Estimating mortality from external causes using data from retrospective survey questions: a validation study in Niakhar (Senegal)
Mortality and causes of death in global South countries are not well known because individuals’ vital information is often incomplete. Specifically, some deaths go unreported. To estimate adult mortality, researchers often use responses from sample respondents asked about their siblings: respondents are asked to give a list of their siblings, including age or, if deceased, at what age. Gilles Pison, Géraldine Duthé, Bruno Masquelier, Cheikh Sokhna, Almany Malick Kante, Valérie Delaunay, Cheikh Tidiane Ndiaye, Stéphane Helleringer and Laetitia Douillotcherché, the authors of “Estimating mortality from external causes using data from retrospective survey questions: a validation study in Niakhar (Senegal)”, set out to determine whether adding questions that would discern violent deaths (by accident, homicide or suicide) of respondent’s siblings might contribute to a reliable estimation of mortality from external causes.
They conducted a study in the Observatory on population, health and environment in Niakhar, Senegal, questioning a sample of 1,189 adults aged 15-59 about their siblings, as is done in national surveys, and adding questions to identify violent deaths. They then compared respondent-reported information with the real situation, apprehended independently by the Observatory, and measured respondent report errors.
Altogether, respondents did not mention 12% of siblings who had died as adults, and a great many siblings who had died at a young age. Adult sibling deaths due to accidents and violence are well reported; those due to other causes are not as well reported.
Cause of death is correctly reported for siblings who died in an accident; for example, a traffic accident. On the other hand, some respondents whose sibling’s death was due to homicide or suicide report a different cause of death. Adding the type of question specified above should therefore facilitate correct estimation of accident-caused mortality and enable researchers to follow trends in it, but the results are less promising for estimates of homicide and suicide-induced deaths.
Source: Gilles Pison, Bruno Masquelier, Almamy Malick Kante, Cheikh Tidiane Ndiaye, Laetitia Douillot, Géraldine Duthé, Cheikh Sokhna, Valerie Delaunay, Stephane Helleringer, 2018, Estimating mortality from external causes using data from retrospective surveys: A validation study in Niakhar (Senegal), Demographic Research.
Online: August 2018