Fig. 2: Increase in viral load for clustered versus nonclustered individuals.

a Betas from the multivariate regression model for the difference in log₁₀ viral load (copies/ml) in clustered individuals versus nonclustered at different stages of infection at diagnosis and different genetic distance thresholds: ≤0.015 substitutions/site, ≤0.010 substitutions/site, and ≤0.005 substitutions/site. Stage of infection is denoted by color. b Betas from the multivariate regression model for difference in log₁₀ viral load for individuals diagnosed at stage 1 infection with increasing node degree centrality (i.e., number of genetically linked partners) and cluster size relative to nonclustered individuals. Error bars represent the 95% confidence intervals for these estimates of beta. Node degree centrality compares individuals with at least that degree versus nonclustered individuals. Hence, node degree ≥1 in b is equivalent to clustered versus nonclustered depicted in a. Cluster size compares individuals in clusters of at least that size versus individuals in clusters of small sizes (i.e., cluster size ≥5 versus cluster size <5). Δ denotes the difference in log₁₀ viral load for each increase in node degree centrality or cluster size. Network constructed at ≤0.015 substitutions/site. Sample sizes (n) for statistical tests are provided in Table 1.