Page 38 - IJPS-2-1

Monitoring adult mortality by type of residence in the absence of death registration: a perspective from Burkina Faso

not significant. In addition, these estimates are not impacted by variation in the quality of data ac-
cording to urban/rural location (Appendix, Table A2). Except, for the 20–24 year-old age groups in
rural areas, proportions of surviving parents reported by men and women are not statistically differ-
ent at the threshold of 5%. The urban/rural mortality differentials derived from data on young
adults collected during the EMUIB survey are probably the most reliable, but they are only available
for the late 80s and early 90s.
Estimates based on sibling histories present the advantage of depicting the trend in adult mortali-
ty by place of residence in a relatively long period. However, the major issue related to these esti-
mates is the approximation of the place of residence of the siblings of interviewed women. In the
absence of migration data, I made the assumption that siblings and interviewed women share the
same place of residence. This approach has been adopted in previous studies (Bicego, 1997; De
Walque and Filmer, 2013); however, it is problematic when migration flows are important. Another
possibility was to limit the analysis to siblings of interviewed women who had never migrated before
the survey as experimented by Günther and Harttgen (2012). This may lead to a selection bias if
siblings of migrant women are affected by lower or higher mortality compared with siblings of
non-migrant women. In contrast with previous research which ignored or did not quantify the impact
of these approximations of sibling’s place of residence on mortality differentials, further analysis
shows that adult mortality is likely overestimated in urban areas and underestimated in rural areas.
First, from the experience I have drawn from the orphanhood method applied on data collected dur-
ing the EMUIB survey, it is clear that a large amount of deaths occurring in rural areas are trans-
ferred to urban areas and the opposite is done with deaths occurring in rural areas, but with a lesser
magnitude. This suspicion is strengthened by the higher sibships reported in urban areas compared to
rural areas on average. Second, from the analysis of the mean number of reported siblings, I saw that
there is a more pronounced underreporting of siblings in rural areas and if related to adult deaths, it
is likely that mortality is underestimated in rural areas. In summary, an urban disadvantage is not
likely, what appeared in early periods is probably spurious and generated by the misclassification of
sibling’s place of residence, and by the poor data quality mainly in rural areas.
By carefully analyzing the case of Burkina Faso, the results presented in this paper showed that
the urban disadvantage in adult mortality put forward by Günther and Harttgen (2012) is improbable,
even though the differences in favor of the urban areas found here are not significant. Such a conclu-
sion is reinforced by estimates derived from the different Health Demographic Surveillance Systems
(HDSS) located in Burkina Faso. By taking the HDSS of Ouagadougou as a proxy for the urban
areas and the other HDSS located in the country (Nouna, Nanoro, Kaya) as proxies for rural areas,
one can again observe that adult mortality (45q15) is lower in urban areas (for the period of
2009–2011) (Sié, Soura, Derra et al., 2015). Taken together, these results support the argument that
urban adults in Burkina Faso still benefit from better health conditions. Although the dynamic of
differentials in urban/rural mortality is still somewhat erratic and difficult to depict based on the
available evidence, the urban environment continues to be negatively associated with adult mortality
in Burkina Faso. Despite the global concern about the rising burden of non-communicable diseases
in least developed countries associated with urbanization, urban dwellers still live longer than their
rural counterparts in Burkina Faso. Efforts made in reducing adult mortality in rural settings should
therefore not be abandoned.
Beyond the case of Burkina Faso, this paper highlighted the need for additional information on
the type of residence of close relatives to obtain a better picture of adult mortality differences by ur-
ban/rural location in countries without death registration. Given that expansion of data collection in
large programs such as DHS will not happen in a near future, the comparative approach taken
here could be implemented in other countries when measuring spatial inequalities in health indica-
tors for all ages in the context of the SDGs. The availability of untapped sources of data particularly
on migration (e.g., EMUIB in Burkina Faso, the 2009 migration and remittances household surveys
in Senegal), may help to correct data on the survival of siblings and parents for a better measurement
of the levels and trends in spatial inequalities in mortality.

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