The aim of this study is derive robust and reliable estimates of level and trend in adult mortality in Zambia. To derive the estimates of the level and trend in adult mortality, the study applies the following techniques: the Census Survival method and Preston-Bennett method to Zambian census data for 1980, 1990 and 2000 to estimate life expectancies at age 5 and above as well as probabilities of dying between ages 15 and 60 years; the orphanhood method to 1992, 1996, 2001/2 and 2007 Zambia Demographic and Health Surveys (ZDHS) and 1996, 1998, 2002/3, 2004 and 2006 Living Conditions Monitoring Survey (LCMS) information on survivorship of parents to estimate probabilities of dying between ages 25 and 35 ( 10 25 q ); and 25 and 40 ( 15 25 q ) for females; and 35 and 45 ( 10 35 q ) for males; the siblinghood method using the 1996, 2001/2 and 2007 ZDHS sibling histories data to estimate the probabilities of dying between ages 15 and 50 years ( 35 15 q ) for both males and females; the Generalised Growth Balance and Bennett-Horiuchi âExtended SEGâ methods using the 1996, 2004 and 2006 LCMS household deaths in the last 12 months to estimate completeness of reporting of deaths relative to the coverage of surveys and hence the probability of dying between ages 15 and 60. The Census Survival and Preston-Bennett method do not produce accurate measures of mortality, or trend for females but does for males. The orphanhood method does capture some of the trend but fails to provide definitive estimates of mortality. The siblinghood method produces an inconclusive pattern of adult mortality. The GGB and âExtended SEGâ methods perform well with the 1996- 2004, 1996-2006 inter-survey periods. The methods also perform well with male LCMS data for 2004-2006 inter-survey periods. The GGB and âExtended SEGâ methods produced a
good fit to age ranges 5+ to 60+. The study finds that adult female mortality is higher than male adult mortality, 69 per cent and 64 per cent, respectively. These adult mortality rates are comparable to estimates from other sources. Further research is needed on how to refine the GGB and SEG method to perform better with survey data. Research is needed to understand why the siblinghood method produced inconclusive estimates of the level and trend of adult mortality. The study recommends that the LCMS survey should add month and year at death to questions on household deaths to deal with the problem of time reference. The 2010 Zambian census should add questions on orphanhood and household deaths.