There are two likely reasons why rates of hospitalization and confirmed cases have not mirrored each other in Canada and other countries. The first explanation is an increased case detection of less severe covid-19 infection as testing capacity increased. In this situation, population transmission of covid-19 overtime is more consistently measured using hospital data compared to reported cases. The second explanation is a reduction in covid-19 transmission in people at high risk of complications compared to the general population. Reduced transmission in high-risk groups would happen if at-risk people more effectively adhered to physical distancing compared to low-risk people. In this situation, short and long-term models are improved with the use of different should transmissibility estimates (effective reproduction number, Rt) for high-risk groups and the general population. Both these explanations – changing case detect or changing transmissibility – argue for covid-19 forecasts and modelling of control measures that consider transmissibility based on hospital data.
Using hospital data to estimate population covid-19 incidence or past infection has limitations. The time between symptom onset and hospitalization for covid-19 is about seven days, which will result in reporting delays for newly acquired disease. Hospitalization occurs in only a fraction of confirmed cases, particularly for younger people. Therefore, the use of hospital data to estimate overall population infection rates and outbreak dynamics is statistically robust only for larger communities.