*Caution non-peer reviewed preprint* US #COVID19 death rates are lower now than in the spring. Is disease less lethal (better treatments, etc), or just younger people getting sick? We examined @nyulangone outcomes over time to explore. /1 @PetrilliMD medrxiv.org/content/10.110… 
We know that risk of mortality is highly dependent on characteristics like age, comorbidity, severity of illness on presentation. We also know that the current epidemic is skewing younger. That alone could explain lower mortality. /2 vox.com/2020/7/18/2132…
But, it's possible we've also gotten better at #COVID19 care. It's also possible people are distancing/wearing masks more, and getting a lower viral innoculum, causing less severe disease. Disentangling these issues is important to help understand expected future outcomes.
We took a look at every hospitalization @nyulangone with lab-confirmed #COVID19 from March to June (now updating with data through August). Our patient population was slightly younger, less sick and lower proportion male over time.
Unadjusted in-hospital mortality dropped a LOT (blue line): from 30% to 3%. Comparing observed deaths to expected number of deaths (accounting for age, sex, race/ethnicity, BMI, smoking, comorbidity, admit vitals and labs), we see less of a drop (orange line), but still present.
We conclude there is some evidence of improvement in mortality over time, perhaps 2/2 better treatment, more experience. We are now updating the data through August and adding some different modeling approaches for eventual publication.
