Assessing the effects of interventions on child and maternal health-related outcomes in Uganda

In this PhD, health-related outcomes studied include the under-five mortality rate (U5MR), the prevalence of fever, diarrhoea, symptoms of acute respiratory infections (ARI) as well as maternal mortality ratio (MMR.
Every year in the world, millions of children die before their fifth birthday. In 2016, an estimated 5.6 million under-five deaths occurred with half of the burden concentrated in the sub-Saharan Africa (SSA) region. In these countries, the U5MR is unacceptably high yet progress is slowed down by the uneven distribution of key determinants of child mortality, for example, child interventions, childhood diseases and the socio-economic factors. Such imbalances lead to substantial variations in the U5MR within countries which may hinder the achievement of Sustainable Development Goal (SDG) target 3.2. In Uganda, the U5MR is much higher than the SDG target of 25 or less deaths per 1 000 live births. In addition, significant differences in the U5MR as well as determinants of U5MR are huge and disproportionately distributed within the country. A better understanding of the determinants of the existing inequalities in the under-five mortality would guide in the prioritization of effective and equitable strategies to realise mortality targets
Another fundamental mortality indicator is the maternal mortality ratio (MMR). MMR measures the quality of the health system and also reflects inequality between sub-groups and, between and within countries. The indicator is also essential for tracking progress in development and for spurring action to improve maternal health.
According to the World Health Organisation, the MMR is highest in SSA and accounts for approximately 66% of the global maternal deaths. In SSA, direct and indirect causes of maternal deaths are the most prevalent conditions yet prevention and treatment measures are hindered by dysfunctional national health systems and a low socio-economic status. This leads to poor maternal health outcomes in SSA, resulting into vulnerable families and increased chances of infant mortality before reaching their second birthday. Furthermore, maternal mortality deteriorates economic development since more women survive with chronic and incapacitating ill health for each maternal death.
Uganda ranks number nine among the top ten high-burdened countries and experiences a MMR far higher than the SDG target 3.1. At the same time, large regional disparities in MMR and its determinants (e.g. maternal interventions) prevail within the country. Therefore, strategies to end maternal mortality need to be implemented, in particular, approaches to address the sources of inequities. This may reduce variations in MMR within Uganda, and thus, quicken the achievement of SDG target 3.1 in the country.
The adoption of the United Nations (UN) Millennium Declaration in the late 2000, established a global partnership of countries and development partners committed to eight voluntary development goals, to be achieved by 2015. Two of the eight Millennium Development Goals (MDGs) focused on U5MR reduction and maternal health improvement. U5MR has fallen by 53% and maternal mortality by 43% since 1990 to 2015. Even though this is a cause for celebration, both declines fell short of the MDG targets of two thirds and three quarters reductions from the 1990 levels. With the end of the era of the MDGs in 2015, the international community agreed on a new framework – the SDGs. The SDG targets for under-five and maternal mortality represent a renewed commitment to the world’s children and mothers. By 2030, end preventable deaths of children under five years of age, with all countries aiming to reduce U5MR to at least 25 deaths per 1 000 live births while maternal mortality should not exceed 140 deaths per 100 000 live births.
Tracking progress towards child and maternal mortality SDG targets requires significant investment in measuring nationally representative data relevant to the estimation of mortality indicators. The implementation of the National Population and Housing Census (NPHC), nationally representative household surveys, that is, Demographic and Health Surveys (DHS), Malaria Indicator Surveys (MIS) and the Uganda Service Delivery Indicator (SDI) Survey has resulted in rich sources of data in Uganda which has made it practical to monitor progress in mortality indicators and their determinants. Censuses collect data for each individual in the country and are therefore an important source of microdata, which enables the study of sub-national differences. The SDI survey data facilitates the assessment of health facility readiness in the country while DHS and MIS data are spatially structured and can be used to identify high risky areas as well as track progress in the distribution of the determinants of mortality such as health interventions and diseases.
Despite the rich data sources, data utilisation remains poor and information extracted by researchers is restricted to national estimates that neither take into account sub-national discrepancies nor assess the effects of interventions and childhood diseases on mortality or morbidity differentials in space. National estimates mask geographical heterogeneities that may exist at a local scale. Therefore, most important interventions at a local scale, areas affected by the disease burden as well as high mortality clusters cannot be identified. This is because the standard frequentist methods commonly employed in the analysis assume independence of observations yet the DHS and census collect mortality and morbidity data at neighbouring locations, and therefore correlated in space. This is because observations at close geographical proximity are likely to share common exposures and thus affected in a similar way. In case of mortality, spatial correlation arises from its determinants such as infectious diseases. An example is malaria which is transmitted by mosquitoes as they fly long distances in surrounding areas. Ignoring spatial correlation in the data results into imprecise effects of covariates and incorrect estimates of mortality risk which are essential for determining most important interventions, areas affected mostly by diseases and high mortality clusters.
Spatial statistical methods fitted via Markov Chain Monte Carlo simulations, are the novel approach developed to incorporate spatial correlation in space. They can estimate high mortality clusters within the country and evaluate the effects of health interventions and childhood diseases on health-related outcomes at the national and sub-national scale for targeted intervention.
The goal of this PhD thesis is to develop Bayesian spatial models to assess the ffects of interventions on child and maternal health-related outcomes at the national and sub-national scale in Uganda, through the following specific objectives; 1) to quantify the effects of childhood diseases on all-cause under-five mortality over space; 2) to estimate the effects of health interventions on all-cause under-five mortality over space; 3) to assess the contribution of childhood diseases on the geographical distribution of fever risk among children less than five years; 4) to quantify the effect of the presence of soap and water at handwashing places in households on the risk of diarrhoea and respiratory infections among children under-five years and 5) to assess the effects of maternal health interventions on all-cause maternal mortality.
In Chapter 2, Bayesian geostatistical proportional hazards models with spatially varying coefficients were applied on the 2011 DHS and 2009 MIS data to estimate the effects of childhood diseases on all-cause under-five mortality at the national and sub-national levels. The models took into account geographical misalignment in the locations of the surveys. Childhood diseases had significant but varying effects on mortality across regions. At national level, the U5M was associated with prevalence of malaria (hazard ratio (HR) = 1.74; 95% BCI: 1.42, 2.16), severe or moderate anaemia (HR =1.37; 95% BCI: 1.20, 1.75), severe or moderate malnutrition (HR = 1.49; 95% BCI: 1.25, 1.66) and diarrhoea (HR = 1.61; 95% BCI: 1.31, 2.05). The relationship between malaria and U5M was important in the regions of Central 2, East-Central, Mid-North, North-East and West-Nile. Diarrhoea was associated with under-five deaths in Central 2, East-central, Mid-Eastern and Mid-Western. Moderate/severe malnutrition was associated with U5M in East-Central, Mid-Eastern and North-East. Moderate/severe anaemia was associated with deaths in Central 1, Kampala, Mid-North, Mid-Western, North-East, South-West and West-Nile.
In Chapter 3, Bayesian geostatistical proportional hazards models with spatially varying coefficients were developed to determine interventions’ effects on under-five mortality at national and sub-national levels, and to predict mortality risk at unsampled locations. The data used in the analysis were obtained from the 2011 DHS. The most important interventions at the national level were artemisinin-combination therapy (HR = 0.60; 95% BCI: 0.11, 0.79), initiation of breast feeding within one hour of birth (HR = 0.70; 95% BCI: 0.51, 0.86), intermittent preventive treatment (IPT) (HR = 0.74; 95% BCI: 0.67, 0.97) and insecticide treated nets (ITN) access (HR = 0.75; 95% BCI: 0.63 0.84). Other important health interventions had more or less comparable effects on mortality. The effects of health interventions on under-five mortality varied by region. In Central 2, Mid-Western and South-West regions, the largest reduction in the under-five mortality burden was associated with ITN access. Improved source of drinking water explains most under-five mortality reduction in Mid-North and West-Nile. Improved sanitation facilities account for the highest decline in under-five mortality in the North-East. In Kampala and Mid-Eastern, IPT had the largest impact on mortality. In Central 1 and East-Central, ORS or RHF and postnatal care were respectively associated with the highest decreases in under-five mortality.
High mortality clusters were found in the North-East, West-Nile, southern of Mid-North, East-Central along the Victoria Nile River, southern of Central 1 stretching to the South-West region and along the country border in Mid-Western between Lakes Albert and Edward. Lowest mortality hazard rates were predicted in Kampala, centre of Mid-North extending to West-Nile, North-East, Mid-Eastern and East-Central regions. Also, areas around Lake George in Mid-Western and a few spots in Central 2 were predicted with low mortality hazard rates.
In Chapter 4, we applied Bayesian geostatistical logistic models on the 2016 DHS data and quantified the contribution of childhood diseases to the geographical distribution of fever risk among children less than five years. At the national level, the population attribution fraction of diarrhoea, ARI and malaria to the prevalence of fever in the under-five was 38.12 (95% BCI: 25.15, 41.59), 30.99 (95% BCI: 9.82, 34.26) and 9.50 (95% BCI: 2.34, 25.15), respectively. The attribution of diarrhoea was common in all regions except Bunyoro, while ARI was more common in Bugisu, Karamoja and West Nile, and malaria was commonest in Bunyoro. In Lango, the attribution of diarrhoea and ARI was similar
In Chapter 5, we analysed the 2016 DHS data and quantified the effect of the presence of soap and water at handwashing places in households on the risk of diarrhoea and ARI among the under-five using Bayesian geostatistical logistic models. The odds of diarrhoea and ARI in children who lived in households having soap and water at handwashing places were 14% and 24% less than those living in households without the intervention (adjusted odds ratio, aOR = 0.86; 95% BCI: 0.77 – 0.96) and (aOR = 0.76; 95% BCI: 0.65 – 0.88) respectively.
In Chapter 6, Bayesian negative binomial CAR models were employed to evaluate the effects of maternal health interventions on all-cause maternal mortality. Data were extracted from the 2016 DHS and 2014 NPHC. The risk of maternal mortality declined with increasing coverage of intermittent preventive treatment (Mortality rate ratio (MRR) = 88%; 95% BCI: 86%, 91%), iron supplements (MRR = 95%; 95% BCI: 93%, 98%), skilled birth attendance (MRR = 96%; 95% BCI: 94%, 98%) and family planning (MRR = 95%; 95% BCI: 92%, 98%).
The results of this thesis will guide prioritization and targeted allocation of high impact and evidence-based interventions to maximize benefits of resources. This will alleviate within country morbidity and mortality discrepancies and consequently accelerate progress towards achieving SDG targets 3.1 and 3.2 in Uganda by 2030.