Indian Journal of Pharmacology Home 

[Download PDF]
Year : 2015  |  Volume : 47  |  Issue : 5  |  Page : 572--573

Author Reply

B Sharanabasayyaswamy 
 Department of Pharmacology, Sri Dharmasthala Manjunatheshwara College of Medical Sciences and Hospital, Dharwad, Karnataka, India

Correspondence Address:
Dr. B Sharanabasayyaswamy
Department of Pharmacology, Sri Dharmasthala Manjunatheshwara College of Medical Sciences and Hospital, Dharwad, Karnataka

How to cite this article:
Sharanabasayyaswamy B. Author Reply.Indian J Pharmacol 2015;47:572-573

How to cite this URL:
Sharanabasayyaswamy B. Author Reply. Indian J Pharmacol [serial online] 2015 [cited 2023 Feb 5 ];47:572-573
Available from:

Full Text


I would like to thank the author for his extensive analysis of the multivariable regression analysis used in my study. While I agree with some errors noted by the author in our study but I wish to clarify my views regarding the EPV to be considered in logistic regression. I agree that as per rule of thumb derived from the simulation study for logistic regression, at least 10 EPV are required for the minimum outcome. EPV for the minimum outcome, as calculated on the basis of 10 subjects per variable, is not correct for logistic regression. The author further explained that "11 subjects out of 31 were survived on the end of the study. Thus, minimum outcome event is survival at end of the study. Therefore, the survival benefits outcome had only 5.5 EPV as author included two risk factors in the model."

It is further stated that the rule of 10 EPV, in fact, is not a well-defined parameter especially in logistic regression; as opined by Vittinghoff and McCulloch.[1] These authors opine that the "problems are fairly frequent with 2–4 EPV, uncommon with 5–9 EPV, and still observed with 10–16 EPV." Further they also opine that "When a statistically significant association is found in a model with 5–9 EPV, only a minor degree of extra caution is warranted, in particular for plausible and highly significant associations hypothesized a priori." However, the author suggests that if even the low risk of problems seen with 5–9 EPV is unacceptable, modern resampling tools can be used to validate the model-based inferences to derive bias-corrected CIs.

Considering our study results which is supposed to have used 5.5 EPV and had very high significant association between use of intravenous fluids and survival benefits, it would have been better that we should have used modern resampling tools and stated that the results of the study should be interpreted cautiously. In fact we stated that this observation of very high significant association between use of intravenous fluids and survival benefits may be controversial in the background pathophysiology of volume overload in cirrhosis of liver, and it needs to be further analyzed in randomized trials. We also stated that the observed strong correlation between improved survival benefits and use of IVFs can also be explained on the drawback of uncertainty over diagnosis of hepatorenal syndrome (HRS) with regard to abiding to IAC diagnostic criteria of "No improvement of serum creatinine level after at least 2 days of diuretic withdrawal and volume expansion with albumin." This means that the patients who were included in study might not be strictly considered as the patients of HRS and it is possible that these patients perhaps would have recovered from renal failure and survived, provided they had undergone optimum volume expansion. Hence, the significant correlation between use of intravenous fluids and survival benefits would be explained on this basis. We would like to believe that our study has directly or indirectly highlighted the role of optimum volume infusion in patients suspected of HRS.


1Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol 2007;165:710-8.