As an emergency physician, there’s very little I do on-shift that isn’t influenced by Covid-19. As a clinical informaticist and data geek, I’ve pored over data from dozens of sources (91-Divoc is my go-to, along with the MA Weekly Public Health Report which has city-specific testing data) to get a handle on where this is all going — and what my family and hospital can do to make it go better. Lately, as a dad with kids in school, I’ve even found myself pondering the significance of wastewater sampling.
So much data! And yet, I was asked a question last week that I realized wasn’t easily answerable: We know how many patients are admitted with Covid-19 at our hospital, but how are the other hospitals in our area doing? Are they full or empty, and what’s the trend?
As the fall surge continues to swell, possibly requiring us to transfer patients out, that sort of information would be very useful to have — but the Massachusetts dashboard doesn’t include it in any charts or tables. However, I discovered that the state does collect and publish each hospital’s daily Covid-19 census in its raw data archive. You have to dig each day’s data out of a separate zip file, but with the help of a script, I was able to download data for most days since June 1st and assemble these charts.
The title chart shows the trend of Covid-19 occupancy rates for the hospitals in Middlesex, the state’s most populous county (and where I live and work!). The one above is for Suffolk County, which includes the various Boston hospitals. A rate of 10% means that, however many other patients the hospital may have admitted, one in ten of their total beds are being used by patients with Covid-19. I used a 14-day rolling average to smooth out the data and make the charts more readable. There are charts for every county in alphabetical order at the bottom of this article.
Taken together, the charts show a consistent pattern. Rates were highest, but falling, in the Spring as the first surge was finishing up. (In April and May, many hospitals, including mine, had occupancy rates approaching 100%, but the state’s data on individual hospitals only goes back to June 1st.) During the summer, rates were low, generally under 5%, but they’ve since crept up, with most between 5–10% and some approaching 15%. Still, the trends show a lot of variability, with some hospitals climbing, a few actually falling, and several others basically flat.
I calculated the occupancy rate by taking each hospital’s census according to the state and dividing it by the number of licensed beds according to public data. I excluded hospitals without any licensed beds or with an average Covid-19 occupancy rate of less than 1%. The vertical scale of each is optimized for readability. The charts were generated using R, and the raw data is available for download.
So…enjoy! Let me know if you have any questions or spot an error. If there is interest, I will update these at least weekly.