Statistics from Altmetric.com
Commentary on the paper by Riordan et al
It has become axiomatic that high quality health care requires application of the best available evidence in the context of the individual patient’s situation. Medical schools and residency training programmes are required to provide training in critical appraisal of the literature, and no self respecting guideline would claim to be other than “evidence based”. In spite of this wide acceptance of evidence based medicine as the right thing to do, it is clear, from studies such as the one by Riordan et al in this issue, that we are just not quite there yet.1 These investigators wondered whether “best paediatric evidence” was accessible and used by on-call doctors working at inpatient paediatric and neonatal units. What they found was perhaps predictable: the sources they defined as “best paediatric evidence” were generally accessible, but they were not often used.
Other studies suggest that the problem is widespread: only a minority of Canadian internists reported using evidence based information sources,2 and similar results were found on a survey of family practitioners in New Zealand.3 Fewer than 5% of Australian general practitioners had ever used the Cochrane Library in 1999.4 Insufficient time, inadequate skills, and limited access to evidence are the most commonly cited reasons that physicians give for not seeking and using evidence more consistently.5
The practice of evidence based medicine has been conceptualised as a five step process: recognising information needs and describing them in well formulated clinical questions; efficiently finding information; critically appraising the information; applying the information to the individual patient; and evaluating the outcomes.6 If this is the EBM process, and it seems like EBM is not being practiced, then what is the rate limiting step? This study and many others address step two: accessing information.
What information do doctors need? Therapeutic interventions dominate the questions that relate to medical knowledge.7 Is information available to support decision making in paediatrics? In spite of dismal predictions about how little evidence there is to support medical practice, studies in paediatric settings show that from 47% (in a community paediatric setting) to 75% (in an inpatient paediatric setting) of interventions are supported by good quality evidence.8,9
What are the best sources of evidence for paediatric practice? Riordan et al asked on-call doctors about access to Medline, the Cochrane Library, paediatric journals, and local guidelines. Although Medline is a rich source of information, and can be tailored (by the use of the “Clinical Queries” feature) to select for higher quality evidence, it is nonetheless a giant database for which searching is time consuming and often frustrating. Information that is already critically appraised and synthesised will provide more rapid access to best quality evidence. The Cochrane Library is one such source; its abstracts are freely available on the internet, and the large majority of subjects in the current study had access to the entire Cochrane Library, at least for some part of the day. For neonatologists, the Cochrane neonatal reviews are available in full text on the NICHD’s website at http://www.nichd.nih.gov/cochrane/. Clinical Evidence, a resource that is question driven, very concise, and readily available to primary care practitioners in the UK and the United States, may provide even more rapid access to high quality synthesised information relevant to primary care. Nearly 100 topics in Issue 9 (June 2003) are relevant to paediatric practice (www.clinicalevidence.com). Other resources, such as UpToDate, Archimedes, and Evidence-based On-Call are also readily available on the internet.
Not having enough time is one of the most consistent reasons given by doctors for not practicing evidence based medicine. Studies of the amount of time it takes to find evidence show that going to the primary literature, in particular, really is very time consuming—ranging from 43 minutes to 4.5 hours for skilled searchers.10,11 However, the increasing number and quality of sources of high quality synthesised evidence are ameliorating the complaint that it takes too much time to find information. Using predigested sources of information, many common questions can be answered very quickly.12
Even if they can find evidence, do doctors recognise the best quality evidence? Studies of information sources that practising clinicians actually use suggest that most do not. They answer their questions by reading review articles or textbooks, or consulting with colleagues.2,13 Medical students and residents can be trained to search for and critically appraise evidence from the medical literature,14 but they do not seem to carry this into clinical practice.
So which is the rate limiting step? Evidence is available, much of it (and this will only increase) on the internet; the internet is progressively more available to anyone at any time. I believe that the rate limiting step is not availability or accessibility of evidence, or even the skills of critical appraisal. The rate limiting step is the first one: practitioners don’t really recognise the extent of their information needs. In 1985, Covell and colleagues compared self reported information needs with information needs of the same physicians when they were observed in practice.13 The study physicians hugely underestimated (by more than a factor of 10) the number of questions that would arise in practice.
This problem will be addressed not by haranguing physicians to ask more questions, but by—painlessly and quickly—providing the information they need at the point of care. In fact, when physicians identify their own information needs, the majority of what they need is patient related clinical information—information from the history, the physical examination, or from diagnostic studies.7 Relevant evidence from scientific studies should be provided along with patient information just at the time that the knowledge could be applied. Christakis et al randomised physicians to receive or not receive an evidence based message about antibiotic use in otitis media at the point of entering a prescription into an electronic system. The doctors who received the information were significantly less likely to prescribe antibiotics than those who did not.15 Clearly, the advance of information technology with the emerging (slowly in some places!) electronic medical record, electronic physician order entry, and intelligent systems that can recognise what information might be needed in a given situation has great potential to increase the use of evidence in practice. This kind of technology is still uncommon in inpatient settings and rare in outpatient settings, due in part to high start-up costs and lack of standardised (computer) language.
Guyatt and colleagues have identified several levels of evidence based practice: the “doer” of evidence based medicine, who has the time and skills to construct appropriate questions, search efficiently, and critically appraise and synthesise information; the “user” of evidence based medicine who recognises information needs and seeks information from secondary sources; and the “unconscious user” of evidence based medicine who practices in an environment of strategies that target specific clinical behaviours.16 It is clearly naïve to assume that all paediatricians will become “doers” of evidence based medicine. The current study suggests that a great deal of time and effort will be required to increase the number of “users” of evidence based medicine.
Medicine is a combination of art and science. We must constantly refine the art by listening to and learning from our patients, and we must constantly update the science by recognising that our knowledge is time limited and seeking updated information every time we need it. Technology can help move the process forward—information offered at the point of care, information sources available and easily searched on the practitioner’s wireless handheld computer, and increasing availability of synthesised information will all help. But underneath it all is our awareness of our limitations and knowledge of the ways to overcome them in order to provide the very best care we can.
Commentary on the paper by Riordan et al