Diagnosis of tuberculosis in patients with pleural effusion in an area of HIV infection and limited diagnostic facilities.

In a prospective study of 118 patients with pleural effusion admitted to four medical wards in Muhimbili Medical Center between January and August 1991 Dar es Salaam Tanzania tuberculosis (TB) was diagnosed in 112. In 84 patients the diagnosis of TB was made by detection of acid-fast bacilli by stain (auramine Ziehl-Neelsen) or by culture of mycobacteria (Lowenstein-Jensen medium) in pleural fluid or pleural tissue obtained by closed biopsy or by the presence of caseating granulomas in histological sections. In 28 patients the diagnosis of TB was considered probable based on good response to anti-tuberculous therapy. In the remaining 6 non-TB patients adenocarcinoma (1) bacterial infection (2) and aspecific inflammation (3) were diagnosed. 58% of the TB and 3 of the non-TB patients were infected with HIV. The diagnostic procedures were evaluated in 75 patients. The highest diagnostic yield was obtained by histology (85%) followed by culture of pleural biopsy (37%) and pleural fluid culture (36%). Pulmonary tuberculosis was found in 8 (4 HIV-positive) patients and dissemination of TB to other sites in 25 patients of whom 20 were HIV-positive. By logistic regression analysis two independent diagnostic markers for TB pleuritis were identified: pleural fluid protein >50 g/l (odds ratio [OR] 12.1) and pleural fluid adenosine deaminase level of >10 U/l (OR 11.08). The sensitivity of these two diagnostic tests was 82% and 97.3% and the specificity was 83.6% and 50% respectively. TB was the underlying cause in nearly all patients who presented with pleural effusion (94.9%). TB was confirmed in 75% of these using the referral hospital. Conventional facilities of a referral hospital are sufficient to diagnose tuberculous pleuritis as well as disseminated tuberculosis irrespective of HIV infection. However in regions with overburdened health facilities and high prevalence of tuberculous pleurisy in patients with pleural effusion a simplified diagnostic approach is suggested based on exclusion of other causes of pleural effusion by simple use of these diagnostic markers.