Study objective: To document the validity of a Hospital Discharge Index (HDI) as a data base on injured patients.
Design: Patient information in trauma registries was compared with information in HDI.
Population: Injured patients admitted to trauma centers.
Methods: Patients in HDI were crossmatched with individuals in one or two trauma registries using deterministic matching techniques. Agreement regarding the presence and severity of injury was assessed.
Results: A comprehensive trauma registry from a level I trauma center and HDI agreed on the presence of an injury in each of 6 body regions over a range of kappa values from 0.17 to 0.71. The severity of injury score assigned by the two data bases demonstrated agreement over a range of intraclass correlation values from 0.12 to 0.82.
Conclusion: HDI provides adequate information concerning injury for the majority of hospitalized patients, but was primarily limited by incomplete information. Efforts to improve HDI should focus on guidelines for data abstraction.