The increasing incidence of fragility fractures has posed an ever-increasing burden on healthcare systems worldwide.1 Low-energy fractures are most common in women, especially in older age, but the highest estimated increase in the next 40 years is expected in men.2 The effective primary prevention of fractures is therefore of high importance.
Physical activity is considered to be an important, modifiable lifestyle factor that can increase or maintain bone mineral density and improve muscle strength and balance in children and adults, thereby reducing the risk of fracture in old age.3–6 Indeed, there is consistent epidemiologic evidence that physical activity is associated with reduced fracture risk in middle-aged and old men.7–9 Also, it is known that the amount of vigorous physical activity and physical fitness are correlated.10, 11 From the perspective of primary prevention, it is thus of interest whether physical fitness in young adulthood influences the risk of fracture later in life. Some studies have investigated whether bone mineral density, a surrogate marker for bone strength, is sustained with reduced physical activity with inconsistent results12–14 and of limited value as a result of short follow-up periods and the lack of information on the primary endpoint of interest; ie, fractures. This issue has been further evaluated in a few retrospective studies investigating the prevalence of fractures in former male athletes,15, 16 but again with inconsistent results. In view of these conflicting results, an adequately powered, prospective study with a sufficient follow-up period is necessary to evaluate the impact of physical fitness on low-energy fracture risk in midlife.
The present study evaluated the association of objective measures of physical fitness, such as muscle strength and aerobic capacity, with the incidence of low-energy fractures in a population-based cohort of 435,000 young men during a median follow-up period of 35 years. The contributions of genetic and environmental factors to the variation in physical fitness were also assessed in a large subcohort of twins.
Subjects and Methods
The cohort considered for inclusion in the present study consisted of all Swedish men (n = 457,262) who were conscripted for compulsory military service between 1969 and 1978 and recorded in the Swedish Military Service Conscription Register (SMSCR). These men represented approximately 97% of the Swedish male population born between 1950 and 1962. The background of the SMSCR has been described in detail.17 Until recently, exemptions for conscription were granted only to incarcerated men and those with severe chronic medical conditions or disabilities documented by a medical certificate. Only 2% to 3% of all Swedish men were exempted from conscription during the period of interest in this study.
Data on deaths occurring during the study period were collected through the National Cause of Death Register, administered by the Center for Epidemiology at the National Board of Health and Welfare in Sweden. Subjects who emigrated were identified using the database maintained by Statistics Sweden. Information on smoking habits was available only for a subcohort of 23,529 men who were conscripted between 1969 and 1970. To reduce errors of random misclassification, exclusion limits were set for weight (<40 or >170 kg), height (<140 or >215 cm), and age (<16 years). These limits led to the exclusion of 8560 (1.9%) men; another 8553 (1.9%) men were excluded because of death and 4704 (1.0%) because of emigration before 40 years of age, leaving a study sample of 435,445 men.
All conscripts underwent a standardized, 2-day examination before receiving an assignment in the Swedish Armed Forces. The conscription tests, which included the evaluation of muscle strength and aerobic capacity, took place at six regional conscription centers. All conscripts were seen by a physician, who diagnosed any disorder according to the Swedish version of the International Classification of Diseases (ICD), 8th edition.
Muscle strength test
Maximal isometric muscle strength was measured in knee extension, elbow flexion, and handgrip tests by dynamometers; each muscle group was tested three times, strength was measured in Newtons (N), and the highest value was used in further analysis. If a subject's final value was highest, testing continued until the value stopped increasing. All instruments were calibrated daily. Right knee extension strength was measured with the subject seated and the knee positioned at a 90-degree angle. The dynamometer strap was secured at the malleolus of the right ankle. Right elbow flexion strength was measured with the subject seated and the arm strapped down; the elbow was flexed to 90 degrees and the forearm was held vertically. The subject's left hand was placed on his left knee to secure a straight shoulder position during the test. The dynamometer strap was secured at the level of the radial styloid process. During the test, the subject was instructed to hold his thumb upward to ensure good reproducibility. Handgrip strength of the dominant hand was measured using a dynamometer with the subject standing, the upper arm held vertically along the body, and the elbow flexed at 90 degrees.
Aerobic capacity test
Aerobic capacity was assessed using an electrically braked bicycle ergometer test.18 In short, each participant underwent a resting electrocardiogram (ECG); if it was normal, he proceeded with 5 minutes of submaximal bicycling at work rates of 75 to 175 Watts (W), depending on body weight. Thereafter, the workload was gradually increased by 25 W/min until exhaustion. During the test, the subject was instructed to maintain a pedal cadence of 60 to 70 revolutions/min and the heart rate was measured continually. The final work load (Wmax) was recorded and used in further analysis. This information was available for 313,122 men who conscripted between 1972 and 1978. These men were slightly younger (mean difference 0.3 years), heavier (mean difference 1.6 kg), taller (mean difference 0.3 cm), and had slightly higher knee extension strength (mean difference 2N), than the rest of the cohort.
Diagnosis of fractures
Information on fracture diagnoses between January 1, 1987 and December 31, 2009 was obtained by record linkage using the subjects' unique personal identification numbers with the National Hospital Discharge Register (HDR), which covers all public inpatient care in Sweden from 1987, and the National Hospital Outpatient Register, which covers all public outpatient care in Sweden from 2001. Both registers are administered by the Center for Epidemiology at the National Board of Health and Welfare. Diagnoses were recorded using the ICD versions 9 (1987–1996) and 10 (1997–2009). Only low-energy fractures that were diagnosed as caused by a fall from standing height, by tripping, or by slipping were included in our analysis. Thus, high-energy fractures, eg, occurring during car accidents or caused by falling from heights, were excluded. Furthermore, we included only major fractures; ie, fractures of the hip, limbs (excluding the fingers and toes), spine, or pelvis. The HDR has previously been found to be valid in identifying cases of fracture.19, 20 Information of deaths in the cohorts was obtained from the Center for Epidemiology at the National Board of Health and Welfare.
Identification of twins within the cohort
The Swedish Twin Registry (STR) was initiated in the late 1950s to study the association of smoking and alcohol consumption with the risk of cancer and cardiovascular diseases, while controlling for the genetic propensities for these diseases. The STR has been regularly expanded and updated since that time.21 In the present conscription cohort, the STR enabled the identification of 4438 male twins, including 411 male monozygotic pairs and 598 male dizygotic pairs. These twin pairs were used to estimate the heritability of muscle strength and physical fitness and the unique (non-shared) environmental effects on these attributes.
Data are presented as means ± SDs. Differences in baseline characteristics as a function of subsequent fracture status were investigated using Student's t test for independent samples. The relationship between the explanatory variables of interest at baseline, ie, muscle strength and physical fitness, and the incidence of fracture after 40 years of age was investigated using Cox's proportional hazard models. If not otherwise indicated, the models were adjusted for baseline age, weight, height, conscription center, year of conscription, and diagnoses found to be associated with the risk of a fracture during follow up (Table 1). For all Cox regression models, each subject's duration of follow-up was determined by the date of a registered fracture, date of death, date of emigration, or December 31, 2009, whichever came first. The proportional hazard assumption was verified graphically by Kaplan-Meier curves. The SPSS software (ver. 18.0; SPSS Inc., Chicago, IL, USA) was used for all statistical analyses.
Table 1. Characteristics of the Subjects at 18 Years of Age According to Fracture Status After the Age of 40 Years
Subjects with fractures during follow-up (n = 8030)
Subjects with no fractures during follow-up (n = 427,415)
Wmax = maximum work load in Watts.
Data available only for 23,529 men that were conscripted from 1969 to 1970.
Data available for 313,122 men that were conscripted from 1972 to 1978.
To assess the heritability of muscle strength and physical fitness, we first computed measures of similarity (intrapair correlations) for monozygotic (MZ) and dizygotic (DZ) pairs. Because MZ twins share all their genes and DZ twins share, on average one-half (of their segregating) genes, the comparison of similarities permits estimation of the relative importance of genes and environment for individual differences in a trait. Quantitative genetic models were used to decompose the phenotypic variance (individual differences) in muscle strength and physical fitness into additive genetic (A), shared environmental (C), and unique (individual specific) environmental (E) components in the subcohort of twins. Additive genetic effects (A) are indicated when MZ twins are more similar than DZ twins. Shared environmental influences C are those that make family members similar to each other regardless of biological relatedness. The E parameter includes influences not shared by a pair of twins, such as physical training, diseases or accidents that affect only one sibling, and measurement error.22, 23 Evidence for E comes from the difference between the MZ correlation and 1.0. Estimates of these three variance components were obtained by fitting structural equation models to the data, using the OpenMx free-source software (ver. 1.0.7)24 in the R programming environment (ver. 2.13.1).
All statistical tests were two-sided. A p value of less than 0.05 was considered to indicate statistical significance.
The cohort studied included 435,445 men with a mean age of 18.5 ± 0.7 years at baseline. During a median follow-up period of 35 years (range, 11–41 years), at least one low-energy fracture was recorded in 8030 men after 40 years of age. Baseline characteristics, grouped by fracture status, are shown in Table 1. Men who sustained a fracture were marginally older, weighed less, and were more often smokers than were men who had no fracture (p < 0.01 for all). They also had lower muscle strength (all measures) and aerobic capacity at baseline (p < 0.001 for all). Several diagnoses at baseline were also associated with a significantly higher risk of sustaining a fracture during the follow-up period (Table 1). Ankle and wrist fractures were most common during the follow-up period, and 523 men sustained a hip fracture (Table 2).
Table 2. Type and Number of Fractures in a Total of 435,445 Men Followed for a Mean of 35 Years
Hazard ratios are presented for knee extension, elbow flexion, hand grip strength, and aerobic capacity (n = 313,078), per decrease of 1 SD. The models were adjusted for age, weight, height, year, and place of conscription and all diagnoses associated with the risk of fractures according to Table 1.
HR = hazard ratio; CI = confidence interval.
Type of fracture
Wrist (n = 1609)
Humerus (n = 1051)
Tibia (n = 1274)
Hip (n = 523)
Lower leg (n = 921)
Ankle (n = 2210)
Other fractures (n = 442)
Total (n = 8030)
Cox regression analyses revealed that, after adjusting for all confounders, the risk of a low-energy fracture increased with every SD decrease in aerobic capacity (hazard ratio [HR], 1.17; 95% confidence interval [CI], 1.13–1.20). Furthermore, the risk of a fracture increased significantly with every SD decrease in knee extension strength (HR, 1.12; 95% CI, 1.09–1.14), after adjusting for the same confounders. Consistent results were also obtained for grip strength (HR, 1.11/SD decrease; 95% CI, 1.08–1.14) and elbow flexion strength (HR, 1.10/SD decrease; 95% CI, 1.07–1.13; Table 2). The associations between the different measures of muscle strength and fractures were also investigated with and without entering smoking as a covariate in a subcohort of 23,529 men. The inclusion of smoking as a covariate in separate models decreased the HRs negligibly by a mean of 0.008 (p > 0.05 for all). Smoking was excluded from consideration in subsequent analyses.
The associations of deciles of muscle strength and aerobic capacity with fractures, after adjusting for all confounders, are shown in Fig. 1. For all measures of muscle strength and physical fitness the risk of a fracture increased in a dose-dependent fashion (p < 0.001 for trend). Men in the lowest deciles of baseline muscle strength showed an increased risk of a future low-energy fracture by factors ranging from 1.37 (elbow flexion strength) to 1.51 (knee extension strength; p < 0.001 for all), compared with men in the highest deciles. Similarly, men in the lowest decile of baseline aerobic capacity showed an increased risk of a future low-energy fracture by a factor of 1.78 (95% CI, 1.57–2.03), compared with men in the highest decile. Finally, men in the lowest decile of aerobic capacity had a 2.7-fold higher risk (HR, 2.73; 95% CI, 1.59–4.71) of hip fracture compared with those in the highest decile.
The cumulative incidence of fractures for deciles of aerobic capacity is shown in Fig. 2. Men in the lowest decile of baseline aerobic capacity reached the cumulative incidence of 125 fractures/10,000 subjects a mean of 4.7 years earlier than did those in the highest decile.
To investigate whether aerobic capacity and muscle strength conferred independent effects with respect to the risk of a low-energy fracture, deciles of aerobic capacity and knee extension strength were entered into the same Cox regression model as independent variables. After adjusting for confounders, men in the poorest deciles of both aerobic capacity (HR, 1.71/SD decrease; 95% CI, 1.50–1.95) and knee extension strength (HR, 1.33/SD decrease; 95% CI, 1.16–1.51) were found to be have a higher risk of a low-energy fracture compared with men in the highest deciles.
The results of our analysis of the heritability of muscle strength and aerobic capacity are presented in Table 3. Our investigation revealed that 35% to 62% of the variation in muscle strength and 78% of the variation in aerobic capacity could be attributed to additive genetic factors, whereas 27% to 39% of the variation in muscle strength and 22% of the variation in aerobic capacity could be attributed to individual specific environments.
Table 3. Associations Within 411 MZ and 598 DZ Twin Pairs for the Different Estimates of Muscle Strength and Aerobic Capacity
The associations are presented before and after adjustment for differences in weight and height between the twin pairs. In addition, unadjusted variation (%) in the different estimates of muscle strength and aerobic capacity attributed to additive genetic factors (A), common environment (C), and unique environment (E) was estimated using structural equation modeling.
rMZ = associations within monozygotic twin pairs; rDZ = associations within dizygotic twin pairs; A = additive genetic factors; CI = confidence interval; C = common environment; E = unique environment.
Finally, 14,308 men died during follow up after 40 years of age. After adjusting for all confounders, any low-energy fracture during the follow-up increased the risk of death about 1.8 times (HR, 1.77; 95% CI, 1.62–1.94), whereas the hip fracture in itself increased the risk almost fivefold (HR, 4.52; 95% CI, 3.66–5.59). If adjusting these models also for physical fitness the associations between fracture and the risk of death were reduced marginally (HR, 1.74 and 4.33, respectively). The most common specified causes of death were from cardiovascular disease (24.1%) and cancer (20.8%).
In the present study, we demonstrated that objective measures of physical fitness in young adulthood were strongly associated with the risk of a low-energy fracture in middle age, several decades later. Twin analyses indicated that a substantial part of the variation in aerobic capacity and muscle strength is attributable to individual specific environmental factors and thus amenable to exercise intervention. Altogether, our results suggest that good physical fitness in young adulthood may substantially reduce or postpone the risk of fracture in midlife in the general male population. These observations also have public health relevance because we found that sustaining a low-energy fracture during the follow-up was associated with an increased the risk of death.
The results showed that men in the lowest decile of aerobic capacity in young adulthood had almost a doubled risk of fracture in midlife. Moreover, of the separate fractures, a low fitness was associated with the highest risk of a hip fracture. To date, no randomized exercise intervention trial has evaluated fractures as a primary endpoint. Given the large number of subjects and lengthy follow-up period required to achieve sufficient statistical power, it is likely that such a study will never be performed. Nevertheless, exercise intervention trials have shown that endurance training can increase maximal oxygen uptake by 20% to 30%25, 26 and strength training can increase muscle strength by 30% to 40%.27, 28 In the present study, a 30% increase in aerobic capacity and muscle strength was associated with reduced fracture risks of about 25% and 15%, respectively, as shown in Fig. 1. For hip fractures, a 30% increase in aerobic capacity was associated with more than a halved fracture risk. Thus, inferences from previous studies and our results suggest that a regular multicomponent exercise in young adulthood has potential to reduce fracture risk later in life at least by one-third, or that a fracture is postponed by several years as illustrated in Fig. 2. Given that the effects of training on muscle strength and aerobic capacity are more pronounced in untrained subjects,29 any preventive efforts should probably be targeted to individuals with the lowest physical fitness.
Our data provide the first evidence that objective measures of physical fitness in young adulthood convey an important risk factor for fracture more than 30 years later. Several mechanisms appear to underlie these associations. Previous studies have shown that physical activity patterns in childhood may persist into older age,30, 31 and have sustained effects on the cardiovascular and musculoskeletal systems. General physical strength likely reduces the risk of falling and better equips a subject to withstand the loads imposed on the skeleton by a fall. High physical fitness in young adulthood may also be sustained later in life independent of the amount of subsequent physical activity. In support of this hypothesis, we previously found a significantly lower risk of fracture in former athletes whose concurrent physical activity level was similar to that of a control group without athletic background.15 It is also possible that estimates of physical fitness and bone strength are influenced by the same genetic factors. In support of this notion we previously found an association between muscle strength in 17-year-old boys and their parents' bone density.32 A few longitudinal studies have suggested that self-reported physical activity is associated with reduced risk of fracture in middle-aged and elderly men.7, 8 Moreover, in a 35-year prospective study, Michaelsson and colleagues9 showed that not only self-reported physical activity but also increases therein reduced the risk of fracture in men after 50 years of age.
It should be noted that since the present cohort included only men, the observed associations may not be directly applicable to women. Furthermore, because this was an observational prospective study, relationships of causality between the different measures of physical fitness and fractures should be further evaluated. In the present study, however, the associations between all objective measures of physical fitness were consistently dose-dependent with respect to fractures and independent of environmental factors and baseline diagnoses. This finding has obvious clinical ramifications because previous studies have shown that muscle strength and aerobic capacity can be improved substantially by physical training.25–28 In addition, our analysis of a subcohort of twins demonstrated that a substantial amount of the variation in muscle strength and physical fitness is attributable to individual environmental factors, such as physical exercise, altogether supporting the likelihood of a causal relationship.
In summary, the present study demonstrated that greater muscle strength and better aerobic capacity in young adulthood were associated in a dose-dependent manner with a lower risk of low-energy fractures in men. Our results suggest that regular strength and cardiovascular training, leading to improved physical fitness, would also lead to a 30% reduced risk of fracture in middle-aged men, or a postponement of the likelihood of sustaining a fracture by several years. These results are also of public health interest because a low-energy fracture was associated with an increased risk of death in this cohort.
All authors state that they have no conflicts of interest.
This study was supported by grants from the Swedish Research Council. The funders had no role in the design, analysis or interpretation of the results as well is in preparation, review or approval of the manuscript.
Authors' roles: Study design: PN and AN. Data collection: PN and NP. Data analysis: PN. Data interpretation: PN, HS, NP, YG, and AN. Drafting manuscript: PN, HS, and AN. Revising manuscript content: PN, HS, NP, YG, and AN. Approving final version of manuscript: PN, HS, NP, YG, and AN. PN takes responsibility for the integrity of the data analysis.