|
|
|
|
|
Andrew Shore, PhD, Department of Health Policy and Management, Johns Hopkins University, 624 North Broadway, Baltimore, MD 21205, 410-955-6547, ashore@jhsph.edu, Melissa McCarthy, ScD, Department of Emergency Medicine, Johns Hopkins University, Suite 6-111, 1830 East Monument Street, Baltimore, MD 21205, Tracey Serpi, MS, Office of Injury Prevention and Health Assessment, Maryland Department of Health and Mental Hygiene, 201 West Preston Street, Room 302, Baltimore, MD 21201, and Melanie Gertner, BS, Maryland Institute of Emergency Medical Services Systems, 653 West Pratt Street, Room 653, Baltimore, MD 21201.
Purpose. To examine the validity of using Maryland Hospital Discharge (MHD) data to identify and characterize traumatic brain injury (TBI)-related hospitalizations. Methods. All TBI-related hospitalizations that occurred in Maryland in 1999 were identified using MHD and Maryland Trauma Registry (MTR) data. In addition, a stratified random sample of 1,002 of the TBI-related hospitalizations that were identified by the MHD database were selected for medical record abstraction to compare agreement between administrative data and chart information regarding the presence of intracranial lesions, skull fractures, neurologic abnormalities and amnesia. Results. The MHD file identified fewer (64%) TBI-related hospitalizations compared to the MTR (86%). Two-thirds of the TBI cases documented only by the MTR were of minor severity (i.e. head Abbreviated Injury Scale (AIS)=2)). There was good agreement between the MHD and the medical record data regarding the presence of skull fractures or intracranial lesions (kappa=0.73 and 0.83 respectively); there was poor agreement for the presence of neurologic abnormality or amnesia. The mean head AIS score of the MHD database was 1.6 compared to 2.5 from the medical record data (p < 0.0001). Conclusions. TBI-related hospitalizations may be underreported by administrative data. MHD data were better at detecting anatomic injuries (e.g. skull fractures or intracranial lesions) compared to symptoms and sequelae of brain trauma (e.g. amnesia or neurologic abnormalities). The MHD file underestimated TBI severity, largely because of the frequent use of non specific diagnosis codes . These shortcomings need to be considered when using hospital discharge data for injury surveillance purposes.
Learning Objectives: The participants in this session will be able to
Keywords: Surveillance, Traumatic Brain Injury
Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.