Association of SARS-CoV-2 viral load distributions with individual demographics and suspected variant type: results from the Liverpool community testing pilot, England, 6 November 2020 to 8 September 2021

Background The PCR quantification cycle (Cq) is a proxy measure of the viral load of a SARS-CoV-2-infected individual. Aim To investigate if Cq values vary according to different population characteristics, in particular demographic ones, and within the COVID-19 pandemic context, notably the SARS-CoV-2 type/variant individuals get infected with. Methods We considered all positive PCR results from Cheshire and Merseyside, England, between 6 November 2020 and 8 September 2021. Cq distributions were inspected with Kernel density estimates. Multivariable quantile regression models assessed associations between people’s features and Cq. Results We report Cq values for 188,821 SARS-CoV-2 positive individuals. Median Cqs increased with decreasing age for suspected wild-type virus and Alpha variant infections, but less so, if not, for Delta. For example, compared to 30–39-year-olds (median age group), 5–11-year-olds exhibited 1.8 (95% CI: 1.5 to 2.1), 2.2 (95% CI: 1.8 to 2.6) and 0.8 (95% CI: 0.6 to 0.9) higher median Cqs for suspected wild-type, Alpha and Delta positives, respectively, in multivariable analysis. 12–18-year-olds also had higher Cqs for wild-type and Alpha positives, however, not for Delta. Overall, in univariable analysis, suspected Delta positives reported 2.8 lower median Cqs than wild-type positives (95% CI: 2.7 to 2.8; p < 0.001). Suspected Alpha positives had 1.5 (95% CI: 1.4 to 1.5; p < 0.001) lower median Cqs than wild type. Conclusions Wild-type- or Alpha-infected school-aged children (5–11-year-olds) might transmit less than adults (> 18 years old), but have greater mixing exposures. Smaller differences in viral loads with age occurred in suspected Delta infections. Suspected-Alpha- or Delta-infections involved higher viral loads than wild type, suggesting increased transmission risk. COVID-19 control strategies should consider age and dominant variant.


Additional study details
In the Cheshire and Merseyside, UK region, 6365 positive tests on 6241 individuals were obtained at laboratories other than Lighthouse laboratories. Since these labs did not test for all three of the genes of interest in this study, we have not considered these positives in any of the analysis in this paper. Pillar 2 testing was available to over 5's. 4208 individuals had age was recorded as 4 or younger, and additionally 13 individuals with age recorded as being over 112. These 4221 individuals were removed from the density plots for Cq by age, and from the regression models and assumed to be data entry errors. Similarly, 220 individuals did not have their sex recorded, and were omitted from the density plots for sex and the regression models.

Is the age effect explained by symptom status, swabbing method or time period?
Supplementary Figure S3 shows that higher Cq values for school age children can be seen in both symptomatic and asymptomatic individuals (except for suspected cases of Delta variant) but is more pronounced in asymptomatic individuals. The numbers of individuals in each category is reported in Supplementary Table S1.
It could be that swabbing technique in young children provides a sample of lower quality due to the difficulties of performing the test. This could be exaggerated in self-administered tests. However, Supplementary Figure S4 shows the Cq distributions by age for self-administered tests and for those done by health care professionals, and in both cases the shifted distributions for 5-11-and 12-18-year-olds remains for both wild-type and alpha variants, and the 5-11-year-old distribution for suspected Delta variant cases. The shift is more pronounced for the health care professionals, possibly suggesting better swabbing quality. Supplementary Table S2 reports the numbers of individuals in each category related to Supplementary Figure S4.
It is known that viral load is associated with increased disease severity. Supplementary Figure  S5 shows that the age effect is reduced in later time periods, corresponding to the point at which lateral flow testing was government policy for secondary school children. It could be that the introduction of twice weekly lateral flow testing has encouraged earlier uptake of PCR testing which is reflected in lower Cq values, recorded earlier in an individual's infection cycle.

Cq differences by Sex
Median regression showed that males had slightly lower Cq values than females (Table 3). However, visual inspection of Figure 2 shows that there is very little difference between males and females in terms of their viral loads, and similar proportions of positive tests were in each of the viral load categories (Table 2). This suggests the statistical significance was largely a result of sample size.

Cq differences by Test Location
Median regression suggested tests taken at home had slightly lower Cq values than tests taken elsewhere (Table 3), although visual inspection of Figure 2 suggests minimal differences.

Cq differences by Administration Method
There were a large number of individuals for whom it was not known how the PCR test was administered. Multivariable quantile regression shows statistically significant differences in median Cq of self-administered tests compared to those administered by health care professionals, although the effect size is small (Table 3 and Figure 2).