Coping and depression in diabetic foot ulcer healing: causal influence, mechanistic evidence or none of the above?

Abstract

To the Editor: We read with interest the article by Vedhara and colleagues on coping, depression and healing of diabetic foot ulcers [1], as we believe the investigation of psychosocial aspects of foot ulceration is important and understudied. However, we were disappointed that claims of ‘influence’ and ‘mechanistic evidence’ in the title and conclusions of the paper do not appear to be supported by the data presented. Although problems regarding the reliability of depression and avoidance coping measures, and small sample sizes for secondary analyses are noted, other important methodological problems are not. First, the presented data provide no mechanistic evidence, but merely demonstrate that indicators of cortisol and matrix metalloproteinase (MMP) are associated with ulcer healing. It is not reported whether these potential mediators were associated with the predictors in question (i.e. depression and confrontational coping), or whether controlling for the purported mechanisms attenuated the relationship between the predictors and outcome [2]. As the authors had these data at their disposal, they could and ought to have evaluated the relationships between depression and coping, and cortisol and MMP, which they allude to in their discussion. Restricting their analyses to those with complete data on these variables would also have addressed the possibility that the reported differences in the associations between confrontational coping and depression, and ulcer healing and change in ulcer area may have been due to chance variations in the relationships between these variables across quite different samples (e.g. n=93 versus 21). It is also unclear why the authors stress a relationship between the pro-form of MMP2 and ulcer healing (the direction of which is misstated in the abstract). Wound fluid levels of the active form of this extracellular protease would be of functional significance, yet crucially no difference in active MMP2 levels was observed. Detailed histological analysis (e.g. in situ zymography) would have clarified this discrepancy. Importantly, depression was not associated with ulcer healing, but with change in ulcer area; the question of whether cortisol or MMP indices were related to change in ulcer area was not addressed. Thus, no data presented link these potential ‘mechanisms’ to depression or change in ulcer area. Second, the analytical approach could have affected important analyses. The authors note that there were no a priori hypotheses regarding which of three coping styles would be associated with their primary or secondary outcomes. There were also no predictions about which of five indicators of cortisol or four indices of MMP would be related to the two ulcer healing measures. Lack of hypotheses or control for familywise error may substantially increase the chance of spurious findings [3]. Over-fitting, or having too many predictor variables for a given sample J. S. Gonzalez (*) Ferkauf Graduate School of Psychology, Yeshiva University, Rousso Building, 1300 Morris Park Avenue, Bronx, NY 10461, USA e-mail: jeffrey.gonzalez@einstein.yu.edu