Administrative and registry databases are useful for researchers given their availability and size, yet their limitations for specific applications remain undefined. We compared injury records from a large administrative database and the National Trauma Data Bank (NTDB) with the goal of furthering the understanding of their respective limitations.
The study hospitals had submitted records to both the NTDB and the Nationwide Inpatient Sample (NIS) for patients admitted during 2002. Record inclusion criteria for comparison included nonelective admissions with a primary diagnosis of injury (excluding isolated hip fractures). Numbers of cases and variables common to both databases were compared.
Twenty-four hospitals had records both in the NTDB (24,619 records) and in the NIS (25,586 records). We found less missing cost and payer information in the NIS compared with the NTDB (0% and 0.1% vs. 30.5% and 24%, respectively), higher mean number of comorbidities per record in the NIS (0.77 vs. 0.18), and a lower crude case fatality rate in the NIS (3.5% vs. 5.2%).
The main differences between the databases reflected the different motives for data collection and the inclusion or exclusion criteria imposed by trauma registries. These differences require consideration when using either database to investigate injury-related questions.
From the Division of Research and Optimal Patient Care (B.P., M.S., A.C.F.), American College of Surgeons, Chicago, Illinois; Department of Surgery (D.E.C.), Maine Medical Center, Portland, Maine; Division of General Surgery and Trauma, St. Michael's Hospital and the Department of Surgery (A.B.N.), University of Toronto, Toronto, Canada; and the Department of Surgery (A.C.F.), Louisiana State University Health Science Center, New Orleans, Louisiana.
Submitted for publication October 25, 2006.
Accepted for publication October 11, 2007.
Address for reprints: Bart Philips, BS, 633 N. Saint Clair St., Floor 25E, Chicago, IL 60611; email: email@example.com.