Introduction: TB is a major health problem in South Africa, with increasing numbers of patients notified, inadequate successful treatment rates and an emerging problem with resistant strains. This study was conducted at a District Hospital in KwaZulu-Natal where the successful treatment rate was as low as 23%. The aim of this study was to identify key factors at the hospital that may affect adherence to TB treatment and to recommend interventions that could improve adherence.

Methods: The study design was a case control study using prospectively collected data. Information was collected over an 8-month period, when the patients started their anti-TB treatment, according to the known factors that influence TB adherence. The patients were then followed up to determine those who did not adhere to their treatment, and those who successfully completed treatment. The two groups involved, therefore, were the controls (those who did adhere) and the cases (those who did not adhere). The data previously collected were then compared for significant associations with the controls and cases.

Results: Data were obtained from 159 TB patients, 105 (66%) were adherent and 54 (34%) non-adherent. The following variables showed a significant association (p < 0.05) with non-adherence: higher level of education; distance from the hospital; time taken to travel; the method of transport; satisfaction with the hospital; food security; income; the smoking of tobacco and/or marijuana; the patients' perspective and beliefs; HIV testing and status; functional status; social support; the relationship with the TB nurse; depression score; and self-rating of confidence. A stepwise logistic regression was performed, and only two variables remained significantly associated: travel time (OR7.9, 95%CI 1.4–44.1) and the relationship with the TB nurse (OR2.6, 95%CI 1.3–5.1).

Conclusions: The most important recommendation is to improve the relationship between patients and TB nurses through training in communication skills. A more holistic assessment of patients would help identify issues such as depression, and a more patient-centred approach would help to understand and address patient's concerns, beliefs and expectations. It may also be important to ensure that management and administrative systems support a more patient-centred approach. It may be important to encourage the recruitment of family physicians who are trained in communication skills and a patient-centred approach, to work in the rural areas, where they can mentor and teach other colleagues and staff. More needs to be done in terms of helping patients to access facilities through transport, or in making services more available at the community level through DOTS (Directly Observed Treatment) supporters and adequate home-based carer support.