Validation of the Swedish inpatient and cause-of-death registers in the context of stroke

Background - Quality follow-up within stroke care is important in times when stroke prevalence is increasing and health care funds are limited. Administrative data, such as data from the inpatient register (IPR) and the cause-of-death register (CDR) are often used for this purpose, but the validity of such data has not been ascertained. Methods - During the year 1999-2000, a community-based stroke register was established in a Swedish municipality. Data from that register was compared with two administrative registers, the IPR and the CDR. Results - Using multiple overlapping data sources, 377 patients with first-ever stroke were found in the community-based register. Forty-four of these (12%) were missing in the IPR/CDR. Non-hospitalized patients were less likely to be registered in the IPR/CDR, as were patients who were not initially treated in a stroke unit. Stroke severity was lower among non-registered patients. Thirty patients (8%) in the IPR/CDR were misclassified as stroke patients. Conclusions - Quality follow-up within stroke care could be biased or have low comparability, when administrative data are used. Great caution should be taken when data derived from the inpatient and cause-of-death registers, and more validation work needs to be carried out in the context of stroke.