Page 31 - IJPS-2-1

Bruno Yempabou Lankoande

place of residence at the time of the survey or at the time of death is usually not collected.

2.3 Methods
The different techniques used to estimate differences in urban/rural mortality levels in Burkina Faso
are presented in this section. The choice of these methods was mainly guided by the availability of
data. For each method, adult mortality rates, and confidence intervals around estimates (if relevant)
were computed according to urban/rural residence for each sex. The datasets used in each estimation
technique, the reference period for the estimates, the sample description, and the approximations of
the place of residence are reported in Table 1.

Table 1. Datasets used, reference period, sample description and approximation of the place of residence for each estimation method

Estimation method Datasets Sample description Approximation of the Reference period for male Reference period for female
place of residence mortality mortality
Growth balance Census 2006 Household heads reporting on deaths Household’s place of 2006 2006
among adults aged 15–60 years old residence
Direct estimation DHS 1999 Women aged 15–49 years old reporting on Women’s place of 1992–1997 1992–1997
from sibling sur- the survival of their siblings residence
vival data DHS 2003 1998–2003 1998–2003
DHS 2010 2004–2009 2004–2009
Orphanhood DHS 1993 Children aged 5–9 years old and 10–14 years Children’s place of 1988 1987,2; 1989,4
old, reporting on the survival of their parents residence
DHS 2003 1998,5 1997,8;2000,0
DHS 2010 2005,6 2004,8; 2007,0
EMUIB 2000 Young adults aged 15–19 years old; 20–24 Parent’s place of 1989,8; 1991,4 1989,4; 1990,9;1992,6
years old and 25–29 years old, reporting on residence
the survival of their parents

2.3.1 Estimating Adult Mortality from the Growth Balance Method
Data collected on the number of deaths in each household were discarded by the National Institute of
Statistics (INSD) when deriving mortality estimates from the 2006 census. The published estimates
of adult mortality were obtained from child mortality rates combined with model life tables. Even
though the estimation method is not entirely clear in the official report, it seems that child mortality
rates were estimated indirectly from reports on the number of children ever born and still alive
(INSD, 2009). This approach is inadequate because trends in child and adult mortality do not always
evolve in the context of Sub-Saharan Africa. In Burkina Faso, DHS estimates indicate that child
mortality rates have declined substantially in recent decades, while adult mortality rates have mostly
stagnated, although the reason for these divergent trends is undetermined (Masquelier, Reniers, and
Pison, 2014).
To move away from the child-mortality matching approach used by the INSD, I estimated for
each place of residence, adult mortality using the Growth Balance Method (GBM) developed by
Brass (1975). Its principle is to estimate the completeness of the reporting of deaths relative to an
estimate of the population under the assumptions that the population is stable, is closed to migration
and that the completeness of underreporting of deaths is constant above a certain age limit. This es-
timate which is an indicator of data quality, is then used to adjust mortality rates upward to account
for incompleteness of death reporting (Moultrie, Dorrington, Hill et al., 2013). To reduce the sensi-
tivity of the method to internal migration, which can be a great concern in a country such as Burkina
Faso, the completeness estimates were obtained solely from reported deaths among adults aged 35
and above, because they are less likely to migrate (Beauchemin, 2011). With this method, there was
no need to compute confidence intervals because estimates were derived from the census data.
2.3.2 Estimating Adult Mortality from Orphanhood Data

The rationale behind the orphanhood method is to convert proportions of respondents classified by

International Journal of Population Studies | 2016, Volume 2, Issue 1 25