. Calcif Tissue Int. 1984; 36 (suppl): S5661.
  • 15
    De Laet C, Kanis JA, Odén A, et al. Body mass index as a predictor of fracture risk: a meta-analysis. Osteoporos Int. 2005; 16: 13301338.
  • 16
    Burton BT, Foster WR, Hirsch J, VanItallie TB. Health implications of obesity: NIH consensus development conference. Int J Obes Relat Metab Disord. 1985; 9: 155169.
  • 17
    Hla MM, Davis JW, Ross PD, et al. A multicenter study of the influence of fat and lean mass on bone mineral content: evidence for differences in their relative influence at major fracture sites. Am J Clin Nutr. 1996; 64: 354360.
  • 18
    Simpson E, Rubin G, Clyne C, et al. Local estrogen biosynthesis in males and females. Endocrine-Related Cancer. 1999; 6: 131137.
  • 19
    Robinovitch SN, Hayes WC, McMahon TA. Prediction of femoral impact forces in falls on the hip. J Biomech Eng. 1991; 113: 366375.
  • 20
    Blair S. Evidence for success of exercise in weight loss and control. Ann Intern Med. 1993; 119: 702706.
  • 21
    Banks E, Reeves G, Beral V, Balkwill A, Liu B, Roddam A, for the Million Women Study Collaborators. Hip fracture incidence in relation to age, menopausal status, and age at menopause: prospective analysis. PLOS Medicine. 2009; 6: e1000181.
  • 22
    The Million Woman Study Collaborative Group. Million Woman Study: design and characteristics of the study population. Breast Cancer Res. 1999; 1: 7380.
  • 23
    Kendrick S, Clarke J. The Scottish Record Linkage System. Health Bull (Edinb). 1993; 51: 7279.
  • 24
    HESonline Hospital Episode Statistics.
  • 25
    World Health Organization. 1992; International Statistical Classification of Diseases and Related Health Problems, vol. Tenth revision, Geneva.
  • 26
    Office of Population Censuses and Surveys. 2000; Classification of Surgical Operations and Procedures - Fourth Revision. Crown copyright, London.
  • 27
    Craig CL, Marshall AL, Sjöström M, et al. International Physical Activity Questionnaire. 12-Country Reliability and Validity. Med Sci Sports Exerc. 2003; 35: 13811395.
  • 28
    Australian Institute of Health and Welfare. The Active Australia Survey - a guide for implementing analysis and reporting, AIHW Canberra Cat vol. No. CVD 22. Australian Institute for Health and Welfare, Canberra. 2003.
  • 29
    Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of Physical Activities: an update of activity codes and MET intensities. Med Sci Sports Exerc. 2000; 32: S498S516.
  • 30
    StataCorp. Statistical Software: Release 10.1. Stata: College Station, TX. 2009.
  • 31
    Easton DF, Peto J, Babiker AG. Floating absolute risk: an alternative to relative risk in survival and case-control analysis avoiding an arbitrary reference group. Stat Med. 1991; 10: 10251035.
  • 32
    MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke, and coronary heart disease. Part 1, prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet. 1990; 335: 765774.
  • 33
    Spencer EA, Appleby PN, Davey GK, Key TJ. Validity of self-reported height and weight in 4808 EPIC–Oxford participants. Public Health Nutrition. 2002; 5: 561565.
  • 34
    Nieto-García FJ, Bush TL, Keyl PM. Body mass definitions of obesity: sensitivity and specificity using self-reported weight and height. Epidemiology. 1990; 1: 146152.
  • 35
    Nicodemus KK, Folsom AR. Type 1 and type 2 diabetes and incident hip fractures in postmenopausal women. Diabetes Care. 2001; 24: 11921197.
  • 36
    Hamilton MT, Hamilton DG, Zderic TW. Role of low energy expenditure and sitting in obesity, metabolic syndrome, type 2 diabetes, and cardiovascular disease. Diabetes. 2007; 56: 26552667.
  • 37
    Banks E, Jorm L, Rogers K, Clements M, Bauman A. Screen-time, obesity, ageing and disability: findings from 91 266 participants in the 45 and Up Study. Public Health Nutr. 2010; 110.
  • 38
    Phongsavan P, Merom D, Marshall A, Bauman A. Estimating physical activity level: the role of domestic activities. Epidemiol Community Health. 2004; 58: 466467.
  • 39
    Brown WJ, Trost SG, Bauman A, Mummery K, Owen N. Test-retest reliability of four physical activity measures used in population surveys. J Sci Med Sport. 2004; 7: 205215.
  • 40
    Bonjour J, Theintz G, Law F, Slosman D, Rizzoli R. Peak Bone Mass. Osteoporos Int. 1994; 1 (suppl): S713.
  • 41
    Conn JM, Annest JN, Gilchrist J. Sports and recreational related injury episodes in the US population, 1997–99. Inj Prev. 2003; 9: 117125.
  • 42
    Williams B, Whatmough P, McGill J, Rushton L. Patients and procedures in short-stay independent hospitals in England and Wales, 1997–1998. J Public Health Med. 2000; 22: 6873.
  • 43
    McColl A, Roderick P, Cooper C. Hip fracture incidence and mortality in an English region: a study using routine National Health Service data. J Public Health Med. 1998; 20: 196205.
  • 44
    O'Neill TW, Marsden D, Adams JE, Silman AJ. Risk factors, falls, and fracture of the distal forearm in Manchester, UK. J Epidemiol Community Health. 1996; 50: 288292.