What Covid Models Get Wrong: Focus on the burden on hospitals, not on the oft-mistaken forecasts

WSJ editorial. Excerpts:

"Take New York, where Gov. Andrew Cuomo locked down the state in mid-March based on dire warnings. His public health experts projected the state would need as many as 140,000 hospital beds and 40,000 intensive care units—two to three times more regular hospital beds and 10 times more ICU beds than were available. The UW model forecast that 49,000 regular beds and 8,000 ICU beds would be needed at the peak.

New York was hit hard, but Covid-19 hospital bed utilization in New York peaked at 18,825 and 5,225 for ICUs in mid-April. Even in New York City, hospital utilization never exceeded 85% of capacity and 89% for ICUs. Government-run hospitals in low-income neighborhoods with the most cases were unprepared, but they were ill-managed before the pandemic."

"The original UW model, which was based on the experiences in Italy and Wuhan, assumed that strict lockdowns would curb infections, reduce hospitalizations and lower deaths faster than they actually did in the Northeast."

"Hospital utilization by Covid-19 patients in New York City has fallen 94% since the peak, which has allowed some non-essential treatments to resume. New York City has 29% of its hospital beds and 34% of its intensive care units now available. New cases have fallen by about 40% and new hospitalizations by a third in the last two weeks, despite the recent protests.

Warnings about reopening states are also overblown so far. While Arizona has had an uptick in hospitalizations, about 59% of its emergency beds and 17% of ICU beds are unused. A month ago, 43% of hospitalized patients with Covid were in the ICU. Now only a third are, suggesting that better and earlier treatment is easing disease severity.

In Texas, hospitalizations have also been climbing, but weekly fatalities are down 40% from a month ago. Covid-19 patients occupy fewer than 5% of all hospital beds, and more than a quarter are available."

"Covid-19 patients take up a small share of ICU beds in most states that have reopened"

"But as Stanford epidemiologist John Ioannidis explains in a new paper, most models have overshot in part by making faulty assumptions about virus reproduction rates and homogenous susceptibility. A Massachusetts General Hospital model predicted more than 23,000 deaths within a month of Georgia reopening but the state had only 896.

“In the presence of strong groupthink and bandwagon effects, modelers may consciously fit their predictions to what is the dominant thinking and expectations—or they may be forced to do so,” Mr. Ioannidis writes. “Forecasts may be more likely to be published or disseminated, if they are more extreme.”"