Study of non-hip fractures, which are aa serious public health problem for persons greater-than-or-equal-to 65 years of age, has been hindered by the absence of an economical method for case identification. We assessed the utility of computerized Medicare inpatient, emergency room, hospital outpatient department and physician claims for identifying fractures in an elderly Tennessee Medicaid population. We used these files for 1987 to identify 3086 possible fractures and reviewed medical records for aa sample of 1440. Using this sample, we developed aa definition of probable fractures that excluded claims unlikely to represent newly diagnosed fractures. For all fractures, this definition had aa positive predictive value of 94%, which for individual fracture sites, ranged from 79% (tibia/fibula) to 98% (hip). Of fractures in the reviewed sample, 91 % were identifed as probable fractures; this upper bound for sensitivity varied between 75% (femoral shaft) and 100% (patella). These data suggest that computerized Medicare files can be used for rapid and economical fracture ascertainment among persons greater-than-or-equal-to 65 years of age. However, further work is needed to obtain better estimates of sensitivity.
