time in the near future. Thus it will be important to consider what Alan Lopez and others referred to as the “best-buy” strategies for health data in sub-Saharan Africa.
Discussing the choice of data sources, the conferees concluded that it would be useful to combine a variety of sources with different strengths. Such an approach will require research on the quality of various data sources in order to understand which sources are complementary and which are viable in a particular national context and over time. In evaluating the data sources researchers should consider such factors as the size and diversity of a country; the nature and coverage of the country’s health system; the local costs of relevant items such as wages, travel, and communications; and which data strategies have been more or less successful in a particular locality.
One particularly useful resource for informing the design of data-gathering systems for tracking epidemiological transition is AusAID’s Health Information Systems Knowledge Hub (HISHUB; http://www.uq.edu.au/hishub/) at the University of Queensland. Although this system was designed primarily for Asian and Pacific countries, the principles it lays out should translate well to sub-Saharan Africa. According to HISHUB, the minimum data set for understanding epidemiological transitions in order to inform health transitions includes:
1. Reliable unbiased documentation of age-and sex-specific mortality, including the major causes of deaths in the population (civil registration with vital statistics and sentinel or sample mortality surveillance systems with verbal autopsy). Several sources of data on causes of death should be considered. Research should focus on means for obtaining physician-certified coverage of all deaths (at least 90 percent) in a country with cause-of-death coding of reasonable quality; the establishment of systems to collect such data has proved elusive for low-income countries as well as for many middle-income countries and will potentially require decades to achieve. In the meantime, the use of the World Health Organization Health Metrics Network and others interim data sources should be explored. Such interim data sources could include sentinel (urban and rural) demographic surveillance sites and, where possible, statistically representative sample registration sites with verbal autopsy on all deaths. It was suggested that designing, funding, and implementing these interim measures within one to two years would be an attainable goal.
2. Periodic documentation of exposure to the top 10 major risk factors of mortality by age and sex (via periodic population-based surveys). Concerning this issue, it was noted that there are standard adapted survey instruments for each risk factor (smoking, nutrition, high blood pressure, obesity, HIV serostatus, solid fuel smoke exposure, etc.) but that these are rarely assembled into an omnibus national sample survey. Future research strategies should include an extension of the Health Metrics Network and the Household Survey Network in order to promote greater integration and more strategic scheduling of national household surveys.
3. Periodic documentation of the effective coverage of key preventive and curative health interventions aimed at the aboce causes and risk factors