Q2 #AnesJC. Some thoughts on the results:

The functional form of the continuous relationship between fasting time and log-odds of hypotension is not what I would have expected. Mechanistically you would expect a more monotonic function with increasing fasting times.

1/n https://twitter.com/Duke_Anesthesia/status/1305536787390042112

I suspect, as the authors raise in the discussion, that there may be some non-observed cofounding by indication that is driving the seemingly neutral effect of a moderate duration of fasting (8-12 hrs). #AnesJC

2/n

I would be curious to see the results from the fractional polynomial modeling, as in general continuous predictors are best modeled as continuous variables rather than discretized categories. Also curious if restricted cubic splines were trialed as well? #AnesJC

3/n

I think the general conclusion would likely be very similar, but the coefficients from the categories of fasting times become difficult to interpret clinically for patients near the boundaries of each category. #AnesJC

4/n

Finally, I would be curious to see the variance estimate for the random intercept in the mixed effects model. How much of the variability in risk for hypotension did attending anesthesiologist account for? #AnesJC

5/5