Low sputum smear positive tuberculosis among pulmonary tuberculosis suspects in a tertiary hospital in Mwanza, Tanzania

Abstract

Early diagnosis of tuberculosis (TB) and prompt initiation of treatment are essential for an effective tuberculosis control programme. In many resource limited settings microscopic diagnosis is still the pivotal tool in the diagnosis of pulmonary TB. This study aimed at evaluating laboratory microscopic diagnosis of tuberculosis in a tertiary hospital in Mwanza, Tanzania. This retrospective hospital based study reviewed consecutively from TB registry and patients' files a total of 5,922 TB suspects who submitted their sputum for examination between January 2007 and May 2010 at Bugando Medical Centre (BMC). Among TB suspects (mean age=36.1±13.6 years) female accounted for 54.1% of the patients. The prevalence of HIV among TB patients was 59.4%. The sputum smear positivity rate among the TB suspects was 6.1%; the rate was higher in HIV positive than in HIV negative patients (9.9 % versus 3.2%, P-value < 0.001). The overall positivity rate for the first smear was 94.2% with an incremental percentage yield of 5.2% and 0.6% for the second and third smears, respectively. The study found that 28.6% of patients who were positive in the first smear did not return for the second smear. The risk factors among smear positive TB patients were co-illness (32.5%), previous history of TB (7.5%) and history of positive TB contact (4.7%). These findings also show that as CD4+ T Cells count increases, the quantity AFB in sputum smear also increase although not statistically significant. The sputum smear positivity rate at Bugando Medical Centre is low and more than a quarter of initial TB suspects who were positive in the first smear were lost to follow up posing a threat of continuous transmission of tuberculosis to the community. The finding of more sputum smear positivity rate among HIV positive than HIV negative patients at BMC requires a prospective study to ascertain whether it is a reality or a coincidence.