|
Abstract
|
This commentary aims to provide a constructive critique of the findings of this important and interesting study. As described, this study determined, in a cohort of pediatric burn patients, whether long-term scarring outcomes are different in those who had surgical treatment vs. those who were treated conservatively (1). The Brisbane Burn Scar Impact Profile (BBSIP) was used to measure scarring outcomes. Mean scores for each of the BBSIP questions were reported for both groups, and the scores were compared between groups. In the penultimate paragraph of their manuscript, the authors wrote that "no difference was found in the long-term scar outcomes." The problem with this claim is that, for many of the BBSIP outcomes, the evidence does not unequivocally support such a conclusion.
What does the evidence seem to indicate? Table 3 presents the means and standard deviations, by treatment group, for the responses to each of the 57 questions of the BBSIP. For 31 of the 57 questions, the mean response was identical for both groups with no within-group variability. For those questions, all respondents in both groups selected the lowest possible category regarding impact ("not at all"). However, for the other 26 questions, there are differences between treatment groups, both in terms of central tendency and variability. Although the authors note that part 7 of the BBSIP had the most variation, which focuses on physical symptoms, there seems to be at least as much variability in questions from part 1 regarding the impact on the life of a child. For example, for the fourth question regarding scar treatments, the mean for the surgical treatment group is 1.46, which is ∼0.4 scale points higher than the mean for the conservative treatment group. In addition, the standard deviation for the surgical treatment group is 1.2, which is almost 1 point more than that for the conservative treatment group. Because an individual score cannot be <1, these results indicate that some respondents in the surgical treatment group indicated "somewhat," the middle of the five possible ordered response categories, for the level of impact. The distribution of responses for this question and, therefore, the observed impact of treatment modality is not identical between groups.
Why is there a discrepancy between the evidence offered in Table 3 and the prose characterization of the results? The reason for the discrepancy is a common error in interpretation. ...
Language: en |