Measuring Coverage in MNCH: Testing the Validity of Women's Self-Report of Key Maternal and Newborn Health Interventions during the Peripartum Period in Mozambique

Background:As low-income countries strive to meet targets for Millennium Development Goals 4 and 5, there is growing need to track coverage and quality of high-impact peripartum interventions. At present, nationally representative household surveys conducted in low-income settings primarily measure contact with the health system, shedding little light on content or quality of care. The objective of this study is to validate the ability of women in Mozambique to report on facility-based care they and their newborns received during labor and one hour postpartum.Methods and Findings:The study involved household interviews with women in Mozambique whose births were observed eight to ten months previously as part of a survey of the quality of maternal and newborn care at government health facilities. Of 487 women whose births were observed and who agreed to a follow-up interview, 304 were interviewed (62.4%). The validity of 34 indicators was tested using two measures: area under receiver operator characteristic curve (AUC) and inflation factor (IF); 27 indicators had sufficient numbers for robust analysis, of which four met acceptability criteria for both (AUC >0.6 and 0.75<IF<1.25). Two of these indicators are considered high demand and are recommended for incorporation into international survey programs: presence of a support person during labor/delivery and placement of the newborn skin to skin against the mother. Nine indicators met acceptability criteria for one of the validity measures. All 13 indicators are recommended for use in in-depth maternal/newborn health surveys.Conclusions:Women are able to report on some aspects of peripartum care. Larger studies may be able to validate some indicators that this study could not assess due to the sample size. Future qualitative research may assist in improving question formulation for some indicators. Studies of similar design in other low-income settings are needed to confirm these results.