Copy the page URI to the clipboard
Macharia, Peter Mwangi
(2020).
DOI: https://doi.org/10.21954/ou.ro.00011fc2
Abstract
All-cause under-five mortality (U5M) forms a key parameter for monitoring and planning of health strategies. Decision-makers need to understand subnational variations in U5M and its determinants to plan equitable and efficient service delivery. To improve our understanding of the variation in U5M and its determinants, the thesis estimates trends of
U5M, its determinants and their association between 1965 and 2015 across 47 counties used for decision making in Kenya.
Twenty household surveys and three censuses (1989-2015) were assembled and aligned to county boundaries. U5M was generated through cohort and period demographic techniques and smoothed via a Bayesian ecological spatio-temporal Gaussian process regression accounting for variation in sample size, demographic methods and spatio-temporal relatedness. The coverage/prevalence of 43 determinants was estimated using small-area estimation and model-based geostatistics approaches. Inequities, summary indices and progress in meeting health development goals were evaluated. Under-five lives saved or lost given the changes in determinants relative to 1993 were computed in a counter-factual framework based on a Bayesian ecological space-time mixed-effects regression model.
Nationally, U5M reduced by 62% characterized by declines and reversal. At the county level, reductions were heterogeneous and inequitable ranging between 20% to 80%. Likewise, there was an uneven improvement in the coverage of interventions and reduction in infection prevalence with disadvantaged counties coinciding with high U5M areas. Development goals were sub-optimally achieved for both U5M and determinants. Nationally, U5M reversal in 1990s was associated with rising HIV infection and maternal
autonomy decline. Post-2006, U5M decline was associated with a decline in HIV, malaria, stunting and improvement in sanitation, early breastfeeding, institutional delivery,
treatment-seeking and maternal autonomy. Declining high parity and fully immunized children were associated with deterioration of U5M. Under-five lives saved or lost were
heterogeneous across counties and over time. The results provide new insights to improve child survival for decentralized health planning in Kenya by prioritizing
marginalized counties.