Comments on “Effect of oral care protocol with versus without chlorhexidine on mortality of mechanically ventilated patients: Prospective randomized controlled trial”

Dear Editor, We would like to provide constructive feedback on a recently published article by Singh et al.[1] The primary outcome of this study was to compare the mortality rate of patients receiving oral care with and without chlorhexidine, but the conclusion drawn emphasizes the incidence of ventilator-associated pneumonia, which was one of the secondary objectives. Even in the results section, the authors have discussed the mortality rate between the two groups after presenting their secondary outcomes. Hence, the conclusion drawn is not in line with the selected primary objective. Furthermore, the sample size was calculated using a reference article that reported an observed mortality rate between the intervention and control groups as 23.5 and 21.2%, respectively.[2] The authors did not mention the probability of α-error and the power of the study used to derive the required sample size. The normal standards are to use an α-error of 5% and the power of 80%.[3] By applying the above standards and mortality proportion from the reference article, the required sample size would be 10,808 (5404 per group). Even if we consider 10% margin of error as the α-value quoted by the authors, the required sample size would be 7934 (3967 per group), which is much higher than the number of patients recruited. Hence, this trial is significantly underpowered to analyze the primary outcome. The second row of Table 1 shows duration since hospital admission to intensive care unit transfer (in days), and the authors have expressed the value as mean with standard deviation (SD). The authors have also acknowledged that they have used the Kolmogorov–Smirnov test to check for the normality of the data. However, when we compare the mean with SD values, we can notice that the SD is exceeding the mean value suggesting that the data were non-normally distributed.[4] In such cases, the data should have been analyzed using Mann–Whitney U test and presented as median with interquartile range. Similarly, pregnancy was one of the exclusion criteria of this study, but the bar diagram comparing the presence of comorbid conditions [Figure 2] includes patients of this cohort in data analysis, which shows protocol deviation. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.