Confronting COVID-19: Surging critical care capacity in Italy

Jose Manuel Rodriguez Llanes; Rafael Castro Delgado; Morten Gram Pedersen; Pedro Arcos Gonzalez; Matteo Meneghini

Our study is not exempt of limitations. Our rate of daily ICU use is obviously affected by conditions such as changes in case definitions, testing capacity, changes in clinical early detection, and changes in disease severity. We were conscious that case definition and testing strategies to optimize resource use were redirected before the end of February in Italy. Most severe patients probably arrived first to hospitals and a majority of them were of advanced age and with varying levels of comorbidities (11, 14) . Equally important, ICU admission criteria may have changed over the course of the emergency to deal with scarce ICU resources in particular locations, with difficult ethical choices to be made between providing ICU treatment versus palliative care in certain patients (26). To counteract these effects we chose a longer period to estimate our daily ICU rates and we corroborated the consistency of these values across days and periods defined. Additionally we chose the period at the beginning of the emergency where decisions were more likely made based on clinical realities and not on allocation of scarce resources. Our models, to a certain extent, depends on the quality of parametrization and its ability to predict future evolution of the epidemic. With this issue in mind, we tested its predictive ability on a six day window and we hope that this is enough to provide a range of good predictions from 25 March on and for at least 10 days, in a context where new drastic measures are introduced almost daily. We are aware that the numbers used in the model parametrization come with their own uncertainty, given that this disease is new, and context relevant data is not always available.