RESUMO
The infectiousness and presymptomatic transmission of SARS-CoV-2 hinder pandemic control efforts worldwide. Therefore, the frequency of testing, accessibility, and immediate results are critical for reopening societies until an effective vaccine becomes available for a substantial proportion of the population. The loss of sense of smell is among the earliest, most discriminant, and prevalent symptoms of COVID-19, with 75-98% prevalence when clinical olfactory tests are used. Frequent screening for olfactory dysfunction could substantially reduce viral spread. However, olfactory dysfunction is generally self-reported, which is problematic as partial olfactory impairment is broadly unrecognized. To address this limitation, we developed a rapid psychophysical olfactory test (KOR) deployed on a web platform for automated reporting and traceability based on a low-cost (about USD 0.50/test), six-odor olfactory identification kit. Based on test results, we defined an anosmia score -a classifier for olfactory impairment-, and a Bayesian Network (BN) model that incorporates other symptoms for detecting COVID-19. We trained and validated the BN model on two samples: suspected COVID-19 cases in five healthcare centers (n = 926; 32% COVID-19 prevalence) and healthy (asymptomatic) mining workers (n = 1, 365; 1.1% COVID-19 prevalence). All participants had COVID-19 assessment by RT-PCR assay. Using the BN model, we predicted COVID-19 status with 76% accuracy (AUC=0.79 [0.75 - 0.82]) in the healthcare sample and 84% accuracy (AUC=0.71 [0.63 - 0.79]) among miners. The KOR test and BN model enabled the detection of COVID-19 cases that otherwise appeared asymptomatic. Our results confirmed that olfactory dysfunction is the most discriminant symptom to predict COVID-19 status when based on olfactory function measurements. Overall, this work highlights the potential for low-cost, frequent, accessible, routine testing for COVID-19 surveillance to aid societys reopening.
