Extended Data Fig. 2: Analysis of ddPCR quantification.

(a) Linear mixed model analysis of estimated difference in log10ddPCR SARS-CoV-2 N copies/ng RNA between all respiratory and all non-respiratory tissues among early, mid, and late cases with SE and p-values from relevant contrasts, (b) graph of linear mixed model analysis comparing linear trends of log10ddPCR SARS-CoV-2 N copies/ng RNA by log10DOI of respiratory and non-respiratory tissues, (c) linear mixed model analysis estimating the linear trends of log10ddPCR SARS-CoV-2 N copies/ng RNA by log10DOI of individual tissue groups with intercept with standard error (SE), and slope with SE and p-values from relevant contrasts (a signification p-value indicates a non-zero slope), (d) Spearman correlation between ddPCR (N copies/ng RNA) and sgRNA (copies/µL RNA) for all tissues tested for sgRNA, with subset analyses of these tissues from early, mid and late cases and all non-respiratory and respiratory samples with 95% confidence intervals (CI), (e) Spearman correlation between ddPCR (N copies/ng RNA) and sgRNA (copies/µL RNA) for tissues jointly positive by both assays, with subset analyses of these tissues from early, mid, and late cases and all non-respiratory and respiratory samples with 95% CI, (f) receiver operating characteristic (ROC) curve of logistic regression using log10ddPCR in tissues to predict the detection of sgRNA in tissues, area under the curve is 0.965 (95% CI 0.953, 0.977), optimal cut-off for ddPCR is 1.47 N copies/ng RNA (sensitivity 93.0%, specificity 91.6%). All p-values were two-sided without adjustment for multiple comparisons.