Aim

To identify what core outcome sets are currently used in the assessment of spinal pain and associated limb, trunk, abdomen, or pelvic pain, and to thematically analyze the domains to identify consistencies and gaps across them.

Objectives

1. To identify, map and examine published core outcome sets intended for use in the assessment of: spinal (cervical, thoracic, lumbar) pain; spinal related lower limb pain; spinal related upper limb pain; and spinal related trunk, abdomen or pelvic pain.
2. To synthesise the consistent domains across core outcome sets for spinal pain and associated limb, trunk, abdomen, or pelvic pain. For easier interpretation, domains are conceptualized and categorized according to anatomical region of relevance, mode of implementation (patient-reported, physical/performance-based, or clinician/observer-rated) and broad construct (e.g., biological, physiological, psychological, social, environmental).

Protocol and registration

The systematic review is registered with PROSPERO (CRD 42023473529), a register of anticipated health and social care systematic reviews held by the University of York Centre for Reviews and Dissemination. Registration ensures transparency, reducing bias caused by potential selective reporting and was essential to avoid duplication [29,33].

Eligibility criteria

An adapted PICOS framework ((P) patient population or the disease being addressed, (I) interventions or exposure, (C) comparator/control group, (O) outcome measure, (S) study design)) was used to inform eligibility criteria [25,34]. PICOS enables specific elements of clinical evidence to be identified in a systematic review [25,34,35]. The eligibility criteria are detailed in Fig 1.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Eligibility criteria.

https://doi.org/10.1371/journal.pone.0338184.g001

Information sources

The literature search employed sensitive topic-based strategies designed for each of the sources identified in Fig 2. The specific databases searched were selected following discussion with subject matter experts, methodologists, and advice from specialist librarians at St George’s University of London (UK) and Western Univerity (Ontario, Canada). The information resources accessed were chosen to enable a holistic literature search [25,29,36]. Grey literature was explored using ProQuest Dissertations and Theses, along with trial registers such as clinicaltrials.gov and the International Clinical Trials Registry Platform of the World Health Organization. In addition, experts in the area were consulted to detect any further studies [25,36–38].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Information sources.

https://doi.org/10.1371/journal.pone.0338184.g002

Search strategy

Pre-defined search terms and combinations, with database specific standardised vocabulary were employed to ensure all relevant studies were retrieved [25,36–38]. This was essential to ensure that no studies were missed due to different words being used to describe the same concepts [25,36–38]. The search of databases encompassed the period over the 25-year window from 01.01.2000 to 01.04.2025 to ensure all contemporary COSs were captured. Fig 3 demonstrates the MEDLINE (Ovid) search strategy that was adapted to meet search terms of other included databases. Where a protocol or pilot study was identified, the definitive study was sought, or the authors were contacted to determine whether further published or unpublished research has been undertaken. The reference lists of the included studies were searched to ensure no studies were missed [36,38].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Medline (Ovid) search strategy.

https://doi.org/10.1371/journal.pone.0338184.g003

Data management

A web-based software platform (COVIDENCE, Veritas Health Innovation, Melbourne Australia) was used for managing various stages of systematic and other literature reviews. This platform facilitated the importation of all citations, assisted in eliminating duplicates arising from the search strategy, and supported the determination of eligibility throughout the screening and review phases [39].

Selection process

Two investigators searched the information sources (TN + CL) and independently assessed studies for inclusion by grading each eligibility criterion. The use of two investigators reduced the possibility of missing relevant literature or excluding relevant research where challenging judgements were necessary for selection or rejection of a study [33]. In the event of a selection disagreement a third reviewer (AR/DW, methodological experts) was consulted [33,37]. Both reviewers independently evaluated studies by title and abstract for potential eligibility. Following discussion between reviewers, if a study could not be explicitly excluded based on its title and abstract, its full text was reviewed [29,31]. All potentially relevant studies proceeded forward to full text review after which the two reviewers made independent judgements as to whether an individual study met the eligibility criteria. Any conflict was resolved with a 3^rd reviewer. The numbers of studies included and excluded at the different stages was recorded using a PRISMA flow diagram [25,33,36].

Data collection process

Data extraction and charting was performed by the primary reviewer (TN) and checked and agreed by the secondary reviewer (CL). Data collection used pre-determined data collection sheets specific to the review objectives which were piloted, refined and agreed by the researchers prior to use [33,36]. Any differences were resolved at a consensus meeting of all researchers [37], and the third reviewers (AR/DW) checked for consistency and clarity.

Data items and outcomes

The following data items and outcomes were extracted from all included studies: authors, date of publication, study aim(s)/objective(s) and design, anatomical region of interest, definition of the condition the COS was used for, recommended outcome measures included within the COS, including verbatim domains or constructs (or specific tools) endorsed and recommendations for the frequency of use for patient assessment [25,27,36].

Quality (including internal validity) assessment of individual studies

As of this review there were no widely-accepted risk of bias (RoB) tools published for critical appraisal of Delphi or consensus research, which was anticipated to be the most commonly used method for development of the COSs. Accordingly, for purposes of this review we adapted the Conducting and Reporting of Delphi Studies (CREDES [40]) best practice framework to evaluate the quality (including internal validity) of the included studies regarding conduct and reporting [40].

While other frameworks exist for evaluating COS development and reporting—namely COS-STAD (Core Outcome Set–Standards for Development) [41] and COS-STAR (Core Outcome Set–Standards for Reporting) [42]—these were not used directly in this review. COS-STAD provides methodological guidance for COS design, and COS-STAR outlines standards for transparent reporting. Despite their distinct focuses, both frameworks share key domains such as clear rationale, stakeholder engagement, consensus methodology, and scope definition. The CREDES checklist was chosen over COS-STAD and COS-STAR because it integrates domains relevant to both frameworks and is specifically tailored to Delphi and consensus methodology [40–42]. This made it the most suitable tool for appraising both the design and reporting quality of the included studies, capturing aspects of internal validity, transparency, and methodological consistency across the consensus-based COS literature.

CREDES was developed by methodological experts to enable quality and consistency in the conduct, reporting and appraisal of Delphi studies, using a 16-item checklist [40] specifically tailored to Delphi and consensus methodology [40–42]. A modified version of the ‘Buchbinder appraisal scale’ (BAS) was used to evaluate each included study against the 16-items within the CREDES [43–45]. The BAS awards one point for fully meeting a criterion, half a point for partially meeting a criterion, or zero points for not meeting a criterion or being unable to score due to a deficit in the information reported [43–45]. Therefore, a quality score between 0–16 was reported of each of the included studies [40]. The BAS was originally developed for the critical appraisal of treatment based classification systems across 7 criteria with an overall Intraclass Correlation Coefficient (ICC) of 0.82 for inter-rater reliability [43,44]. The scale has now been adopted successfully across a range of studies where quality assessment tools are yet to be developed and validated [43,44].

Synthesis of results

A three-step process was used to synthesise the data.

Stage 1: Identification and mapping of COSs

An explanation of each included study’s characteristics and COS outcomes/recommendations were tabulated.

Stage 2: Synthesis of outcome measures across regions of the spine

For each anatomical region, data across the studies were synthesised to identify the assessment meaures recommended in the following categories: patient reported measures (PROMs), physical/ performance-based measures, biological measures, clinician-assessed psychological measures, social measures, environmental/ ecological measures [26–28]. A narrative synthesis was used to synthesise the measures within spinal pain and spinal related limb, chest, abdominal, and pelvic pain ‘regions’ [27,29,37,46].

Stage 3: identification of recurrent outcome domains across COSs

The grouping of data across studies and spinal region into core domains, was completed by using the tabulated data from stages 1 and 2. The constructs endorsed within each core outcome measure set were grouped through thematic content analysis [47]. This process started by reviewing and re-reviewing the verbatim constructs included within outcome measures within each COS, and where specific PROMs were endorsed, returning to the literature as needed to define the construct(s) being measured by each. Domains were interpreted verbatim per the COS authors, while one-off or customized COS-specific questions or questionnaires were interrogated for the domain(s) with which they most aligned based either on the original development publications of those tools or, where the question or tool lacked adequate theoretical grounding in its initial development, interpreted against contemporary domains of health and wellness with the International Classification of Functioning, Disability, and Health (ICF) as the primary framework. Each domain, question, or PROM were therefore aligned with one or more dominant domains for thematic analysis.

Following extraction and interpretion of the most accurate construct from all measures included in the COSs, the thematic grouping moved from specific to broad, wherein groups of like constructs were clustered together to form ‘Sub-Domains’ that continue to be clustered into conceptually-distinct ‘Domains’ with consensus reached by all authors. A domain was considered to be in “common use” if it was reported in more than two thirds (66.6%) of the included core outcome sets. This threshold was selected based on its use in previous quality studies assessing the frequency of domains within patient-reported outcome measures (PROMs) [48].

Certainty of evidence

Assessment of the overall certainty or confidence in the evidence was not conducted in this systematic review. Established frameworks such as GRADE (Grading of Recommendations Assessment, Development and Evaluation) [49,50] and GRADE-CERQual (Confidence in the Evidence from Reviews of Qualitative Research) [51] were not appropriate for the included studies, as they are primarily designed for assessing certainty in intervention studies and qualitative evidence, respectively. The studies included in this review were primarily Delphi and consensus-based, which lack a suitable, validated framework for assessing certainty of evidence [49–51]. As such, no formal assessment of certainty was applied.

Study selection

The search returned n = 8,046 potentially relevant studies. Following screening for duplicates, n = 5,790 citations remained. One additional study was identified from the Internet searches, and one from hand-searching the citation lists of the remaining articles, both of which were subsequently screened. Reviewing by title and abstract excluded n = 5,751 studies that clearly did not meet the eligibility criteria. No relevant unpublished studies were found, and no further studies were identified from reviews of the national research register or via experts in the field. The full texts of the remaining n = 39 studies were examined in detail and evaluated against the eligibility criteria.

Two studies—White et al (2004) [52] and Rebbeck et al (2007) [53]—were ‘whiplash associated disorder’ (WAD)- and ‘mechanical neck pain’-specific adaptations of a COS developed by Deyo et al (1998) [21] for low back pain. While the Deyo publication (1998) pre-dated publication cut-off of 2000, it was retained in the final selection of sources as the two subsequent adaptations were otherwise uninterpretable without the context of the original. All three were individually rated for quality and included separately in stages 1 and 2 of the data synthesis. In stage 3, they were considered a single COS (with region-specific adaptations) for data extraction and synthesis, as their constructs were consistent. Similarly, Verburg et al (2019) [24] and Verburg et al (2021) [54] were two studies from one programme of research. Both met eligibility criteria and were retained. Again, they were rated and synthesised separately in stages 1 and 2, but treated as a single COS in stage 3 as both retained the same outcome domains although the later study introduced outcome-based quality indicators for patient-reported outcome measures (PROMs). For the same reasons Chen et al (2019) [55] and Sterling et al (2023) [23] were also considered a single study in stage 3, as Sterling et al (2023) [23] provided specific measurement instruments for the COS constructs initially defined by Chen et al (2019) [55].

Thirteen studies were included in the systematic review. 100% inter-rater agreement was achieved following open discussion at each stage. Third reviewer mediation was not required. Fig 4 presents the number of studies at each stage of the selection process.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. PRISMA flow diagram.

https://doi.org/10.1371/journal.pone.0338184.g004

Stage 1: Identification and mapping of core outcome measure sets

Study characteristics

Studies were published between 1998–2023. Study designs included Delphi technique (n = 5) [22,24,55–57], other consensus methods (n = 6) [21,23,54,58–60], and adaptations and evaluation of existing COS (produced through consensus methods) to produce region-specific COS (n = 2) [52,53].

Four studies described COSs for the cervical spine region [23,52,53,55]: three focused on whiplash-associated disorders [23,53,55] and one on mechanical neck pain [52]. Eight studies described COSs for the lumbar spine region exclusively [21,22,24,54,56,58–60]. Of these, four addressed ‘nonspecific’ or ‘mechanical’ low back pain [21,24,54,58], two were intended for all low back pain excluding cases with sinister pathology, inflammatory disease, infection, or fracture [22,56], one focused exclusively on chronic low back pain [59], and one was inclusive of both acute and chronic low back pain [60]. One study investigated the thoracolumbar region (including both thoracic and lumbar regions) [57], specifically targeting adolescent and young adult patients with spinal deformity.

Six studies developed COSs specifically for research purposes [21–23,55,59,60], seven for clinical practice [24,52–54,56,57], and one for both research and clinical settings [58]. COSs developed for clinical practice were generally shorter, including fewer domains or measures, whereas those designed for research settings tended to be more comprehensive, incorporating a broader array of domains or measures.

In terms of scope, two studies aimed to define core outcome domains only [55,58], one focused on defining appropriate patient-reported outcome measures (PROMs) [24], and ten defined both outcome domains and the PROMs to assess them [21–23,52–54,56,57,59,60].

Only three studies explicitly defined the frequency with which the COS should be used [24,56,60]. One study recommended specific application strategies tailored to individual studies [59], while the remaining nine did not specify frequency of use [21–23,52–55,57,58].

Quality assessment

Quality assessment data can be found in Supplementary file 1. Scores ranged from 0.5 to 15 out of a maximum of 16 points. The most commonly met quality domains across studies included clear justification for using the Delphi or consensus technique, well-defined consensus procedures, and adequate interpretation of results. High-scoring studies consistently addressed these criteria. In contrast, the least commonly met domains were prevention of bias, external validation, and detailed reporting of expert panel and participant recruitment.

Results of individual studies

The data and quality assessment outcome from individual studies (COSs) are presented in Table 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Data from individual studies.

https://doi.org/10.1371/journal.pone.0338184.t001

Results of syntheses

Stage 2: Synthesis of outcome measures.

Table 2 presents a synthesis of outcome measures across spinal regions, structured according to the outcome domains reported within each included study. Each outcome is categorised under the domain as defined by the original authors of the respective core outcome set.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Synthesis of outcome measures across regions of the spine.

https://doi.org/10.1371/journal.pone.0338184.t002

Cervical spine COSs

Three studies [23,52,53] with a quality rating of 0.5 to 14/16 reported outcome meaures in COSs for the cervical spine. All three COSs endorsed standardized PROMs. None included physical/performance-based, biological, non-patient-reported psychological, social, or environmental/ecological measures.

Pain was the most commonly included domain, appearing in all three COSs [23,52,53]. The Visual Analogue Scale (VAS) was included for use in all three. Two [52,53] (quality 0.5/16) included a custom symptom scale, while one [23] (quality 14/16) included the Numeric Pain Rating Scale (NPRS).

Disability was included in two COSs [52,53] (both quality 0.5/16) using tailored questions about daily activity limitation and time lost from work or school due to symptoms.

Percieved Recovery (global percieved change) was included in one COS [23] (quality 14/16) using a 7- or 11-point Global Rating of Change (GROC) scale.

Physical function was included in one COS [23] (quality 14/16) using both the Neck Disability Index (NDI) and the Whiplash Disability Questionnaire (WDQ).

Psychological functioning was included in one COS [23] (quality 14/16), which recommended the use of validated instruments including the Pictorial Fear of Activity Scale-Cervical (PFActS.-C.), the Pain Self-Efficacy Questionnaire, and the Pain Catastrophising Scale.

Satisfaction with care was included in one COS [52] (quality 0.5/16) using a single-item measure regarding the participant’s overall impression of their medical care.

Social function was included in all three COSs [23,52,53]. Two C.O.S. [52,53] (both quality 0.5/16) included a ‘PROM question’ that asked how pain interfered with work and home responsibilities. One C.O.S. [23] (quality 14/16), included two ‘PROM questions’, asking about employment status (e.g., modified duties, disability pension) and the percentage of usual work hours performed.

Wellbeing assessment was included in two COSs [52,53] (quality 0.5/16) using a single global item asking participants how they would feel if their current symptoms persisted for the rest of their lives.

Thoracic spine COSs

One study [57] with a quality rating of 12.5/16 reported a COS for the thoracic spine comprised entirely of standardized PROMs. The Scoliosis Research Society–22 revised questionnaire (SRS-22r) was the primary PROM included, capturing multiple outcome domains relevant to thoracic spine conditions. These included pain, physical function, recreation and leisure, self-image, and overall outcome of surgery. In addition, satisfaction with the cosmetic result was assessed post-operatively using the same instrument. The COS also included the EuroQoL-5D (EQ-5D) to assess overall quality of life. An extended version of this COS specific to surgical outcomes was also presented that included clinician-assessed outcomes such as re-operation rates, post-operative complications, and changes in spinal deformity. No other outcome types—such as physical or performance-based measures, non-patient-reported psychological or social measures, or environmental/ecological measures—were included,

Lumbar spine COSs

Nine studies (exclusively lumbar n=8, thoracolumbar n=1) [21,22,24,54,56–60] with a quality rating range of 2 to 15/16 reported COSs for the lumbar spine. All nine COSs included PROMs to assess multiple outcome domains. One COS [61] (quality 5/16) included physical/performance-based measures and two COSs [60,61] (both with quality rating 5/16) included biological measures (blood tests, MRIs and radiographs). None included non-patient-reported psychological or social outcomes, or environmental and/or ecological measures.

Pain was the most frequently reported outcome domain, included in seven COSs (quality 2.5-15/16). The Numeric Pain Rating Scale (NPRS) was included in four COSs [22,24,54,56] with quality scores ranging from 2.5 to 15/16. One C.O.S. [21] (quality 2/16) included the Visual Analogue Scale (VAS), one COS [60](quality 5/16) included the Numeric Pain Scale (NPS), and two COSs [21,60] (quality 2-5/16) included custom symptom-specific questions. One thorocolumbar scoliosis-specific COS [57] (quality 12.5/16) included the SRS-22r which includes a pain subscale as part of a multidimensional PROM.

Disability was included in five COSs [21,24,54,56,60] (quality 2 to 13/16). These included: The Oswestry Disability Index (ODI) in two [24,56] (quality 7.5-13/16), the Quebec Back Pain Disability Scale (QBPDS) in two [24,54] (quality 2.5-13/16), and the Roland-Morris Disability Questionnaire (RMDQ) in one COS [60] (quality 5/16). One C.O.S. [21] (quality 2/16) included single-item disability questions, asking about days spent in bed, missed work, or restricted normal activity due to back or leg pain.

Physical function assessment was included in six COSs [21,22,24,54,57,61] (quality 2-15/16), using both validated PROMs and COS-specific structured questions. The O.D.I. 2.1a was included in three COSs [21,22,61] (quality 2-15/16). The Patient-Specific Functional Scale (PSFS) was included in two COS [24,54] (quality 2.5-13/16), while the RMDQ was included in three [21,22,61] (quality 2-15/16). One COS [57] (quality 12.5/16) included the SRS-22r to assess physical functioning as part of a multidimensional PROM. One COS [21] (quality 2/16) also included a single question on pain interference with work and home responsibilities.

Quality of life was included in four COS [22,56,57,61] (quality 5-15/16). The E.Q.-5D. was included in three COSs [22,56,57] (quality7.5-15/16), while the EuroQol visual analogue scale (E.Q.-VAS) was included for use by one [56] (quality 7.5/16). The S.F.-12 was recommended by one COS [22] (quality 15/16). Two C.O.S.s [22,61] (quality 5-15/16) included the P.R.O.M.I.S.-GH-10.

Wellbeing was included in two COSs [21,60] (quality 2-5/16). One C.O.S. [21] (quality 2/16) included the EuroQol, the S.F.-12, and included a question asking how patients would feel if their current symptoms persisted for life. One C.O.S. [60] (quality 5/16) included the S.F.-36.

Patient-reported Psychological Function were included in two COSs. Both the Pain Catastrophising Scale and Fear-Avoidance Beliefs Questionnaire (FABQ) were included in one COS [60] (quality 5/16). Two C.O.S.s [60,61] (quality- both studies 5/16) included measures to evaluate depression: one [61] included the Beck Depsression Inventory (BDI-II) and Center for Epidemiological Studies – Depression scale (CES-D), while the other [60] included the Patient Health Questionnaire-9 (PHQ-9).

Satisfaction with care was included in four COSs [21,24,54,60] (quality 2.5-13/16). Two C.O.S.s [21,60] (quality 2-5/16) included a single-item satisfaction question. Two others [24,54] (quality 2.5-13/16) included the Global Perceived Effect scale.

Work status was recommended for inclusion in two COSs [56,60] (quality 5-7.5/16). One [56] (quality 7.5/16) included a general work status question, while one [60] (quality 5/16) included more detailed questions about return-to-work outcomes and days missed due to back or leg pain.

Pre and post-operative outcomes were included in one COS [57] (quality 12.5/16) recommending the SRS-22r to assess the overall outcome of surgery, satisfaction with cosmetic results, recreation and leisure, and self-image.

Physical and performance-based measures were included in one COS [61] (quality 5/16), being clinician-assessed hip medial rotation, lower limb muscle strength, deep tendon reflexes, and straight leg raise (SLR) for patients with radicular symptoms.

Biological measures were included in two COSs [61]. Both included metrics extracted from diagnostic testing including MRI. One [60] (quality 5/16) also included use of blood tests and radiographs where indicated.

Reporting of adverse events was included in one [61] (quality 5/16).

No included studies reported non-patient-reported psychological, social, or environmental/ecological measures.

Stage 3: identification of recurrent outcome domains.

A total of n=82 domains, questions, or PROMs were extracted from across all included COSs detailed in Table 1. Thematic content analysis categorised the 82 recommendations into 22 conceptual domains (Table 3). Of those, four domains were endorsed by >6 (66.66%) of the 9 COSs meeting the threshold for ‘common use’ status: patient rated physical function (100%), pain symptoms/Pain intensity (88.9%), participation - work or school (77.8%), and participation - activities of daily living(66.7%).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Presence of each of the 22 thematically-classified Domains endorsed by each of the COSs.

https://doi.org/10.1371/journal.pone.0338184.t003

Reporting biases

For all eligible studies included in this review, complete data were available from the published reports. No additional data were required, and no contact with study authors was necessary to obtain missing information. As such, the risk of reporting bias related to unavailable or selectively reported data was considered minimal.