Fig. 3

Forest plot of point estimates from the association between nasopharyngeal ACE2 transcription of the primary case and the SARS-CoV-2 secondary cases. Poisson and Negative Binomial (NB) regression models were used for the primary analysis and sensitivity analysis of the assumptions used to construct the infection tracing networks. (A) βeta-coefficients (β) and 95% confidence intervals for the association between nasopharyngeal ACE2 transcription and secondary cases within health care workers, adjusting for viral load. The best fit NB model shows no association in the primary analysis or sensitivity analysis (ii, iv or v). A significant association between nasopharyngeal ACE2 transcription and the number of SARS-CoV-2 secondary cases was found for healthcare workers in sensitivity analysis iii, where the serial interval of 5.6 days was not used to build the transmission network. (B) βeta-coefficientsand 95% confidence intervals for the association between nasopharyngeal ACE2 transcription and secondary cases within patients, adjusting for viral load. The best fit NB model finds a positive association between in the primary analysis, and sensitivity analysis iii, and v. No association between nasopharyngeal ACE2 transcription and the number of SARS-CoV-2 secondary cases was found for patients in sensitivity analysis ii and iv, where assumptions of transmission occurring within hospital units and a 15.6-day infectious period were excluded from constructing the transmission network