OBJECTS: In a comparative analysis, the effect of risk factors for childhood mortality in Burkina Faso, sub-Saharan Africa, were computed from Demographic and Health Survey (DHS) and Demographic Surveillance System (DSS), two very different sources. While most analyses so far determined levels of risk factors, this analysis focuses on the effects of those risk factors. METHODS: Mortality levels are often calculated by strata of risk factor levels separately for each factor. In this publication, their effect was modelled jointly by survival analysis, using all-cause childhood mortality as outcome variable. All live births in rural Burkina Faso in the period of 1994-1998 recorded by the DHS survey Burkina Faso 1998-1999 (5018 births) and the Demographic Surveillance System based in Nouna, western Burkina Faso (6196 births) were included. RESULTS: A simultaneous estimation of hazard rate ratios by a Cox regression model yielded similar estimates for the DHS and DSS data, in line with previous findings based on the Nouna DSS alone. Although DHS surveys do not include children whose mothers had died, ignoring the strong risk factor "death of the mother" does not strongly affect DHS-based results. CONCLUSIONS: Despite the different nature of survey (DHS) and longitudinal (DSS) data, these findings demonstrate that, despite some limitations, results derived from DHS surveys are broadly comparable to DSS data. Both are valuable tools for assessing the importance of risk factors for childhood mortality in sub-Saharan Africa, and they could be combined for better predictions.
OBJECTS: In a comparative analysis, the effect of risk factors for childhood mortality in Burkina Faso, sub-Saharan Africa, were computed from Demographic and Health Survey (DHS) and Demographic Surveillance System (DSS), two very different sources. While most analyses so far determined levels of risk factors, this analysis focuses on the effects of those risk factors. METHODS: Mortality levels are often calculated by strata of risk factor levels separately for each factor. In this publication, their effect was modelled jointly by survival analysis, using all-cause childhood mortality as outcome variable. All live births in rural Burkina Faso in the period of 1994-1998 recorded by the DHS survey Burkina Faso 1998-1999 (5018 births) and the Demographic Surveillance System based in Nouna, western Burkina Faso (6196 births) were included. RESULTS: A simultaneous estimation of hazard rate ratios by a Cox regression model yielded similar estimates for the DHS and DSS data, in line with previous findings based on the Nouna DSS alone. Although DHS surveys do not include children whose mothers had died, ignoring the strong risk factor "death of the mother" does not strongly affect DHS-based results. CONCLUSIONS: Despite the different nature of survey (DHS) and longitudinal (DSS) data, these findings demonstrate that, despite some limitations, results derived from DHS surveys are broadly comparable to DSS data. Both are valuable tools for assessing the importance of risk factors for childhood mortality in sub-Saharan Africa, and they could be combined for better predictions.
Authors: Laina D Mercer; Jon Wakefield; Athena Pantazis; Angelina M Lutambi; Honorati Masanja; Samuel Clark Journal: Ann Appl Stat Date: 2015-12 Impact factor: 2.083
Authors: Abdoulaye Hama Diallo; Nicolas Meda; Halvor Sommerfelt; Germain S Traore; Simon Cousens; Thorkild Tylleskar Journal: BMC Public Health Date: 2012-09-05 Impact factor: 3.295
Authors: Daniel Kadobera; Benn Sartorius; Honorati Masanja; Alexander Mathew; Peter Waiswa Journal: Glob Health Action Date: 2012-11-09 Impact factor: 2.640