A site-dependent association between obesity and fracture has been reported in postmenopausal women. In this study we investigated the relationship between body mass index (BMI) and fracture at different skeletal sites in older men (≥65 years). We carried out a population-based cohort study using data from the Sistema d‘Informació per al Desenvolupament de l‘Investigació en Atenció Primària (SIDIAPQ) database. SIDIAPQ contains the primary care and hospital admission computerized medical records of >1300 general practitioners (GPs) in Catalonia (Northeast Spain), with information on a representative 30% of the population (>2 million people). In 2007, 186,171 men ≥65 years were eligible, of whom 139,419 (74.9%) had an available BMI measurement. For this analysis men were categorized as underweight/normal (BMI < 25 kg/m2, n = 26,298), overweight (25 ≤ BMI < 30 kg/m2, n = 70,851), and obese (BMI ≥ 30 kg/m2, n = 42,270). Incident fractures in the period 2007 to 2009 were ascertained using International Classification of Diseases, 10th edition (ICD-10) codes. A statistically significant reduction in clinical spine and hip fractures was observed in obese (relative risk [RR], 0.65; 95% confidence interval [CI], 0.53–0.80 and RR, 0.63; 95% CI, 0.54–0.74, respectively), and overweight men (RR, 0.77; 95% CI, 0.64–0.92 and RR, 0.63; 95% CI 0.55–0.72, respectively) when compared with underweight/normal men. Additionally, obese men had significantly fewer wrist/forearm (RR, 0.77; 95% CI, 0.61–0.97) and pelvic (RR, 0.44; 95% CI, 0.28–0.70) fractures than underweight/normal men. Conversely, multiple rib fractures were more frequent in overweight (RR, 3.42; 95% CI, 1.03–11.37) and obese (RR, 3.96; 95% CI, 1.16–13.52) men. In this population-based cohort of older men, obesity was associated with a reduced risk of clinical spine, hip, pelvis, and wrist/forearm fracture and increased risk of multiple rib fractures when compared to normal or underweight men. Further work is needed to identify the mechanisms underlying these associations.
A site-dependent association between obesity and fracture has been reported in postmenopausal women. Ankle,[1, 2] lower and upper leg, vertebral, and proximal humerus fractures[3, 4] have been reported to be more frequent in obese than non-obese women, whereas obesity appears to be protective against hip, clinical spine, wrist/forearm, and pelvic fractures.
Little is known about fracture site and obesity in men. Nielson and colleagues reported that, after adjustment for age, race, and bone mineral density (BMD), obesity was associated with increased risk of nonvertebral fractures and of hip fracture in men aged ≥ 65 years, in contrast to the protective effect demonstrated in postmenopausal women. The aim of our study was to evaluate the relationship between body mass index (BMI) and fracture site in Spanish men ≥65 years old.
Materials and Methods
We carried out a population-based retrospective cohort study using data from the Sistema d‘Informació per al Desenvolupament de l‘Investigació en Atenció Primària (SIDIAPQ) database. SIDIAPQ contains the primary care computerized medical records of >1300 general practitioners (GPs) in Catalonia (Northeast Spain), with information on a representative 30% of the population (>2 million people). It comprises the clinical and referral events registered by primary care health professionals (GPs and nurses) and administrative staff in electronic medical records, comprehensive demographic information, prescription and corresponding pharmacy invoicing data, specialist referrals, primary care laboratory test results, hospital admissions, and their major outcomes. Health professionals gather this information using International Classification of Diseases, 10th edition (ICD-10) codes, and structured spreadsheets designed for the collection of variables relevant for primary care clinical management, including country of origin, gender, age, height, weight, BMI, smoking and drinking status, blood pressure measurements, preventive care provided, and blood and urine test results. Only GPs who meet quality control standards can contribute to the SIDIAPQ database. Encoding personal and clinic identifiers ensures the confidentiality of the information in the SIDIAPQ Database. SIDIAPQ is further linked to the official hospital discharge database (CMBD-AH; its acronym in the Catalan language) to improve data completeness.
We identified all men aged ≥65 years who were permanently registered in the database in 2007, and who had at least one BMI measurement in the year prior to the fracture.
Ascertainment of fractures
Incident fractures in the study period (2007–2009, both inclusive) were ascertained within both primary care and hospital episodes data using ICD-10 codes. Fracture sites considered for these analyses were those defined by Center and colleagues as major fractures, based on their associated mortality (hip, clinical spine, pelvis, tibia, multiple rib, and proximal humerus), and the most prevalent minor osteoporotic fracture in our data (wrist/forearm). Multiple rib fractures were defined as more than one at a single time point (ICD-10 code S22.4). Data on fracture have been validated in the SIDIAPQ database using both prospective cohort and hospital admission databases as a reference: hip, clinical spine, and wrist/forearm fracture coding have been shown to be highly specific (99%, 99%, and 98% respectively) in SIDIAPQ.
Only men with at least one BMI measurement in SIDIAPQ were included in these analyses. If more than one BMI measurement was registered, we used the value recorded closest to the date when the fracture occurred. Subjects were classified according to the WHO BMI categories into four groups: underweight/normal weight (BMI < 25 kg/m2), overweight (25 ≤ BMI < 30 kg/m2), and obese (BMI ≥ 30 kg/m2).
Age-specific fracture incidence rates for each fracture site were calculated assuming a Poisson distribution, and plotted against age (numerical estimates are available from the corresponding author). Adjusted Poisson regression models (relative risk [RR]; adjusted for age, smoking, alcohol intake, use of oral glucocorticoids, and Charlson comorbidity index) were used in order to calculate site-specific fracture incidence rates per 100,000 person-years at risk (pyar) and 95% confidence intervals (CIs) for each BMI category, and to provide RRs and corresponding p values to compare fracture rates among overweight and obese men versus underweight/normal weight men (reference group). All these analyses were two-sided, and were carried out using Stata SE version 12 for Mac (StataCorp LP, College Station, TX, USA).
In 2007, 186,171 men ≥65 years old were eligible, of whom 139,419 (74.9%) had an available BMI measurement (Fig. 1). The follow-up was (median [interquartile range]) 2.998 (2.995–3.000) years. Only 0.6% (n = 806) of these men were underweight (BMI < 18.5 kg/m2); therefore, they were categorized as underweight/normal (BMI < 25 kg/m2, n = 26,298), overweight (25 ≤ BMI < 30 kg/m2, n = 70,851), or obese (BMI ≥ 30 kg/m2, n = 42,270). The mean (SD) age of participants was 75.8 (7.8) years. Baseline characteristics for study participants are shown in Table 1. Men who completed follow-up were younger, had a higher BMI, a lower Charlson Index (less comorbidities), and were less likely to be users of oral glucocorticoids than those who died or were lost to follow-up during the study period.
|Whole population||Completed follow-up (Reference group)||Lost to follow-up||Dead|
|n (%)||186,171 (100)||164,319 (88.3)||3,347 (1.8)||18,505 (9.9)|
|Age (years), mean (SD)||75.8 (7.8)||75.0 (7.3)||77.0 (9.4)***||82.6 (8.5)***|
|BMI (kg/m2), mean (SD)||28.3 (3.9)||28.4 (3.9)||28.0 (4.3)**||27.4 (4.3)***|
|BMI, n (%)|
|<25 kg/m2||26,298 (14.1)||22,703 (13.8)||262 (7.8)||3,333 (18.0)|
|25 to <30 kg/m2||70,851 (38.1)||64,963 (39.5)||530 (15.8)||5,358 (29.0)|
|30 to <35 kg/m2||42,270 (22.7)||38,971 (23.8)||310 (9.3)||2,989 (16.1)|
|Not available||46,752 (25.1)||37,692 (22.9)||2,245 (67.1)||6,825 (36.9)|
|Current smoker||31,239 (16.8)||65,940 (40.1)||584 (17.5)***||6,259 (33.8)***|
|Ex-smoker||39,417 (21.2)||27,942 (17.0)||417 (12.5)||2,880 (15.6)|
|Never smoked||72,783 (39.0)||35,173 (21.4)||370 (11.0)||3,874 (20.9)|
|Not available||42,732 (23.0)||35,264 (21.5)||1,976 (59.0)||5,492 (29.7)|
|None/mild||58,535 (31.4)||53,176 (32.4)||227 (6.8)***||5,132 (27.7)***|
|Moderate||53,843 (28.9)||51,559 (31.4)||130 (3.9)||2,154 (11.6)|
|Severe||5,575 (3.1)||5,307 (3.2)||14 (0.4)||254 (1.4)|
|Not available||68,218 (36.6)||54,277 (33.0)||2,976 (88.9)||10,965 (59.3)|
|Oral glucocorticoids, n (%)||9,853 (5.3)||7,315 (4.5)||73 (2.2)***||2,465 (13.3)***|
|Charlson comorbidity index|
|0||124,307 (66.8)||112,757 (68.6)||2,278 (68.0)||9,272 (50.1)|
|1||34,180 (18.4)||30,663 (18.7)||589 (17.6)||2,928 (15.8)|
|2||15,875 (8.5)||12,939 (7.9)||252 (7.5)||2,684 (14.5)|
|3||6,756 (3.6)||4,865 (2.9)||139 (4.2)||1,752 (9.5)|
|≥4||5,053 (2.7)||3,095 (1.9)||89 (2.7)***||1,869 (10.1)***|
Overall, the most commonly affected sites were hip, followed by clinical spine and wrist/forearm. Adjusted hip fracture incidence rates decreased with each higher BMI category: 52.45/100,000 pyar (95% CI, 47.42–58.00) in underweight/normal men, 25.77 (95% CI, 23.64–28.09) in overweight men, and 22.43 (95% CI, 19.91–25.28) in obese men (Table 2). Age-specific fracture site rate ratios in the different BMI groups are shown in Table 2.
|Fracture site||Normal/underweight: BMI < 25 kg/m2 n = 26,298 (18.9%)||Overweight: 25 ≤ BMI < 30 kg/m2 n = 70,851 (50.8%)||Obese: BMI ≥ 30 kg/m2 n = 42,270 (30.3%)|
|IR (95%CI)||IR (95%CI)||Age-adjusted RRa (versus Reference group)||Adjusted RRb (versus Reference group)||IR (95%CI)||Age-adjusted RRa (versus Reference group)||Adjusted RRb (versus Reference group)|
|Hip (n = 1718)||52.45 (47.42–58.00)||25.77 (23.64–28.09)||0.61 (0.54–0.67); p < 0.0001||0.63 (0.55–0.72); p < 0.001||22.43 (19.91–25.28)||0.61(0.52–0.72); p < 0.0001||0.63 (0.54–0.74); p < 0.001|
|Spine (n = 899)||25.74 (22.29–29.72)||18.08 (16.31–20.04)||0.77 (0.65–0.92); p = 0.004||0.77 (0.64–0.92); p = 0.004||14.84 (12.82–17.19)||0.66(0.54–0.82); p < 0.0001||0.65 (0.53–0.80); p < 0.001|
|Multiple ribs (n = 54)||0.42 (0.13–1.29)||1.25 (0.84–1.85)||3.06(0.92–10.15); p = 0.067||3.42 (1.03–11.37); p = 0.045||1.50 (0.95–2.38)||3.67(1.08–12.48); p = 0.037||3.96 (1.16–13.52); p = 0.03|
|Wrist/forearm (n = 737)||18.54 (15.66–21.96)||15.93 (14.27–17.78)||0.90 (0.74–1.10); p = 0.31||0.91 (0.74–1.11); p = 0.34||13.34 (11.43–5.58)||0.77 (0.61–0.98); p = 0.025||0.77 (0.61–0.97); p = 0.03|
|Pelvis (n = 238)||6.64 (5.01–8.81)||5.09 (4.20–6.18)||0.89(0.63–1.25); p = 0.51||0.88 (0.63–1.25); p = 0.48||2.34 (1.61–3.38)||0.45(0.28–0.71); p = 0.001||0.44 (0.28–0.70); p < 0.001|
|Proximal humerus (n = 459)||11.21 (9.02–13.94)||9.44 (8.18–10.88)||0.92(0.71–1.19); p = 0.53||0.93 (0.72–1.21); p = 0.61||7.84 (6.40–9.60)||0.80(0.59–1.07); p = 0.14||0.81 (0.60–1.10); p = 0.17|
|Tibia (n = 93)||2.21 (1.36–3.61)||1.90 (1.38–2.61)||0.87(0.49–1.56); p = 0.64||0.88 (0.49–1.59); p = 0.67||1.75 (1.14–2.69)||0.80(0.42–1.53); p = 0.50||0.82 (0.42–1.58); p = 0.55|
When compared with men of normal weight, obese men had a higher prevalence of myocardial infarction, heart failure, and type 2 diabetes mellitus but a lower prevalence of stroke and dementia. All these comorbidities were included in the Charlson comorbidity index, for which the results were adjusted.
A statistically significant reduction in clinical spine and hip fractures was observed in obese (RR, 0.65; 95% CI, 0.53–0.80 and RR, 0.63; 95% CI, 0.54–0.74, respectively), and overweight men (RR, 0.77; 95% CI, 0.64–0.92 and RR, 0.63; 95% CI, 0.55–0.72, respectively) when compared to underweight/normal men (Figs. 2, 3). Additionally, obese men had less wrist/forearm (RR, 0.77; 95% CI, 0.61–0.97) and pelvic (RR, 0.44; 95% CI, 0.28–0.70) fractures than underweight/normal men (Fig. 3). Conversely, multiple rib fractures were more frequent in overweight (RR, 3.42; 95% CI, 1.03–11.37) and obese men (RR, 3.96; 95% CI, 1.16–13.52). No association was observed between BMI and either proximal humerus or tibia fractures (Table 2). The effect sizes observed and directions of the associations were similar when the analysis was limited to participants who completed the 3 years of follow-up (data not shown).
In this population-based cohort of older men, obesity was associated with a significantly lower risk of hip, clinical vertebral, pelvis, and wrist/forearm fractures than underweight/normal men, whereas the risk of multiple ribs fracture was almost fourfold higher. Proximal humerus and tibia fracture rates were similar across BMI groups.
Interpretation/comparison with previous studies
Our previous results in postmenopausal women in this cohort also demonstrated a decreased risk of hip and pelvis fracture in overweight and obese women when compared with normal and underweight women, and in other studies obesity in postmenopausal women has been associated with a decreased risk of incident wrist fracture. In contrast, although the risk of proximal humerus fractures was almost 30% higher in obese women in the SIDIAPQ cohort, we found no association between proximal humerus fractures and BMI categories in older men. Furthermore, the significant increase in risk of multiple rib fractures and reduced risk of clinical spine fractures in obese men was not observed in obese postmenopausal women from this cohort. Some previous studies have suggested that obesity may be associated with increased risk of vertebral fracture in postmenopausal women, although this has not been observed in men.[11, 12] Increased risk of ankle fracture has also been reported in obese postmenopausal women, but was not seen in men in this study. Whether or not there are true gender-related differences in the predisposition to fracture at different sites in obese individuals remains to be established.
In men, as in women, the relationship between BMI and fracture risk is nonlinear, with the highest risk of hip fracture seen in underweight individuals.[13-20] In the meta-analysis reported by De Laet and colleagues, no additional increase in the risk of hip fracture was seen in men with higher than normal BMI. However, in the Osteoporotic Fractures in Men (MrOS) study, a population-based study of men aged ≥ 65 years, obesity was associated with a higher incidence of nonspine and nonhip fractures when compared with normal weight men after adjustment for hip BMD. In this study the majority of men (72%) were overweight or obese, and fewer than 1% were underweight. The results are therefore consistent with studies that have shown the highest risk of hip fracture in underweight individuals, but suggest that obesity may increase the risk of nonspine and nonhip fractures independently of BMD. In the present study, BMD measurements were not available and thus it was not possible to investigate whether the observed relationship between BMI and fracture was modified by adjustment for BMD.
A number of factors have been implicated in the pathogenesis of fractures associated with obesity. Increased risk of falling has been reported in obese individuals[22, 23] and a higher impact of falling together with impairment of the normal protective responses to falling may increase the trauma sustained. The risk of falls may be related to reduced physical mobility; Friedmann and colleagues reported that both men and women with BMI >40 kg/m2 had significantly increased risk of functional limitation; and in the MrOS study, the increase in risk of nonspine fractures in obese men was attenuated after adjustment for limitations in mobility. In addition, because of reduced mobility, obese individuals may be more likely to fall backward or sideways as opposed to forward, thus protecting the wrist from impact.
Strengths and limitations
The main strengths of our study are that the SIDIAPQ database is highly representative of the whole population of our region. The population structure for the SIDIAPQ database matches that of Catalonia, and the prevalence and incidence rates for a list of conditions in SIDIAPQ match those from the existing literature.[25-31] The database covers >30% of the Catalonia population and the prevalence of obesity in this database is similar to that described in the literature for the source population. Aranceta and colleagues published a population-based survey in 2003 and reported that about 52% of men aged 55 to 60 years were overweight and 22% were obese, and the prevalence of obesity increased with age. In our study, among older men (aged ≥65 years), the prevalence of obesity was somewhat higher (about 30%) whereas 51% were overweight. Interestingly, in our population, obesity was more common in men than in women. Similar findings have been reported in a small cross-sectional study from Madrid, in which it was reported that 73% of men were overweight versus 42.3% of women (BMI ≥ 25 kg/m2) In addition, the prevalence of overweight and obesity in SIDIAPQ was close to the latest population-based data published from Spain in 2003. The sample size of our study population (139,419 subjects) and the fact that these data were collected during routine clinical practice give high generalizability to our findings. Even so, differences in the prevalence of obesity in our population and that of other studies (eg, MrOS, National Health and Nutrition Examination Survey [NHANES]) might explain the differences between the results obtained. Oversampling of the obese population could be an issue in our study if BMI was measured preferentially in obese/overweight people; nevertheless, it is routine practice to measure BMI yearly in primary care in all elderly individuals, and the prevalence of obesity in our study population is similar to that found in the abovementioned studies. The main limitation of this study is the nonvalidation of fractures at an individual level. Similar to other electronic medical records databases, SIDIAPQ does not provide access to X-ray images, and fractures are identified using ICD-10 codes. Nevertheless, hip, clinical spine, and wrist/forearm fracture coding has been shown to be highly specific (99%, 99%, and 98%, respectively) in SIDIAPQ. It is possible that there are some missing fractures in our dataset. However, underreporting is likely to be random and hence would not bias our estimates. Finally, differences in baseline characteristics between patients who were lost to follow-up and those who completed the study might introduce some degree of survival bias, but the exceptionally low proportion of participants who did not complete the follow-up period (<2% for a 3-year follow-up) seen in this study should minimize this issue. Our findings of no association between proximal humerus, tibial fractures, and BMI might be limited by reduced statistical power, and require replication in further studies.
Conclusions and Clinical Implications
Obesity in men was protective against hip, wrist/forearm, pelvic, and clinical spine fractures but appeared to increase the risk of multiple rib fractures. Tibia and proximal humerus fracture rates were not related to BMI. Further studies are needed to identify the mechanisms underlying these associations.
All authors state that they have no conflicts of interest.
Authors’ roles: Study Design: MP, XN, JC, ADP, and DPA. Study conduct: FFA, APC, MP, and DPA. Data management: FFA. Data interpretation: MP, APC, XN, JC, ADP, and DPA. Drafting manuscript: DPA, MP, JC, and ADP. Revising manuscript content: All the authors. Approving final version of manuscript: All the authors.