Understanding Reporting Errors of Population Survey–Based Neonatal Mortality: A Validation Study From Guinea-Bissau

Li Liu , Johns Hopkins University
Yue Chu, The Ohio State University
Amabelia Rodrigues, Bandim Health Project
Ane Fisker, Statens Serum Institut
Stephane Helleringer, Johns Hopkins University

In low-income countries, neonatal mortality rates (NMR) are usually estimated using population survey, specifically through full birth history (FBH). However, reporting errors, including omission, time transference, age transference and misclassification, threaten the validation of FBH based NMR. In this study, we validated FBH against data in the health demographic surveillance site in Bissau (BHDSS) to examine the impacts of reporting errors on NMR. Using BHDSS as the sampling frame, we applied a stratified random sampling strategy to identify study women, and then administered a standard FBH. We found moderate sensitivity (0.79, 95% confidence interval (CI) [0.72, 0.85]) and imperfect specificity (0.993, 95% CI [0.989, 0.996]). The moderate sensitivity was largely due to misclassification between pregnancy terminations and neonatal deaths in FBH, whereas the imperfect specificity was driven by omissions. Collectively, the reporting errors would most likely lead to overestimation of NMR in high mortality settings. Study implications are discussed.

See paper

 Presented in Session 109. Measurement Challenges and Innovations in Infant and Child Health and Mortality