Transmissibility of COVID-19 depends on the viral load around onset in adult and symptomatic patients

Abstract

Objective: To investigate the relationship between viral load and secondary transmission in novel coronavirus disease 2019 (COVID-19).

Methods: Epidemiological and clinical data were obtained from immunocompetent laboratory-confirmed patients with COVID-19 who were admitted to and/or from whom viral loads were measured at Toyama University Hospital. Using a case-control approach, index patients who transmitted the disease to at least one other patient were analysed as "cases" (index patients) compared with patients who were not the cause of secondary transmission (non-index patients, analysed as "controls"). The viral load time courses were assessed between the index and non-index symptomatic patients using non-linear regression employing a standard one-phase decay model.

Results: In total, 28 patients were included in the analysis. Median viral load at the initial sample collection was significantly higher in symptomatic than in asymptomatic patients and in adults than in children. Among symptomatic patients (n = 18), non-linear regression models showed that the estimated viral load at onset was higher in the index than in the non-index patients (median [95% confidence interval]: 6.6 [5.2-8.2] vs. 3.1 [1.5-4.8] log copies/μL, respectively). In adult (symptomatic and asymptomatic) patients (n = 21), median viral load at the initial sample collection was significantly higher in the index than in the non-index patients (p = 0.015, 3.3 vs. 1.8 log copies/μL, respectively).

Conclusions: High nasopharyngeal viral loads around onset may contribute to secondary transmission of COVID-19. Viral load may help provide a better understanding of why transmission is observed in some instances, but not in others, especially among household contacts.

Fig 1. Trends in viral loads in the symptomatic patients.

(A) Viral load time courses of index (red) and non-index patients (blue). (B) Nonlinear regression models of index (red) and non-index patients (blue). The models were calculated by using all the data of the index or non-index patients. When no virus was detected, the data were hypothetically plotted as –0.5 log copies/μL. Solid curves are best-fit models and dotted lines indicate the 95% confidence interval of each model.