On the need of validating inpatient registers

Abstract

Study design: Register study. Objectives: To design and implement a validation process to check the completeness of the Hospital Discharge Register (HDR) held by the Swedish National Board of Health and Welfare. Setting: Sweden. Methods: An accurate traumatic spinal cord injury prevalence group (n=495) was acquired from the Swedish Spinalis Clinic. A register control was performed on the group by raising three questions to check the validity of the HDR: Is an inpatient stay registered in association with the injury date? Is the reported first length of stay plausible, given the level and extent of injury? Are all the anticipated care and/or rehabilitation providers represented in the HDR? Results: For 62% (of 413 cases) the first registered hospitalization date correlated with the injury date. For the other 38%, hospitalization was reported to start between 2 and 8651 days after injury. Considering the level and extent of injury, individuals were reported to have unrealistically short initial hospitalization. The prevalence group visited 42 different hospitals and 47 clinics. Five rehabilitation clinics, though, were not reported. Conclusions: The HDR is a valuable source when conducting epidemiological and health services research. However, using the register without any validation process could, as detected in the investigated diagnosis group, lead to a severe underestimation of the inpatient usage. The study showed that systematic errors could be detected by means of extensive knowledge of the diagnosis group. © 2008 International Spinal Cord Society All rights reserved.