Fig. 2

The analysis steps: a CO-CIN is inflated to match total COVID-19 hospitalised cases in SUS. b The same weekly adjustment is used to estimate the number of identified hospital-onset, hospital-acquired (HOHA) cases. c The length of stay for non-COVID-19 hospital patients and incubation period distribution is used to generate estimates of the proportion of hospital-acquired infections that would be identified (Fig. 1). This proportion (p) is used to estimate how many unidentified hospital-acquired infections there would be for each identified hospital-onset hospital-acquired infection by assuming a Binomial distribution and calculating the number of “trials” or “unidentified” hospital-acquired infections there were. d The unidentified hospital-acquired infections with symptom onset after discharge (“missed”) may return to hospital as a COVID-19 case: the trajectory of their disease is calculated to determine their contribution to hospitalised cases. e These “missed” unidentified hospital-acquired infections are assumed to contribute to onward transmission in the community: here we capture four generations of transmission to estimate the number of hospital-linked infections and subsequent hospitalised cases under different R estimates