Evidence-based medicine helps health care professionals and patients decide best care, drawing on research about effectiveness and safety of interventions. Systematic reviews summarise the evidence; guidelines report consensus between experts (including patients) on interpreting it for everyday practice. Although guideline recommendations are only one component of shared decisions that will vary patient-to-patient, the hoped-for outcome is health benefit for each individual. Guidelines also inform starker decisions by policymakers and health care leaders—for example, when to withdraw approval or funding for a poorly evidenced or harmful intervention. To assess whether all this research-driven activity is useful, 2 questions need answering: how well are guidelines followed in real-life practice and do patients benefit in the long-term?
In a new systematic review, Kamper et al. tackle the first question in relation to first-contact care for patients with low back pain provided by family practice and emergency department physicians. As the authors state, low back pain has major significance for the international pain community. It is the leading single cause of years lost to disability globally, and there is good evidence for what constitutes best first-contact treatment. The review selected best-quality studies of routine health care data to investigate whether first-contact physicians are putting back pain guidelines into practice (“usual care”). The results paint a bleak picture: only a minority of patients apparently receive simple positive messages to stay active and exercise, while inappropriate use of analgesia and imaging persists. The review adds to evidence that the care doctors give patients with low back pain is dominated by guideline-discordant interventions that are unnecessary, expensive, and “low-value” (ie, harm is more likely than benefit).
These findings are not entirely unexpected nor, as the authors point out, should blame be placed entirely at the door of front-line physicians. Qualitative studies reveal family practitioners agreeing with guideline conclusions but frustrated in implementing them by factors such as patients with complex needs or lack of resources for behavioural interventions and rehabilitation. Furthermore, physicians will point to the nuances of individual patients who do not neatly fit the template for guideline recommendations. The review's authors acknowledge this when discussing the difficulty of measuring “appropriateness” of physician requests for imaging.
The review did not study nonphysician primary care providers. Many such practitioners deliver care aligned with guidelines, such as advice about activity and exercise. More allocation of first-contact low back pain care to professionals such as physical therapists could help to address the problem of inappropriate care. However, although some back care, such as imaging requests by physical therapists, may be more guideline-concordant, a recent systematic review indicated physical therapists have low adherence generally to guidelines for musculoskeletal pain, and a comparison of nurse practitioners and physicians in primary care revealed similar rates of unnecessary imaging requests.
It is increasingly clear that simply expecting individual clinicians to adhere more closely to guidelines is not going to close the evidence-practice gap. Recent articles have argued that effective “high-value” care for patients with low back pain (“benefit more likely than harm”) will only be achieved through large-scale top-down changes across health and social systems (“system strengthening”). Such changes include engaging policymakers, politicians, and profession leaders to change laws and reimbursement practices; addressing counterproductive commercial pressures; creating incentives for optimal care; providing adequate resources for delivery of guideline care (including digital innovation); shifting professional education and training toward high-value care, including nonpharmacological approaches; dissolving boundaries between health care professions to create unified programmes of care; and changing culture and pain literacy among patients and populations to encourage prevention and positive health activity. The success of such innovations, however, cannot be taken for granted and will need evidence that they bring long-term patient benefit.
There is a second important issue highlighted by the authors of the review—the scarcity of relevant information to answer their question “how well are back pain guidelines implemented?” The review focused on studies reporting actual clinical care because it was recorded in health care or insurance databases. A challenge for data collection systems, particularly in family practice and emergency departments, is how to ensure such records include all aspects of care. Prescription medicines, imaging requests, and referrals to other services tend to be routinely recorded, whereas the nature and detail of patient assessment, and the content of advice and information given, often go unrecorded. Absence of data may mean either care has not been delivered—or it has been delivered but not recorded. Thirty years after randomised controlled trials reassured patients that an average low back pain episode did not require a week in bed, there is no high-quality data on what primary care physicians have been advising about bed rest.
This absence of information about relevant areas of the primary care consultation should concern us as much as the size of the evidence-practice gap or how to close it. Having the right information will itself help change policy and training and behaviour about guideline treatments, for example, by easing the path to audit as a means to improve practice. The “finely grained” information the review calls for (such as linking prescription records with the indications for prescribing) would address concerns that health care databases do not currently convey all the nuances of decision-making—not all imaging requests are wrong, not all opioid prescriptions are wrong-headed.
The authors of the review could find no high-quality studies using actual practice data from first-contact low back pain care in low- and middle-income countries (LMICs). This is of particular concern because LMIC populations are being exposed to technology and treatments for low back pain (such as opioids) that high-income countries have popularised and made profitable but that are often inappropriate and harmful. Low- and middle-income countries need appropriately strong systems to support delivery of high-value care, including guideline-concordant approaches to low back pain prevention and care that align with local practices and cultures of health and wellbeing. Achieving data collection systems and routine analysis to document low back care is a plausible investment target for LMIC populations such as Nepal that are digitally equipped even when geographically remote but will be hindered while pain research has no priority and remains focused on low-value items of care.
It is a strange situation to contemplate. In an information-dominated world where people's purchasing actions are instantly known, retrievable, linked to other data, and acted upon, the health care most people are receiving for the world's leading cause of long-term disability is not known because it is either not recorded or the data are not accessible or reported. The challenge to digital science and modern informatics is how to record and retrieve this information without overloading already busy clinicians or saturating patients with data collection. But, along with the challenges and the risks, there is the potential for new technology to change and improve delivery of health care globally for conditions such as back pain. The case revealed by Kamper et al. for having more accessible, better organised, well-resourced, easily collectable information about the daily content and outcomes of consultations for this most common of disabling pain conditions is clear. The international pain community, in collaboration with health care data scientists, should get behind it.
The authors have no conflicts of interest to declare.