A SIMPLE forecasting tool for critical care demand has been updated to better reflect changing COVID-19 pandemic conditions, offering improved predictive power for clinicians managing healthcare capacity.
Originally developed in Ontario, Canada, the model used early pandemic data to estimate critical care occupancy based on SARS-CoV-2 case numbers, average patient age, and testing volume. In this new study, researchers validated and recalibrated the model using comprehensive provincial data from March 2020–September 2022, spanning six pandemic waves. The initial model, based on waves 1 and 2, was then compared to updated versions incorporating data from wave 3, dominated by N501Y+ variants, and adjusted for widespread vaccination. Model accuracy was assessed using Spearman’s rho, and counterfactual modelling was employed to estimate the effect of vaccination on ICU admissions.
The initial model remained well calibrated (Spearman’s rho=0.85) but showed limited predictive validity for future waves (rho=0.46). However, predictive accuracy improved significantly with updated models incorporating wave 3 data, yielding rho values of 0.60 without vaccination data and 0.68 when vaccination effects were included. The improvement with vaccination was statistically significant (P=0.013). Modelling suggested that, in the absence of vaccination, critical care admissions would have reached 22,017, compared with the 9,020 observed: an estimated 144% reduction attributable to vaccination.
These findings reinforce the continued value of regression-based models in critical care planning, while highlighting the need for regular recalibration to reflect evolving immunity and variant profiles.
Reference
Grima AA et al. Derivation and validation of a point-based forecasting tool for SARS-CoV-2 critical care occupancy: a population-based modeling study. Lancet Reg Health Am. 2025; DOI:10.1016/j.lana.2025.101143.
