What should clinicians do when faced with conflicting recommendations?

Clinical practice guidelines sometimes make conflicting recommendations.1 2 3 4 5 6 7 8 For example, a sore throat may be managed differently in North America, France, and Finland—where guidelines recommend that a diagnostic test should be performed and that treatment should be conditional on its result—than it would be in England, Scotland, the Netherlands, and Belgium—where guidelines recommend that the decision to prescribe penicillin should depend mainly on the severity of symptoms, with no testing.1

Such disagreements occur for both valid and non-valid reasons. Valid reasons include honest differences in the many judgments that go into a recommendation—judgments about which research is relevant; the risk of bias in that research; the applicability of the research findings to the question at hand; and the relative importance of the anticipated benefits, adverse effects, and costs. Non-valid reasons include conflicts of interest, lack of awareness of relevant evidence or ignoring such evidence, failure to appraise the relevant research critically, failure to consider outcomes that are important to patients, and inappropriate valuations of outcomes.

Conflicting recommendations can be bewildering for clinicians and patients. What then should clinicians do when faced with conflicting recommendations?

Simply ignoring guidelines altogether is one solution, but this would be hazardous. Clinicians need good quality clinical practice guidelines. Busy clinicians try to provide the best care they can for their patients but find it almost impossible to keep up to date with the deluge of new information crossing their desks. Good guidelines can synthesise all the research that is relevant to practice and can help with the complex judgments needed to translate that evidence into practice.

Clinicians should be wary of recommendations that have not been developed systematically and transparently. Well informed clinical recommendations require evidence and judgments. Evidence is needed to estimate the consequences of alternative management strategies and, ideally, as a basis for judgment about how patients value those consequences. Judgments need to be made about the evidence (table 1⇓) and about the balance between the desirable and undesirable consequences of adhering to a recommendation (table 2⇓). These judgments are complex and may be difficult, and if they are made informally and non-systematically errors may occur. Such errors include oversights, introduction of extraneous information, too much weight being given to some evidence (for example, personal experience) and too little to other evidence (for example, well designed research that conflicts with personal experience), conflicting interests, heuristics, and biases. Hence, particularly when benefits and harms need to be balanced, guidelines should quantify—in terms of relative and absolute risks—the important benefits and harms of the management options. If such summaries are missing and judgments are not made transparently, it is impossible for others to appraise the soundness of the judgments that were made.

Clinicians must therefore be able to identify guidelines that are systematically developed—that is, those that describe the methods that were used—and that provide the essential estimates of treatment effects. They can take several shortcuts when doing this, such as ignoring guidelines where a clear conflict of interest exists and prioritising guidelines that are linked to systematic reviews. Clinicians should be cautious about non-systematically developed recommendations. They should use or adapt evidence based guidelines for decisions that they make often and that are particularly important. They may also wish to appraise the primary studies themselves.

Although clinicians cannot appraise every guideline that they use, clinicians need criteria to determine whether a guideline has been developed systematically so they can make informed judgments about what to do when recommendations conflict with one another or current practice,11 and why such conflicts exist. Reasons for conflicting recommendations include differences in what evidence was considered, judgments about that evidence (table 1),12 and differences in judgments about the desirable and undesirable consequences of adhering to a recommendation (table 2).9

Recommendations can disagree for many reasons, but usually there are only one or two key ones. In the sore throat example, the main reason for conflict is that the North American, French, and Finnish guidelines consider the prevention of rheumatic fever important enough to prescribe antibiotics, whereas the other European guidelines consider rheumatic fever to be rare and sore throat self limiting and therefore recommend that the severity of symptoms should be balanced against the disadvantages of giving antibiotics. Australian guidelines give different recommendations for different settings according to the risk of rheumatic fever, which is still high in aboriginal communities.

Clinicians can develop the skills they need to appraise guidelines critically in several ways. These include self study, online and non-online courses, journal clubs, and workshops. These skills are an essential clinical tool. As with other clinical skills, practice is necessary and it is probably more effective and efficient to do this in a group than individually.

The bottom line is that clinicians need guidelines and use them all the time, but they should not accept recommendations uncritically. To serve their patients well, they must be able to make informed judgments about which guidelines are appropriate, and what to do when recommendations conflict with one another.