Monaz: Hello, I'm Monaz Mehta, editor in the Cochrane Editorial and Methods department. Most Cochrane Reviews look at the effects of interventions on health, but a growing number provide evidence on how to diagnose a disease. In June 2019, one of these, on a test used to detect tuberculosis, was updated. The lead author, David Horne from the University of Washington in Seattle in the USA, sets the scene and tells us what the review found.

David: Tuberculosis, or TB, causes more deaths than any other infectious disease: 1.6 million people died from it in 2017. Most of these deaths could have been averted with earlier detection. The early diagnosis of TB is also important to prevent disease progression and long-term consequences, alleviate symptoms, and to protect public health by interrupting its transmission.
An important aspect of the global TB epidemic is the threat from drug-resistant TB. Unlike most other infectious diseases, TB treatment requires multiple drugs given for at least 6 months to prevent the emergence of drug-resistance and cure TB. But, some patients’ TB is resistant to the most potent drugs in a standard treatment regimen, rifampicin and isoniazid. Treatment of this multi-drug resistant or MDR-TB takes longer and requires drugs that are much more expensive and toxic than standard therapy. Every year, over half a million people develop MDR-TB or rifampicin-resistant TB; but, in 2017, only one in four of these started appropriate treatment and just over one-half of those were cured.
Culturing TB in the laboratory is the best test for diagnosing TB and is the first step in detecting drug resistance. Unfortunately, though, this complex and slow procedure is not widely available in many resource-constrained settings. 
As an alternative, in 2010, the World Health Organisation recommended GeneXpert MTB/RIF as a rapid test that can simultaneously detect both TB and rifampicin resistance in sputum and other specimens. It’s designed to be used outside of reference laboratories in settings nearer to patients, such as hospitals and clinics in low-resource settings. The test has been a giant leap forward and called a “game changer” but it struggles if the patient has something called smear-negative pulmonary TB, when their sample contains too few TB bacteria. To try to overcome this limitation, the manufacturer introduced a new version called Xpert MTB/RIF Ultra.
We performed our first systematic review of GeneXpert in 2014 and this update allows us to bring in the more recent studies of both GeneXpert and Xpert Ultra for detecting TB in the lungs and for detecting rifampicin resistance in adults. We now have 95 studies, with 68 that were not previously available for our 2014 review, and data from more than 37,000 people. 
Looking first at the detection of TB, we found that GeneXpert’s sensitivity was 85% and its specificity was 98%. This means that it correctly identifies TB in 85 of every hundred people with it, but suggests that TB is present in two in every hundred people without it. The sensitivity of GeneXpert was significantly lower for people with smear-negative pulmonary TB at 67%, compared to 98% in those with smear-positive disease. Similarly, sensitivity was lower in people living with HIV at 81%, compared to 88% for those who were HIV-negative.
The accuracy of GeneXpert was high for the detection of rifampicin resistance: sensitivity was 96% and specificity was 98%.
There is much less evidence for Xpert Ultra. We could include just one study, which compared Xpert Ultra to GeneXpert in over 1400 participants. It found that Xpert Ultra was more sensitive than Gene Xpert, 88% versus 83%, but its specificity was slightly lower, 96% versus 98%.
In summary, GeneXpert is sensitive and specific for the detection of both TB and rifampicin resistance. For TB detection, it has higher sensitivity in people who are smear-positive or HIV-negative. Based on a single study, Xpert Ultra had higher sensitivity and lower specificity for TB detection, but was similar for detecting rifampicin resistance. Therefore, we can conclude that both GeneXpert and Xpert Ultra provide accurate results and can allow rapid initiation of treatment for MDR-TB. However, there is still a need for additional studies of Xpert Ultra in diverse patient populations, including those with a prior history of TB.

Monaz: If you would like to read more about the use of these tests and to watch for future updates of the review if those studies are done, you can find the full review at Cochrane Library dot com. Just go to the website and search 'Xpert for TB in adults'.