Abstract

Objectives: Estimate the detection limits of the COVID-19 surveillance system (SS) in Chile, by estimating the SARS-CoV-2 true prevalence (TP) and the reported official positivity prevalence (OPP) gap. Study design: Randomized cross-sectional. Methods: Two sampling campaigns (SC) were conducted (October-November 2020 and December 2020-January 2021) in the cities of Temuco, Valdivia, and Osorno. Blood was collected from adults from randomly selected households. Sera were analyzed using a commercial later flow test (LFT). A meta-analysis was performed to estimate LFT-performance in asymptomatic-cases. Data were analyzed using a Bayesian latent class model (BLCM) to estimate TP. Finally, BLCM outputs were compared with the OPP, by calculating the TP/OPP rate. Results: 1124 and 1017 households were visited during the 1st and 2nd SC, respectively. The BLCM rendered TP estimates of 6.5 %, 3.2 %, and 6.6 % for the cities of Temuco, Valdivia, and Osorno, respectively (1stSC), increasing to 9.4 %, 5.0 %, and 7.5 %, 60 days later (2ndSC). Depending on the city and SC, TP/OPP rates varied between 2.3 and 5.7. Conclusion: The national SS was unable to detect 70-79 % of all infected cases, suggesting that mild and asymptomatic cases were scarcely detected.