Serial testing for latent tuberculosis using QuantiFERON-TB Gold In-Tube: A Markov model

Healthcare workers (HCWs) in low-incidence settings are often serially tested for latent TB infection (LTBI) with the QuantiFERON-TB Gold In-Tube (QFT) assay, which exhibits frequent conversions and reversions. The clinical impact of such variability on serial testing remains unknown. We used a microsimulation Markov model that accounts for major sources of variability to project diagnostic outcomes in a simulated North American HCW cohort. Serial testing using a single QFT with the recommended conversion cutoff (IFN-g > 0.35 IU/mL) resulted in 24.6% (95% uncertainty range, UR: 23.8–25.5) of the entire population testing false-positive over ten years. Raising the cutoff to >1.0 IU/mL or confirming initial positive results with a (presumed independent) second test reduced this false-positive percentage to 2.3% (95%UR: 2.0–2.6%) or 4.1% (95%UR: 3.7–4.5%), but also reduced the proportion of true incident infections detected within the first year of infection from 76.5% (95%UR: 66.3–84.6%) to 54.8% (95%UR: 44.6–64.5%) or 61.5% (95%UR: 51.6–70.9%), respectively. Serial QFT testing of HCWs in North America may result in tremendous over-diagnosis and over-treatment of LTBI, with nearly thirty false-positives for every true infection diagnosed. Using higher cutoffs for conversion or confirmatory tests (for initial positives) can mitigate these effects, but will also diagnose fewer true infections.