The aim of evidence-based medicine (EBM) from its early days was to provide the appropriate means for making effective clinical decisions, not only for avoiding habitual practice but also for enhancing clinical performance. It is, however, unrealistic to simply assume that the results of research will soon evolve into practice. In this article, when aiming to translate results from research into practice, we focus on certain challenges, which can be broadly categorised into two: how the evidence is generated, and how the evidence is implemented. When focussing on generating evidence, a major barrier to the rapid passage of research into clinical practice is that the ‘practice’ in trials or research settings could be a long way from the setting, circumstances, patient groups and resources available in the daily practice of many clinicians. This is the consequence of the several choices that researchers have to make in designing a trial regarding population, measurement tools and interventions. For implementing the evidence, clinical guidelines appear to be one of the most promising and effective tools for improving the quality of care. Although the importance of implementing clinical guidelines is widely recognised, little is known about the optimal implementation strategy. We present two examples of implementing low back pain guidelines, illustrating that it remains difficult to develop effective implementation strategies. Finally, we discuss some future directions that have been proposed for EBM, which aim to overcome the essential tension between internal validity and external validity (generalisability), specifically in fields where complex syndromes and complex interventions are the focus of attention.
Evidence-based medicine
The aim of evidence-based medicine (EBM) from its early days was to provide appropriate means for making effective clinical decisions, not only for avoiding habitual practice but also for enhancing clinical performance . Since then, the influence of EBM has steadily grown. Although recent definitions of EBM have specifically included patient preferences and clinical expertise , at first sight, the primary focus seems to be on scientific evidence. EBM advocates the use of the best available evidence, which is often defined by a ‘hierarchy of evidence’ in which randomised controlled trials (RCTs) and systematic reviews are considered to be the most valid forms of evidence; opinions of respected authorities are considered to be the least valid form of evidence (see Fig. 1 ). In other words, apparently RCTs and summaries of RCTs constitute the apex of the scientific evidence that supports and underpins the practice of EBM.
From research to practice: a problematic journey
If health-care practitioners claim to base their practice on scientific principles, then on the face of it there should be no problem. Results of original research are peer reviewed by scientific journals, and those that conform to expected standards are published, which raises the question: Can these results be simply taken up by the readers and put into clinical practice? One of the first RCTs for Viagra, for example, was published in 1996 and the drug was licensed for use in clinical practice in 1998. Extensive advertising, costing more than $400 million in 2004, has made Viagra and its newer competitors, Cialis and Levitra, among the best-known drug brands in the United States, and their combined global sales reached about US$2.5 billion per year .
Such a combination of widespread public demand and marketing, geared to meet that demand, is rarely repeated outside the occasional spectacular pharmaceutical or technological advance. Therefore, it is unrealistic to simply assume that the results of research will promptly translate into practice. A review of hand washing as a means to prevent hospital-acquired infections was able to draw on nine RCTs, all of which confirmed benefits . However, audits of hand washing in hospitals continue to identify major shortfalls in such a simple and well-known intervention . Back pain provides another typical example. The first RCT to establish that a 7-day bed rest brought no benefits to people with an uncomplicated back pain episode was published in 1986 . In a survey of UK general practitioners (GPs) and practitioners in 2008, Bishop et al . reported that more than 25% would not advise activity or return to work when presented with a vignette of a patient with simple back pain. Nevertheless, the impact of EBM on the field of low back pain (LBP) is clearly evident. The quantity and quality of research into management of this condition have dramatically increased during the past decade. There are several hundred RCTs of treatment for back pain, and the evidence from these trials has been summarised in numerous systematic reviews. For example, the Cochrane Back Review Group has published over 46 Cochrane reviews pertaining to interventions for back and neck pain (see http://www.cochrane.iwh.on.ca/rev_comp.htm ). These systematic reviews have, in turn, served as the basis for a number of primary care LBP clinical guidelines (see, e.g., http://www.guideline.gov/ ). In the past decade, national evidence-based guidelines for the management of acute LBP in primary care have been published in the United States, the Netherlands, the United Kingdom, New Zealand, Israel, Australia and other countries; however, there still remain many treatments and strategies for which evidence is scarce or uncertain. Consequently, further research is often advocated .
Hence, EBM is rightly at the core of current medicine, in general, as well as in the field of LBP. This holds not only from a scientific or clinical point of view, but also from an ethical point of view: if patients and the society rely on the competency of medical professionals and, based on this competency, delegate various responsibilities to the medical profession, these professionals must ensure that their competency reflects the state of the art of medical science . This may not, however, be so straightforward in practice. Focussing on LBP, for example, challenges can be identified that hamper the overall aim of EBM. In this article, we firstly consider the issue of current clinical practice not reflecting evidence in general. We then argue that there are a specific set of challenges when generating evidence in the field of LBP, which relates to implementing this evidence. Secondly, as evidence-based guidelines are perceived as the link between generating and implementing evidence, we focus on the development, dissemination and implementation of these guidelines. Finally, we focus on the future directions for EBM methodology that have the potential to overcome some of the current problems.
Challenges in using evidence to make informed clinical decisions
Current clinical practice does not reflect evidence
A major barrier to the rapid passage of research into clinical practice is that the ‘practice’ in trials or research settings could be a long way from the setting, circumstances, patient groups and resources available in the daily practice of many clinicians. In an important overview, Rothwell identified this issue of ‘external validity’ as a frequent and major reason why results of research are not taken up in clinical practice. In addition, in designing RCTs for LBP, internal validity often conflicts with external validity, or the generalisability, of the results. This is illustrated in the several choices that researchers have to make in designing a trial regarding population, measurement tools and interventions.
Population
In the quest for homogeneous populations, strict inclusion and exclusion criteria are often defined. This increases internal validity. However, rigorous inclusion and exclusion criteria could be the greatest challenge to generalisability of results. For example, specific subgroups who typically attend clinical practice are often underrepresented in RCTs. A recent systematic review assessing the eligibility criteria of RCTs, published in high-impact general medical journals, concluded that exclusion criteria were not always clearly reported. Moreover, women, children, the elderly and those with common medical conditions are frequently excluded from RCTs. Trials with multiple centres and those involving drug interventions are most likely to have extensive exclusions . For LBP RCTs, in addition to these subgroups, patients with previous back surgeries and co-morbidities are regularly excluded. Patients with strong treatment preferences will not consent to randomisation, a potential threat to the generalisability of trials of specific treatments such as injections, medication and surgery. These findings highlight a need for careful consideration and transparent reporting and justification of exclusion criteria in clinical trials; as such, exclusions could impair the generalisability and, consequently, the implementation of RCT results. This has to be balanced against the need for logical, common-sense extrapolation of results of RCTs when that appears reasonable to do so.
Interventions
Scientific rigour also prescribes that interventions be strictly standardised. However, as a result, interventions might be standardised to such a degree that they no longer represent typical community practice. For example, health-care providers could be trained in specific treatment protocols in a study, or specialised high-tech equipment used in a trial might not be widely available in community practice. Another challenge is complex interventions, where the co-operation and delegation of responsibilities among care providers are key differences between the trial and usual care. Interventions often cannot be blinded, possibly resulting in contamination. Advances in study design are making efforts to tackle these issues. For example, a pre-randomisation design could be considered. The term ‘pre-randomisation’ implies that randomisation takes place before seeking informed consent. Owing to this deviation from the usual informed-consent procedures, pre-randomisation is controversial among medical ethicists, health lawyers, methodologists and clinicians, with opinions varying from total to acceptance only unacceptability under specific conditions .
Outcomes
Scientific rigour also prescribes the use of valid and reliable outcome measurement. As a result, RCTs often include outcomes with no relevance for clinical practice, and little relevance for the patient. An example in Rothwell’s overview is the use of surrogate outcomes in osteoporosis: fluoride has been found to have a beneficial effect on increasing bone density, but when evaluated using a relevant clinical outcome, fracture rate, it turned out that there was not a major decrease in fractures. In LBP, and certainly within the primary care setting, differences between scientifically validated outcome measures and patient-specific priorities might be less dramatic, but still important. For LBP studies, the International Classification of Functioning, Disability and Health (commonly known as ICF) is a classification of health and health-related domains, which might guide in selecting the appropriate outcomes. The ICF, for example, illustrates that work participation can be viewed as an extension beyond clinical recovery alone. In essence, the most patient-relevant outcome should be selected given the goals of the intervention.
Challenges in generating evidence summaries
Systematic reviews base their conclusions on RCTs; therefore, the certainty of these conclusions largely depends on the quality and generalisability of the included RCTs. In general, most systematic reviews on back pain conclude that the majority of the included RCTs have a high risk of bias and, consequently, the conclusions are often not supported by strong evidence. This is frustrating for clinicians who may find it difficult to decide on what is to be done in the absence of strong evidence, distinct from the situation of being provided with clear evidence that something should or should not be done.
Furthermore, the methodology of performing systematic reviews is still under development. The Cochrane Back Review Group, for example, recently adopted the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system to summarise the evidence (see http://www.gradeworkinggroup.org/index.htm for description of the GRADE Working Group and system). Within the GRADE system, the quality of the evidence for a specific outcome is based upon five principal factors: (1) limitations (e.g., due to study design), (2) inconsistency of results, (3) indirectness (e.g., generalisability of the findings), (4) imprecision (e.g., sufficient data) and (5) other considerations, such as reporting bias. In general, the overall quality is considered to be high when multiple RCTs, with a low risk of bias, provide consistent, generalisable and precise data for a particular outcome. The quality of the evidence is downgraded by one level when one of the above-described factors is not met. Although there are still issues that need to be developed further (when to downgrade the evidence, what exactly is considered to be ‘direct evidence’ versus ‘indirect evidence’, etc.), the GRADE system provides a more transparent, positive and promising approach to grading evidence.
Because of the many choices that have to be made in a systematic review, different reviews on the same topic might arrive at different conclusions. For example, Assendelft et al . identified a total of 51 reviews (!) on spinal manipulation for LBP, 17 of which were neutral and 34 positive. The overall methodological quality of the systematic reviews was low. Of the 10 reviews, nine with the best methodology were positive. Other factors associated with a positive reviewer’s conclusion were review of spinal manipulation only, inclusion of a spinal manipulator in the review team and a comprehensive literature search. Strong conclusions, however, were precluded by the overall low quality of the reviews. This example clearly illustrates that using systematic reviews for clinical decision making is certainly not without challenges.
In summary, a first barrier regarding the implementation of scientific evidence into daily practice is the fact that RCTs and systematic reviews do not always reflect daily practice – the issue of ‘external validity’ referred to earlier in the article. Future research methods must therefore find ways of accommodating clinical reality .
Using evidence in daily practice
Despite the increasing number of RCTs and systematic reviews, and a growing awareness of evidence-based practice among health-care providers, implementation of EBM has proved to be difficult. There are several reasons for this, including obvious practical reasons why busy health-care professionals cannot generally be expected to analyse and synthesise original research evidence as a continuous part of their daily clinical activity. Many health-care professionals lack the requisite skills and knowledge to discriminate between low- and high-quality trials or to correctly interpret trial findings. A survey of GPs revealed that most did not understand the fundamental terms used in EBM, such as ‘absolute risk reduction’ and ‘number needed to treat’ . A recent survey by Jette and co-workers of a random sample of 1000 Australian physiotherapists suggests similar problems. Less than 20% of respondents to the survey reported that they completely understood the terms ‘relative risk’, ‘absolute risk’, ‘odds ratio’ and ‘meta-analysis’, and only 55% reported that they were confident in their critical appraisal skills . Therefore, publishing research findings in the scientific literature is important to inform research colleagues, but it is unlikely to have a great impact on daily clinical practice.
Secondary sources of information may be useful for busy health-care providers who lack the time, inclination or expertise to approach and appraise the original literature. Examples are the Critically Appraised Papers (CAPs) sections that are published in several journals, usually in one page or less; hence, they potentially provide a significant time-saving mechanism for busy health-care providers. These CAPs distil the key findings of high-quality RCTs (and other types of studies; e.g., Cochrane systematic reviews); however, as they often cover only one specific topic their impact and utility for clinicians are limited. Moreover, underpinning clinical practice on a single study is at odds with the basic principles of EBM.
Evidence-based guidelines
The development and implementation of evidence-based guidelines appear to be one of the most promising and effective tools for improving the quality of care . The basic theory underpinning guidelines is the idea that if the research is synthesised and disseminated by respected national authorities, it will lead to changes in clinical behaviour . However, development and dissemination of guidelines do not necessarily imply that health-care providers will read, understand and use the guidelines. Passive dissemination of information is generally ineffective, and specific implementation strategies are necessary to establish changes in practice. Systematic reviews have shown that a clear and strong evidence base, clear messages, consistent messages across professions, clear sense of ownership, communication with all relevant stakeholders, charismatic leadership, continuity of care, continuous education and continuous evaluation are all required ingredients for implementation of guidelines . Furthermore, some programmatic evaluations suggest that economic and behavioural incentives might also be required to significantly impact guideline compliance . These measures are probably more effective at an individual provider level than at an institutional level .
Guidelines for the management of LBP in primary care have been developed in various countries and, recently, the European guidelines for the management of LBP have been issued. To ensure an evidence-based approach, the recommendations of the European guidelines were based on Cochrane reviews (and on other systematic reviews if a Cochrane review was not available), additional trials published after the Cochrane reviews and existing national guidelines. As such, these guidelines can be seen as a summary of the most recent evidence. Despite the fact that these recommendations seem to be widely endorsed, there is evidence that, in general, the adherence to these recommendations is rather disappointing. For example, in the Netherlands, the physiotherapy guideline for acute LBP recommends a restricted number of sessions; however, a national registration-based study concluded that this was only met in 17% of the patients. Treatment goals were formulated with a focus on activity level, with interventions fostering this activity, in only 53% of patients . Bishop et al . studied adherence to the guidelines by general practitioners and physiotherapists in the UK. They used a vignette of a patient with non-specific LBP (NSLBP) asking questions about recommendations for work, activity and bed rest. Although the majority gave advice that was more or less in line with guideline recommendations about activity and bed rest, over 25% of professionals recommended that the vignette patient should remain off work. They also found that the attitudes and beliefs of health-care practitioners were significantly associated with the work advice they offered for the vignette patient. Practitioners with stronger biomedical and weaker behavioural treatment orientations were more likely to report advice, regarding work, which was ‘not in line with clinical guidelines’. Bishop et al. discuss the complex nature of the clinical consultation, and the fact that general practitioners view sickness certification as a potential threat to the doctor–patient relationship. This may be a potential barrier for following the guidelines regarding work.
In summary, adherence to evidence-based guidelines is certainly suboptimal, with room for improvement. One challenge is to identify the optimal performance indicators for measuring adherence to guidelines. Moreover, one can question what the realistic ‘room for improvement’ really is. What are the norms/benchmarks that one should use in measuring guideline adherence, knowing that the benchmark will never (and maybe never should) meet 100%?
Implementation of clinical guidelines on LBP
Although the importance of implementing clinical guidelines is widely recognised, little is known about the optimal implementation strategy. In the Netherlands, two studies were conducted that assessed the effectiveness of tailored interventions (multifaceted strategies) to implement LBP guidelines .
Engers et al . randomised general practitioners to an intervention or a control group. The general practitioners in the intervention group received tailored interventions, consisting of the Dutch LBP guideline for general practitioners, a 2-h educational and clinical practice workshop, two scientific articles on LBP management, the guideline for occupational physicians, a tool for patient education as well as a tool for reaching agreement on low back care with physical exercise and manual therapies. The control group received no intervention. The results showed that the multifaceted intervention strategy only modestly improved implementation for certain portions of the recommendations in the Dutch LBP guideline by general practitioners and produced only small concomitant changes in patient management. The implementation strategy produced fewer referrals to therapists during follow-up consultations.
Another implementation study in the Netherlands focussed on physiotherapists . They assessed the effectiveness of an active strategy, which consisted of a multifaceted programme, including education, discussion, role playing, feedback and reminders, for the implementation of the clinical guidelines. Physiotherapists in the control group received the guidelines by mail only, with no further intervention. The results showed that the active strategy improved adherence to the guidelines. Physiotherapists in the intervention group more often: (1) correctly limited the number of treatment sessions for patients with a normal course of back pain, (2) set functional treatment goals, (3) used mainly active interventions and (4) gave adequate patient education. However, the magnitude of the effects was only moderate. The evidence that back pain guidelines are not yet successfully impacting a wide audience, including all aspects of clinical practice, underlines the evidence from these trials of implementation – namely that it remains difficult to develop effective implementation strategies. There is no magic bullet.
In the literature, several barriers have been identified: the practice behaviour of health professionals can be influenced by a lack of knowledge, a shortage of time, disagreement with the guideline content or reluctance from colleagues to adhere to the guideline. Furthermore, health professionals might ‘get lost’ in the large number of different guidelines available. One useful framework for identifying components of guideline implementation, where barriers might be occurring, was proposed by Pathman et al. : awareness, agreement, adoption and adherence. Recent work in primary care on LBP guidelines has highlighted that most general practitioners are aware of their national guidelines and also most agreed with their scientific content (e.g., refs. ). The problem with adoption was that the practitioners felt that they were faced with more complex problems in daily practice than was assumed by the guidelines; that patients’ priorities might not coincide with those assumed by the guideline; and that there was a tension between the standardisation of care required by the guidelines and the individualisation of their problem often required by patients consulting with back pain . Moreover, lack of resources often reduced practical capacity to adhere to the guidelines. A priority in translating evidence into practice is identifying barriers to the behavioural change of health professionals. However, patients also have specific views or beliefs and these may not correspond to the care proposed in the guidelines. Therefore, recently, consideration of preferences and values of patients has been advocated for the development and implementation of clinical practice guidelines. It should be noted, however, that there are profound differences in the meaning of ‘considering preferences’ and the underlying rationale to support such policy. Recent work by Boivin et al . starts to shed some more light on this issue.
Another initiative that aims to improve the implementation of guidelines are the recommendations for guideline development that have been published by the Appraisal of Guidelines Research Evaluation (AGREE) collaboration (for more information see http://www.agreecollaboration.org/ ). In short, guidelines should have a clear and strong evidence base and be based on systematic reviews. Guidelines that are not based on sound scientific evidence might effectively implement the wrong evidence. In addition, there should be an explicit link between recommendations and evidence. Messages should be clear, specific and unambiguous; inconsistent recommendations across health professions can be confusing. Therefore, messages of the various health-care providers involved in the management of LBP should be consistent. Communication with all relevant stakeholders (patients, professional organisations and policymakers) is also important for successful implementation. Guideline committees should include representatives of all relevant stakeholders. In addition, all stakeholders should have the opportunity to comment on the guidelines before publication, such that stakeholders have a clear sense of ownership of the guidelines. However, this is only a start as economic incentives may also be required to achieve compliance with guidelines.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
