Ismael Segura-Ulate ^1,2* Navilla Apú ^1,2 Bernal Cortés ³ Yamitzel Zaldívar ⁴ Jordi Querol-Audi ^5,6 Carlos A. Ortega ^7,8 Fernando Flores-Mora ^1,2 Germán Madrigal-Redondo ^1,2 Andrés Gatica-Arias ^1,2,9

• ¹ University of Costa Rica, San José, Costa Rica
• ² Instituto de Investigaciones Farmacéuticas (INIFAR), San Jose, Costa Rica
• ³ Agencia Costarricense de Investigaciones Biomédicas (ACIB) - Fundación INCIENSA (FUNIN), San Jose, Costa Rica
• ⁴ Gorgas Memorial Institute of Health Studies, Panama City, Panama
• ⁵ Laboratory of Experimental and Applied Microbiology, Fluminense Federal University, Niterói, Brazil
• ⁶ University of Panama, Panama City, Panama
• ⁷ Other, San Salvador, El Salvador
• ⁸ University of El Salvador, San Salvador, San Salvador, El Salvador
• ⁹ School of Biology, Faculty of Sciences, University of Costa Rica, San Pedro, Costa Rica

Background: the gold-standard diagnostic protocol (GSDP) for COVID-19 consists of a nasopharyngeal swab (NPS) sample processed through traditional RNA extraction (TRE) and amplified with retrotranscription quantitative polymerase chain reaction (RT-qPCR). Multiple alternatives were developed to decrease time/cost of GSDP, including alternative clinical samples, RNA extraction methods and nucleic acid amplification. Thus, we carried out a cross comparison of various alternatives methods against GSDP and each other. Methods: we tested alternative diagnostic methods using saliva, heatinduced RNA release (HIRR) and a colorimetric retrotranscription loop-mediated isothermal amplification (RT-LAMP) as substitutions to the GSDP. Results: RT-LAMP using NPS processed by TRE showed high sensitivity (96%) and specificity (97%), closely matching GSDP. When saliva was processed by TRE and amplified with both RT-LAMP and RT-qPCR, RT-LAMP yielded high diagnostic parameters (88-96% sensitivity and 95-100% specificity) compared to RT-qPCR. Nonetheless, when saliva processed by TRE and detected by RT-LAMP was compared against the GSDP, the resulting diagnostic values for sensitivity (78%) and specificity (87%) were somewhat high but still short of those of the GSDP. Finally, saliva processed with HIRR and detected via RT-LAMP was the simplest and fastest method, but its sensitivity against GSDP was too low (56%) for any clinical application. Also, in this last method, the acidity of a large percentage of saliva samples (9-22%) affected the pH-sensitive colorimetric indicator used in the test, requiring the exclusion of these acidic samples or an extra step for pH correction. Discussion: our comparison shows that RT-LAMP technology has diagnostic performance on par with RT-qPCR; likewise, saliva offers the same diagnostic functionality as NPS when subjected to a TRE method. Nonetheless, use of direct saliva after a HIRR and detected with RT-LAMP does not produce an acceptable diagnostic performance.