I have been curious about the changes to ICU capacity capacity. The weekly data captured at healthdata.gov/sites/default/……… now shows data through the week of 12/11. At first blush, the data doesn't appear to have changed much since the prior week. 1/n
2/n The aspect I am most curious about is the reporting of the data. An example of this is in reporting of ICU beds available and used. See below for one PA example: It is confusing to see the beds at this facility change during the weeks of 10/2 and 9/18 #covid#hospitals
3/n In contrast, note the reporting at this alternative #Pennsylvania hospital: While the average total #ICU beds increase slightly during the two most recent weeks, in contrast with the prior example, this hospital shows a percentage used < 100%. #LehighValley#ABE#Bethlehem
4/n [There is a data-library found at: healthdata.gov/covid-19-repor……… that indicates ICU beds are, "[t]otal number of staffed inpatient ICU beds" whereas occupancy, "[t]otal number of staffed inpatient ICU beds that are occupied." Ref. FAQ 5/6. hhs.gov/sites/default/………
5/n Here is the ICU capacity utilization distribution reported in this dataset for Pennsylvania hospitals for recent weeks, as well as from July (incl. median). I imagine I share in hope/optimism with others as #vaccines begin rollout, including in our local community. #mrna
#Healthcare is just an interesting business to be a part of. I saw a fellow post details from the PSU Health website that included a bunch of information in json format. I was curious to see what the data revealed. 1/n
2/n The most interesting surprise, aside from specific breakdowns of variation in reimbursement per service by payer, was the specific classification for "Amish/Mennonite OP Rate".
I was discussing #healthcare cost variation (& #EHP#RBP models) across geography with a colleague earlier today. I have followed the work of @zackcooperYale and thought it would be interesting to, in very simple terms, visualize some of the cost variation. A short thread. 1/n
2/n I started looking for a datasource. While our #benefitconsulting and #insurance clients engage networks of private insurers, that data is subject to different rules... I wondered what variation "looks like" for Medicare claims. I landed here: dartmouthatlas.org
3/n First, there is a ton of #Medicare data. It can be downloaded by state (see below) as well as to a county level (which I pulled into Python.) To consider the amount of cost attributable to different segments, I looked at the largest categories with observations by county.