We thank Dr. Chen and Dr. Mpody1 for their interest in our article2 and for highlighting important aspects of our study for readers. We concur with the authors regarding percentile categorizations, the nonlinear dose–response relationship, and the identification of cutpoints based on the distribution of the primary predictor. This point was raised during the process of manuscript review, and in response, we provided a sensitivity analysis using a 2-h fasting duration interval to complement the 4-h quartile analysis. Analysis with smaller time intervals could certainly have been performed, yet we eschewed that option because of word count limitations and because the results would likely have remained consistent with the nonlinear relationship shown in the locally estimated scatterplot smoothing curve on the scatterplot.2 As mentioned in the article, we weighed fractional polynomial transformation versus fasting time categorization during the analysis. Our goal was to take into account the nonlinearity of the relationship between clear fluid fasting duration and low blood pressure while providing an interpretable result. While fractional polynomial modeling might have been a more appropriate approach to the nonlinearity, the challenge of a straightforward clincial interpretation remained, as the regression coefficient became difficult to interpret after the transformation.
The authors raise a valid point concerning the open-ended categorization of clear fluid fasting time at a cutpoint of 12 h. Insight might have been gained by extending the sensitivity analysis to greater than 16 or 18 h as long as the sample size in the category remained adequate to provide a reliable estimate of the odds ratio. This relationship between extreme clear fluid fasting durations and low blood pressure may be a topic that we examine more closely in future work.
With regard to anesthesia medications, we agree that propofol, opioid, and neuraxial anesthesia were potential confounders of the association between the fasting duration and low blood pressure during induction. To control for the confounding, anesthesia medications were included in the final multivariable model in table 2 of our article.2 This was done to ensure independence of the primary exposure–outcome relationship from the controlled confounders, such as propofol. A sensitivity analysis separating patients who received propofol from their counterparts who did not receive propofol would indeed allow us to assess whether the association of fasting with blood pressure varied across these two groups of patients. With two different stratum-specific odds ratios, propofol would likely be an effect modifier, suggesting a potential interaction between fasting duration and propofol.3,4 We again thank the authors for sharing their insightful comments, and we look forward to their and other researchers’ efforts to improve our understanding of the relationship between clear fluid fasting duration and blood pressure in children undergoing anesthesia.
We also thank Dr. Sharma and Dr. Naik5 for their interest in our article. We included the American Society of Anesthesiologists (Schaumburg, Illinois) physical status III and IV patients in our analysis to enhance our study’s generalizability, and the results of this analysis are shown in table 2 of our article.2 However, the authors make an excellent point regarding the possible impact of the differences between our study population and the population that was used to generate the blood pressure reference normograms.6 We thank the authors for sharing their astute observation of the discrepancy between the two study populations, which is a limitation that is worth highlighting to readers of our article.
The authors declare no competing interests.