While I commend them for this work, a few things jump out at me right away looking at the data tables. The sample of the patients was heavily skewed towards a few demographics. 1/ https://twitter.com/BrennanSpiegel/status/1280527527325265920

74.5% of the COVID-19 patients were 30-39 yo (27.1% in the control group). 69.7% were Latinx (15.3% in control). 73.1% smoke every day (22.5% in control). 2/

But here's where I raise an eyebrow. 69.6% of the COVID-19 patients had a high school education or less (28.7% in control), but 63.5% of the COVID-19 made 200k+ per year (7.4% in the control group). Ultra-wealthy non-college 30-39yos, who mostly smoke. Huh? 3/

I'm not sure how with that data sampling, you can pick out PPIs as the culprit, regardless of what regression you do. We're going to need far better evidence than this before we throw PPIs under the bus yet again based on an observational study.