INTRODUCTION: Absolute 10-year fracture risk is the preferred method for fracture risk assessment. The validity of applying published fracture rates from one population to another population is uncertain. METHODS: 20,579 women age 47.5 years or older at the time of baseline femoral neck bone mineral density (BMD) were identified in a database containing all clinical DXA results for the Province of Manitoba, Canada. Individual 10-year fracture risk was predicted from age-only and age plus femoral neck T-score using published 10-year fracture risk for Swedish women. Health service records were assessed for the presence of non-trauma 'osteoporotic' fracture codes (hip, clinical spine, wrist, humerus) subsequent to BMD testing (86,447 person-y follow up, 1173 patients with osteoporotic fractures). Fracture rates were derived for subgroups stratified by age (5-year strata) and estimated risk (5% strata). 10-year fracture rates were computed directly by the Kaplan-Meier method (10-year continuous data) and by the actuarial method (two 5-year periods with adjustments for aging, death and expected BMD loss). RESULTS: Direct and actuarial methods gave nearly identical point estimates, but the latter were more precise. There was a strong linear correlation between predicted and observed fracture rates based upon age-only (r = 0.95) and age plus BMD (r = 0.99). For age strata 50 to 75, and for estimated risk strata from 0-5% to 20-25%, the confidence intervals overlapped the line of identity. For women age >77.5 or estimated risk >25%, observed exceeded estimated fracture rates. This is explained by healthy selection bias whereby elderly women referred for BMD testing have lower mortality than expected, hence more years at risk for fracture. Corrected for survival bias, women age >77.5 had observed fracture rates no different than predicted. CONCLUSION: Swedish 10-year fracture risk data are generally applicable to the Canadian female population referred for clinical BMD testing, though fracture rates were underestimated in the oldest and highest risk subgroups due to healthy selection bias.
INTRODUCTION: Absolute 10-year fracture risk is the preferred method for fracture risk assessment. The validity of applying published fracture rates from one population to another population is uncertain. METHODS: 20,579 women age 47.5 years or older at the time of baseline femoral neck bone mineral density (BMD) were identified in a database containing all clinical DXA results for the Province of Manitoba, Canada. Individual 10-year fracture risk was predicted from age-only and age plus femoral neck T-score using published 10-year fracture risk for Swedish women. Health service records were assessed for the presence of non-trauma 'osteoporotic' fracture codes (hip, clinical spine, wrist, humerus) subsequent to BMD testing (86,447 person-y follow up, 1173 patients with osteoporotic fractures). Fracture rates were derived for subgroups stratified by age (5-year strata) and estimated risk (5% strata). 10-year fracture rates were computed directly by the Kaplan-Meier method (10-year continuous data) and by the actuarial method (two 5-year periods with adjustments for aging, death and expected BMD loss). RESULTS: Direct and actuarial methods gave nearly identical point estimates, but the latter were more precise. There was a strong linear correlation between predicted and observed fracture rates based upon age-only (r = 0.95) and age plus BMD (r = 0.99). For age strata 50 to 75, and for estimated risk strata from 0-5% to 20-25%, the confidence intervals overlapped the line of identity. For women age >77.5 or estimated risk >25%, observed exceeded estimated fracture rates. This is explained by healthy selection bias whereby elderly women referred for BMD testing have lower mortality than expected, hence more years at risk for fracture. Corrected for survival bias, women age >77.5 had observed fracture rates no different than predicted. CONCLUSION: Swedish 10-year fracture risk data are generally applicable to the Canadian female population referred for clinical BMD testing, though fracture rates were underestimated in the oldest and highest risk subgroups due to healthy selection bias.
Authors: Lisa Langsetmo; Suzanne Morin; Christopher S Kovacs; Nancy Kreiger; Robert Josse; Jonathan D Adachi; Alexandra Papaioannou; David Goltzman; David A Hanley; Wojciech P Olszynski; Jerilynn Prior; Sophie A Jamal Journal: Menopause Date: 2010 Sep-Oct Impact factor: 2.953
Authors: Alexander S Dash; Sanchita Agarwal; Donald J McMahon; Felicia Cosman; Jeri Nieves; Mariana Bucovsky; X Edward Guo; Elizabeth Shane; Emily M Stein Journal: Bone Date: 2019-12-20 Impact factor: 4.398