We thank Drs. Kumar and Jain1  for their interest in our study on the association between intraoperative use of phenylephrine, compared to ephedrine, and delirium after noncardiac surgery under general anesthesia.

We appreciate the authors’ remarks emphasizing the importance of our secondary and sensitivity analyses from a clinical perspective: Drs. Kumar and Jain raise concerns about variability in provider choice of phenylephrine versus ephedrine. In our institutions as in many others, some providers may use ephedrine, and others use phenylephrine as their “weapon of choice” when treating hypotension. We have emphasized this in our secondary analysis investigating provider variability in vasopressor preference—adding the primary anesthesia provider as a random effect in a mixed-effects logistic regression model did not alter the primary findings. To further characterize provider variability, we have now used random-effect variances and model residuals to calculate intraclass correlation. This measure quantifies variability in the use of ephedrine versus phenylephrine that may be attributed to preference of the individual provider. Intraclass correlation demonstrated that in our model, anesthesia providers accounted for an additional 11.5% (95% CI, 10.4 to 12.8%) of the total variance in using ephedrine versus phenylephrine.

Second, Drs. Kumar and Jain note that while both drugs are typically administered as bolus, phenylephrine, as opposed to ephedrine, may also be given as an infusion. In addition, the authors write that based on a previously reported dose potency ratio of 81:1 in patients undergoing elective cesarean delivery under spinal anesthesia and given the observed doses of phenylephrine and ephedrine in our study, much higher equivalent doses of vasopressors were required in patients receiving phenylephrine, potentially introducing bias due to a higher degree of hypotension requiring treatment. This notion emphasizes the importance of our sensitivity analysis excluding patients who received doses of phenylephrine that are in line with an infusion, which confirmed the robustness of our primary analyses. In addition, when given as a bolus, the duration of action of ephedrine is three times as long as that of phenylephrine and higher equivalent doses in the phenylephrine group may further be explained by more required repetitive doses due to a shorter duration of action. We would like to emphasize that blood pressure levels, as shown in figure 2 of the article, were actually slightly higher among patients receiving phenylephrine than in those receiving ephedrine. While the notion stated by Drs. Kumar and Jain that ephedrine itself may cause delirium and psychosis is indeed provocative, it is important to note that the cited article does not contain information as to whether this may occur at clinically relevant doses rather than at high-dose intoxication. 

Most importantly, as elaborated by Dr. Gaskell et al. in their accompanying editorial discussed with Dr. Lam in an associated podcast, and stated in our conclusions, we agree with the authors that our study provides a strong justification for a randomized controlled trial as basis for treatment recommendations. In the meantime, our findings argue for careful consideration of vasopressor choice depending on specific clinical context and patient characteristics.