The Xpert MTB/RIF (Cepheid) test improved tuberculosis (TB) diagnosis and reduced time to treatment, but not long-term TB-related morbidity, according to the results of a multicenter, randomized, controlled trial.

“Despite already being rolled-out in many countries, our study is the first to look at the feasibility of the Xpert test in a real-life clinical setting in southern Africa,” lead author Keertan Dheda, MBBcH, from the Department of Medicine, University of Cape Town, South Africa, said in a news release. The study results were published online October 28 in the Lancet.

The researchers enrolled adult patients with symptoms suggestive of active TB at 5 primary care facilities in South Africa, Zimbabwe, Zambia, and Tanzania. They randomly assigned patients to either Xpert MTB/RIF testing, performed at the clinic by a nurse who received 1 day of training, or to sputum smear microscopy. On the basis of local World Health Organization–compliant guidelines, participants with a negative test result were managed empirically.

The main study endpoint, analyzed by intention to treat, was TB-related morbidity, measured with the TB score and Karnofsky performance score at 2 months and 6 months after randomization in culture-positive patients who had started anti-TB treatment.

Of 758 assigned patients to smear microscopy between April 12, 2011, and March 30, 2012, 182 were culture positive, as were 185 of 744 patients assigned to Xpert MTB/RIF. Among culture-positive patients, median TB score and median Karnofsky performance score in culture-positive patients did not differ between groups at 2 or at 6 months.

Diagnostic Performance of Point-of-Care MTB/RIF

Compared with microscopy, point-of-care MTB/RIF had higher sensitivity (83% vs 50%; P = .0001), but similar specificity (95% vs 96%; P = .25). Compared with laboratory-based MTB/RIF, point-of-care MTB/RIF had similar sensitivity (83%; P = .99), but higher specificity (92%; P = .0173).

Of 744 tests with point-of-care MTB/RIF, 34 (5%) failed, as did 82 (6%) of 1411 with laboratory-based MTB/RIF (P = .22). More patients in the MTB/RIF group than in the microscopy group had a same-day diagnosis (24% vs 13%; P < .0001) and same-day treatment initiation (23% vs 15%; P = .0002).

Because of the lower dropout rate, more culture-positive patients in the MTB/RIF group were receiving treatment by study end (8% untreated in the MTB/RIF group vs 15% in the microscopy group; P = .0302). By day 56, however, the proportions of all patients receiving treatment were similar (43% vs 42%, respectively; P = .6408).

“Although earlier diagnosis by the Xpert test did not reduce overall severity of TB-related illness, and moreover did not reduce the overall number TB cases treated over the course of the study, it has substantial benefits over smear microscopy including improved rates of same-day diagnosis and reducing treatment drop-out,” Dr. Dheda said in the release.

Cost-Effectiveness May Be a Concern

Limitations of this study include about 20% loss to follow-up of patients with culture-confirmed TB, mostly resulting from staffing problems at 1 site, and possible lack of generalizability to seriously ill patients or those with extrapulmonary TB.

“Whilst Xpert may not be the ideal point of care TB test in particularly poorly resourced settings, in countries like South Africa where the clinic infrastructure is relatively good, rates of drug-resistant TB are high, and patient drop-out are significant problems, within clinic placement of Xpert in TB hotspots might be appropriate and enable earlier diagnosis of drug-resistant TB thus likely reducing community-based transmission,” Dr. Dheda noted. “Nevertheless, prevention of TB and adherence to TB treatment is critical and remains a major priority.”

In an accompanying comment, Christian Wejse, MD, PhD, associate professor, GloHAU, Center for Global Health, Department of Public Health, Aarhus University, Denmark, wonders about the cost-effectiveness of Xpert testing.

“At a cassette cost of US$10 (reduced price for low-resource settings), testing large numbers of people with suspected tuberculosis will put substantial pressure on already resource-limited tuberculosis programmes in which the drugs for treatment might not always be available,” Dr. Wejse writes. “Hence, the provocative question raised by this study is whether tuberculosis elimination is most likely to be advanced by distributing GeneXpert machines to all peripheral health facilities in the world, or by investing the same amount in ensuring that health facilities have the set-up available in this study—ie, well trained and paid staff, electricity, and reagents.”