Making sense of consensus statements.

We live in an age of ever-increasing quantities of information. In 2014 there were 211 veterinary journals listed on the SCImago Journal & Country Rank Database (SCImago Journal & Country Rank 2014). This is a 50% increase on the listing for 2004, which contained 141 veterinary journals. The number of veterinary publications is increasing at a faster rate. In the biomedical literature search website PubMed.gov, there were 22,770 citations that contain the word “veterinary” in 2014, which is an increase of 87% in the number of citations for 2004. Considering that the average adult learner may take an hour to read eight pages of complex information (Welch 1998), how does any practitioner keep up to date with developments in veterinary medicine? A common answer to this is the literature review. Literature reviews exist in non-systematic and systematic forms. A non-systematic review is typically a summary of a subject, written by an expert in the subject being reviewed. This has been referred to as the “narrative”, “journalistic” or “authoritative” review (Oxman and Guyatt 1993, Greenhalgh 2010). For the purposes of this editorial, I will use the descriptor “authoritative”. Authoritative reviews can be in the form of journal articles, book chapters or continuing education lectures. Typically the expert (or experts) will summarise the relevant literature in combination with his or her own experience in the subject. The method for how studies are identified and selected for inclusion in the review is not presented. The biggest issue with authoritative reviews is that they are prone to bias. If multiple experts are each asked to compose an authoritative review on a subject, it is likely that the content of each review will differ. This difference will depend on the method used for identifying and selecting studies and his or her individual experience and beliefs. For example, even for management of a disease as common as hypertrophic cardiomyopathy in cats, anecdote and beliefs have been shown to result in very different clinical practices among cardiologists and clinicians with interest in cardiology (Rishniw and Pion 2011). This issue of bias in authoritative reviews is not restricted to veterinary medicine. Over 20 years ago, a study by one of the forerunners of evidence-based medicine, Gordon Guyatt, demonstrated that subject experts in human medicine are less likely to produce a consistent and reliable literature review on their subject of expertise than non-experts with training in research methodology (Oxman and Guyatt 1993). The authors of this study concluded that, compared with research methodologists, subject experts “write reviews of inferior quality”, that “the greater the expertise the more likely the quality is to be poor” and that “the poor quality may be related to the strength of their prior opinions”. Furthermore, a different study published at the same time showed that the recommendations of clinical experts in authoritative reviews did not always agree with the results of objective studies (Antman et al. 1992). Faced with these problems, the science of systematic reviews arose as a solution (Mulrow 1994). Experts in research methodology usually undertake systematic reviews, possibly in conjunction with a subject expert. A systematic review is essentially a type of scientific research study in which the data being studied are other published studies. In common with scientific studies, a systematic review contains both a statement of objectives and materials and methods in which a predetermined systematic process for identifying and including a study in the review is described in sufficient detail for the methods to be replicated by another investigator. Included studies are then assessed for methodological quality and weighted accordingly. Statistical methods may be used for systematic reviews that include meta-analyses of primary studies (Greenhalgh 2010). Systematic reviews are now well established in human medicine as the highest level of evidence [Oxford Centre for Evidence-based Medicine – Levels of Evidence (2009)]. They are used to guide health care practitioners, researchers and policy makers (Mulrow 1994) and guidelines exist to ensure optimal reporting (Moher et al. 2009). There are examples of systematic reviews in veterinary medicine, including those in the Journal of Small Animal Practice (Beauvais et al. 2012a, 2012b). However, as highlighted in a previous editorial in the Journal of Small Animal Practice, a systematic review may be unable to provide sufficient evidence to make a recommendation about a subject if there are insufficient clinical studies of adequate quality (Holmes 2012). It is therefore of no surprise that, despite their inherent limitations, authoritative, non-systematic reviews are still the dominant form of subject review in veterinary medicine. This brings us onto consensus statements. In this issue of the Journal of Small Animal Practice, a panel of nine veterinary cardiologists and seven nephrologists (the Cardiorenal Syndrome Consensus Group) report a consensus statement on cardiovascular-renal axis disorders in dogs and cats (Pouchelon et al. 2015). Consensus statements are a form of authoritative review in which multiple subject experts are invited to write a set of statements or guidelines regarding a topic of interest to a profession. Recent examples in veterinary medicine include consensus statements on chronic valvular heart disease (Atkins et al. 2009) and glomerular disease in dogs (IRIS Canine GN Study Group Standard Therapy Subgroup 2013). Consensus statements differ from most authoritative reviews in that multiple subject experts are asked to contribute and there is a predetermined method for developing recommendations and determining whether a consensus exists. In the statement by Pouchelon et al., the method used is the modified Delphi method in which a literature review was undertaken and 13 summary statements drafted by 5 of the 16 members of the panel, the statements rated, discussed and revised by all 16 members and a final round of rating for the revised 13 summary statements undertaken, with the results forming the final consensus statement.