Coronavirus models also overestimated hospital beds required

John Solomon:

A Web site that tracks actual hospital beds in use suggests the model used by top White House health officials to project the trajectory of the coronavirus has so far overestimated the number of Americans hospitalized by the disease by tens of thousands.

Those projections, popularly known as the "Murray" model after the model's lead author, University of Washington professor Christopher Murray, were explicitly cited by Dr. Deborah Birx, the response coordinator for the White House's Coronavirus Task Force, at a press conference in the last week.

Birx told reporters that Murray's model, which predicts a shortage of tens of thousands of hospital beds throughout the country by the middle of April, underscored the task force's "concern that we had with the growing number of potential fatalities" based on the model's projections.

Yet a comparison of actual hospitalized patients by state and nationally suggests the model has so far overestimated the number of beds needed to treat pandemic patients.

The forecast predicted, for example, that the United States would need around 164,750 hospital beds for COVID-19 patients on Saturday. Yet the COVID Tracking Project, a team of journalists and data analysts who collect and tabulate coronavirus data from state tallies around the country, reported only around 22,158 currently hospitalized coronavirus patients nationwide on Saturday.

The discrepancies are also stark when looked at on a state-by-state basis. The model estimated that 65,434 patients would need hospital beds in New York State on Friday. In reality, there were 15,905 hospitalizations in that state by Sunday morning, according to the COVID Tracking Project.

Notably, the model touts its predictions as occurring under "full social distancing" through May of this year, meaning the projected hospitalizations are meant to occur even with significant quarantine measures.
...

Were the models overstating the needs in order to get people's attention?   That is certainly the effect whether it was intended or not.  But it had other effects including responses that did great damage to the economy and to workers.  Whenever you have projections that are inconsistent with observed results it is because one or more assumptions were invalid.

Sean Trende makes the case that people's experience varies widely between those in high-density urban areas and those who do not.  Power Line is not impressed with the models either.  But the models tend to have real-world consequences because of the reaction to them by governments.  From governors to Washington DC, the modelers are making an impression and it is not always the right one.