Diagnostic Value of Cholinesterase Activity for Postoperative Delirium

Diagnostic Value of Cholinesterase Activity for the Development of Postoperative Delirium after Cardiac Surgery

With a great interest we read the recent article by Saha et al[1] assessing the diagnostic value of cholinesterase activity for postoperative delirium (POD) after cardiac surgery, published online in the Thoracic and Cardiovascular Surgeon in November 2020. By the multivariate logistic regression analyses, they showed that both a decrease in postoperative cholinesterase activity of more than 50% and an early postoperative cholinesterase activity below 4,800 U/L were the independent risk factors for the occurrence of POD. Their findings have the potential implications, but there are several methodological issues in this article on which we wish to invite them to comment.

First, in statistical analyses, the authors described that the multivariate analysis were performed with the binary logistic regression using a forward stepwise (conditional) model, where significance for entry was set at p < 0.05 and significance for exit was set at p < 0.10. However, the readers were not provided the details of multivariate modeling. Generally speaking, the modeling of multivariate analysis includes several important steps: first, the bivariate analyses of demographics and perioperative variables between patients with and without a POD are performed by the Student's t-test for continuous variables and chi-squared test for categorical variables, as described in Tables 1 and 2 of Saha et al's article. Second, the variables with statistical significance in the bivariate analyses, defined as p < 0.05, are incorporated into the univariate model to assess the multicollinearity among candidate covariate variables. Finally, the covariate variables with large p-values (p < 0.2) in the univariate analyses are included into the multivariate model using POD as the dependent outcome variable for identifying the independent factors of POD, with their p-values, adjusted odds ratios, and 95% confidence intervals.[2] As there are the lack of the step regarding the univariate analysis examining the multicollinearity among candidate covariate variables in Saha et al' study, we are concerned that their results of multivariate analysis would be subject to bias due to multicollinearity. Furthermore, it was also unclear whether the Hosmer-Lemeshow test was used to determine the appropriateness of the multivariate model. We cannot determine whether the multivariate model built in this study has a good fit.

Second, the authors did not clearly describe how they determined the two cutoff values of a decrease in postoperative cholinesterase activity of more than 50% and an early postoperative cholinesterase activity below 4,800 U/L for prediction of POD. We believe that addressing this issue would improve the transparency of this study.

Third, a main purpose of this study is to determine the diagnostic value of cholinesterase activity for the occurrence of POD after cardiac surgery. In fact, this study only confirmed the associations of postoperative cholinesterase activity of more than 50% and an early postoperative cholinesterase activity below 4,800 U/L with the development of POD, but did not assess the diagnostic values of two interested variables. To determine the diagnostic values of two interested variables, after the associations of two interested variables with the occurrence of POD were demonstrated and their adjusted odds ratios for POD were obtained with multivariate analysis, the receiver operating characteristic curve analysis should further be performed. By the receiver operating characteristic curve analysis, the areas under the curve, specificities, sensitivities and positive and negative predictive values of two interested variables for the prediction of POD can be obtained.[3] Based on these results, the readers can determine whether the two interested variables have good diagnostic values for the occurrence of POD.

Note

We have screened our manuscript for plagiarism using the Plagiarism Checker (www.duplichecker.com) and no any plagiarism is found.

Authors' Contributions

All authors had carefully read the manuscript of Saha et al, analyzed their methods and data. BH suggested the comment points and drafted the manuscript. FSX critically revised the comment points and authored the manuscript. LJZS and TT revised the comment points and the manuscript. All authors had seen and approved the final manuscript.

Publication History

Article published online:
17 August 2021

• References

• 1 Saha S, Karaca K, Jebran AF. et al. Diagnostic value of cholinesterase activity for the development of postoperative delirium after cardiac surgery. Thorac Cardiovasc Surg 2020; ••• DOI: 10.1055/s-0040-1716897.
• 2 Katz MH. Multivariable analysis: a primer for readers of medical research. Ann Intern Med 2003; 138 (08) 644-650
• 3 Yin J, Mutiso F, Tian L. Joint hypothesis testing of the area under the receiver operating characteristic curve and the Youden index. Pharm Stat 2021; 20 (03) 657-674