Diagnostic performance and clinical implications of rapid SARS-CoV-2 antigen testing in Mexico using real-world nationwide COVID-19 registry data

Abstract

Background: SARS-CoV-2 testing capacity is important to monitor epidemic dynamics and as a mitigation strategy. Given difficulties of large-scale quantitative reverse transcription polymerase chain reaction (qRT-PCR) implementation, rapid antigen tests (Rapid Ag-T) have been proposed as alternatives in settings like Mexico. Here, we evaluated diagnostic performance of Rapid Ag-T for SARS-CoV-2 infection and its associated clinical implications compared to qRT-PCR testing in Mexico.

Methods: We analyzed data from the COVID-19 registry of the Mexican General Directorate of Epidemiology up to April 30th, 2021 (n = 6,632,938) and cases with both qRT-PCR and Rapid Ag-T (n = 216,388). We evaluated diagnostic performance using accuracy measures and assessed time-dependent changes in the Area Under the Receiver Operating Characteristic curve (AUROC). We also explored test discordances as predictors of hospitalization, intubation, severe COVID-19 and mortality.

Results: Rapid Ag-T is primarily used in Mexico City. Rapid Ag-T have low sensitivity 37.6% (95%CI 36.6-38.7), high specificity 95.5% (95%CI 95.1-95.8) and acceptable positive 86.1% (95%CI 85.0-86.6) and negative predictive values 67.2% (95%CI 66.2-69.2). Rapid Ag-T has optimal diagnostic performance up to days 3 after symptom onset, and its performance is modified by testing location, comorbidity, and age. qRT-PCR (-) / Rapid Ag-T (+) cases had higher risk of adverse COVID-19 outcomes (HR 1.54 95% CI 1.41-1.68) and were older, qRT-PCR (+)/ Rapid Ag-T(-) cases had slightly higher risk or adverse outcomes and ≥7 days from symptom onset (HR 1.53 95% CI 1.48-1.59). Cases detected with rapid Ag-T were younger, without comorbidities, and milder COVID-19 course.

Conclusions: Rapid Ag-T could be used as an alternative to qRT-PCR for large scale SARS-CoV-2 testing in Mexico. Interpretation of Rapid Ag-T results should be done with caution to minimize the risk associated with false negative results.

Fig 1

A) Number of rapid antigen tests per 100,000 population across different Mexican states. Figure also shows the percentage of rapid antigen tests amongst all SARS-CoV-2 tests administered in Mexico City and the rest of Mexico (B) and the curve of confirmed cases according to date from symptom onset in Mexico City and the rest of Mexico (C, D). Abbreviations: qRT-PCR: Reverse transcription polymerase chain reaction.

Fig 2

Mixed effects logistic regression models assessing predictors of cases with false negative compared to true positive test results (A) and false positive compared to true negative test results (B) using qRT-PCR as reference tests. Abbreviations: qRT-PCR: Reverse transcription polymerase chain reaction; CKD, Chronic Kidney Disease; CVD: cardiovascular disease; COPD: Chronic Obstructive Pulmonary Disease; OR: Odds Ratio; 95%CI: 95% Confidence interval, ICU: Intensive Care Unit.

Fig 3

Models assessing risk associated to confusion matrix categories in rapid antigen test results compared to qRT-PCR for COVID-19 outcomes including hospitalization (A), requirement for intubation (B), risk of adverse outcomes (C) and lethality (D). Abbreviations: qRT-PCR: Reverse transcription polymerase chain reaction; CKD, Chronic Kidney Disease; CVD: cardiovascular disease; COPD: Chronic Obstructive Pulmonary Disease; OR: Odds Ratio; HR: Hazard ratio; 95%CI: 95% Confidence interval; HC: Healthcare.