The meta-analysis included 6 studies, of which 5 were conducted in sub-Saharan Africa, showed that the diagnostic accuracy for Xpert testing was similar to the former symptom screening and rapid test format.
The WHO recommends conducting molecular diagnostic testing for tuberculosis in all HIV-positive patients in high-burden settings; these guidelines were informed by a paper published in The Lancet HIV.
A team of worldwide investigators conducted a systematic review and meta-analysis in order to inform updated WHO tuberculosis screening guidelines, which since 2011 have included a two-step process of a symptom screening and if positive, a subsequent rapid test. The WHO recommends that HIV-positive patients be routinely screened for tuberculosis, the study authors noted. The study authors compared the existing process to an alternative screening test, which included symptom screening and Xpert testing.
The study authors searched medical databases for studies of adult and adolescent HIV-positive inpatients enrolled regardless of signs and symptoms of tuberculosis. They found 6 eligible studies which encompassed 3660 participants. All of the studies recruited patients from medical wards, 5 were conducted in sub-Saharan Africa and 1 was conducted in Asia. The study authors observed that the median age of participants was 37 years, 58 percent of the participants were women, and two-thirds of the participants were receiving antiretroviral therapy (ART). The study authors noted they did not collect data on ethnicity.
The study authors also explained that they assessed the proportion of inpatients eligible for Xpert testing, assessed the accuracy of the symptom screening called W4SS and alternative screening tests or strategies to guide diagnostic testing, and compared the accuracy of the WHO Xpert algorithm (defined as W4SS followed by Xpert) to Xpert for all.
For the 6 studies, the study authors learned the pooled proportion of inpatients with a positive W4SS was 90 percent with an estimate ranging from 85 to 100 percent. The pooled tuberculosis prevalence was 20 percent in 4 studies that collected sputum for culture as a reference standard, but jumped to 25 percent with culture or Xpert as the reference standard, the study authors wrote.
When the WHO Xpert algorithm of W4SS plus Xpert was used, sensitivity was 76 percent and specificity was 93 percent, the study authors said. Diagnostic accuracy for Xpert for all was similar to the algorithm at 78 percent sensitivity and 93 percent specificity.
“In a hypothetical cohort of 1000 HIV-positive inpatients at 20 percent tuberculosis prevalence, the WHO Xpert algorithm would result in 940 Xpert tests, but miss 48 tuberculosis cases; Xpert for all 1000 HIV-positive inpatients would miss 44 tuberculosis cases,” the investigators wrote.
Across all 6 studies, the study authors said, Xpert was positive in 6 of 251 patients who had available Xpert results but were ineligible for Xpert testing according to the WHO Xpert algorithm.
“Our findings have informed the 2021 WHO recommendation to do molecular rapid diagnostic testing (eg, with Xpert) in all HIV-positive inpatients in high-burden settings (>10 percent tuberculosis prevalence),” the study authors wrote, but said that more accurate initial screening tests to guide further diagnostic testing need to be developed. The current screening tests are not accurate enough, they wrote, and hospitals with limited resources may not be able to do systematic diagnostic testing for all inpatients that are HIV-positive.
“Although routine molecular rapid diagnostic testing might reduce the current diagnostic gap, a negative result still does not rule out tuberculosis,” they continued. “Xpert Ultra could additionally bridge the diagnostic gap and requires evaluation in unselected HIV-positive inpatients.”