An assessment of changes in infant and under-five mortality in demographic and health survey data for Madagascar

To do this we make use of the 1997 and 2003 DHS data as well as the 1992 DHS, and where possible resort to information from other sources. The plan of the analysis is as follows. In the next section we consider the presence (in each DHS) of birth year displacement problems and of age of death misreporting, and investigate the implications for estimates of infant and under-five mortality. This analysis also highlights the discrepancies between the 2003 DHS and earlier surveys in mortality rates calculated for similar periods of calendar time. In Section 3, we undertake a detailed comparison of sample characteristics in each survey, to see if problems in sampling might explain the discrepancies. In Section 4, we examine whether the problems instead may be attributable to the performance of specific interviewers. In Section 5 we consider two distinct approaches to correcting the mortality estimates for biases. For this we estimate hazard regression models of child survival. Section 6 considers whether the measured reductions in mortality may be explained, in whole or in part, by recent program activity that was centered in two of the six provinces of the country. A final section summarizes the findings of the analysis. (Author abstract)