A Comparison of Sub-national Survey Data and Health Management Information System Data in Zambia

Abstract

Child health outcomes and coverage of key maternal and child health interventions vary widely across districts in Zambia. Sub-national data is necessary to empower health decision makers to target informed strategies to improve service delivery and population health. Some efforts have been made to produce representative estimates of coverage and health outcomes at the district level in Zambia. This investigation explores the feasibility of using data from the Health Management Information System (HMIS) to inform the estimation of trends in child mortality (5q0), skilled birth attendance (SBA), and antenatal care (ANC). District-level trends in 5q0, SBA, and ANC were constructed using HMIS data from 2000-2008. HMIS-based trends were then compared to survey-based data using visual analysis and pearson's correlation. Resulting correlations ranged from strongly-correlated (r >0.6) to negligible association (r=0) to strong negative correlation (r <-0.6). In some districts and indicators, HMIS data mirrored the trend produced using survey data and statistical modeling, while in others the two datasets failed to agree. Further investigation is necessary, but preliminary results suggest that HMIS data is a powerful and neglected resource for subnational estimation.