Comparison of household socioeconomic status classification methods and effects on risk estimation: lessons from a natural experimental study, Kisumu, Western Kenya

• DOI: 10.1186/s12939-022-01652-1
• Publication Date: 2022-04-09T17:02:39Z
• Authors: Vincent Were; Louise Foley; Eleanor Turner-Moss; Ebele Mogo; Pamela Wadende; Rosemary Musuva; Charles Obonyo
• Keywords: Household income; Cross-sectional study; Biostatistics

Abstract Introduction Low household socioeconomic status is associated with unhealthy behaviours including poor diet and adverse health outcomes. Different methods leading to variations in SES classification has the potential generate spurious research findings or misinform policy. In low middle-income countries, there are additional complexities defining SES, a need for fieldwork be conducted efficiently, dearth of information on how could impact estimation disease risk. Methods Using cross-sectional data from 200 households Kisumu County, Western Kenya, we compared three approaches classifying into low, middle, high SES: fieldworkers (FWs), Community Health Volunteers (CHVs), Multiple Correspondence Analysis econometric model (MCA). We estimated sensitivity, specificity, inter-rater reliability misclassification using MCA as comparator. applied an unadjusted generalized linear determine prevalence ratios assess association self-reported diagnosis diabetes hypertension one member. Results Compared MCA, FWs successfully classified 21.7% (95%CI = 14.4%-31.4%) households, 32.8% 23.2–44.3) middle no households. CHVs 22.5% 14.5%-33.1%) 23.2%-44.3%) The level agreement was similar between but particularly SES. None differed estimating risk diabetes. Conclusions FW CHV assessments community-driven classification. these appeared biased towards not sensitive did differ hypertension. A mix further evaluation refine methodology recommended.