Urho M. Kujala, Jaakko Kaprio, Pekka Kannus, Seppo Sarna, Markku Koskenvuo. Physical Activity and Osteoporotic Hip Fracture Risk in Men. Arch Intern Med. 2000;160(5):705–708. doi:10.1001/archinte.160.5.705
Physical activity has been related to reduced risk of osteoporotic hip fractures, but the evidence among men is weak.
To determine the association between baseline leisure physical activity and future risk of osteoporotic hip fracture in men.
At baseline in 1975 our prospective study cohort included 3262 men who were 44 years or older and did not have chronic disease restricting their ability to exercise. At baseline, physical activity was assessed by a questionnaire. Hip fractures were followed for 21 years, or from the age of 50 years for subjects who were initially younger than 50 years.
The hazard ratio of osteoporotic hip fracture, adjusted for other possible predictors (height, body mass index, baseline diseases, smoking, use of alcohol, work-related physical activity, and occupational group), in men participating in vigorous physical activity compared with men not participating was 0.38 (95% confidence interval, 0.16-0.91) (P=.03).
These results provide further evidence that there is an inverse association between baseline physical activity and future hip fracture risk among men.
SOME PROSPECTIVE studies suggest that leisure time physical activity prevents osteoporotic hip fractures, but the scientific evidence among men is weak and based on rather short follow-up periods.1- 8 Baseline selection bias cannot be ruled out. Case-control studies support the hypothesis that lifetime leisure physical activity is protective against hip fractures among men, but these studies may suffer from classification errors due to recall bias: the case subjects may remember their level of activity differently from the control subjects because they have experienced a fracture.8
We investigated leisure physical activity as a predisposing or preventive factor for hip fractures in men. Our focus was on participation in vigorous activity, as we consider that this may best be associated with the neuromuscular ability to reduce the risk of falls or make the falls less injurious.
The Finnish Twin Cohort was compiled from the Central Population Registry of Finland via procedures described elsewhere.9 In brief, the baseline sample comprised all same-sex twin pairs born in Finland before 1958 and with both cotwins alive in 1967. The cohort also included other subjects who were not twins, but who had the same family name and home parish, and were born on the same day. The target group for the present study consisted of men born in 1931 or earlier (aged ≥44 years on January 1, 1976).
These subjects were mailed a questionnaire in autumn 1975 that included items on physical activity, occupation, body weight, height, alcohol use, smoking, and physician-diagnosed diseases (response rate, 77.8%).
We included men who supplied complete questionnaire data on physical activity, were alive on January 1, 1976, and did not have a hip fracture before January 1, 1976. In addition, subjects who died before the age of 50 years were excluded. The final study cohort included 3262 men.
Because chronic disease or subjects' medications may restrict the ability to exercise and change the risk of hip fracture, we constructed an index to assess if the subjects had certain chronic diseases at baseline. Based on the questionnaire, subjects who reported physician-diagnosed angina pectoris, myocardial infarction, stroke, or diabetes, or who had angina pectoris according to standard chest pain history items were determined to have chronic disease.10 Using the reliable nationwide hospital discharge register maintained by the National Board of Health,11 subjects with inpatient admissions for diabetes (International Classification of Diseases, Eighth Revision, [ICD-8] code 250), cardiovascular disease except for hypertension or venous diseases (codes 390-399 or 410-449), or chronic obstructive pulmonary diseases (codes 490-493) between 1972 and December 31, 1976, were determined to have chronic disease at baseline. In addition, all subjects who had been granted reimbursable medication for selected chronic diseases other than hypertension before December 31, 1976, based on data obtained from the Social Insurance Institution of Finland,12 were classified as having disease at baseline, as were those who had had incident malignant cancer before 1977 according to the Finnish Cancer Registry.13
Using the guidelines from previous epidemiological studies of osteoporotic fractures,14 we defined an osteoporotic hip fracture as a proximal femur fracture occurring in individuals aged 50 years or older as a consequence of a minimal trauma only, ie, a fall from standing height or less. All 50-year-old or older subjects who were admitted to Finnish hospitals (1976-1996) for primary treatment of hip fracture were selected from the National Hospital Discharge Register (ICD-8 or ICD-9 code 820 except for cases with major trauma [ie, cases with codes E807-846 for traffic accidents and codes E916-928 for other external causes/accidents in the ICD-8 until the end of 1986, as well as cases with codes E800A-830A and E920A-928A, respectively, in the ICD-9 from the beginning of 1987 and onward, were excluded]). The Finnish National Hospital Discharge Register has been reported to cover severe injuries such as hip fractures adequately (annual coverage ≥97%) and record them accurately (accuracy ≥97%).11,15,16
At baseline we asked the subjects about their leisure physical activity (monthly frequency, mean duration and mean intensity of physical activity sessions, and their opinion of their overall physical activity level) and physical activity to and from work using a series of structured questions (questionnaire available from the authors).17,18 In brief, first we classified our men according to their participation in vigorous leisure physical activities (yes/no) in which the intensity was greater than walking (ie, at least the equivalent of jogging or running), without any qualifications on frequency or duration. Second, in the categorical classification of leisure physical activity, those who reported exercising at least 6 times per month for a mean duration of at least 30 minutes and with a mean intensity corresponding to at least vigorous walking to jogging were classified as conditioning exercisers. Those who reported not partaking in leisure physical activity were considered sedentary. Other subjects were classified as occasional exercisers (for percentage of subjects in each category see Table 1). Third, for volume of activity, we calculated an activity metabolic equivalent (MET) index by assigning a multiple of resting metabolic rate (MET score) to each activity and by calculating the product of intensity × duration × frequency of activity.18,19 According to the categorical physical activity, the mean (SD) leisure MET index was 0.33 (0.83) in sedentary subjects, 2.57 (2.32) in occasional exercisers, and 8.42 (5.80) in conditioning exercisers. A questionnaire with identical physical activity items was readministered in 1981 showing that the physical activity habits in the cohort were rather stable.18
Using the subjects' self-reports at baseline, we obtained information on other (in addition to baseline diseases as described) potential predictors of hip fractures using validated methods presented earlier18: work-related physical activity, occupational group, height, body mass index (reported weight in kilograms divided by the square of reported height in meters), cigarette smoking, and alcohol consumption (Table 1).
The hip fracture follow-up began on January 1, 1976, and continued to December 31, 1996, the date of occurrence of hospitalization for hip fracture (first hip), or death. Among subjects who were younger than 50 years at baseline, the follow-up began when they turned 50. We first studied occurrence of hip fractures in the whole study cohort by calculating hazard ratios (HRs) during the follow-up by physical activity, and adjusting them for age, diseases at baseline, occupational group, work-related physical activity, height, body mass index, smoking, and use of alcohol at baseline using the Cox proportional hazards model.20 All significance tests were 2-tailed. The analyses were performed with the SAS statistical package version 6.12 (SAS Institute, Cary, NC).
During the follow-up, 60 men were hospitalized for hip fracture. The age-adjusted HR of hip fracture in men participating in vigorous physical activity compared with men not participating was 0.42 (95% confidence interval, 0.19-0.94) (Table 2). After adjustment for other possible predictors of hip fracture (height, BMI, baseline diseases, smoking, use of alcohol, work-related physical activity, and occupational group), the HR was 0.38 (95% CI, 0.16-0.91) (Table 2). In this final model, in addition to vigorous physical activity, only BMI was a statistically significant predictor of hip fractures (P=.01), with the risk decreasing by 12% (95% CI, 3%-20%) per unit increase in BMI. Subjects with diseases at baseline tended to have increased risk (P=.15; HR, 1.51; 95% CI, 0.86-2.67), and the risk tended to decrease with category of increasing work-related physical activity (P=.16; HR, 0.79; 95% CI, 0.57-1.10).
Despite the low number of observations and consequently low statistical power, the hip fracture risk decreased with increasing physical activity category (Table 2), which classification took into account both the intensity and volume of exercise. However, in the case of MET index quartiles, the second quartile compared with lowest quartile tended to give maximal benefit (Table 2). Compared with the lowest quartile and after adjustment for all covariates, subjects from the MET index quartiles 2 through 4 combined tended to have lower risk (P=.06; HR, 0.58; 95% CI, 0.32-1.03).
Our results agree with earlier findings that there is an inverse association between baseline physical activity and future hip fracture risk,1- 8 which has been previously more clearly documented among women.7,8 Physically active persons differ from others in many respects, but our results accord with those of earlier studies on physical activity and hip fracture risk; adjustment for factors other than age and sex does not influence the estimates much.8
Based on our analyses it seems that in terms of volume of activity, even moderate levels of baseline leisure time physical activity, compared with sedentariness, protects men from future hip fracture, and a further reduction in risk is associated with more vigorous activity at baseline. This protection may be mediated by the better neuromuscular performance level of the vigorously exercising subjects. However, the limited number of subjects, and limited number of cases, in our study does not allow any detailed final conclusions on the volume and intensity of exercise associated with reducing the risk of hip fracture. Finally, it is noteworthy that, in accordance with the hypothesis, an increase in work-related physical activity also tended to decrease the hip fracture risk in our study.
Accepted for publication June 29, 1999.
This work was supported by grant 42044 from the Academy of Finland and the Finnish Ministry of Education, Helsinki.
Corresponding author: Urho M. Kujala, MD, Unit for Sports and Exercise Medicine Mannerheimintie 17 (Töölö Sports Hall), FIN-00250 Helsinki, Finland (e-mail: Urho.Kujala@helsinki.fi).