Researchers Develop New Algorithm for Diagnosing Tuberculosis in Patients With HIV

Researchers from South Africa have conducted a new study in which they propose a clinical prediction rule to make quicker, earlier tuberculosis (TB) diagnoses in seriously ill HIV-infected patients.

In 2016, TB was the top infectious killer worldwide, according to a report released by the World Health Organization (WHO) in October, and it remains the leading cause of death for individuals with HIV. Although worldwide efforts to reduce the impact of the disease have saved an estimated 53 million lives since 2000, a large number of TB infections go undetected each year. The two epidemics of TB and HIV continue to compound each other, particularly in sub-Saharan where about 86% of all HIV-associated TB deaths occur, and where individuals living with HIV are 16 to 27 times more likely to develop the disease than those without HIV.

In 2007, WHO developed guidelines for diagnosing smear-negative TB cases, particularly needed in HIV-prevalent and resource-constrained settings. Smear microscopy is recommended by WHO to detect acid-fast bacilli and identify the Mycobacterium tuberculosis bacteria that cause TB, and though the test can be quick, one WHO report noted that just 57% of new cases of pulmonary TB reported come up smear-positive. The 2007 guidelines proposed a diagnostic algorithm including a cough lasting 2 to 3 weeks and more than one of the other danger signs including low respiratory rate, elevated heart rate, and high temperature.

To attempt to improve the WHO algorithm for diagnosing TB in seriously ill HIV-infected patients, a team led by University of Cape Town researchers conducted a study recently published in the journal Clinical Infectious Diseases. They aimed to develop a clinical prediction rule — in other words, determine which set of symptoms best predicts the probability of a TB diagnosis. They included evaluation of cough of any duration, classic tuberculosis symptoms, chest radiographic features, hemoglobin, white cell count, and the newer Xpert MTB/RIF assay.

“The problem of undiagnosed TB in HIV is complex and not only related to poor access to care,” explained the study’s lead author, Gary Maartens, MMed, in a recent interview with Contagion ^®. “In many resource-limited settings, access to new rapid diagnostic tests, like the Xpert MTB/RIF assay, remains limited. They still rely on sputum smear for diagnosis, which is not very sensitive in HIV-infected patients so delayed diagnoses are common, which is a problem as TB progresses more rapidly in HIV-infected patients.”

The 484 participants enrolled in the study were HIV-infected, 18 years of age, had a cough (of any duration) and exhibited one or more of the WHO danger signs for TB. Using the six selected variables, the researchers developed a clinical prediction rule with a score chart; TB diagnosis was made in 52.7% of participants. The most significant predictors of culture-positive tuberculosis were the inability to walk unaided, a radiologist assessment of likely tuberculosis on chest radiograph, and anemia. Raised white cell count was a significant negative predictor of tuberculosis. While cough duration of 14 or more days was predictive of tuberculosis, 28.6% of culture-positive tuberculosis participants had cough duration of fewer than 14 days.

“We developed a clinical prediction rule for diagnosing TB in seriously ill HIV-infected patients, which could guide decisions about whether to start empiric therapy for TB,” said Dr. Maartens. “The clinical prediction rule uses simple clinical and laboratory features, which are available in almost all resource-limited settings. Low scores in our clinical prediction rule are associated with a very low probability of TB and could be used as a ‘rule out’ test.”

The research team conducted this study between 2011 and 2014, and, since then, WHO has updated their guidelines, said Dr. Maartens. The new guideline’s features were already incorporated into the study, so the researchers were able to evaluate WHO’s new algorithm. “The 2016 WHO guidelines for seriously ill patients are improved in several respects: the guidelines are no longer restricted to pulmonary involvement, cough of any duration is allowed, and the rapid diagnostic test Xpert MTB/RIF assay—which wasn't available in 2007—has replaced sputum smear,” he explained. “This is a big advance, as the Xpert MTB/RIF assay is much more sensitive than sputum smears, which are often false-negative in HIV-infected patients.”