Abstract
In the article by Yersiz et al. (1) published in the Journal on 27 October 1995, the authors report a study on the effect of donor age on immediate and long-term liver allograft function, as well as eventual outcome. Their study consisted of two cohorts, one including 95 transplants from donors older than 50 years (group I), and another composed of 50 recipients transplanted with grafts from donors aged 20-30 years (group II) (1). Although the text mentions that group II was a matched cohort, there is no information regarding what variables were used to match these groups, nor on the method employed. In their analysis of posttransplant graft function, the authors performed pairwise comparisons between several variables (serum glutamic oxaloacetic transaminase, serum glutamic pyruvic transaminase, lactate dehydrogenase, prothrombin time, and bilirubin) that were measured on multiple occasions (postoperative days 1, 3, 10, and 30, and at 1 year). These comparisons were done by multiple applications of the two-sample t test, without any procedure to control the family-wise error rate. This is inappropriate. Not only is there a problem with unadjusted multiple comparisons, but the data are also correlated; a repeated-measures analysis of variance should have been used (2). Figure 2 is a plot of the expected proportion of graft failures as a function of donor age, based on a logistic regression model, and shows that the risk increases continuously throughout the range. If we examine the χ axis we can see that the plotted range goes from 30 to 62 years. We find this rather curious since the study, by design, does not include any donors aged 30-50, so there are no observations on which to base 63% of the plotted range. To fit a model using two arbitrarily selected, discontinuous sets of observations, and then use it to make predictions within the range for which there are no data, is to stretch statistics well beyond the breaking point. The authors seem to believe that it is permissible to interpolate back to the younger donor group, presumably making the assumption that this variable is linear in the logit scale. In logistic regression, or any linear model for that matter, this is not an assumption that can be made lightly (3), even in the presence of data; the authors would be well advised to refrain from doing so. Donor age just happens to be an example of this pitfall. We recently described how the effect of donor age on the outcome of liver transplantation is definitely nonlinear, being negligible until after age 45 (4). Although we agree with the main message of the article-using older donors is indeed a calculated risk-we are puzzled by the way the authors arrived at this conclusion. It is stated in the text that after performing stepwise logistic regression, using both donor and recipient variables, the only independent predictor was postoperative bilirubin (on postoperative day 10). If this is the case, what is the significance of donor age? Howard R. Doyle1; Ignazio R. Marino School of Medicine; Pittsburgh Transplantation Institute; Pittsburgh, Pennsylvania 15213
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.