Can tuberculosis relapse be predicted?

Abstract

Objectives: The main objectives were to determine the characteristics of TB patients with relapse, identify the risk factors for TB relapse so that we can prevent future episodes.Design/setting: This was a case control study conducted at five urban clinics in Lusaka district. We recruited 184 patients, 92 were patients who had had TB, successfully treated but relapsed (the cases), and these were compared with those who had previously been treated for TB in full but had not relapsed for at least six months (controls). Results: The age distribution between the cases and controls was significantly different (p = 0.048). Older patients were less likely to relapse than younger patients. No sex difference was observed between the cases and the controls. A positive sputum smear had a high sensitivity in predicting relapse but a low positive predictive value. The sensitivity of DOTS in predicting relapse was relatively high at 70.7 percent though it also had poor positive predictive value. Having an HIV infection had a sensitivity of 64.1 with lower positive predictive value. Conclusions: HIV infection, sex, area of residence and severity of initial illness were all poor predictors of relapse. Living in overcrowded communities and having an HIV infection are well recognised risk factors for tuberculosis aetiology but they may not be important in predicting which patients are more likely to relapse. Recommendations: We need to do more research into more reliable and cost effective ways of identifying which patients are more likely to relapse and find better ways of preventing relapse. We also need to work towards ensuring that all centres providing tuberculosis services in this country also provide DOTS to sputum positive patients.