As Tattevin and Bouvet point out, the clinical and radiological presentation of TB among HIV-infected patients may be atypical, and thus the prediction rule may perform differently according to the prevalence of having HIV-positive individuals among patients suspected of TB. However, the model is composed of predictors, such as upper lobe infiltrate on chest radiograph or a positive tuberculin skin test result, that are usually considered to be “typical” of TB among HIV-negative patients. Thus, our finding that the rule had a high sensitivity in a population with a high prevalence of HIV-infected patients further supports the generalizability of the model, regardless of the patient’s HIV status.
Wisnivesky JP, McGinn TG. A Prediction Model for Isolating Inpatients With Suspected Pulmonary Tuberculosis—Reply. Arch Intern Med. 2005;165(15):1795. doi:10.1001/archinte.165.15.1795-a