Estimating Mortality Hazard Ratios Using Cross-Sectional Data from Maternal Birth Histories

• Activity Number: 700
• Type: Contributed
• Date/Time: Thursday, August 4, 2016 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Philimon Gona* and Janet Dzangare and Vasco Chikwasha and Sowmya Rao
• Companies: University of Massachusetts and Ministry of Health and Child Welfare and University of Zimbabwe College of Health Sciences and Boston University
• Keywords: Mortality hazards ; Birth histories ; Cross-sectional study ; Complex Surveys

Abstract

The Demographic and Health Surveys (DHS) are conducted by USAID approximately every 5 years to "provide data for a wide range of monitoring and impact evaluation indicators for population, health, and nutrition". By design, cross-sectional health surveys preclude time-to event analyses. In a 2010 DHS conducted in Zimbabwe, mothers were queried about their histories of live births and deaths of their children. We considered age (survival time) to be censored on survey date for surviving children; otherwise age at death was calculated. Age and vital status were used as time and censoring variables, respectively. Weighted crude, sex- and region-specific mortality rates were calculated using Cox models accounting for household factors and the complex survey design. Of the 6725 livebirths, there were 473 deaths, i.e., crude mortality rate was 6.6/1000 person-years, 95% CI (6.0-7.2). There were increased hazards in urban areas, among Pentecostals; a positive trend over time and regional variation in hazard ratios (HRs). HRs did not show regional heterogeneity. This approach is useful for monitoring mortality trends over time and comparing them across countries where DHS is administered.