Evaluation of the implementation of health interventions and their impact on child survival in Tanzania

It is widely accepted that achieving the highest and most equitable levels of health of populations through the most rational use of resources is the ultimate goal of national and international policymakers, public health officials and health professionals at large. However, doing this depends upon understanding the burden of disease, its distribution and causes in a given population and the effectiveness of different preventive, curative and palliative interventions that can reduce these burdens.
Demand for comparable cause-specific mortality data of high quality has grown due to increased pressure to meet ambitious short-term goals and targets set by the international donor community. Robust data are urgently needed to assist policy makers and health planners in setting intervention priorities, the allocation of resources, and the analysis of the equity and effectiveness of health interventions and systems.
The counting of births, deaths by age and sex, and documentation of causes of death is the norm for all routine vital registration systems implemented throughout the developed world. But in most developing countries, routine empirical data on population burden of disease are usually missing, or at best, grossly incomplete due to the lack of systems and resources to support their collection and documentation.
Mortality surveillance systems or surveys using verbal autopsy have the potential to provide invaluable data for informing the health system on the burden of disease, and for monitoring and evaluating of the impact of different health and health system interventions as they are being implemented. The sentinel surveillance platform that includes the Ifakara, Rufiji and AMMP Demographic Surveillance System sites in Tanzania offers a great opportunity to examine this potential.
The goal of this thesis was to explore a variety of innovative approaches to evaluating the implementation of health interventions and their impact on child survival in Tanzania.
Ths was pursued by analyzing the burden of disease for the period from 2000-2002 in the rural areas of Kilombero and Ulanga district in which a population of approximately 65,000 people is under continuous surveillance. I also examined health systems access for pregnant women and children younger than five in a rural area in Rufiji district by combining demographic surveillance systems with geographic information systems in a population of approximately 70,000 people in 12,000 households. Using a sentinel surveillance platform in a non-randomised "plausibility" design across the four districts of Kilombero, Ulanga, Morogoro Rural and Rufiji, the thesis also examines the child survival effectiveness, cost and impact of the integrated management of childhood illness (IMCI).
The main findings were:
- 42% of mortality in children younger than five years of age occurred due to conditions that are well known and for which Districts have the technology to prevent or treat.
- Spatial access to health care by children and pregnant women in Rufiji District was similar with an average travel time to a health facility of less than 1 hour.
- Facility based IMCI improved quality of care and was associated with a 13% reduction in mortality in children younger than five in intervention districts.
- The costs of child health care in districts implementing IMCI was similar to or lower than those in comparison districts.
- Introduction of IMCI led to improvements in child health that did not occur at the expense of equity.
- Changes in the programmatic delivery strategy of vitamin A supplementation improved coverage in Tanzania and has been sustained for more than three years.
- Delivery of high dose of vitamin A supplementation in mothers and children less than six months of age was well tolerated, but did not confer any important absolute effect on morbidity.
Experience gained from the studies documented in this work can contribute to the body of knowledge on the estimation of causes of death, inform future evaluations, and help to shape child health policy in Tanzania and other similar settings. The need for robust, representative routine demographic and health statistics is critical for the monitoring and evaluation of health interventions and systems. The model recently proposed by the Health Metrics Network provides this opportunity for more countries. Investing in the strengthening of health systems, including health information sub-systems such as sentinel surveillance, is necessary if strategies like IMCI are to be prioritized and implemented effectively. IMCI implementation was successful in Tanzania because of the strong health system support that existed. Although results from the DHS 2004 and from sentinel surveillance indicate dramatic improvements, overall, infant and under five mortality rates are still unacceptably high. Achieving the Millennium Development Goal of reducing the 1990 level of child mortality by two-thirds by 2015 will require infensified efforts and new interventions to prevent deaths from major killers of children in Tanzania which include malaria, pneumonia, diarrhoea, under nutrition and perinatal causes. Wider and more equitable coverage is required, especially for the districts that are still lagging behind in the implementation.
This thesis shows that important progress can be made with a practical mix of population based evidence used in a health systems approach.