Hospitalizations were down yesterday but still net positive. ICU admissions are down today — so much so that New York State is now discharging more people from intensive care than it’s taking in. Is that (further) proof of the curve flattening or something else?

Here’s the ICU chart in larger form:

There are two possibilities that might explain the decline. One is some sort of systemic change in how doctors are treating COVID patients that might account for why fewer need critical care. Remember that Stat article from a few days ago about ventilators being overused, and possibly doing more harm than good? Maybe that thinking has penetrated among New York doctors to a greater extent than anyone realizes right now. They might be getting more conservative about placing patients with low blood oxygen levels on ventilators, opting for nasal cannula or BiPAP machines instead. Check out the dramatic change here since April 5:

If this were a pure matter of declining case counts as transmission slows in New York, I would think we’d see more of a gradual downward slope over the last week. Instead the number of intubations dropped by more than half literally overnight and hasn’t increased much. Are doctors doing something differently?

The other possibility, of course, is that social distancing really has flattened out case counts and is beginning to drive them downward. It makes sense that we might see that effect in ICU admissions most dramatically since ICU cases are a tiny fraction of overall infections; if the latter number is still positive but in decline, the former might end up net negative sooner rather than later. Relatedly:

Deaths haven’t fallen off yet in New York — they were at 777 today, slightly fewer than yesterday — but it makes sense that that number might take longer to fall than hospitalizations and ICU admissions would. The time from infection to death is presumably longer on average than the time from infection to hospitalization or ICU admission. Many of the people dying right now were probably infected before the cascade of shutdown orders in New York began. That is, the population of people “likely to die” might not have begun to shrink yet (or is only beginning to shrink now) whereas the population of people likely to be hospitalized or admitted to the ICU has been shrinking for weeks.

Trevor Bedford, the scientist in Seattle who first spotted “hidden” community spread of coronavirus in Washington by looking at the genomes of different strains, thinks the decreasing rates of transmission in the U.S. and abroad are “entirely” due to social distancing:

Seattle’s done a good enough job limiting contact between people that a field hospital that was built to care for patients who couldn’t be taken by overloaded hospitals is being dismantled without taking a single patient. California, which moved towards lockdowns more quickly than New York did, is seeing an especially big payoff:

California caught the coronavirus earlier in its trajectory. That fact is reflected in the case count (20,000 to New York’s 150,000) and the death count (550 to New York’s 7,000). But the hospitalization rate may be the most telling metric. Since the state began reporting hospitalization data at the end of march, total confirmed COVID-19 admissions — they currently stand at 2,825, with another 2,803 suspected cases — have been rising by only one to four percent per day. That’s the roughly the same slow and steady rate of increase Cuomo is now touting in New York as a plateau.

Except so far, California has been all plateau. Meanwhile, on Wednesday, the number of COVID-19 patients requiring intensive care dropped — by 1.9 percent, to 1,132 — for the first time since California began tracking them.

Earlier is better. Because New York didn’t move sooner, this is now the grim reality facing the entire country: Coronavirus is now the leading daily cause of death in the United States.

The good news is that that should change soon. The bad news is that it might take longer than we think/ Italy has been locked down for many weeks and had seemed to turn a corner, with case counts and deaths in decline, but both numbers suddenly crept up again yesterday. It’s unclear why. Maybe some Italians, encouraged by the progress, began socializing again too soon? It’d be hard to explain the uptick in deaths that way, though, given the time horizon for “likely to die” cases noted above.

I’ll leave you with this nifty visualization from Ohio’s health department, which has done a bang-up job with Mike DeWine in holding down deaths.