Symptom and digital chest X-ray TB screening in South African prisons: yield and cost-effectiveness.

BACKGROUND: Correctional inmates are at a high risk of tuberculosis (TB). The optimal approach to screening this population is unclear.METHODS: We retrospectively reviewed records from TB screening in 64 correctional facilities in South Africa between January 2015 and July 2016. Inmates received symptom screening (any of cough, fever, weight loss, or night sweats) combined with digital chest X-ray (CXR), when available. CXRs were assessed as 'abnormal' or with no abnormalities. Inmates with either a symptom or an 'abnormal' CXR were asked to provide a single spot sputum for Xpert® MTB/RIF testing. We estimated the incremental cost-effectiveness ratio (ICER) per additional TB case detected using CXR screening among asymptomatic inmates.RESULTS: Of 61 580 inmates, CXR screening was available for 41 852. Of these, 19 711 (47.1%) had TB symptoms. Among 22 141 inmates without symptoms, 1939/19 783 (9.8%) had an abnormal CXR, and 8 (1.2%) were Xpert-positive among those with Xpert tests done. Of 14 942 who received symptom screening only and had symptoms, 84% (12 616) had an Xpert result, and 105 (0.8%) were positive. The ICER for CXR screening was US$22 278.CONCLUSION: Having CXR in addition to symptom screening increased yield but added considerable cost. A major limitation of screening was the low specificity of the symptom screen.

Prevalence and risk factors of tuberculosis disease in South African correctional facilities in 2015.

SETTING: Sixteen South African correctional facilities.OBJECTIVE: To determine the prevalence of and risk factors for tuberculosis (TB) in South African correctional facilities using data collected during a TB screening program in South African correctional facilities in 2015.DESIGN: Inmates in 16 South African correctional facilities were screened for TB from January to December 2015. Inmates reporting ≥1 TB symptom or having an abnormal computer-assisted digital chest X-ray (CXR) provided sputum. Abnormal CXRs were interpreted by a radiologist. Sputum was tested for Mycobacterium tuberculosis using Xpert® MTB/RIF. Data from 16 South African correctional facilities were used in regression analysis, and prevalence estimates calculated for 12 South African correctional facilities with >30% screening coverage.RESULTS: In 12 South African correctional facilities included in the prevalence estimates, 837 inmates had TB disease (2653/100 000) as indicated by current TB treatment or screening-identified TB by radiologist or Xpert. Previous TB was associated with increased odds of screening-identified TB in HIV-positive inmates (OR 4.3, 95%CI 2.5-7.3). For HIV-negative inmates, previous TB (adjusted OR [aOR] 4.9, 95%CI 1.7-14.1) and self-reported symptoms vs. none (1 symptom, aOR 8.8, 95%CI 1.2-67.7; >2 symptoms, aOR 21.7, 95%CI 3.0-158.8) were independently associated with increased odds of screening-identified TB.CONCLUSIONS: Routine TB screening, including CXR, is needed in South African correctional facilities to identify and refer inmates with active TB.

Pooling sputum from multiple individuals for Xpert® MTB/RIF testing: a strategy for screening high-risk populations.

SETTING: Symptom-based screening for tuberculosis (TB) disease is limited by poor performance of symptom screening in several key populations. We tested the hypothesis that pooling sputum from multiple individuals for Xpert(®) MTB/RIF testing would reduce the number of tests required while retaining an acceptable sensitivity, thus allowing the use of Xpert for TB screening. METHODS: We compared pooling ratios that would require the least number of assays using Xpert and determined that for a population with a TB prevalence of approximately 3%, a 1:5 pooling ratio is optimal. To evaluate sensitivity, we generated pools of one specimen with known Mycobacterium tuberculosis culture positivity (smear microscopy-positive or -negative) with four culture-negative specimens. RESULTS: All 20 of the pools generated from a smear- and culture-positive sputum sample were positive using Xpert. Of the 22 pools with a smear-negative, culture-positive sample, we included 17 in the analysis, of which 13 (76%) were Xpert-positive. CONCLUSIONS: Pooling of sputum samples using Xpert achieved reasonable sensitivity and warrants further evaluation of the systematic screening of high TB prevalence populations.