 |
|
In addition to not reading what you posted, you also didn't
Posted on: May 4, 2021 at 10:11:31 CT
CulturedDan MU
Posts:
93745
Member For:
15.09 yrs
Level:
User
M.O.B. Votes:
0
read my post. Read point #2, and explain to me how I don't understand that it was a summary of "available" evidence. (It's actually a summary of evidence the CDC employee wanted to present, not all available evidence). In that point I pointed out the poor quality of evidence used to make the authors argument, therefore, by the evidence of what I posted, I clearly understand what the article was. In addition, I also have much more experience with medical literature than you can imagine, but since you are throwing stones and think that you know, maybe you can help me with something I'm struggling with.
I'm studying the effect of inhaled epopoprostenol in patients with moderate to severe ARDS. My primary endpoint is looking for patient characteristics that are associated with a good response (response being a 10% improvement of SPO2 to FIO2 ratio at 6 hours). To do so we are running a risk regression analysis over 24 factors, including SARS-CoV-2 status.
So according to what came through the regression, the only statistically significant patient characteristics were, that for every 10 decrease in SPO2:FIO2 the OR was 0.78 (CI 0.62 - 0.87) and SARS-CoV-2 status. OR 0.42 (CI 0.28-0.57). So, it looks like that 1. If your ARDS is worse, the more likely you are to respond and You're almost 50% more likely to respond if you have COVID-19.
However, after looking at the raw data, 67% of patients without covid had a positive response to epoprostenol, whereas on 55% of patients with COVID-19. So the regression doesn't make sense. Turns out, it's just the way the analysis works, in that it's saying that there is a statistically significant difference between response if the COVID status is positive or negative - in other words, the odds that there's a difference between the groups is a greater than a 50% probability and not likely to be a false correlation, however, as reader of the paper will see that Odds Ratio and conclude that patients with COVID-19 were more likely to have a positive response. So there's our dilemma.
Since you're really good at this science stuff, can you help us figure out how to accurately portray the Odds Ratio without confusing the reader?
I'm studying the effect of inhaled epopoprostenol in patients with moderate to severe ARDS. My primary endpoint is looking for patient characteristics that are associated with a good response (response being a 10% improvement of SPO2 to FIO2 ratio at 6 hours). To do so we are running a risk regression analysis over 24 factors, including SARS-CoV-2 status.
So according to what came through the regression, the only statistically significant patient characteristics were, that for every 10 decrease in SPO2:FIO2 the OR was 0.78 (CI 0.62 - 0.87) and SARS-CoV-2 status. OR 0.42 (CI 0.28-0.57). So, it looks like that 1. If your ARDS is worse, the more likely you are to respond and You're almost 50% more likely to respond if you have COVID-19.
However, after looking at the raw data, 67% of patients without covid had a positive response to epoprostenol, whereas on 55% of patients with COVID-19. So the regression doesn't make sense. Turns out, it's just the way the analysis works, in that it's saying that there is a statistically significant difference between response if the COVID status is positive or negative - in other words, the odds that there's a difference between the groups is a greater than a 50% probability and not likely to be a false correlation, however, as reader of the paper will see that Odds Ratio and conclude that patients with COVID-19 were more likely to have a positive response. So there's our dilemma.
Since you're really good at this science stuff, can you help us figure out how to accurately portray the Odds Ratio without confusing the reader?
Please explain why this message is being reported.
MESSAGE THREAD
In addition to not reading what you posted, you also didn't - dangertim MU - 5/4 10:11:31