Clinical pathways for the management of low back pain from primary to specialised care: a systematic review

Clinical pathways for low back pain (LBP) have potential to improve clinical outcomes and health service efficiency. This systematic review aimed to synthesise the evidence for clinical pathways for LBP and/or radicular leg pain from primary to specialised care and to describe key pathway components. Electronic database searches (CINAHL, MEDLINE, Cochrane Library, EMBASE) from 2006 onwards were conducted with further manual and citation searching. Two independent reviewers conducted eligibility assessment, data extraction and quality appraisal. A narrative synthesis of findings is presented. From 18,443 identified studies, 28 papers met inclusion criteria. Pathways were developed primarily to address over-burdened secondary care services in high-income countries and almost universally used interface services with a triage remit at the primary-secondary care boundary. Accordingly, evaluation of healthcare resource use and patient flow predominated, with interface services associated with enhanced service efficiency through decreased wait times and appropriate use of consultant appointments. Low quality study designs, heterogeneous outcomes and insufficient comparative data precluded definitive conclusions regarding clinical- and cost-effectiveness. Pathways demonstrated basic levels of care integration across the primary-secondary care boundary. The limited volume of research evaluating clinical pathways for LBP/radicular leg pain and spanning primary and specialised care predominantly used interface services to ensure appropriate specialised care referrals with associated increased efficiency of care delivery. Pathways demonstrated basic levels of care integration across healthcare boundaries. Well-designed randomised controlled trials to explore the potential of clinical pathways to improve clinical outcomes, deliver cost-effective, guideline-concordant care and enhance care integration are required.


Introduction
Low back pain (LBP) is a leading cause of disability globally [1] and exerts a considerable burden on the individual, healthcare systems and the wider socio-economic milieu. As worldwide health expenditure continues to grow [2], healthcare systems are challenged with containing costs whilst ensuring quality of care. Despite numerous clinical guidelines for LBP that synthesise the best available scientific evidence with generally consistent recommendations, the evidence-practice gap remains. A demedicalised approach that emphasises self-management, physical and psychological therapies in primary or community care settings, with a minority of patients requiring referral to specialised services, where management may include injections or surgery, is generally recognised as optimal [3,4]. LBP accompanied by radicular leg symptoms is considered to have a favourable natural course, with most people also responding to conservative management in primary care [5]. Despite the consistency of numerous guideline recommendations, implementation remains problematic with many patients receiving guideline discordant care [3,4]. Clinical pathways have the potential to enhance both quality of care by facilitating translation of research into practice and cost-effectiveness through optimisation of resource use [6]. They are structured multidisciplinary care plans describing essential steps in care provision to patients with a specific condition and endeavour to optimise clinical outcomes and efficiency through linking clinical practice and best evidence [7]. In 2018, the Lancet Low Back Pain Series highlighted clinical pathway redesign from first contact through to specialised care as a potential measure to support healthcare systems and organisations in best-practice implementation with improved outcomes for patients and service providers [3]. Such care integration across the primary-secondary care continuum aims to address care fragmentation, reduce duplication, and improve clinical and cost effectiveness [8].
A previous systematic review completed ten years ago outlined examples of clinical pathways in LBP management and the evidence for their success; pathways that did not include all types of mechanical LBP were excluded and only two studies provided effectiveness data [9]. Given the growth in research regarding LBP clinical pathways and integrated care, a further systematic review is now warranted and will include all studies with outcome data of clinical pathways for LBP and/or radicular leg pain that captured the patient journey from first contact through to specialised care. This review, therefore, aims to: (1) identify and describe pathway components; (2) determine how LBP care is managed and coordinated across different levels and traditional boundaries of healthcare provision; (3) summarise and evaluate the evidence for the identified pathways.

Methodology
This review was reported in accordance with Preferred Reporting Information for Systematic reviews and Meta-Analyses (PRISMA) guidelines [10]. The protocol was prospectively registered in PROSPERO (CRD42021237824). Four databases, CINAHL (via EBSCO), MEDLINE (via Pubmed), Cochrane Library, and EMBASE, were searched to identify all relevant publications between January 2006 and February 2021, a timeframe considered adequate to reflect international LBP guideline publication as circa 15 primary care clinical LBP guidelines were published between 2008 and 2017 [11] and the European Guidelines for the Management of LBP were published in 2006 [12,13].
The search strategy used three groups of keywords, 'clinical care pathway', 'low back pain' and 'lower limb radicular pain' (Online Resource 1). Additional publications were identified through manual searching of reference lists and electronic searching for citations associated with included studies. Search results were exported to Covidence (systematic review software). All full-text studies published in English pertaining to evaluation of clinical pathways for LBP and/or radicular leg pain for adults ≥ 18 years that address health care provision from first contact in primary care through to specialised care were included. Studies describing more general musculoskeletal pathways were excluded.
Titles and abstracts were considered against inclusion/ exclusion criteria by two review authors (CM/CC), followed by full-text screening by two independent review authors (CM and CC/HF). Methodological quality of included studies was independently evaluated by two review authors (CM and CC/HF) using the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool for Quantitative Studies [14] or the Critical Appraisal Skills Programme (CASP) qualitative studies checklist [15]. A bespoke data extraction template was devised by the authors based on identified outcome domains in clinical pathway literature: (1) healthcare resource use, (2) patient flow, (3) clinical outcomes, (4) patient and clinician satisfaction and experience, with an additional section for any outcomes outside of these domains. Data were extracted using this template by one review author (CM) and subsequently verified for accuracy by a second review author (CC/HF). At all stages, betweenauthor discrepancies were resolved initially by discussion, and if required, by consultation with a third author (GM). Extracted data were collated in summary tables to facilitate analysis. A narrative synthesis of findings is presented as meta-analysis was precluded by the heterogeneity of the research.

Literature Search
The PRISMA flow diagram (Fig. 1) shows that initial database searching yielded 18,443 studies, with 13 studies identified through hand and citation searching. Following removal of duplicates and application of inclusion/exclusion criteria, 27 full text published papers and one service evaluation report were included in the final analysis. Three publications related to the same research study [16][17][18], including a Health Technology Assessment (HTA) report [16], RCT [17] and a qualitative study [18].

Design and Quality
Study designs varied with 18 observational cohort designs [19, 20, 22, 24, 25, 27-32, 34-36, 40-43], two audits [38,39], two pre-test post-test designs [26,33], one RCT [17], one mixed methods study [21], one clinical case series [37], and two qualitative studies [18,23] included in the review (Table 1). On quality assessment (EPHPP), the quantitative studies were typically rated as weak evidence; one was rated strong overall [17] and six studies were rated moderate [19,20,22,26,29,35]. Using the CASP checklist, both of the standalone qualitative studies [18,23] were considered to have appropriate methodological quality to address the stated aims, whereas the qualitative methods of the mixed methods service evaluation report [21] were not adequately described. Table 2 summarises the main limitations of the individual studies. Table 2 outlines the key features of each pathway, whilst Fig. 2 graphically illustrates the key patient contact points. The reasons stated for pathway introduction are outlined in Table 3. Pathways were commonly accessed via the patient's GP, although all Canadian pathways began with a more overarching 'primary care provider' category of health professional and Whedon et al. [42] used a primary spine practitioner, in this instance a chiropractor, as the first point of contact. Whilst a substantial number of studies provided minimal detail regarding primary care management, others  [19,20] Mixed methods:

Pathway Features
(1) Same cohort as [19] (1) Determines return rate to pathway post discharge from triage & treat (1) and (2)  Reports on design, implementation and initial evaluation of BAC, a pathway providing community-based specialist review of neck referred to the use of stratification or subclassification in primary care or provided some detail on treatment/onward referral guidelines [17-21, 23, 27-30, 33, 39]. All but one study [33] used non-consultant clinics with a triage remit between first contact in primary care and consultant-led clinics in secondary care. These clinics are consistent with the NHS definition of 'interface services', namely 'any service, excluding consultant-led services, that incorporates any intermediate levels of triage, assessment and treatment between traditional Primary Care and Secondary Care' [44]. Physiotherapists were the most common professional group in these interface services, working independently [17, 18, 22-29, 35, 38, 40], or alongside other professionals [19-21, 30, 34, 37, 41, 43]. Figure 2 illustrates that interface services were community-based [23-26, 30, 39-42] or hospital-based [19-21, 34-37, 43], although the location was not always clear, and three studies used hospital-led telehealth-assisted triage [31,32,38]. Interface services were principally accessed via direct referral from primary care, although in a small number of pathways the referral was first sent to a secondary care consultant service and subsequently triaged to an interface clinic [31,32,34,36,38,41,43] or both direct and indirect referral was available [24][25][26]35]. Twelve pathways (17 studies) [17-21, 23, 27-30, 34, 35, 37-39, 41, 43] provided guidance regarding referrals appropriate for interface assessment, but this guidance varied considerably from detailed criteria [41] to broad statements, such as 'non-urgent' referrals [38]. Subclassification, guiding interface service management, featured in four pathways [27-30, 37, 39]. Interface services differed in the reported clinical management options available, although most had access to advanced imaging ( Table 2). The delineation between primary care and interface services was blurred in two pathways [37,42]; for example, the primary spine practitioner was a first contact practitioner but also accepted referrals from other primary care providers in an interface capacity [42].
The detail provided regarding guidance for appropriate referral for consultant opinion varied across studies; some studies cited ill-defined criteria, such as clinical judgement, urgent referrals or patients potentially requiring surgery [35,36,38,43], whilst others provided more specific guidance [22, 23, 27-30, 33, 34, 37, 39, 41]. In one study all routine referrals to consultant secondary care clinics had first to be reviewed by a community-based spine specialist team [39]. The sole study without an interface service used a fast-track consultant appointment pathway if referral from GPs was preceded by reported adherence with conservative care guidelines [33]. The management of 'red flag' presentations was not documented by all studies. Some authors described direct referral pathways for patients presenting in primary care with red flag symptoms to urgent/emergency specialised review [27-29, 33, 39], or on referral triage in secondary  [31,32,34,41,43]. In another study, all triage clinicians were trained in red flag identification and these patients were escalated to an appropriate medical appointment at the first level of triage [37].

Outcome Domains
A narrative synthesis utilising the four pre-specified outcome domains, healthcare resource use, patient flow, clinical outcomes and patient/clinician satisfaction and experience, is presented; Table 2 contains the main results of the individual studies.
1. Healthcare resource use Outcomes related to this domain were the most frequently examined, reported in 21 studies. Rates of independent management or discharge from in-person interface services were consistently high (69-92%) [27,34,40,43], with consultant appointments post-triage required in 1-30% of patients [24,26,27,30,34,35,37,40,41,43]. Conversion rates from consultant appoint-ment to an intervention of between 40 and 97% were reported, with conversion rates to surgery most frequently assessed [24,27,28,30,34,35,40,43]. Few studies provided comparative data, although the higher proportion of surgical candidates amongst patients referred for surgical consultation through SSP triage clinics than the conventional route, supported the use of interface clinics to triage appropriately to surgical clinics [28]. Data regarding investigation use did not provide useful comparisons due to cohort variability; for example, studies reported on investigation use in triage clinics [37,40,43], on MRI use in patients referred for surgical consultation [30], in interface service compared to consultant clinic [41], or in primary spine care model compared to usual care [42]. Two studies reported the effect of specific measures on pathway fidelity [33,39], for example, a 'bounce back' process from secondary care consultants to community interface for referrals not following the local spinal pathway [39]. Six studies considered cost of care with variable methodology, including costs of telehealth triage compared to rural outreach clinic [38]; implementation costs of a shared care guideline [33]; patient cost savings from telehealthassisted triage [32]; MRI costs [41]; per-person primary care expenditure compared with usual care [42]. The SCOPiC RCT presented a comprehensive cost analysis, reporting that stratified care for sciatica was more costly than usual care [17]. Four studies reported use of healthcare services, but in various ways: likelihood for ED visits and hospitalisations [42]; use of non-surgical treatments [29]; percentage of patients having no further treatment after one year [22] and rate of return to pathway following discharge [21] (Table 2).

Patient Flow
Outcomes relevant to patient flow primarily concerned wait-times and were reported in eight pathways [22, 28-34, 38, 41] (Table 2). The SSP reported both no difference [29] in wait-times for surgeon appointment for pathway patients and shorter wait-times [28] when compared to conventional referrals; both studies reported significantly shorter wait-time for MRI scan for pathway patients. Five studies compared wait-times before and after the introduction of a pathway, reporting shorter wait-time for interface assessment than pre-pathway wait-time for consultant appointment [22,30,41], and decreased wait-times for consultant appointments fol- lowing introduction of physiotherapy-led triage [34] and telehealth-assisted triage [31]. One study reported that pathway changes may have resulted in care delays for standard referrals but not for 'fast-track' patients [33].

Clinical Outcomes
Nine pathways reported clinical outcomes, mostly pertaining to interface services and using patientreported outcome measures (PROMs) [17, 19-22, 26, 27, 29, 36-38, 41] (Table 2). Studies did not provide comparative data [22,26,36,37,41], and some lacked methodological [21,36,37] or statistical detail [37]. NERP associated better clinical outcomes for the 'triage and treat' pathway service at 6 and 12-month follow-up with shorter pain duration [19,20], and reported greater improvement in quality of life (EQ-5D) than a comparator non-pathway service [21]. The SCOPiC Trial [17] reported no difference in time to first resolution of symptoms nor in PROMs between stratified and usual care, although stratification was slightly less effective on quality-adjusted life-year (QALY) analysis (Table 2). Wu et al. [29] found no difference in surgery type nor outcome between patients referred for surgical assessment via SSP interface clinic and conventional referrals.

Service-User Satisfaction and Experience
Outcomes within this domain were reported by eleven pathways [18-21, 23, 25, 29, 31, 34, 36, 37, 41, 43] ( Table 2). Mostly high levels of patient satisfaction with care were reported using ordinal instruments [19,20,25,29,31,36,37,41,43]. High referrer satisfaction rates with physiotherapy-led triage [25,34] and fair to good referrer satisfaction with telehealth-assisted pathway [31] were reported, albeit with low response rates (18-40%). Four studies reported qualitative data [18,21,23,25], and key findings are summarised in Table 2. Martin et al. [21] highlighted key findings from a range of stakeholders for consideration in preparing to implement their pathway in new areas. Ryan et al. [23] explored the experiences of patients with sciatica managed on a NELBPP-consistent pathway and identified common challenges and pathway issues. The SCOPiC trial [18] reported on both patient and clinician experiences of their stratified care pathway.

Discussion
This review identified and reported key features of published clinical pathways for LBP/radicular leg pain, how pathways coordinated care across healthcare boundaries and summarised and evaluated the evidence for the identified pathways.
There were clear similarities across pathways: all originated from high-income, western countries; initial patient access was generally via GP with multiple disciplines contributing to care. Pathways were most commonly developed to address over-burdened health care systems, aiming to address long wait times for surgeon appointments. Interface services, primarily physiotherapist-led, were used almost universally, serving a clear purpose in managing the boundary between primary and secondary care by triaging patients for surgical opinion or conservative management. It follows that the included studies mainly examined outcomes pertaining to the effect of interface services on markers of patient flow and healthcare resource use, such as wait-times and referral rates to surgical clinics. Interface services were associated with improved service efficiency, with high rates of independent management, high conversion rates to intervention from subsequent consultant appointments, reduced wait times for care, and high levels of patient satisfaction ( Table 2). The wide ranges of values reported for rates of independent management in interface services and conversion rates from consultant appointment to intervention may reflect differing access criteria for interface services and diverse scopes of practice of triage clinicians in different services.
Interface services were evenly represented across community (seven pathways) and hospital settings (eight pathways), and although clinical guidelines consistently recommend demedicalising LBP, with care provision in community settings for the majority of patients, this was not a commonly stated driver for pathway development. Although integrated care is advocated, pathways demonstrated limited integration, perhaps reflecting that care integration was not a stated priority in pathway development, with only four pathways citing the promotion of integrated, shared or interprofessional care as a motive. Using Ahlgren and Axelsson's model that conceptualises the degree of integration on a fivelevel continuum ranging from 'full segregation' to 'full integration', the included clinical pathways demonstrated either 'full segregation' or, at best, 'linkage' between primary and secondary care, the two most basic levels of integration [45]. 'Linkage' between primary and secondary care was mainly demonstrated by regional pathways or those considered local manifestations of national pathways, suggesting that higher levels of care co-ordination were achieved by pathways with broader geographic and organisational scopes of implementation. A recent comparative study of LBP care pathways similarly reported that no pathways demonstrated full integration of primary and secondary care [46].
The evidence presented is undermined by the typically low methodological quality of included studies, usually observational service evaluations with retrospective data collection and limited statistical analysis. The pragmatic study designs may reflect challenges in implementing high-quality research in real-world clinical settings, particularly involving clinical pathways where the research intervention crosses different levels of healthcare provision with multiple providers and professions. The cost-effectiveness of clinical pathways in the provision of LBP care has not been adequately evaluated. Future research of economic impact should employ comprehensive cost evaluation pre-and post-pathway implementation and include implementation costs. Despite the clear need to improve clinical outcomes in LBP care to address the associated disability burden, this was seldom a stated motive for pathway development and has not been sufficiently evaluated in the literature. To examine clinical effectiveness, future clinical pathway evaluations require robust methodologies with comparative data and prospective data collection to strengthen findings, and the SCOPiC trial is an excellent example of how high-quality research in this area can be achieved with a well-designed RCT. Improving LBP outcomes has been linked to implementation of best-practice recommendations, one of the most stated reasons for pathway development in the included studies, but the literature did not examine if clinical pathways can bridge the evidence-practice gap in LBP care. Determining the ability of clinical pathways to optimise guideline-concordant care, as well as the most effective strategies to enhance pathway and LBP guideline fidelity should be research priorities. The richness of the findings presented by the qualitative studies emphasises the merit of hearing the patient's voice and experience in pathway development and evaluation, and the need for future research of these complex interventions to move beyond the use of limited ordinal satisfaction questionnaires, which tend to be biased towards eliciting positive ratings [47].
This study benefitted from the robust approach of the systematic review methodology, including the use of independent review authors. The review authors found limitations in using the EPHPP quality assessment tool across the wide range of study designs, but this was offset by the use of independent reviewers to reach consensus. Some studies provided details primarily for the portion of the pathway being evaluated in their paper. Study heterogeneity precluded meta-analysis and the diverse range of outcomes assessed limited comparisons between studies. Recommended outcome measures for the evaluation of spinal clinical pathways, agreed amongst the international spinal research and clinical communities, may facilitate meta-analysis in future systematic reviews with stronger conclusions on effectiveness and guidance for optimal implementation.

Conclusion
A limited volume of research has evaluated clinical pathways for LBP/radicular leg pain that cross the traditional primarysecondary care boundary. Most of the research studies were pragmatic, retrospective evaluations of interface services that were primarily introduced to assist over-burdened secondary care resources in high-income countries. This is reflected in the predominance of outcomes concerning patient flow and healthcare resource use with triage services associated with increased efficiency of care delivery. Pathways demonstrated basic levels of care integration across primary and secondary care. The potential of clinical pathways to enhance LBP clinical outcomes has not been adequately evaluated, nor has their ability to deliver cost-effective or guideline-concordant care or to enhance care integration across the primary-secondary care divide. In addressing these knowledge gaps, future research should prioritise RCTs and be more cognisant of the patient experience.
included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.