CONSORT 2010: progress and challenges

Abstract

State of the literature Poor reporting of the medical literature has been recognized as a serious problem for decades [1,2]. This has a multitude of economic, policy and clinical implications [3]. Worldwide, each year over US$100 billion is spent on health research. This research, in turn, is relied on heavily by clinicians and policy-makers alike to make decisions about patient care. Randomized controlled trials (RCTs) have been the focus of particular scrutiny because of their direct impact on health care. For instance, systematic reviews of RCTs, the widely-recommended vessel for determining efficacy for health care interventions, are often severely and frequently hampered in what they conclude because of inadequate reporting of included RCTs [4]. When authors fail to adequately describe RCT methods (i.e., how treatments should be administered or what outcomes were measured, when and how) and results, their data are of limited use to reviewers, and in turn, cannot be used to make the decisions for which they were intended. Ultimately, the money spent on funding such research goes to waste when their reports do not accurately and/or completely reflect what was done. This is not only unfortunate; some argue, it is unethical, immoral and unacceptable [3,5]. In order to implement successful treatments and thereby, replicate successful outcomes in practice, clinicians need to be presented with complete descriptions about treatments, know which outcomes will be affected and have all the information available to help determine whether bias was present that may affect study findings. As obvious as this may seem, particularly in today’s age of evidence-based healthcare, the majority of RCTs are still largely incompletely reported. In a sample of 80 RCTs and systematic reviews published over a 1-year period (October 2005– October 2006), Glasziou et al. found that only 49% of trials sufficiently described the details of the interventions (i.e., timing, dose, duration and frequency of treatment, route of administration and any monitoring used) [6]. More recently Duff et al., found that the methodological reporting of a sample of RCTs from major oncology journals was inconsistent and deficient in describing at least ten items about cancer therapies deemed essential by oncologists applying them [7]. Only 11% of 262 trials sampled reported all ten items completely. Due to the complex nature of cancer and other treatments, it essential that studies evaluating them provide all the necessary information a clinician might need to implement successful treatments in practice. Reporting of trial outcomes has also been found to be incomplete and inconsistent. A recent systematic review, based on 16 studies, each comparing a median of 54 RCTs to their available protocols or registry entries, found that up to 50% of trial reports omitted, introduced or changed at least one primary outcome from what was stated in the protocol [8]. The review also identified a number of other common