Are Procalcitonin Measures a Reliable Predictor of Stopping Antibiotics Among Patients With Sepsis?—Reply
In Reply We designed the ADAPT-Sepsis trial as a superiority study, addressing a prioritized research funding brief and international evidence gaps. Our sample size estimate, type I error rate, and power were largely driven by a clinical effectiveness outcome (total antibiotic treatment duration to 28 days after randomization) and not safety (28-day all-cause mortality). The resulting sample size estimate allowed us to measure safety with a noninferior 28-day all-cause mortality margin of 5.4%, the narrowest reported to date, as indicated by Dr Bosch and colleagues. We primarily reported trial safety for a reduction in total antibiotic duration for the PCT group compared with standard care using this approach. The 95% CI for the difference in 28-day mortality did not exceed 5.4% and crossed 0. Therefore, we have shown that PCT is not worse than standard care and that there is no difference between PCT and standard care. Furthermore, there were no differences in the survival curves to day 28 (Figure 2B in our article) or day 90 (eFigure 7 in the article’s Supplement 3). Unlike Bosch and colleagues, we do not believe that implementing daily PCT monitoring will result in excess mortality given the totality of our reported trial data and the body of open-label trial evidence. However, given our concealed trial intervention methods, we suggest that future systematic open implementation of our PCT protocol should be informed by evidence (inducing adherence data) to be presented from an in-trial process evaluation.
In Reply We designed the ADAPT-Sepsis trial as a superiority study, addressing a prioritized research funding brief and international evidence gaps. Our sample size estimate, type I error rate, and power were largely driven by a clinical effectiveness outcome (total antibiotic treatment duration to 28 days after randomization) and not safety (28-day all-cause mortality). The resulting sample size estimate allowed us to measure safety with a noninferior 28-day all-cause mortality margin of 5.4%, the narrowest reported to date, as indicated by Dr Bosch and colleagues. We primarily reported trial safety for a reduction in total antibiotic duration for the PCT group compared with standard care using this approach. The 95% CI for the difference in 28-day mortality did not exceed 5.4% and crossed 0. Therefore, we have shown that PCT is not worse than standard care and that there is no difference between PCT and standard care. Furthermore, there were no differences in the survival curves to day 28 (Figure 2B in our article) or day 90 (eFigure 7 in the article’s Supplement 3). Unlike Bosch and colleagues, we do not believe that implementing daily PCT monitoring will result in excess mortality given the totality of our reported trial data and the body of open-label trial evidence. However, given our concealed trial intervention methods, we suggest that future systematic open implementation of our PCT protocol should be informed by evidence (inducing adherence data) to be presented from an in-trial process evaluation.
