| Sumario: | BACKGROUND: The main function of a trauma registry is to assess quality assurance and performance improvement (QA/PI) in an individual institution. Nonvalidated registry data may produce unreliable reports and QA/PI information. This study examines the types of data entry errors in a trauma registry database; the effect of errors on time variable estimates, case ascertainment and statistical measurement; dynamics of error occurrence; and data validation (DV) scheme for a trauma registry. METHODS: Query and cross-tabulation techniques were used to expose a variety of data entry errors. Conceptual aspect for each type of error in DV, especially with respect to QA/PI, is given. RESULTS: Findings of different errors are provided: out-of-range time values; false positive and false negative errors; errors of commission and omission; duplication errors; errors in demographics; and errors because of inconsistent and incongruent coding. Error rates were less than 3% in commonly occurring data, such as scene time, demographics, hospital discharge and transportation, and greater in less commonly occurring but important data, such as thoracic aorta injury (9.5%) and audit filter for admit Glasgow Coma Scale in emergency department (55.6%). Dynamics of error occurrence that can prevent or minimize errors is described. The main features of a data validation scheme are displayed. CONCLUSIONS: Errors in a trauma registry database cause invalid frequencies, rates, time estimates and statistical measures and affect QA/PI in trauma care. Every functioning trauma registry should develop an on-going program for DV.
|
|---|