Pregnancy Reporting Completeness in HDSS: Results from a Record Linkage Study in South Africa

Hallie Eilerts-Spinelli, Johns Hopkins Bloomberg School of Public Health
Georges Reniers, London School of Hygiene and Tropical Medicine (LSHTM)

Health and Demographic Surveillance Systems (HDSS) provide valuable empirical insights in low-resource regions for which data is often lacking. However, estimates of under-five mortality are often unrealistically low. In this manuscript, we evaluate the completeness of pregnancy reports in the HDSS through individual-level record linkage with data from medical facilities that serve the HDSS population. Routine programme data from 8 primary health care facilities were probabilistically linked to the Africa Health and Research Institute HDSS. The clinical data were crosschecked with reported pregnancies and pregnancy outcomes in the HDSS. Attributes associated with pregnancy under-reporting in the HDSS were investigated with multivariable logistic regression models. 2,040/3,165 (64.45%) of individuals were matched from the clinical data to the HDSS. A substantial proportion of linked records (29.09%) did not have a pregnancy or pregnancy outcome reported in the HDSS. This is cause for concern, as incomplete pregnancy reporting can bias under-five mortality estimates.