Background: Bone mineral density (BMD) measurement is less precise amongst those with body mass index (BMI) > 30 kg/m2. We hypothesized that the association of BMD with incident hip and major osteoporotic fractures (MOF; hip, clinical vertebral, forearm, or humerus) becomes weaker with increasing BMI.
Methodology: Our study population was 75,391 individuals age ≥ 50 years who had a bone density test in the province of Manitoba 1998 to 2018. BMD of the total hip was assessed on GE Lunar densitometers. Incident MOF and hip fractures were ascertained using linked health claims data over a mean (SD) follow-up of 8.6 (5.3) years. The associations of total hip BMD with incident major osteoporotic and hip fractures were estimated with Cox proportional hazards models including an interaction term between BMI category and BMD to test if the association of BMD with incident fractures varies by BMI.
Result: The multivariable adjusted associations of total hip BMD with incident MOF did not vary by BMI (hazard ratio [HR] 1.56, 95 % C.I. 1.30, 1.85 for BMI ≥ 40 kg/m2; HR 1.36, 95 % C.I. 1.17, 1.58 for BMI <18.5 kg/m2; p-value for interaction 0.14). However, the association of total hip BMD with incident hip fracture was stronger for those with BMI ≥ 35 kg/m2 (HR 2.16, 95 % C.I. 1.71, 2.74) compared to those with BMI <18.5 kg/m2 (HR 1.48, 95 % C.I. 1.19, 1.84, p-value 0.001 for interaction).
Conclusion: The associations of total hip BMD with incident major osteoporotic and hip fracture are as strong for those with very high BMI as for those with normal BMI. However, total hip BMD may have a weaker association with incident hip fracture among underweight individuals. Further studies to confirm and explain this finding are warranted.