The challenge of using randomized trials in wound healing

In industrialized countries, about 1 per cent of the population suffers from a non-healing wound, which counts for a significant part of the healthcare budget1,2. There is, consequently, increasing interest in the quality of available evidence for the effectiveness of specific interventions, technologies and dressing materials used in wound healing3,4. Evidence-based medicine uses current best information to make decisions about patient care. The highest-quality evidence is generally thought to come from randomized controlled trials (RCTs) and meta-analyses of these RCTs. This concept has been challenged in the management of chronic non-healing wounds5–8. These articles indicate major controversy about the best way to evaluate the effectiveness of interventions in this complex patient population. The situation is compounded by the way in which regulatory bodies in various countries differentially advise on study design and how results are interpreted. Many of the difficulties and controversies associated with using RCTs in wound healing research are highlighted by a recent publication in BJS comparing two treatment strategies in patients with venous ulceration9. The key elements of rigour associated with a RCT were all fulfilled. Randomization was achieved, there was a power calculation, inclusion and exclusion criteria were defined, a CONSORT diagram included and endpoints identified. The authors introduced pragmatism into the study by allowing some elements of local clinical practice, such as the use of a range of silver dressings, different types of compression and variation in key elements of condition management such as debridement. Despite this, the study still failed to reflect common practice, because of the restrictions of RCT design. The intervention remained the same throughout the 12 weeks of study, regardless of the progress of the wound towards healing. Wounds were not infected or colonized despite manufacturers' recommendations for the use of silver dressings under these circumstances. Routine use of silver dressings also requires that exudation must be present for silver to be released, but what happened when the clinical condition of the wound changed and the amount of exudate decreased? How pragmatic can a trial in wound management become and still result in confidence that an intervention has been evaluated rigorously without a significant methodological flaw? There are further issues relating to appropriateness of the population and endpoints chosen to test the value of a specific intervention under specific circumstances that are particularly relevant to wound healing. To evaluate antimicrobial therapy, the population must have either infection or a high probability of developing infection. The condition of the patients and their wounds needs to be described in such terms. It could be argued that endpoints should relate to the supposed mechanism of action of the intervention under investigation, such as resolution of infection, episodes of infection and/or recurrence of infection. Is intact skin the best endpoint to evaluate the outcome of antimicrobial therapy? Because of the economic burden associated with chronic wound management, healthcare costs are frequently considered. When studying health economics in a RCT, a basic problem is that the studies cover the resources that are fixed and based on the protocol (these are predefined by the protocol) as well as those that are flexible, based on the choice of management dictated by the clinical team. The specific items to be included can result in an underestimation of total resource use as the analysis focuses on limited items such as dressings or visits. In the recent study9 alluded to, the unit cost of some dressings was higher and the frequency of dressing change (analysis not included in the report) would inevitably influence cost. Items included in the analysis will clearly influence the estimate of cost. How much resource use reflects routine care, compared with the resources used as dictated by the protocol, could not be assessed in the paper and is a common feature in wound healing studies. Another problem with the RCT approach relates to ‘intention-to-treat’ analysis for clinical outcomes but use of only a subsample for other analyses. In the case of the cited article9, cost analysis related to those patients who completed a EuroQol—Five Dimensions questionnaire, although the authors recognized that there was over-representation of healed patients who completed the questionnaire resulting in an underestimation of costs. This highlights the difficulty of economic analysis in the context of a RCT. These issues reflect the ongoing challenges that exist for those working in wound healing research. Researchers need to address the requirements of a high-quality RCT as demanded by regulatory authorities, yet produce evidence that is of relevance and generalizable. In response to this confusion, the European Wound Management Association (EWMA) has established a working group (Patient Outcome Group) with the objective of implementing revised pan-European evidence recommendations for clinical data collection in this patient group10. In the context of non-healing wounds, it is hoped that this will be achieved by identifying barriers related to creating and implementing evidence-based recommendations, creating consensus on the best way of evaluating the effectiveness of treatment strategies, proposing revised recommendations for evaluating treatment strategies and routine clinical data collection, and, finally, by implementing this consensus by participation in public debates and influencing national policy. These goals may not, however, be achieved solely on the basis of RCTs. The authors thank Professor Patricia Price, Cardiff, executive member of the EWMA Patient Outcome Group, who made an invaluable contribution to finalizing this manuscript. The authors declare no conflict of interest.