Increased transmissibility of SARS-CoV-2 lineage B.1.1.7 by age and viral load

New lineages of SARS-CoV-2 are of potential concern due to higher transmissibility, risk of severe outcomes, and/or escape from neutralizing antibodies. Lineage B.1.1.7 (the Alpha variant) became dominant in early 2021, but the association between transmissibility and risk factors, such as age of primary case and viral load remains poorly understood. Here, we used comprehensive administrative data from Denmark, comprising the full population (January 11 to February 7, 2021), to estimate household transmissibility. This study included 5,241 households with primary cases; 808 were infected with lineage B.1.1.7 and 4,433 with other lineages. Here, we report an attack rate of 38% in households with a primary case infected with B.1.1.7 and 27% in households with other lineages. Primary cases infected with B.1.1.7 had an increased transmissibility of 1.5–1.7 times that of primary cases infected with other lineages. The increased transmissibility of B.1.1.7 was multiplicative across age and viral load.


Supplementary Note 1: Descriptive Statistics
From December 21, 2020 (week 52) to February 21, 2021 (week 7), Denmark had 68,169 SARS-CoV-2 cases identified with RT-PCR, of which, 35,684 (52%) were selected for WGS and 28,383 (42%) came back with a genome (Table S1). The proportion of cases being sampled varied over time depending on whether the cases occurred in TCDK or in hospitals ( Figure S2). The proportion of cases selected for WGS and the proportion that came back with a genome is dependent on the Ct value ( Figure S3). For positive tests with a Ct value of 18, 85% of the samples were selected for WGS (purple) and 76% came back with a genome (green). Thus, the success rate was 89% (76/85). Similarly, for positive tests with a Ct value of 38, 37% of the samples were selected for WGS and 5% came back with a genome. Thus, the success rate was 14% (5/37). The success rate starts to decline for tests with a Ct value ≥30.  The proportion of cases being sampled for WGS dependent on the Ct value varies over time ( Figure S4). In week 2 TCDK started to sample systematically and to sample on Ct values. From Figure S4, we see that in week 2, TCDK used a Ct value cut-off of 30, 32, and 35. In weeks 3-6, TCDK used a Ct value cut-off of 35. Samples with higher Ct values (35<Ct≤38) were included, when WGS capacity allowed for it. Figure S4: Proportion of positive RT-PCR tests sampled for WGS and with a genome, by Ct value and calendar week Figure S5: Proportion of positive RT-PCR tests sampled for WGS and with a successfully sequenced genome, by age. Figure S6: Proportion of positive RT-PCR tests sampled for WGS and with a genome, by age and calendar week.  Figure S7).
WGS was mainly obtained for samples with low Ct values compared with the distribution of Ct values from the whole population (gray dotted line in Figure S7). We found that the Ct value distribution for B.1.1.7 and other lineages were approximately similar from week 1 to week 7 ( Figure S7). We see a clear shift in the distribution of cases without a successfully sequenced genome from week 2, when SSI started to systematically select case samples on Ct values. The distribution of the age of the cases stratified by B.1.1.7 (red), other lineages (blue) are relatively similar, although B.1.1.7 seems to mainly infect younger people in weeks 2-4 ( Figure   S8).      Figure S9: Age structured transmissibility stratified by lineage in five-year age groups.

Household
Notes: The transmission rate describes the proportion of potential secondary cases within the household that were infected. The transmission risk describes the proportion of infected primary cases that infected at least one secondary case. This figure is the same as Figure 1, except that it shows five-year age groups. The markers show the estimates of the mean. The shaded areas show the 95% confidence bands of the estimates clustered on the household level.
Primary cases infected with B.1.1.7 generally had a higher transmissibility compared with cases infected with other lineages, across Ct values ( Figure S8 and S9).     This table is comparable to Table 3, but excluding households with co-primary cases, i.e., more than one (primary) case identified on day zero. Columns I-IV provide odds ratio estimates for the increased transmission rate of B.1.1.7 compared with other lineages. Columns V-VII show the same for the transmission risk. Column I provides the crude estimates, i.e., only with a constant and without any controls. Column II further includes fixed effects for ten-year age groups of the primary cases. Column III further includes the age of potential secondary cases. Column IV further includes fixed effects for Ct values in bi-value groups. This sample is further restricted to only include primary cases identified in TCDK, as we only have Ct values on those. Column V provides the crude estimates, i.e., only with a constant and without any controls. Column VI further includes fixed effects for tenyear age groups of the primary cases. Column VII further includes fixed effects for Ct values in bi-value groups. This sample is further restricted to only include primary cases identified in TCDK, as we only have Ct values on those. All effects are included as fixed effects. Pot. Sec. Case = Potential Secondary Cases. Only primary cases identified in TCDK are included in models with Ct values. 95% confidence intervals clustered on the household level.
(a) Transmission Rate (b) Transmission Risk Figure S12: Age structured transmissibility stratified by lineage of the primary case, , excluding households with co-primary cases Notes: This figure is comparable to Figure 1, but excluding households with co-primary cases, i.e., more than one (primary) case identified on day zero. The transmission rate describes the proportion of potential secondary cases within the household that were infected. The transmission risk describes the proportion of infected primary cases that infected at least one secondary case. Figure S7 provides the same graphs for five-year age groups. The markers show the estimates of the mean. The shaded areas show the 95% confidence bands of the estimates clustered on the household level.
(a) Transmission Rate (b) Transmission Risk Figure S13: Transmissibility stratified by lineage and Ct value quartiles, excluding households with co-primary cases Notes: This figure is comparable to Figure S8, but excluding households with co-primary cases, i.e., more than one (primary) case identified on day zero. The transmission rate describes the proportion of potential secondary cases within the household that were infected. The transmission risk describes the proportion of infected primary cases that infected at least one secondary case. The markers show the estimates of the mean. The shaded areas show the 95% confidence bands of the estimates clustered on the household level.
where Agep,10 is the age (in ten-year groups) of the primary case. β measures the transmission rate for each ten-year age group of the primary cases. εp denotes the error term, clustered on the household (event) level.
while varying the link function to compare the model fit of an additive versus a multiplicative effect.
As the two models include the same parameters, the model fits can be compared using the Akaike Information Criterion (AIC). Furthermore, reduced versions of the linear predictors were tested. Across all three model specifications and for both transmission rate and transmission risk, we found that the logit model had a lower AIC and, thereby, was a better fit compared with the identity model, implying that the increased transmissibility is multiplicative and not additive (Table S7 and S8).