Pooling of six respiratory samples for the detection of SARS-CoV-2: A validation and cost study in a cohort in Lima, Peru

Background: The continuous evolution of the SARS-CoV-2 pandemic has led to a high demand for diagnostic testing and major shortages in testing materials, especially in low- and middle-income countries. As an alternative to testing individual samples, pooling of respiratory samples has been suggested. Previous studies have assessed performance of pooling, mainly using nasopharyngeal samples for the detection of SARS-CoV-2, but few studies have examined the performance of pooling the more practical nasal swabs or saliva samples. Objective: To evaluate the sensitivity, specificity, and potential cost reduction of pooling of nasal swab (NS) and saliva (SL) samples for detection of SARS-CoV-2 in a community-based cohort study in Lima, Peru. Study design: A prospective cohort study was conducted in a community setting in San Juan de Lurigancho, Lima-Peru. NS and SL samples were collected from 132 participants twice-a-week for a 2-month period. Pools of 2 to 12 samples of the same type, from participants of the same household, were tested by RT-PCR. After pooled testing, all individual samples from positive pools and all individual samples from randomly chosen negative pools were evaluated. For assessment of diagnostic performance, pool testing results were compared with results from individual testing, which served as reference, and concordance in pooled and individual test detections was evaluated. Laboratory costs for both types of samples and testing were compared. Results: A total of 2008 NS and 2002 SL samples were collected from 132 study participants. We tested 329 NS and 333 SL pools. The mean pool size for NS and SL pools was 6.22 (SD = 0.92) and 6.39 (SD = 1.71), respectively. Using individual testing as reference, NS pooling of 6 had a sensitivity and specificity of 94% and 100%, respectively, with kappa of 0.97 (CI 95%: 0.93-1.00). The corresponding values for SL pooling of 6 were 83%, 100%, and 0.90 (CI 95%: 0.83-0.97). Compared with individual testing, pooling resulted in a cost reduction of 74.8% for NS and 72.4% for SL samples. Conclusions: Pooling easy-to-collect respiratory samples, especially NS, demonstrated very high diagnostic performance for detection of SARS-CoV-2 with substantial cost savings. This approach could be considered in large population screening programs, especially in LMIC.