Background

In 2012, the Danish Finance Act appropriated a total of €10.8 mio for the preparation of clinical guidelines. The Danish Health Authority (DHA) was subsequently commissioned to formulate 47 national clinical guidelines to support evidence-based decision making within health areas with a high burden of disease, a perceived large variation in practice, or uncertainty about which care was appropriate [1]. Two of these areas were low back pain (LBP) and lumbar radiculopathy (LR). Consequently in 2014, two working groups were formed with the aim of developing national clinical guidelines for non-surgical interventions for recent onset (<12 weeks) LBP and for recent onset (<12 weeks) LR. The primary target groups for these guidelines were primary sector healthcare providers, i.e., general practitioners, chiropractors, and physiotherapists, but also medical specialists or others in the primary or secondary healthcare sector handling patients with LBP or LR.

An estimated 15% of the Danish population suffers from low back pain (LBP) [2], and most will experience LBP during their lifetime [3], which is in accordance with estimates globally [4]. Both globally [5] and in Denmark [6], LBP with or without LR is a leading cause of years lived with disability, and consequently has major socioeconomic impact on society. For example, out of the 2.9 million Danes in the workforce, those with LBP have 5.5 million more days off work annually when compared to those without LBP, which accounts for 20% of all sick days, and LBP with or without LR is the most common reason for seeing a general practitioner, accounting for almost one in ten visits [2]. In addition, Danes with LBP visit their general practitioner 3.3 times more often compared to Danes without LBP, and they consult approximately 30% more often chiropractors and physiotherapists [2]. Once you have had an episode of LBP, most will experience recurrences [7], and only a minority will stay pain free for longer periods of time [8]. Additionally, 1–10% of patients with LBP will experience LR, which is associated with a poorer prognosis compared to LBP without LR [9].

This paper summarises the two Danish national clinical guidelines, which were published in 2016 as full reports in Danish [10, 11]. The mandates for the two working groups were to make recommendations based on a maximum of ten clinical questions for LBP and LR each. The working groups were not asked to make recommendations for diagnostic procedures or care pathways.

Methods

Study design

The guidelines were based on systematic reviews of the scientific literature and subsequent meta-analyses. The evidence of effect was balanced against the risk of harms and patient preferences to make recommendations related to each of the clinical questions. The method followed international standards for clinical guidelines [12], which were operationalised in a handbook from DHA [13] and briefly summarised below. This method was based on the Grades of Recommendations, Assessment, Development, and Evaluation (GRADE) approach [14]. The full clinical guidelines are available with all supportive material, including a description of the methods on the DHA webpage (in Danish) [13].

Organisation of the work

For each guideline, a project group within DHA consisting of a chairman, a project manager, a lead reviewer, a search specialist, a methodologist, and a multidisciplinary working group with 10 (LBP) and 12 (LR) members was set up. Working group members were appointed by invitation from professional organisations and scientific societies. They were involved in all parts of the process including formulating the clinical questions, selecting literature, data extraction, rating the quality of evidence, and formulating recommendations. Reference groups with representatives from stakeholders from the Danish healthcare system (municipalities, regions and hospitals), and patient organisations discussed and gave suggestions to the clinical questions and feedback on the recommendations. The lead reviewers coordinated the tasks of the working groups and drafted the reports. Potential conflicts of interest were declared by all involved partners and made publicly available on the DHA webpage (in Danish) [10, 11].

Finally, drafts of each of the clinical guidelines were in a public hearing and reviewed by two external peer-reviewers. The comments and feedback were considered by the working groups and taken into consideration when formulating the final versions of the guidelines.

Formulating the clinical questions

The clinical guidelines addressed a maximum of ten clinical questions, which were structured according to the Patient, Intervention, Comparison, and Outcome approach (PICO) framework [14].

Populations

The target populations were: (1) patients above the age of 16 years suffering from non-specific LBP with or without associated leg pain, but no signs of LR, and (2) patients with symptoms and clinical signs of LR above the age of 18 years. The symptoms had to have lasted less than 12 weeks for both populations. It was assumed that the differentiation between non-specific LBP and LR could be based on anamnestic information and a clinical examination without diagnostic imaging. Therefore, no distinction was made between LR caused by disc herniation and other degenerative conditions. Studies were eligible for the clinical guideline on LBP if at least 75% of the participants in a study matched the inclusion criteria. No such cut point was used in the LR guideline.

Interventions and comparisons

The mandate restricted the clinical guidelines to non-surgical interventions, and for the clinical guideline on LR only to non-pharmacological interventions. The choice of clinical questions was based on the working group’s perceived frequency of use, uncertainty about effectiveness, or uncertainty about whether one intervention was superior to another. Because it was assumed that all patients would receive basic information regarding disease progression, prognosis and danger signals, advice on activity and possible medical pain management when seeking care, it was decided to make recommendations about the interventions as a supplement or add-on to this basic treatment with no further specification, hereafter named ‘usual care’. Thus, trials were eligible for inclusion when usual care was provided in both the intervention and the control group, and the intervention in question was added in the intervention group. By doing so, the effects of adding the interventions to usual care were reviewed; if this was not possible, a comparison of treatments or sham-controlled trials was accepted. Some clinical questions addressed a head-to-head comparison of two interventions when it was assumed that clinicians often will choose between the two.

Outcome measures

For each of the clinical questions, two or more primary outcome measures and their timing were chosen a priori by the working groups. For most LBP questions, back pain intensity and back pain-related activity limitation were deemed primary outcomes. Back pain intensity, leg pain intensity, back pain-related activity limitation, and neurological deficits were considered primary outcome measures for the LR questions. For all outcome measures, the absolute differences between intervention and control groups on generally accepted and validated instruments such as a visual analogue scale (VAS), a numeric pain rating scale (NRS), Roland-Morris Disability Questionnaire (RMDQ) or Oswestry Disability Index (ODI) should be available. In the LBP guideline, the pharmacological questions also included as primary outcomes were serious adverse events. Secondary outcomes measures included fear-avoidance, work status, health-related quality of life, study drop outs, recurrence, and surgery rates.

The working groups defined minimally clinically relevant effects as a difference of 15 mm on a 100 mm on a VAS-scale, two points on an 11-point NRS, and 10 points on a 100-point scale of back pain-related disability [15].

Literature searches and inclusion of literature

For each of the clinical questions, the literature was systematically searched in three-steps: Firstly, Embase, Medline, Cinahl (LBP only), Psycinfo (LBP only), PEDRO (LBP only) as well as national and international guideline databases were searched for clinical guidelines 10 years back (2004 and 2005 included, respectively, for LR and LBP). Secondly, Medline, Embase, Cinahl, PsycInfo, and Pedro (LBP only), and the Cochrane Library were searched for systematic reviews 10 years back, and thirdly, the same databases were searched for randomised clinical trials (RCTs) with no lower limit for publication year. In case a high-quality clinical guideline or systematic review would have covered earlier studies, the date for the last search for this review was used as the lower limit for the new search for randomised trials. All literature searches included studies published until and including December 2014 (LR) or March 2016 (LBP) published in English, German (LBP only), Norwegian, Swedish, or Danish. The search terms and strategies are available at the DHA website [10, 11].

Where no RCTs dealing with recent onset LR could be identified, indirect evidence from LR populations with symptoms lasting for more than 12 weeks informed consensus recommendations.

The lead reviewer screened and retrieved titles and abstracts. Potentially eligible papers were then collected in full text. Subsequently, the lead reviewer and a member of the working group independently screened the full text papers for inclusion or exclusion. Disagreements were resolved by discussion until consensus was reached.

Data extraction and quality assessment

The lead reviewer and a member of the working group or a scientific methods advisor independently extracted data for each clinical question and assessed all included papers for quality. If a high-quality clinical guideline or systematic review was available, data were extracted from these. The quality was assessed using the AGREE-II tool [16] for clinical guidelines, the AMSTAR tool [17] for systematic reviews, and the Cochrane risk of bias tool for RCTs [18]. When a risk of bias assessment was available from a Cochrane review, it was transferred directly to the clinical guideline. The handling of references and data extractions were performed using the web-based software Covidence [19], from which data were exported to the RevMan software [20] for meta-analyses; the results of which were further transferred to either GradePro [21] (LR) or MAGIC [22] (LBP) for GRADE assessment. Disagreements in data extraction and quality assessment were solved by consensus between the two evaluators in all instances. The quality of evidence was graded from very low to high according to the GRADE definitions (Table 1) for each outcome. Downgrading was done following standard definitions of risk of bias, inconsistency, indirectness, imprecision, and publication bias [23]. The overall level of evidence supporting the recommendation for each clinical question was determined based on the quality for the primary outcome with the lowest quality evidence.

Table 1 Definitions of Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) adapted from Balshem et al. 2011 [24]
Full size table

From evidence to recommendations

Finally, the evidence was summarised in evidence tables, and forest plots were constructed when meta-analyses were feasible. Based on the available evidence, strong or weak recommendations for or against an intervention were proposed following the criteria outlined in Table 2. Each recommendation was annotated with the strength of the recommendation and the level of evidence according to GRADE. In case no evidence was available from RCTs, a good practice recommendation was formulated based on clinical experience and consensus in the working group. The recommendations were based on weighing the quality of evidence, positive versus negative effects, patient values and preferences as well as the perception and experience of the working groups.

Table 2 Recommendations and their definitions by the Danish Health Authority (DHA)
Full size table

Results

The guidelines considered ten clinical questions concerning LBP and ten concerning LR. Six interventions were covered by both clinical guidelines, two of which were stand-alone interventions (advice to stay active vs rest; routine use of Magnetic resonance imaging [MRI] and/or X-ray vs. no imaging), and four were evaluated as an add-on to usual care vs. usual care (individualised patient information, supervised exercise, acupuncture, and manual therapy). In addition, the clinical guideline for LBP covered three questions addressing pharmacological interventions [paracetamol, opioids and non-steroidal anti-inflammatory drugs (NSAIDs)] as add-on to usual care, and targeted group-based care vs. non-targeted care. For LR, three head-to-head comparisons of exercise and manual therapy interventions (directional exercise vs. neuromuscular control training; directional exercise in combination with neuromuscular control training vs directional exercise alone; supervised exercises vs. manual therapy) were performed. An overview of the clinical questions and recommendations are provided in Table 3.

Table 3 Overview of recommendations and their level of evidence
Full size table

A short description of eligible papers, primary outcomes, recommendations, and levels of evidence are provided in Tables 4 and 5. Forest plots of all outcomes and risk of bias assessments are provided in Supplementary Appendix 1 and 2. Evidence tables are available in the complete clinical guidelines following each clinical question at the DHA website (in Danish) [10, 11].

Table 4 PICO questions, recommendations, definitions of interventions, supporting evidence and comments regarding recent onset low back pain
Full size table
Table 5 PICO questions, recommendations, definitions of interventions, supporting evidence and comments regarding recent onset lumbar radiculopathy
Full size table

Generally, recommendations from the two guidelines endorse patient enablement through information and education, advice to remain physically active and supervised exercise in addition to usual care. For pain relief, manual therapy including joint mobilisation and manipulation in addition to usual care was recommended, whereas the expert groups recommended only using pain medication in the form of paracetamol, NSAID, and opioids in addition to usual care after careful consideration in patients with LBP. No recommendations were made for the use of pain medication in relation to LR because this was outside of the mandate of the group.

Acupuncture was not endorsed for routine use in the two conditions. The groups recommended against routine imaging, i.e., X-ray or MRI, in patients presenting with both recent onset LBP and/or LR, and against the use of extraforaminal glucocorticoid injections in addition to usual care in patients with LR. Finally, it was recommended that patients with LR are referred for surgical consultation within 12 weeks if severe and disabling pain persists despite non-surgical treatment.

Of the 20 clinical questions, none could be answered by any of the clinical guidelines or systematic reviews that were retrieved. Recommendations were based on RCTs when available (16 out of 20 questions) and the remaining on professional consensus (four questions). Flow charts of included literature, quality assessments of clinical guidelines, systematic reviews, and RCTs and evidence tables are available at the DHA website (in Danish) [10, 11].

Discussion

Multidisciplinary expert groups formulated two national clinical guidelines for the DHA covering non-surgical treatment of recent onset LBP and LR in adults and found a striking lack of evidence for the effectiveness of the interventions examined. Thus, commonly used interventions like information and guidance, medication, mechanical diagnosis and therapy, massage, acupuncture, motor control exercise, and spinal manual therapy had either no or limited quality supporting evidence. Consequently, guideline recommendations are to a large extent based on consensus between members of the working groups; therefore, new high-quality trials focusing on LBP and LR patients are likely to impact future guideline recommendations greatly.

Wong et al. reviewed clinical practice guidelines for non-surgical management of LBP with or without LR published between 2005 and 2014 and found that advice and education about self-management and reassurance as well as advice for staying active, supervised exercise, and manual therapy were universally recommended for people presenting to health care professionals with these conditions [108]. They also found that paracetamol, and NSAID were recommended as treatment options in all guidelines reviewed, whereas muscle relaxants, and a short course of opioids were recommended in some but not all guidelines [108]. In 2016, new guidelines for non-invasive treatments for LBP and sciatica were published from The National Institute for Health Care Excellence (NICE) in the UK [109]. These guidelines are more comprehensive than the Danish national guidelines because they also deal with chronic LBP, clinical examination and surgical treatments. However, for recent onset LBP and sciatica they also recommend providing people with advice and education as well as encouragement to stay active and continue with normal activities, to consider group exercises, and to consider manual therapy treatments as part of treatment that also include exercise [109]. With respect to pharmacological treatment, the NICE guidelines are similar to the Danish guidelines when they recommend against routine use of paracetamol as a stand-alone treatment, that NSAID is only to be used after careful consideration of co-morbidities and other risk-factors for side-effects and if used, then only in the lowest effective dose. Finally, they recommended that opioids should not be given routinely for managing LBP or sciatica [109].

Expert groups have used lack of evidence for benefit or harm for a particular intervention as an argument for not putting forward a recommendation [110]. Such interpretation of the evidence; however, has been met with frustration by health care professionals and professional societies who look to expert groups and task forces for guidance [111]. Fortunately, the GRADE methodology accommodates these circumstances as it classifies evidence as either strong or weak and provide interpretations for patients, clinicians, and policy makers [112]. Faced with either no or weak evidence, it is important that patients know that their particular preference among the various therapies should guide choice of intervention. Clinicians must, therefore, acknowledge that different choices may be appropriate for different patients and must help each patient choose a management option consistent with his or her values or preferences. Finally, policy makers must involve all relevant professional groups and stakeholders when determining how best to design care pathways [112]. Importantly, guideline panels should not refrain from making recommendations because individual patients and clinicians will make different choices when faced with a weak recommendation. In fact, this is to be expected. Consequently, the GRADE Working Group encourage panels to make recommendations wherever possible whether they are based on solid evidence or not [113].

Strengths of this national clinical guideline include the chairmanship by the DHA and the rigorous adherence to relevant scientific standards. Furthermore, the guideline working groups were composed of clinicians and academics with a range of professional backgrounds, as well as relevant professional societies and agencies were consulted during the process, which together aims to ensure buy-in by relevant stakeholders in the country. The guideline working groups were assisted by expert research librarians and guideline methodologists. Finally, the guidelines were peer-reviewed by international experts who provided detailed comments which resulted in revisions and clarifications prior to release of the final report. The main weakness of this work relates to the lack of clinical trials in some areas; therefore, the weak recommendations are mostly based on consensus in the guideline working groups. The DHA recommend that the guidelines are updated three years after the publication unless new developments warrant an earlier update.

Conclusion

Two multidisciplinary working groups developed two national clinical guidelines for non-surgical treatment in adult patients with LBP and LR of less than 12 weeks’ duration under the Danish Health Authority. The recommendations are based on limited evidence or on consensus, but are well aligned with recommendations from similar international guidelines. The guideline working groups strongly recommend that research efforts in relation to all aspects of the management of LBP and, in particular, LR be intensified.