John: Hello, I'm John Hilton, editor in the Cochrane Editorial and Methods department. Most Cochrane Reviews look at the effects of interventions on health, but a growing number are providing g evidence on the accuracy of different ways of diagnosing a disease. These were added to in August 2018 by a new review on a test used in the management of tuberculosis. The review’s first author, Mikashmi Kohli from McGill University in Montreal Canada, sets the scene and tells us what they found in this podcast.

Mikashmi: 2018 is a landmark year for the global fight against tuberculosis, or TB. In May, the World Health Organization published the first Essential Diagnostics List. Then, in September, the United Nations held its first high-level meeting on TB, galvanizing a global commitment to end the disease by 2030. 
TB is the leading cause of infectious disease-related death and a top ten cause of death worldwide. When it’s detected early and effectively treated, TB is largely curable. But about 1.6 million people still die of TB each year, including 300,000 who are living with HIV. Also, in 2017, around half a million people developed MDR-TB/rifampicin-resistant TB, which is important because rifampicin is one of the most important TB drugs. 
For one-fifth of people with active TB, the site of disease is outside the lungs. This is known as extrapulmonary TB and some forms, such as TB meningitis, are extremely dangerous, meaning that a rapid diagnosis can make all the difference to a patient. We assessed the diagnostic accuracy of a test called Xpert MTB/RIF for detecting extrapulmonary TB and rifampicin resistance and, for ease, I’ll call it simply Xpert. 
Xpert is a World Health Organization-recommended test that simultaneously detects Mycobacterium tuberculosis complex and rifampicin resistance, a good proxy for MDR-TB. The test is automated and rapid, providing results within two hours. 
In our review, we evaluated Xpert for eight common forms of extrapulmonary TB and measured its results against those obtained from culture, which means taking a sample and letting it grow in the laboratory. 
We included 66 studies that evaluated more than 16,000 specimens but only one study evaluated the newest test version, Xpert Ultra, for TB meningitis. 
In some body tissues and fluids, we found that Xpert was more than 80% sensitive, meaning that it registered positive in 8 out of 10 people who actually had TB, and it was more than 98% specific, meaning that it registered positive in very few people who were actually negative. We also found that Xpert was both very sensitive and very specific for detecting rifampicin resistance.
Focusing on Xpert testing in cerebrospinal fluid, the pooled sensitivity and specificity for Xpert against culture were 71% and 98%, based on moderate-certainty evidence from 29 studies. What this means is that if 1000 people are tested, of whom 100 would be diagnosed with TB meningitis using culture, 71 of the 100 would register as Xpert-positive and 18 of the 900 without TB would also register positive and be what we call 'false positives'. 911 of the 1000 people would register as Xpert-negative and 29 of these would actually have TB and, therefore, be 'false negatives'.
In summary, we found that in people thought to have extrapulmonary TB, Xpert may be helpful in obtaining the diagnosis. It rarely yields a positive result for people without TB by culture, but the test’s ability to detect TB varies when different specimens are used. Xpert is also accurate for diagnosing rifampicin resistance. However, for people thought to have TB meningitis, a cautious approach should be taken and treatment should be based on clinical judgement and not withheld solely on an Xpert result, as is common practice when culture results are negative.

John: If you would like to read more about the use of the Xpert test for TB outside the lungs, the full review is available at the Cochrane Library dot com. You can find it there with a search for 'Xpert for extrapulmonary TB'.