Setting: Thirteen primary health care (PHC) facilities in the Stellenbosch District, South Africa.
Objective: To assess the use of a sputum register to evaluate the tuberculosis (TB) diagnostic process and the initiation of TB treatment in selected PHC facilities in a country with a centralised laboratory system.
Design: This prospective study was conducted between April 2004 and March 2005. The names of all individuals submitting sputum samples for TB testing were noted in a newly introduced sputum register. We classified all TB suspects with two positive smears as TB cases and consulted TB treatment registers until 3 months after sputum submission to determine how many had started treatment.
Results: A total of 4062 persons aged > or =15 years submitted sputum samples, of whom 2484 were TB suspects. There were 2037 suspects with at least two results, 367 (18%) had at least two positive smears and 64 (17%) of these did not start treatment (initial defaulters). Over the entire diagnostic process, up to 5% of TB cases were missed, and up to 26% did not start treatment and were not reported.
Conclusion: By correcting diagnostic weaknesses identified in the sputum register, PHC facilities will be able to detect, treat and cure a higher percentage of TB patients.