A PREDICTION model for tuberculosis (TB) in people living with HIV has been suggested by the results of a recent study. Researchers from the Department of Medicine at the University of Cape Town, Cape Town, South Africa, found that HIV patients showing symptoms of TB who did not respond to antibiotics were more likely to have the disease than patients who did have a clinical response.
The team set out to discover independent predictors of TB at the initial visit and following a trial of empiric antibiotics in patients who sought clinical care. In addition, a multivariable prediction model for TB was sought, which uses independent predictors that can be identified at initial and return visits. They enrolled 207 adults on the programme, all of whom were living with HIV and had had a positive World Health Organization (WHO) symptom screen. The participants were treated with antibiotics and were sputum tested using Xpert MTB/RIF Ultra. The patients were evaluated over 2 or 3 visits, and the researchers then assessed multivariable prediction models for TB.
They found that 75 (36%) of the total cohort had confirmed tuberculosis, and that clinical response to the antibiotics gave a good indication of disease likelihood after 2 days, and a multivariable prediction model, which was based on two visits but without C0reactive protein provided the team with “acceptable discrimination (c-statistic: 0.75) and calibration (goodness of fit test, p=0.07).” The team added: “A clinical response to antibiotics is useful for diagnosing TB in people with HIV who are identified by passive case finding and is a strategy that could be implemented immediately.” “CRP after antibiotics adds value but is not essential. CRP at index presentation is of limited value in this population.”