Overall a skim makes this seem like a good effort to me and potentially informative. However, I want people to understand a couple things: 1) This article is n

Overall a skim makes this seem like a good effort to me and potentially informative. However, I want people to understand a couple things: 1) This article is not an experiment and doesn't establish that COVID-19 is the causative agent here. They do a variety of statistical tricks to get a little closer to causality. I am particularly worried about this part: "To adjust for the difference in baseline characteristics ... variables were selected based on prior knowledge1,7,24,25. The pre-defined covariates included age, race (White, Black and Other), sex, ADI, body mass index, smoking status (current, former and never) and healthcare use parameters, including the use number of outpatient and inpatient encounters and use of long-term care. We additionally specified..." That's nice, but it involves a lot of experimenter degrees of freedom and doesn't control for the main thing that strikes me as relevant here: behavior. It's not like people were randomly selected to get COVID or not, they were getting COVID or not largely as a function of their behavior. People who are more risk-tolerant and / or less norm-compliant are going to have different health outcomes. EDIT: point 2 is wrong, I misread the data 2) The effects aren't large. For individual cardio conditions, max 1% increase. Overall, max 5% (many of which aren't serious). EDIT: begin the part I still believe again: I'd like to see an analysis that looks at these outcomes merely on the basis of care setting. I.e., if you end up in the ICU for anything at all, how does your increased risk of $CARDIO_PROBLEM change relative to COVID specifically? It's fun to talk anecdotes, but the data isn't ever as exciting as people would like to believe. I think it's very important to work against the hype machine because that's the only way we'll build real evidence-based progress.