Validation of tuberculosis notification in RSA : an epidemiological analysis of the reported tuberculosis cases and deaths in the period 1993 to 2003

Abstract

Tuberculosis (TB) remains one of the major public health problems in South Africa. The overall aim of the research project was to evaluate the completeness of TB notification data. A descriptive study design was used. The TB data from the Disease Notification System for the period 1993 to 2003 were analysed to describe 11 year trends by province, sex and population group. The levels of under-reporting of tuberculosis were estimated by comparing the annual numbers and the rates of notified cases and deaths per 100 000 population with the data of registered cases in the electronic TB register and registered TB deaths from the Statistics South Africa’s metadata on causes of deaths in South Africa. A total of 768896 cases and 39052 deaths were recorded in the Disease Notification System for the period 1993 to 2003. The annual case load declined from 42099 cases in 1993 to 36081 in 1996, then peaked to 89111 in 1998. The peak in 1998 resulted mainly from two-fold increases in notified cases in the provinces of Eastern Cape, KwaZulu-Natal and Western Cape. There was also a three-fold increase in Western Cape in 2001 and four-fold increase in Northern Cape in 2002. The lowest numbers of notified cases were for Mpumalanga in the years 1993 to 1996. In Limpopo, a total of 13 cases only were notified between the years 1999 to 2003 inclusively. Nationally and provincially, the annual numbers of notified cases and deaths and rates per 100 000 population were consistently higher among males than females. The Wilcoxon signed rank test comparing the medians between male and female cases showed a p-value of 0.003 indicating that the difference exists between the two medians. Nationally the lowest number of deaths was 1967 notified deaths in 1994 and the highest number of deaths was 6085 notified in 2002. The number of deaths notified varied between the provinces and fluctuated between the years. It was the highest for the years 1993 to 1996 in Western Cape and the highest for the years 1997 to 2003 in Eastern Cape. It was the lowest in Mpumalanga for the years 1993 to 1997 and the lowest in Limpopo for the years 1999 to 2003 and KwaZulu-Natal in the years 2002 to 2003. The Disease Notification System was found to have lesser numbers of notified cases in comparison to registered cases recorded in the TBSYS or electronic TB register. The percent difference between notified and registered cases ranged between 28% in 2001 to 69% in 1996. Comparison of notified and registered TB deaths for the period 1997-2003 showed that the annual numbers and rates of registered deaths in the Statistics South Africa’s metadata were higher for all the years than the notified deaths in the disease surveillance system. It is recommended that the disease surveillance system is evaluated periodically, facility data assessment tools are introduced and capacity for surveillance is strengthened at all levels of the national health systems. Copyright