Validity of administrative data for characterizing traumatic brain injury-related hospitalizations

Abstract

Primary objective: To examine the validity of Maryland Hospital Discharge (MHD) data for identifying and characterizing traumatic brain injury (TBI)-related hospitalizations. Methods: All TBI-related hospitalizations in 1999 were identified using MHD and Maryland Trauma Registry (MTR) data. In addition, a sample of records were abstracted to compare agreement between MHD and chart data. Results: The MHD file identified fewer TBI cases (61%) compared to the MTR (95%). Overall, TBI-related hospitalization rates based on MHD were significantly fewer (95; 95% CI 92, 98) vs MHD and MTR (144; 95% CI 140, 147). There was good agreement between the MHD and chart data regarding skull fractures or intracranial lesions (κ = 0.73 and 0.83, respectively), but poor agreement for neurologic abnormalities or amnesia. The MHD significantly underestimated TBI severity. Conclusions: TBI cases, especially mild ones, were under-reported by MHD data. MHD data are better at detecting anatomic injuries compared to TBI symptoms and sequella.