Misclassification of income quintiles derived from area-based measures. A comparison of enumeration area and forward sortation area.

Abstract

Census-based methods are often used to estimate socioeconomic status. We assessed the agreement between Forward Sortation Area (FSA) and Enumeration Area (EA) derived income levels for all patients undergoing cardiac catheterization in Alberta, Canada, from 1995-1998. Income quintiles were calculated from census data for FSA and EA level. FSA- and EA-derived income measures were compared for misclassification. Both methods were then applied to the data to determine 4-year survival by income grouping in 21,446 patients following catheterization. The variability in EA-derived incomes for any given FSA-derived income is large. Only 40% of income quintiles are in agreement between the methods. For EA-based analyses, there is a linear relationship between higher income and lower mortality across all quintiles, while for FSA-based analyses, only the lowest income quintile had significantly higher mortality. Assuming that FSA-based methods are more likely to misclassify income compared to EA-based measures, the results for the FSA-based analyses are more likely to be erroneous. EA-derived measures should therefore be used when individual data are not available.