Predicting Airborne Infection Risk: Association Between Personal Ambient Carbon Dioxide Level Monitoring and Incidence of Tuberculosis Infection in South African Health Workers

Abstract

Abstract Background High rates of tuberculosis (TB) transmission occur in hospitals in high-incidence countries, yet there is no validated way to evaluate the impact of hospital design and function on airborne infection risk. We hypothesized that personal ambient carbon dioxide (CO2) monitoring could serve as a surrogate measure of rebreathed air exposure associated with TB infection risk in health workers (HWs). Methods We analyzed baseline and repeat (12-month) interferon-γ release assay (IGRA) results in 138 HWs in Cape Town, South Africa. A random subset of HWs with a baseline negative QuantiFERON Plus (QFT-Plus) underwent personal ambient CO2 monitoring. Results Annual incidence of TB infection (IGRA conversion) was high (34%). Junior doctors were less likely to have a positive baseline IGRA than other HWs (OR, 0.26; P = .005) but had similar IGRA conversion risk. IGRA converters experienced higher median CO2 levels compared to IGRA nonconverters using quantitative QFT-Plus thresholds of ≥0.35 IU/mL (P < .02) or ≥1 IU/mL (P < .01). Median CO2 levels were predictive of IGRA conversion (odds ratio [OR], 2.04; P = .04, ≥1 IU/mL threshold). Ordinal logistic regression demonstrated that the odds of a higher repeat quantitative IGRA result increased by almost 2-fold (OR, 1.81; P = .01) per 100 ppm unit increase in median CO2 levels, suggesting a dose-dependent response. Conclusions HWs face high occupational TB risk. Increasing median CO2 levels (indicative of poor ventilation and/or high occupancy) were associated with higher likelihood of HW TB infection. Personal ambient CO2 monitoring may help target interventions to decrease TB transmission in healthcare facilities and help HWs self-monitor occupational risk, with implications for other airborne infections including coronavirus disease 2019.