No, I don't follow.

By model you are referring to your subtraction of two percentages taken from two different studies conducted at two different times, using different data sets (people: varying by age and characteristics), different assumptions, different methods and likely controlling for different covariates? And of note there are absolutely zero SARS-CoV-2 patients in the Hopkins study.

I get that you are saying that there is a 8.5% probability that hospitalized covid patients will die during their stay (during the first 6-months of the pandemic). That is what you are saying correct?

Next, where I am looking for clarification, you are saying an average person will be hospitalized 40 days and have a 10% probability of medical error death and COVID patients are hospitalized 20 days so they will have a 5% probability of medical error death. Is this correct? And you are assuming the 8.5% covid death rate is the union of medical errors and covid deaths so you can just subtract out 5%? Side note, so if I am hospitalized 400 days I should have a 100% probability of medical error death?

Another issue you are also coming against is transportability. If you are generalizing results to similar samples in the same population the assumptions are weaker, but you are generalizing results to a whole new population not used to generate all of the estimates, so additional assumptions are needed such as are their any new confounders or effect modifiers.

I am not trying to troll you, just trying to get on the same page.

By model you are referring to your subtraction of two percentages taken from two different studies conducted at two different times, using different data sets (people: varying by age and characteristics), different assumptions, different methods and likely controlling for different covariates? And of note there are absolutely zero SARS-CoV-2 patients in the Hopkins study.

I get that you are saying that there is a 8.5% probability that hospitalized covid patients will die during their stay (during the first 6-months of the pandemic). That is what you are saying correct?

Next, where I am looking for clarification, you are saying an average person will be hospitalized 40 days and have a 10% probability of medical error death and COVID patients are hospitalized 20 days so they will have a 5% probability of medical error death. Is this correct? And you are assuming the 8.5% covid death rate is the union of medical errors and covid deaths so you can just subtract out 5%? Side note, so if I am hospitalized 400 days I should have a 100% probability of medical error death?

Another issue you are also coming against is transportability. If you are generalizing results to similar samples in the same population the assumptions are weaker, but you are generalizing results to a whole new population not used to generate all of the estimates, so additional assumptions are needed such as are their any new confounders or effect modifiers.

I am not trying to troll you, just trying to get on the same page.

Last edited:

Next, where I am looking for clarification, you are saying an average person will be hospitalized 40 days and have a 10% probability of medical error death and COVID patients are hospitalized 20 days so they will have a 5% probability of medical error death. Is this correct? And you are assuming the 8.5% covid death rate is the union of medical errors and covid deaths so you can just subtract out 5%? Side note, so if I am hospitalized 400 days I should have a 100% probability of medical error death?

I just pulled up the HLOS rate for COVID patients in a subset of hospitals I am associated (~20K patients), the avg HLOS is around three days. I will note that avg is a garbage metric for LOS since it is a notoriously right skewed variable. In addition the all-cause mortality rate was under 5% for these patients.

Though, I will note that I did not read the original source documents you referenced - in that those links did not take me to an actual PubMed archive manuscript.