Optimizing the sensitivity of detection of respiratory syncytial virus infections in longitudinal studies using the combination of weekly sample testing and biannual serology

Cohort studies are often challenged by incomplete adherence to sampling regimens, limiting the full capture of disease burden. We describe the detection of respiratory syncytial virus (RSV) infections achieved in a birth cohort using a combination of weekly nasal sample testing and serology. The PREVAIL Cohort followed 245 maternal-child dyads from birth to age 18-24 months. Weekly mid-turbinate nasal swabs were tested for RSV using real-time polymerase chain reaction (RT-qPCR). Serum was tested for RSV pre-fusion F IgG and IgA antibody at age 6 weeks and biannually from 6-24 months. Mixed effects classification and regression trees (CART) identified antibody thresholds consistent with a RT-qPCR-identified RSV infection using a subset of participants having ≥90% weekly sample adherence (n=53, 21%). Resulting thresholds were applied to participants with either ≥70% of weekly samples or serum at age 18-24 months (n=194, 79%) Incidence rates were compared using Fisher's exact test. CART identified a log10 change in IgG>0.32 or IgA>0.20 as indicative of an RSV infection. Comparing RT-qPCR-only to a combination of RT-qPCR and serology, RSV cumulative incidence (49% vs 75%, p<0.001) and incidence rate (0.33 vs 0.71 infections/child-year, p<0.001) increased; these rates did not differ from those calculated in those with ≥90% sample adherence.