What causes symptoms suggestive of tuberculosis in HIV-positive people with negative initial investigations?

OBJECTIVE: To identify the causes of symptoms suggestive of tuberculosis (TB) among people living with the human immunodeficiency virus (PLHIV) in South Africa. METHODS: A consecutive sample of HIV clinic attendees with symptoms suggestive of TB (1 of cough, weight loss, fever or night sweats) at enrolment and at 3 months, and negative initial TB investigations, were systematically evaluated with standard protocols and diagnoses assigned using standard criteria. TB was 'confirmed' if Mycobacterium tuberculosis was identified within 6 months of enrolment, and 'clinical' if treatment started without microbiological confirmation. RESULTS: Among 103 participants, 50/103 were pre-antiretroviral therapy (ART) and 53/103 were on ART; respectively 68% vs. 79% were female; the median age was 35 vs. 45 years; the median CD4 count was 311 vs. 508 cells/mm³. Seventy-two (70%) had 5% measured weight loss and 50 (49%) had cough. The most common final diagnoses were weight loss due to severe food insecurity (n = 20, 19%), TB (n = 14, 14%: confirmed n = 7; clinical n = 7), other respiratory tract infection (n = 14, 14%) and post-TB lung disease (n = 9, 9%). The basis for TB diagnosis was imaging (n = 7), bacteriological confirmation from sputum (n = 4), histology, lumbar puncture and other (n = 1 each). CONCLUSION: PLHIV with persistent TB symptoms require further evaluation for TB using all available modalities, and for food insecurity in those with weight loss.

Tuberculosis among adults starting antiretroviral therapy in South Africa: the need for routine case finding.

OBJECTIVE: To investigate the prevalence of and evaluate screening modalities for undiagnosed tuberculosis (TB) in antiretroviral therapy (ART) eligible adults in South Africa. METHODS: Individuals were screened for TB using symptoms, chest radiograph (CXR) and two sputum specimens for microscopy and culture, and were then followed for <6 months to determine TB diagnoses. RESULTS: Among 361 participants (67% female, median age 38 years, median CD4 count 120 cells/mm(3)), 64 (18%) were sputum culture-positive; 114 (32%) fulfilled any TB case definition (culture- and/or smear-positive, or improvement on specific treatment). Symptom screening comprising any of cough, appetite loss or night sweats > 2 weeks had a sensitivity and specificity of respectively 74.5% and 50.8%. Sensitivity was increased by CXR (to 96.1%), but not by smear microscopy. The World Health Organization symptom screen had a sensitivity and specificity of respectively 96.1% and 5.2% in our study population; the addition of CXR increased sensitivity to 100%. Median time to TB treatment was 8 days for diagnoses based on CXR (n = 72) vs. 37 days for diagnoses based only on sputum culture (n = 14). CONCLUSIONS: The very high prevalence of undiagnosed TB among patients presenting for ART mandates their routine investigation. CXR improved sensitivity substantially, allowed rapid treatment initiation and should be routine, where available, pending better point-of-care diagnostics.

Prevalence of latent tuberculosis infection among gold miners in South Africa.

SETTING: South African gold mines. OBJECTIVE: To determine the prevalence of latent tuberculosis infection (LTBI) and risk factors for a positive tuberculin skin test (TST) among gold miners. DESIGN: Cross-sectional survey. Human immunodeficiency virus (HIV) status was determined by self-report and medical records. TST positivity was defined by the mirror method to estimate the prevalence of LTBI, and by the US Centers for Disease Control and Prevention definitions to explore risk factors at the individual level. RESULTS: Among 429 participants (105/130 subjects aged <30 years, 324/390 > or = 30 years), the estimated prevalence of LTBI was 89%; 45.5% of HIV-positive participants had a zero TST response compared to respectively 13% and 13.5% in the HIV-negative and status unknown participants. In participants with TST > 0, there was no significant difference between size of response by HIV status: the mean (standard deviation) widths for HIV-positive, HIV-negative and HIV status unknown were respectively 11.84 (2.75), 12.03 (2.75) and 12.52 mm (3.04) (analysis of variance P = 0.28). Factors independently associated with a TST < 10 mm were positive HIV status (aOR 0.41, 95%CI 0.17-0.96) and not working underground (aOR 0.25, 95%CI 0.09-0.71). CONCLUSIONS: The prevalence of LTBI is very high in gold miners in South Africa. HIV-infected individuals are more likely to have a negative TST, but HIV infection does not affect the size of TST response.

Evaluation of the Arkansas method of urine testing for isoniazid in South Africa.

SETTING: A South African hospital serving gold mine employees. OBJECTIVE: To determine the sensitivity and specificity of the Arkansas method for detecting isoniazid (INH) metabolites among South African adults and to examine the effect of smoking status on positive results. DESIGN: Urine specimens were collected from in-patients taking INH as part of tuberculosis treatment at 6, 12 and 24 h after a directly observed 300 mg oral dose. As a control group, a single urine specimen was collected from surgical in-patients not taking INH. Specimens were tested for INH using a commercially available dipstick. RESULTS: A total of 153 patients on INH and 60 controls were recruited. The sensitivity of the test was 93.3% (95%CI 88.1-96.8) at 6 h post INH, 93.4% (95%CI 88.2-96.8) at 12 h and 77% (95%CI 69.1-83.7) at 24 h. The specificity of the test was 98.3% (95%CI 91.1->99.9). There was no association between smoking status and colour change of positive results. CONCLUSIONS: This test is a useful method of monitoring adherence to TB treatment or preventive therapy among South Africans. However, it is less than 100% sensitive, especially with increasing time post dose, which should be taken into consideration when interpreting results for individual patients.