Body Fat Percentage and the Long-term Risk of Fractures. The EPIC-Norfolk Prospective Population Cohort Study

Association of a Combined Body Mass Index and Regional Body Fat Percentage Metric With Fragility Fracture Risk: Evidence from a Large Observational Cohort

BACKGROUND

Evidence suggests that high body fat and low muscle mass may increase the risk of fragility fractures. However, current fracture risk models, which largely rely on body mass index (BMI), may not fully capture these compositional factors. We recommend integrating additional body composition variables into fracture risk calculators to improve accuracy. Previously, we described partial body fat percentage (PBF%), a novel measure that is routinely available and calculated as the proportion of fat at the lumbar spine and hip during DXA scans. We hypothesize that a combined BMI and PBF% approach (BMI/PBF%) could be associated with fragility fracture.

METHODS

Patients were referred to our DXA scanner between June 2004 and February 2024 and had combined lumbar spine and bilateral femoral scans. Patients were initially categorized by BMI (underweight, normal weight, overweight and obese) and then divided into tertiles of PBF%. Based on each patient's unique combination of BMI and PBF% tertile, they were stratified into 12 binary BMI/PBF% groups for analysis. Multivariable logistic regression models, reporting odds ratios (OR), with BMI/PBF% groups as the independent variables and fragility fractures as the dependent variable were fit, with all results adjusted for known fracture risk factors.

RESULTS

We analysed 36 235 patients (83.4% female, 16.6% male), of whom 14 342 (39.5%) reported fragility fractures. The median (IQR) age was 67.7 (57.5-75.0) years, with a BMI of 26.4 (23.3-30.2) kg/m2 and PBF% of 30.6% (25.5% - 35.4%). In females, those in the lowest PBF% tertile had reduced odds of fragility fractures across all BMI categories (e.g., obese low PBF%: OR 0.70, 95% CI 0.64-0.78), whereas in males, this reduction was observed only amongst overweight and obese individuals (e.g., obese low PBF%: OR 0.71, 95% CI 0.57-0.88). No association was found for patients in the middle PBF% tertile across any BMI group. In contrast, females in the highest PBF% tertile exhibited increased odds of fractures across all BMI categories except underweight (e.g., obese high PBF%: OR 1.31, 95% CI 1.22-1.42), and a similar pattern was seen in males, but limited to the overweight and obese groups (e.g., obese high PBF%: OR 1.27, 95% CI 1.04-1.55).

CONCLUSION

High or low PBF% within BMI categories is associated with fragility fractures, challenging the traditional notion that high BMI protects against fractures. This study highlights the importance of body composition measures beyond BMI in fracture risk assessment.

Partial body fat percentage as a predictor of fragility fractures in a large cohort: a cross-sectional study

Objectives

BMI is a component of fracture risk calculators; however, it may be too simplistic to predict fracture risk. There is emerging evidence for the role that fat plays as a predictor of fracture. Partial body fat percentage (PBF%) may be a novel way to predict both hip and non-hip fractures. The aim of this study is to evaluate PBF% as a predictor of fragility fractures.

Methods

A multivariate logistic regression analysis was conducted looking at PBF% as a predicter of both non-hip and hip fractures in an observational cohort. Our results were adjusted for age, biological sex, gender, smoking status, excess alcohol consumption (>3 units/day), current steroid therapy and the T-scores in both femurs. To allow for comparison, the same model was used with BMI, height and weight as the primary predictor of fracture. A subgroup analysis was conducted stratified by fracture site. A sensitivity analysis using a negative binomial regression was conducted.

Results

A total of 31 447 patients were included in our analysis [mean age 64.9 years (s.d. 12.9)]. PBF% was shown to predict all non-hip fractures after adjustment [odds ratio (OR) 22.14 (95% CI 15.08, 32.50)]. Hip fractures were not predicted by our model [OR 4.19 (95% CI 0.43, 41.46)]. Sensitivity analysis demonstrated a lack of predictive capability for hip fracture but not non-hip fractures.

Conclusion

PBF% may be a suitable predictor for all non-hip fractures, independent of confounding variables. More research is needed on whether it can predict hip fractures.