Grant: Operating Grant

Tuberculosis (TB) infects a third of the world's population, and nearly 9 million cases occur every year. Health care workers (HCWs) are at higher risk of TB and therefore are repeatedly tested (for example, every year) with a skin test to identify latent TB infection (not yet an active disease). Over the past few years, blood-based tests called interferon-gamma release assays (IGRAs) have been increasingly used instead of the skin test. While IGRAs are more accurate than skin tests, evidence is showing that in some cases, if you test someone more than once (as is the case with HCWs) their IGRA result could be different each time: some tests change from negative to positive (conversion) and some from positive to negative (reversion). Because of this variation of IGRA test results, hospitals face a challenge: how to interpret a HCW's IGRA test results if these differ every year? In this context, it's important to understand how likely these variations are to occur and how to interpret them clinically (for example, what is the risk of disease in a patient who is positive one year and negative in the next year without intervention). To do this, we will construct a mathematical model with 100,000 hypothetical individuals that will be representative of North American HCWs. Within this model, we will assess the different factors that can affect the variability of the test. The data for the model will come from a systematic review of all available literature on IGRA testing reproducibility and from experiments of one of our collaborators. This model should allow us to understand how many conversions are true new TB infections, and how many are random (not new cases of TB), and to define new guidelines on how to interpret results from IGRAs in serial testing. We hope that the results from this project can inform policies on TB diagnosis in HCWs in North America.