Read on Twitter
Today we released a new brief on COVID-19 in NC. This one focuses on hospitalizations and available bed capacity. We also add regional analyses in this brief. https://twitter.com/nccovid/status/1273237409061310465?s=20
Key takeaways: 1) As many have noticed, hospitalizations have increased considerably lately. This chart is complicated but shows number of hosps (right axis) and weekly growth (left axis). The flat to neg. growth of early May has become higher since.
2) Meanwhile, the number of empty beds has shrunk. It is not straightforward to estimate the number of hospital beds, especially those that are available but empty, but our best guess is that there are about 4000 fewer empty beds now than early April.
Why? Resuming elective procedures, care that was avoided in March/April. But the net effect is less capacity. That sounds bad, but hospitals have “pressure relief” valves they can use to potentially open this up – cx electives, increase staffing, use vacant beds, etc.
3) Regional effects are considerable. We used PHRST regions. It’s a story of three North Carolinas, really. a) the regions surrounding Charlotte and the Triangle b) regions surrounding Fayetteville, Greenville, and Asheville, and c) Triad and Wilmington. (our names)
a) Charlotte/Triangle: Cases have been growing considerably in these regions, and the hosp trend is similar. Charlotte region’s hosps have increased by 50% in two weeks; Triangle by 39%. These are remarkable, and if trend continues we could see pressure on acute beds shortly.
b)Fay/Green/Ashe: these regions have been flat to slow growth, w/periods of increase. For the last week in our analysis, it is Greenville (43 percent increase) which portends tight supply if that continues, but the less stable trend give us less confidence that will continue.
c)Triad / Wilmington: these have seen flat to decreasing hospitalization rates. Triad saw a very rapid increase through May but that has subsided.
Hospitals can surge beds, adding a couple more weeks to the “runway” in the typical region. Note that the pessimistic view of Greenville region and optimistic view of Fayetteville should be interpreted in light of the above tweet about growth stability in those regions.
Some methods: we use a Poisson regression over last 7 days to model growth (more stable than point estimates). We don’t adjust for response, which may bias growth rates downward (response is falling). Estimating bed capacity is….complicated.
What I hope you takeaway: 1) Earlier we said “can consider opening, but need to be vigilant.” This is the kind of thing to watch for – some regions are showing signs of concern. 2) Strong regional variation (cont.)
3) Some have been skeptical of using cases as a measure (“is it just capturing more testing?”); the results here are largely consistent with those trends -- going up in certain regions https://twitter.com/gmarkholmes/status/1271464950360412160?s=20
4) remember it takes a couple of weeks to turn the ship – the hospitalization rate in two weeks is largely already determined; another two weeks of double digit growth in some regions would be quite concerning.
Great collaboration with @A_McKethan and @PharmDJD. Errors in this thread are my own. /fin
