Syed Ahsan Ali ( Department of Medicine, The Aga Khan University Hospital, Karachi. )
Thank you Dr. Yakoob for taking interest in our article. Having received your feedback and critique, we would like to respond to the comments.
Firstly, we agree that our data had both before and after values of ALT for patients with severe dengue hepatitis. However to clarify, we did not do a matched before-after analysis as the objective was instead to compare the before Paracetamol ALT values of those who improved with the before values of those who worsened, and similarly the ALT values after Paracetamol administration values for both those who improved with those who worsened in order to demonstrate a statistical relationship. In such an un-matched situation, we believe the Mann Whitney U test is valid.1 Also to mention that our choice of the non-parametric test was based on the visual assessment and statistical test for data normality. This non-parametric nature of data is also reflected in the descriptive statistics, where we reported the median instead of the mean. We regret that this detail was not included in the final manuscript.
Secondly, we appreciate your suggestion for a multi-variable analysis to account for an enriched statistical analysis, however, the authors resorted to a descriptive analysis of this retrospective data due to a small sample size, which could only be obtained after a comprehensive screening of over 2000 relevant files. Any original associations that may exist are likely to be masked as a function of the small sample size, reflecting a type II error.2
Our patient data had limited number of pertinent variables for a robust analysis, but presents a descriptive picture that provides groundwork for larger, prospective and preferably multi-center studies that are adequately powered to detect associations. The authors encourage this way forward as an objective approach for future research.
1. McKnight PE, Najab J. Mann?Whitney U Test. Corsini Encyclopedia of Psychology. 2010.
2. Hosmer Jr DW, Lemeshow S, Sturdivant RX. Applied logistic regression: John Wiley & Sons; 2013.