Epidemiological Estimates


Although the measurement of the burden of disease is essential to set priorities, most countries struggle to come out with reliable estimates of incidence, prevalence and mortality by cause.

I can provide technical assistance on how to identify data sources and correctly interpret them to estimate epidemiological indicators.

Examples include:

A hospital discharge survey in all the hospitals of KwaZulu-Natal.   In the early 2000s, no information was available on the causes of hospital discharge because the medical records were not electronically processed. In 2003, I managed a discharge survey team to extract the medical history from a representative sample of medical records pertaining the period 1998-2001. For each medical history, the underlying discharge diagnosis was assigned in double-blind fashion by two physicians. In the case of disagreement on the diagnosis, the two physicians discussed the case until they agreed, otherwise the diagnosis was left as undefined. The method of assigning the underlying cause of death allowed to quantify for the first time the burden of HIV/AIDS on the hospital system. Although hospitalizations do not reflect population morbidity and mortality, the estimates of the causes of hospitalization provided a strong proxy for the burden of disease.

I applied the burden of disease methodology to estimate the causes of mortality in KwaZulu-Natal. The datasets of the recorded death certificates for 2001 was obtained from the South African National Statistics Office. The causes of death were reassigned according to the Burden of Disease methodology to identify the underlying cause of death. This allowed for the first time to assess the impact of HIV/AIDS on mortality in the population. Triangulation was also carried out with the data originating from a demographic surveillance site and the hospital discharge survey.