An analysis of SARS-CoV-2 viral load by patient age

Abstract As children are under-represented in current studies aiming to analyse transmission of SARS-coronavirus 2 (SARS-CoV-2), their contribution to transmission is unclear. Viral load, as measured by RT-PCR, can inform considerations regarding transmission, especially if existing knowledge of viral load in other respiratory diseases is taken into account. RT-PCR threshold cycle data from 3303 patients who tested positive for SARS-CoV-2 (out of 77,996 persons tested in total, drawn from across Germany) were analysed to examine the relationship between patient age and estimated viral load. Two PCR systems were used. In data from the PCR system predominantly used for community and cluster screening during the early phase of the epidemic (Roche LightCycler 480 II), when such screening was frequent practice, viral loads do not differ significantly in three comparisons between young and old age groups (differences in log 10 viral loads between young and old estimated from raw viral load data and a Bayesian mixture model of gamma distributions collectively range between −0.11 and −0.43). Data from a second type of PCR system (Roche cobas 6800/8800), introduced into diagnostic testing on March 16, 2020 and used during the time when household and other contact testing was reduced, show a credible but small difference in the three comparisons between young and old age groups (differences, measured as above, collectively range between −0.43 and −0.83). This small difference may be due to differential patterns of PCR instrument utilization rather than to an actual difference in viral load. Considering household transmission data on influenza, which has a similar viral load kinetic to SARS-CoV-2, the viral load differences between age groups observed in this study are likely to be of limited relevance. Combined data from both PCR instruments show that viral loads of at least 250,000 copies, a threshold we previously established for the isolation of infectious virus in cell culture at more than 5% probability, were present across the study period in 29.0% of kindergarten-aged patients 0-6 years old (n=38), 37.3% of those aged 0-19 (n=150), and in 51.4% of those aged 20 and above (n=3153). The differences in these fractions may also be due to differences in test utilization. We conclude that a considerable percentage of infected people in all age groups, including those who are pre- or mild-symptomatic, carry viral loads likely to represent infectivity. Based on these results and uncertainty about the remaining incidence, we recommend caution and careful monitoring during gradual lifting of non-pharmaceutical interventions. In particular, there is little evidence from the present study to support suggestions that children may not be as infectious as adults.