Search | Search by Center | Search by Source | Keywords in Title
LaFleur J, Nelson RE, Yao Y, Adler RA, Nebeker JR. Validated risk rule using computerized data to identify males at high risk for fracture. Osteoporosis international : a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA. 2012 Mar 1; 23(3):1017-27.
Absolute risk assessment is now the preferred approach to guide osteoporosis treatment decisions. Data collected passively during routine healthcare operations can be used to develop discriminative absolute risk assessment rules in male veterans. These rules could be used to develop computerized clinical decision support tools that might improve fracture prevention. INTRODUCTION: Absolute risk assessment is the preferred approach to guiding treatment decisions in osteoporosis. Current recommended risk stratification rules perform poorly in men, among whom osteoporosis is overlooked and undertreated. A potential solution lies in clinical decision support technology. The objective of this study was to determine whether data passively collected in routine healthcare operations could identify male veterans at highest risk with acceptable discrimination. METHODS: Using administrative and clinical databases for male veterans = 50 years old who sought care in 2005-2006, we created risk stratification rules for hip and any major fracture. We identified variables related to known or theoretical risk factors and created prognostic models using Cox regression. We validated the rules and estimated optimism. We created risk scores from hazards ratios and used them to predict fractures with logistic regression. RESULTS: The predictive models had C-statistics of 0.81 for hip and 0.74 for any major fracture, suggesting good to acceptable discrimination. For hip fracture, the cut-point that maximized percentage classified correctly (accuracy) predicted 165 of 227 hip fractures (73%) and missed 62 (27%). All hip fractures in patients with prior fracture were identified and 67% in patients without. For any major fracture, the maximal-accuracy cut-point predicted 611 of 987 (62%) and missed 376 (38%); the rule predicted all 134 fractures in patients with prior fracture and 56% in patients without. CONCLUSION: Data collected passively in routine healthcare operations can identify male veterans at highest risk for fracture with discrimination that exceeds that reported for other methods applied in men.