Socioeconomic misclassification in Ontario's health care registry

Background: Addresses in some provincial health care registries are not systematically updated. If individuals are attributed to the wrong location, this can lead to errors in health care planning and research. Our purpose was to investigate the accuracy of socioeconomic classification based on addresses in Ontario's provincial health care registry. Methods: The study setting was Toronto's inner city, an area with a population of 799,595 in 1996. We ordered enumeration areas by 1996 mean household income and divided them into five roughly equal income groups by population. We then assigned an income quintile to each individual using both the address from Ontario's provincial health care registry and that from hospital discharge abstracts. We compared these two sets of income quintiles and also used them to generate quintile-specific rates of medical hospital admissions in the year 2000. Results: Provincial registry and hospital-based addresses agreed on the exact enumeration area for 78.1% of individuals and for income quintile for 84.8% of individuals. Disagreement by more than one income quintile occurred for 7.4% of individuals. The two methods of assigning income quintiles yielded income-specific medical hospitalization rates and rate ratios that agreed within 1%. Interpretation: Although address inaccuracy was found in Ontario's health care registry, serious socioeconomic misclassification occurred at a relatively low rate and did not appear to introduce significant bias in the calculation of hospital rates by socioeconomic group. Updating of addresses at regular intervals is highly desirable and would result in improved accuracy of provincial health care registries.