Created by Nathaniel Hupert MD, MPH, Weill Cornell Medicine with Peter L. Jackson PhD, Singapore University of Technology and Design, Michael G. Klein PhD, San Jose State University, and John A. (“Jack”) Muckstadt PhD, Cornell University
For physicians and other COVID-19 responders who would like a tool to help hospital leaders think about hospital caseload projections in a structured way, here is the latest version of one that we have been developing at Cornell (by Nathaniel Hupert at Weill Cornell Medicine with Peter L. Jackson at Singapore University of Technology and Design, Michael G. Klein at San Jose State University, and John A. (“Jack”) Muckstadt at Cornell University) since mid-January, 2020.
For questions about the Cornell COVID Caseload Calculator C5V, please contact Walt Beadling at email@example.com or 610-841-1618.
This tool is designed to make an easily modifiable projection of med/surge and ICU bed requirements over an outbreak of specified type, for a specified catchment area (and market share of that area, to model your hospital system).
This model requires some things that are very hard to pin down, but may be informed by the work of any dynamics of infectious disease modelers in your midst (for those of you fortunate enough to have some).
The inputs are:
In this new version with capacity and ventilators (C4 à C5V), there are additional inputs:
All inputs are now on 1-C5V Scenario, and all outputs are on 2-C5V OUTPUT.
The distributed version is called “hypothetical” because I put in a purely fictional population.
The Excel sheet takes a second or two to update itself after any cell is changed—sorry about that!
The Cornell COVID Caseload Calculator C5V combines mathematical calculation and deterministic simulation to provide estimates of both
A) the rate at which patients in a designated catchment area may present for hospitalization due to the initial 2020 wave of SARS-CoV-2 causing COVID-19 disease, and
B) the simulated hospital load caused by those patients to both medical/surgical and intensive care units, with specific attention paid to identifying the magnitude and timing of the peak daily hospital census for regular and critical care beds throughout the catchment area.
Critical user input to run the model includes actual or estimated age-structured catchment population; overall (final) infection rate; percent (a)symptomatic cases; symptomatic case hospitalization and critical care ratios (both starting from U.S. Centers for Disease Control and Prevention (CDC) estimates); day of peak of epidemic curve; and shape of epidemic curve. The latter two may be estimated from epidemiological models (e.g., the Oxford-Cornell COVID INTERNATIONAL MODEL), and may be checked by day-to-day correlation with actual hospital admission rates for medical/surgical and critical care beds.