Background: When performing a systematic review, it is possible that a study could be included even if it has a serious methodological flaw which can render the results implausible. These 'fatal flaws'can include excessive attrition or a large proportion of drop-outs and withdrawals. However, there is no consensus regarding which methodological flaws can be considered 'fatal’, and what we should do with trials which have them.
Objectives: The aim of this study is to identify, via consensus among a panel of experts, which features of randomised controlled trials (RCTs) should be considered 'fatal flaws' and to develop recommendations for dealing with them. The focus for the study will be RCTs on interventions for low-back pain.
Methods: A modified-Delphi technique using three survey rounds was used to select and reduce a list of methodological characteristics. Experts in low-back pain, systematic reviews, and methodology were invited to participate in the study. These experts were asked to individually rate the list of characteristics according towhether they consider the item to be a serious methodological flaw that could render the results of a study implausible. During three surveys, participants were given the opportunity to provide reasons for their decision, to add items, or to modify the wording of items. Consensus was considered to be reached when the rating of at least two-thirds (67%) of the panel members indicated agreement with an item on a five-point scale.
Results: The respondent panel consists of 68 experts. At present, the first two rounds of the Delphi technique have been completed and the third round is underway. Final results will be presented at the colloquium.
Conclusions: The identification of serious methodological flaws and how to deal with trials which have them may lead to improved validity of the results in systematic reviews.