Helen Macdonald: Should clinicians spend more time thinking about guidelines than evidence?

Old evidence based medicine (EBM) is dead, and clinicians need to get to grips with guidelines as the power tools of EBM. Digitially structured data could improve guidelines. And, with the help of patients and the public, perhaps we will do a better job of improving people’s lives with them. Helen Macdonald sums up what she heard in the opening session from the Guidelines International conference in Manchester, where four speakers talked about why they do what they do.


Gordon Guyatt is a celebrity of the EBM world. So when he suggested that the original model of EBM (where clinicians appraised evidence) is dead, everyone paid attention. (He is a founding father of EBM, a clinician, researcher, and distinguished professor at McMaster University.) The time and skills involved in appraising evidence renders it unfeasible for clinicians. So EBM training is shifting away from critical appraisal, and towards solid capabilities in finding evidence that has been appraised for you. Enter guidelines (complete with a recommendation on what you might do). What the majority of clinicians really need to do is to understand how to assess a guideline, and what say to a patient based on the information that they find.

For example, clinicians need to spot and appreciate the difference between strong and weak recommendations. Most recommendations are weak. Weak recommendations mean that clinicians should anticipate substantial variability in what patients prefer. Whereas in strong recommendations patients are likely to opt for similar choices.

Guyatt also talked about GRADE (a well known and respected methodology for dealing with research and guidelines). It helps you to understand why a recommendation is strong or weak by splitting it into parts; benefit and harms, the quality of the evidence, and the values and preferences of patients. Further, development means that GRADE can also present and consider cost, resources, feasibility, and equity.

But as with any tool, GRADE is only as good as its operators. And no system is perfect. It is also adapting. For example, GRADE rates the certainty of the evidence. But does high certainty mean a) that we are confident a treatment is effective (because the confidence intervals do not touch the line of no difference)? Or does it mean b) we are confident that a treatment is sufficiently effective (for example the confidence intervals do not cross a particular threshold which is deemed clinically important)?

Fergus Macbeth (retired oncologist and guideline developer at NICE) asked what guidelines are really for; quality improvement, education, rationing, coercion? He returned to guidelines in the 1990s. Then, the Institute of Medicine said that they were statements that helped clinicians and patients in particular situations. Scottish developers SIGN suggested that they were optimum care, within resources. Later guidelines were shaped by concepts of evidence based guidelines, and more recently transparency, and patient and public involvement. But whatever their aim, or their methods, guidelines are power tools, he said. And they need to be handled carefully.

Today, most recommendations are based on low quality evidence. So the nub of the issue is how to balance evidence and opinion. This puts people on the guideline committee centre stage. And we understand relatively little about how they do what they do, in the (black) box of a guideline committee. What are the characteristics of a good committee? Who should be on it? Need they meet face to face? How should they make their decision? It is a field ripe for research: “When guidelines are being developed bloody struggles are waged over medical knowledge, all protagonists claim that the ‘evidence’ supports their views, and final recommendations always require fine judgments and compromise,” Macbeth said quoting Gene Feder.

But who is compromising? Who is around the guideline table? In recent times patients and carers have gained a greater role. One of them is Jonathan Senker (carer and chief executive of VoiceAbility, which advocates for people with disabilities). He developed a guideline with people with learning difficulties. Put very simply, he said, “we do what we do to improve people’s lives.”

Guidelines in partnership with patients, carers, and the public are better because they are grounded in people’s lives. It democratises decisions. And patients bring different expertise. For example, people working on a guideline for self harm suggested that good analgesia in emergency departments was important. People undergoing dialysis said, don’t assume that people want it delivered at home.

Having the right people involved is a key part of guidelines. But they also need good information. Evidence is time consuming and costly to prepare. The concept of waste in research is well known, but there is also waste when research is turned into guidelines. One problem is that evidence, particularly data, is hard to move around and repurpose efficiently. It gets stuck in formats and places where it is hard to access.

Linn Brant (researcher and general internist from Oslo) explained how a circular simplification of an evidence ecosystem can help reimagine the journey of data. The problem is that many of the people and organisations that work in this area are in silos. When you try and move data from one silo to another, even if people are keen to help, it is hard. “Text blobs” of evidence are the norm. Technologically minded people imagine a world where these data can be structured digitally, allowing them to be recognised by computers, shared, and recycled. Guidelines are expensive. And reducing the time and effort spent digging for data could be used elsewhere.

Helen Macdonald, clinical editor, The BMJ, and general practitioner.

Competing interests: Helen is The BMJ lead for BMJ Rapid Recommendations. Two of the speakers in this first session, Gordon Guyatt and Linn Brandt, are also involved in this work. The project uses GRADE methodology.