As was clearly outlined in the article, our study was designed to define the minimal effective doses of succinylcholine and rocuronium for a series of ECT treatments. Dr Daoud makes a valid point that for within-subjects comparisons the examined conditions should remain unchanged during the course of a study. To achieve this goal, we used a priori defined crossover design to minimize the carryover effect of the applied neuromuscular blocking agents (NMBAs) and account for within-subject difference between the treatments. In our study of the motor response to the induced seizures (and not the seizure threshold), we ensured that ECT sessions were separated by ≥2 days under the same condition, eliminating the risk for differential carryover effect from previously administered NMBAs and other potential confounding effects. Seizure threshold was not the focus of our study and is unlikely to have affected the intensity of the neuromuscular response to the ECT-induced seizure, and hence the required NMBA dose. Moreover, contrary to Dr Daoud’s statement, seizure threshold does not necessarily increase over the course of ECT treatment. Indeed, in a recent study, Fink et al3 showed that it remains unchanged in the majority of patients (70%), increases in 21% and decreases in 9% at remission. These findings are consistent with earlier publications by, among others, Coffey et al4 that only a fraction of the patients may show a change in seizure threshold over time.
Yet, to verify the absence of any potential confounding effects in our study, we analyzed the administered energy and ECT duration in all patients and all ECT treatments. As was anticipated, our analysis showed no significant difference in the administered energy or its duration between the ECT sessions (Figure, A and B, P values = .36 and .76, respectively). Furthermore, we measured the intrasubject correlation using mixed-effects modeling of all applied doses in subjects for each NMBA (optimal and suboptimal doses) and found very low intrasubject variability, validating the appropriate application of our crossover design.
We thank Dr Daoud for pointing out the typographical error of the pulse width unit in Table 1 (“mA” instead of “ms”). Also, the correct range of this parameter for both succinylcholine and rocuronium is “0.3 to 2.0 (ms),” which was unintentionally rounded by the statistical software to “0 to 2.”
Sample size estimation for the secondary end point of a difference in the duration of block and time to recovery between succinylcholine and rocuronium was based on the assumption of a correlation between the recovery times from the applied NMBAs. Furthermore, mixed modeling was used to examine the correlations within and between subjects’ measurements of recovery over time, which confirmed the independence of observations, most probably because of an appropriate washout interval between treatments. Differences in the mean values for measured variables (eg, recovery time, duration of seizure, and hemodynamic variables) obtained with each NMDA were compared using Welch t test. The Welch test was used to account for the possibility of unequal variances between groups that appeared to be present in the sample data (see Table 1 in original article1). Unless it is known that the variances are equal, a test for comparing means that accounts for the unequal variances t test, such as the Welch test, has been recommended in preference to the Student t test.