Abstract
Iread the study by Zhong et al. with great interest as the study addresses one of the most important complications of flexible ureteroscopic lithotripsy (FUL)—systemic inflammatory response syndrome (SIRS). 1 However, the study seems to have an important statistical flaw on performing multivariate logistic regression analysis.
The authors identified sex (p < 0.001), stone size (p = 0.001), irrigation flow rate (p < 0.001), and irrigation volume (p < 0.001) as factors associated with development of SIRS in the univariate analysis, whereas the groups of patients with and without SIRS were similar for rest of the parameters including use of small caliber ureteral access sheath (UAS) (p = 0.907). 1
The authors further performed a multivariate logistic regression analysis to determine the independent predictors of development of SIRS and stone size (odds ratio [OR] = 1.691; 95% confidence interval [CI], 0.879–3.255), small-caliber UAS (OR = 2.293; 95% CI, 0.730–7.200), irrigation flow rate (OR = 1.161; 95% CI, 1.096–1.230), and struvite calculi (OR = 3.331; 95% CI, 0.971–11.426) were identified as independent risk factors for SIRS after FUL. Normally, for multivariate analysis, it is suggested to test all variables by using models with just one variable at a time (univariate models) and further involve variables with a significant p value in the multivariate model. 2 This was also mentioned by the authors in the methods section.
Small-caliber UAS and presence of struvite stone were not associated with development of SIRS in the univariate analysis, but they were further included in the multivariate analysis and reported to be significant. The p values for these parameters in the multivariate analysis results were not mentioned in the article and 95% CI levels for the two parameters were distributed in both sides of 1, which occurs in case of statistical insignificance. Therefore, I believe that the methodology and interpretation of the results are inappropriate and should be clarified.