English  Français

Using parental survival histories in low- and middle-income countries to estimate mortality over age 50: Assessing selection bias

Ashira Menashe-Oren, Universite catholique de Louvain (UCL)
Benjamin-Samuel Schlüter, Louvain University (UCL)
Bruno Masquelier, Louvain University (UCL)
Stephane Helleringer, Johns Hopkins University

As child mortality has declined in developing countries, more people are reaching older ages, yet little is known about old age mortality in these settings, where civil registration and vital statistics is lacking. Surveys can be a good alternative, but existing data collection instruments do not cover deaths above age 60. We propose using parental survival histories (PSH) which build on four simple questions – vital status of biological parents, age of parent, or age at death, and timing of death. We examine sample selection biases in mortality estimates from PSH. We employ data from Health and Demographic Surveillance Systems (HDSS) to assess the possible correlation between mortality rates above age 50 and the time-varying number of surviving children of “respondent age”, using semi-parametric Cox models. Preliminary analysis, on three HDSS in Senegal suggests that women who have never had children have higher survival than women who have had children, but there seem to be no differences in female survival according to the number of female surviving children. These results are encouraging, suggesting that selection bias in PSH may only be an artifact of having children or not, and not the number of children alive to report on their parent’s survival.

Keywords: Older adults, Mortality, Demographic and social surveys

See extended abstract.

  Presented in Session 118. Estimating and Modelling Mortality