Letter to “Predictors of ischemic bowel in patients with incarcerated hernias”
- 166 Downloads
I read the manuscript entitled “Predictors of ischemic bowel in patients with incarcerated hernias ” with interest. But, I think that there is a major flaw in the presented article.
Multivariate analysis is a tool for determining the relative contributions of different causes to a single event. A common problem in the analysis is the selection of those independent variables in a multivariate analysis that might influence the outcome variable. In most cases, researchers are cognizant of the effects of leaving out an important variable from the model, so they will collect data on any variable they think might be important. With moderate numbers of variables, fitting all possible models is not computationally feasible. The most popular algorithms are stepwise methods. Stepwise methods are easy to compute and, therefore, are extremely popular. There is no agreement on the best criterion for the addition and deletion of variables in a stepwise procedure. Ideally, such a threshold will exclude noise variables and include relevant ones. There is no consensus on what that level should be, and suggestions range from using a p value of 0.01–0.50. Different strategies will have important effects on the final model selected [2, 3].
It is not easy to understand why do the authors choice skin changes and hyponatremia as independent variables for multivariate analyses? Other variables such as WBC count (p = 0.001), bicarbonate level (p = 0.04), and creatinine level (p = 0.005) were not included the multivariate analysis in the study: I want to ask: Why?
In conclusion, due to the major flaw, results of the study must be interpreted cautiously.
Compliance with ethical standards
Conflict of interest
There is no conflict of interest to declare.
For this type of study, Ethical approval was not necessary.
Human and animal rights
This article does not contain any studies directly involving human participants.
For this type of study, formal consent was not necessary.
- 3.Thompson WR (2009) Variable selection of correlated predictors in logistic regression: investigating the diet-heart hypothesis. The Florida State University, TallahasseeGoogle Scholar