In the preceding article, Glynn et al. (1) report the outcomes of treatment of tuberculosis (TB) patients in Karonga District, Malawi, between 1986 and 1994 and analyse the determinants of outcome (e.g. patient characteristics and type of TB). The most important determinant of the outcome of treatment of TB patients is access to reliable diagnosis and treatment services organized within a national tuberculosis programme (NTP) as part of DOTS, the name of the WHO-recommended TB control strategy (2).
The performance of the Malawi NTP is one of the best in sub-Saharan Africa (3) and the programme in Karonga District has benefited since the early 1980s from the substantial contribution of the Karonga Prevention Study (KPS) of leprosy and TB. In addition to the standard TB control activities of the NTP in Karonga District, the KPS has contributed enhanced case-finding and case-holding activities. Karonga District therefore represents a highly developed system for TB control activities, with benefits for TB patients in the district, who have access to excellent control services, and for epidemiologists, who have access to data on TB case finding and treatment outcomes for the analysis of the determinants of treatment outcomes.
There are three types of implications arising from this article for TB control, as discussed below.
* First, well-organized TB control services in line with the DOTS strategy are necessary to ensure that TB patients have the best chance of successful disease detection and treatment outcome. Patients not registered by the Ministry of Health NTP had much higher mortality and default rates than did registered patients.
* Second, accurate evaluation of treatment outcomes depends on how the NTP measures these outcomes. Failure to include all patients diagnosed (i.e. ignoring those who are diagnosed but not registered) results in an over-optimistic evaluation of treatment outcomes. In Karonga, the inclusion of patients who were diagnosed but not registered gave a true case fatality rate of 16%, compared with a rate of 13% when only those patients who were registered were considered. The proportion of patients diagnosed but not registered was 8% of the total in a district with excellent TB control services. Since in urban areas of Malawi, 10-25% of sputum-smear-positive patients are diagnosed but not registered (A.D. Harries personal communication, 1997), the bias in underestimating the TB case fatality rate is likely to be much greater than in Karonga. Part of routine NTP management should be regular cross-checks between laboratory diagnostic registers and NTP treatment registers by district tuberculosis officers and during regular supervisory visits by regional tuberculosis officers. NTP staff should use the number of patients diagnosed rather than the number registered for treatment as the denominator in compiling the statistics for treatment outcomes.
* Third, the organization of TB control services has to take into consideration the determinants of outcome relating to the type of TB, the presence of concomitant infection with human immunodeficiency virus (HIV), and patient characteristics. The outcome measure which provides the earliest indicator of NTP performance is the sputum smear conversion rate when a whole cohort of patients has completed the 2 months' initial phase of treatment. …