Statistics from Altmetric.com
The study by Simanowski et al described some important features of rectal proliferation and alcohol abuse (Gut 2001;49:418–22). However, there are some methodological issues pertaining to the study which need clarification. Firstly, when performing multiple linear regression, it is essential to perform and report sample size and power estimate calculations. This omission, especially with a sample size of only 39 patients, leaves the reader wondering if this sample is sufficient in size and power to adequately support the conclusions drawn from their regression analysis. Furthermore, by not reporting a r2 or an adjusted r2 value, the accuracy of the model is also not addressed. Possible correlations between independent variables should be investigated and discussed when reporting multiple regression results to further support the validity of the analysis.
Secondly, clarification of their patient populations is also required. They originally reported a cohort of 27 heavy drinkers (23 males, four females) and 12 control patients (five males and seven females) in the early paragraphs of the materials and methods section. Later, the authors discuss “rectal biopsies of 17 alcohol abusers (10 males, seven females) and 14 age matched controls (six males, eight females)”. Obviously not from the original cohort based on the different number of female patients and not referred to in any of the figures, the origin of this second group is unclear.
In summary, clarification regarding the above mentioned omissions would greatly solidify the conclusions of their research.
We appreciate the interest of Dr Filion which gives us the opportunity for additional clarification.
As the effect of alcohol on colonic cell proliferation was found to be significant (p<0.05), no type 2 error with respect to the effect of alcohol has to be considered. In this context it should be noted that in case of statistically significant findings, only type 1 errors may occur. The effect of alcohol on cell regeneration was the primary question which was investigated in the study. As stated in the methods section of the paper, a multiple regression analysis was performed to assess possible confounders due to sex and smoking. Thus the p values reported for sex and smoking should only be interpreted in a descriptive manner.
On the basis of numerous epidemiological studies it is generally accepted that the independent variables alcohol, smoking, and sex do correlate. This is in fact the reason for performing an adjusted analysis on the impact of alcohol on cell regeneration.
In 27 heavy drinkers and 12 controls, statistics on proliferative cell nuclear antigen (PCNA) expression were performed. In a second group of 17 alcoholics and 14 age matched controls, various staining procedures were performed, including Ki67, Rb1, Bcl2, p53, and cytokeratins, without statistical analysis.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.