Operational phenotyping for chronic low back pain: a narrative review and methodology for developing routing and escalation pathways across spine care

Introduction

Background and supportive-care relevance

Low back pain (LBP) is not a single disease but a heterogeneous symptom syndrome shaped by variable contributions from spinal pathology, morphology, motion and loading, nervous system sensitivity, psychosocial context, occupational exposure, and previous care experience (1). For patients and care systems, this heterogeneity is operationally consequential because it influences who can be managed with supported self-management, who requires structured rehabilitation, who needs multidisciplinary symptom and behavioral support, and who should be evaluated for interventional or surgical options (2,3).

This topic is relevant to supportive and palliative medicine because chronic LBP (CLBP) is a persistent symptom-management problem that affects quality of life, function, sleep, emotional well-being, work participation, and care utilization. The central supportive-care question is not only whether a structural lesion is present, but how symptom burden, functional risk, patient goals, access barriers, and safety findings should be translated into the least burdensome effective pathway while preserving urgent escalation for neurologic or serious-pathology presentations (4,5).

Phenotyping has been proposed as a response to LBP heterogeneity. However, many studies produce descriptive labels, clusters, imaging signatures, or model-derived profiles without specifying what clinicians should do differently when a phenotype is assigned (6-14). Labels without decisions rarely change care, and pathways without measurement cannot be improved.

Rationale and knowledge gap

Patients with LBP enter spine systems through multiple doors, including primary care, urgent care, emergency departments, physical therapy (PT), pain clinics, and direct-to-surgeon referral. When these entry points use different defaults for imaging, injections, conservative-care duration, or surgical consultation, patients may receive prolonged low-yield care, premature escalation, repeated diagnostic workups, or inconsistent counseling (15-17).

The practical gap is therefore not simply the absence of better subgroup labels. Spine programs need a method for converting feasible clinical inputs into transparent routing decisions, specifying the pathway and dose of care that follows, defining when that pathway has succeeded or failed, and measuring the outcomes and utilization consequences of those decisions. This revision uses the term operational phenotyping for that workflow.

Novelty relative to existing models

The novelty of the present framework is its explicit linkage of phenotype assignment to pathway delivery and outcome audit. Subgroups for Targeted Treatment Back Screening Tool (STarT Back) and related risk-stratified models classify patients by prognostic risk and can help match conservative-care intensity (18-22). Deep phenotyping programs characterize multidomain mechanisms and candidate biomarkers but are often too burdensome for universal clinical intake (10,12,23-26). Machine-learning phenotyping may identify latent profiles or predict outcomes, but clinical implementation depends on calibration, transportability, interpretability, and whether the output changes a deliverable action (13,14,27-32).

Operational phenotyping differs by requiring four elements in every deployable phenotype: feasible inputs, explicit decision logic, an actionable pathway, and measurable outcomes. The framework is shown schematically in Figure 1 and used throughout the program catalog; Table 1 provides the corresponding operational phenotype quality checklist. It is therefore a methodology for building a field guide that can be adapted to local capacity, not a universal prescription for one fixed care pathway.

Figure 1 Three-layer architecture of operational phenotyping. The framework links routing, execution, and learning. Routing assigns patients to pathways using a minimal input set. Execution applies rules or model outputs consistently at the point of care. Learning uses reassessment checkpoints, dashboards, and audit feedback to refine thresholds over time.

Objective

The primary objective of this narrative review was to define operational phenotyping as a pathway-development methodology for CLBP and to provide a simplified implementation structure that can be used across conservative, interventional, and surgical spine settings. The secondary objective was to translate lessons from stratified care, deep phenotyping, digital phenotyping, and spine pathway studies into a concise Inputs-Decision-Pathway-Outcome (IDPO) program catalog that can be prospectively tested. We present this article in accordance with the Narrative Review reporting checklist (available at https://apm.amegroups.com/article/view/10.21037/apm-2026-1-0026/rc).

Methods

Review design and article type

This manuscript is a narrative review conducted with structured search and synthesis methods. It is not a formal consensus statement, practice guideline, systematic review, or meta-analysis. The manuscript uses narrative synthesis because the relevant literature spans prognostic tools, subgrouping studies, implementation trials, imaging stewardship, interventional selection, digital phenotyping, occupational health, and surgical prediction, with substantial heterogeneity in populations, inputs, and outcomes.

Information sources and search strategy

Searches were conducted for the period 1 January 2000 through 28 January 2026. MEDLINE via PubMed was the primary database. Embase, Cumulative Index to Nursing and Allied Health Literature (CINAHL), PsycINFO, Cochrane Central Register of Controlled Trials (CENTRAL), and Scopus/Web of Science were used when available to broaden coverage across rehabilitation, psychology, pain medicine, surgery, and health-services implementation. Additional sources included reference-list screening, guideline and consensus documents, ClinicalTrials.gov and World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP), and publicly available cohort or program materials. The representative PubMed search strings are provided in Appendix 1 and summarized in Table 2.

Eligibility criteria and selection

We included adult studies of acute LBP, subacute LBP, or CLBP, including radicular pain or stenosis when the findings informed imaging, injection, surgical evaluation, or routing decisions. Eligible evidence included randomized and pragmatic trials of stratified or stepped care, observational phenotype or trajectory studies, prediction models with validation, implementation evaluations, pathway or decision-support studies, and high-quality reviews or guidelines. We excluded preprints, retracted articles, pediatric studies, case reports, very small case series, non-LBP pain populations without separable LBP findings, purely technical studies without patient-level linkage to outcomes or decisions, and biomarker-only studies without near-term operational relevance.

Across three PubMed searches, 8,225 records were exported and de-duplicated by PubMed identifier (PMID), yielding 5,965 unique references. Title/abstract screening narrowed the set to 235 candidate records relevant to adult LBP and the IDPO thesis. Structured IDPO screening and full-text prioritization excluded 145 records because they lacked operational decision logic, a defined pathway, measurable pathway outcomes, adult LBP relevance, peer-reviewed publication status, or actionability for the program catalog. Ninety records were retained for synthesis and inclusion in the curated reference list.

Data extraction and IDPO synthesis

For each retained source, we extracted inputs, decision logic, pathway implications, outcomes, and evidence tier. Inputs included questionnaires, examination elements, imaging features, function tests, digital measures, work exposures, and collection burden. Decision logic included thresholds, rules, model outputs, congruence requirements, and missing-data handling. Pathway items included care setting, treatment bundle, conservative-care dose, reassessment timing, and escalation or de-escalation rules. Outcomes included pain, disability, function, work participation, imaging, injections, surgical referral, cost, and durability of response.

Berlin deep-phenotyping as a contextual exemplar

The Berlin Spine Study and related DFG Research Unit FOR 5177 materials were used as a contextual exemplar of deep phenotyping, not as evidence that any Berlin-derived pathway improves outcomes (23-25). Figure 2 summarizes this deep-phenotyping pipeline. Berlin illustrates the upstream discovery problem: multidomain cohorts can combine symptom history, psychosocial context, clinical and neurologic examination, functional and sensor-based motion assessment, imaging, and longer-term activity monitoring. Such programs are valuable for discovering candidate signals, but operational implementation requires distillation into a minimal universal battery plus targeted add-ons.

Figure 2 Deep-phenotyping pipeline used as a contextual exemplar. Multidomain research programs such as the Berlin workflow combine history, questionnaires, clinical and neurologic examination, psychosocial and cognitive assessment, functional and sensor-based motion evaluation, MRI-based structural mapping, and optional longitudinal activity monitoring. The operational task is to distill these domains into feasible clinic inputs. LBP, low back pain; MRI, magnetic resonance imaging.

Methodological limitations

Because this article is a narrative review, it did not include formal risk-of-bias scoring, meta-analysis, certainty grading, or an exhaustive systematic adjudication of every candidate study. The synthesis is therefore best interpreted as a structured implementation framework derived from representative and operationally relevant evidence rather than as a definitive evidence-based guideline. This limitation is important because several pathway decisions in spine care remain supported by mixed or incomplete evidence, requiring locally governed, general pathways rather than rigid universal rules.

Findings and operational synthesis

The retained literature supported five findings that guided the revised framework. First, prognostic stratification is clinically useful only when the output changes the care that can actually be delivered (18-22,33-36). Second, multidomain phenotyping is more feasible when universal intake is brief and deeper assessment is triggered only when it would change management (10,12,23-26,37-42). Third, pathway outcomes must include utilization and value measures, not only pain intensity, because the framework is intended to reduce low-value variation in imaging, procedures, and referral timing (15,43-49).

Fourth, interventional and surgical escalation require separate handling. Some injections are therapeutic trials, but others are diagnostic procedures that help define a phenotype. Diagnostic medial branch blocks, for example, may support a facet-mediated pain diagnosis even when there is no definitive imaging correlate; imaging in that setting may be used primarily to exclude alternative pathology or confirm procedural safety rather than to prove the pain generator (17,45,50-53). Fifth, prediction tools for surgery and procedures can support counseling and shared decision-making, but their performance is context-dependent and should not be treated as a rigid gatekeeper (28-32,51-62).

Operational methodology

Definition

Operational phenotyping is a workflow unit: a small, feasible input set produces a decision, the decision assigns a pathway, and the pathway is evaluated with measurable outcomes. In short, operational phenotyping = inputs → decision → pathway → outcome. Figure 1 presents this as a three-layer architecture: routing, execution, and learning. Routing assigns the pathway; execution applies rules or model outputs consistently; learning uses reassessment checkpoints and dashboards to refine the pathway over time.

Minimal implementation version

A minimal implementation can be built without advanced imaging, sensors, or machine-learning infrastructure. The starting battery should contain five items: safety screen, brief neurologic screen, pain duration and prior care, a disability or function measure, and a psychosocial or complexity screen. These inputs are sufficient to route most patients into three initial pathways: supported self-management, PT-first structured rehabilitation, or higher-complexity multidisciplinary care. The minimum governance requirement is one early reassessment at 2–4 weeks and one pathway success/failure checkpoint at 6–12 weeks.

A minimal rule set might operate as follows. Patients with red flags, cauda equina symptoms, infection or malignancy concern, fracture concern, or new/progressive objective neurologic deficit bypass routine conservative dosing and receive urgent evaluation with appropriate imaging or specialty referral. Patients without safety findings and with low disability/low psychosocial complexity receive education, activity advice, and timed follow-up. Patients with moderate disability or movement limitation enter structured PT. Patients with high disability, high fear/distress, work instability, sleep disruption, or repeated utilization enter psychologically informed PT or multidisciplinary care, depending on local resources.

Four-stage funnel and reassessment

The staged funnel is shown in Figure 3. Stage 0 screens for serious pathology and neurologic urgency. Stage 1 assigns an initial pathway using the minimal battery. Stage 2 adds targeted testing only when the result would change management, such as magnetic resonance imaging (MRI) for concordant radicular symptoms after an appropriate conservative trial, occupational assessment for delayed return to work, or digital activity monitoring when guarded behavior or boom-bust cycling is suspected. Stage 3 defines the response phenotype at a prespecified checkpoint, usually 2–4 weeks.

Figure 3 Four-stage operational funnel for spine systems. Stage 0 screens for serious pathology and neurologic urgency. Stage 1 assigns the initial pathway using a minimal battery. Stage 2 adds targeted assessments only when they change management. Stage 3 reassesses early response at a defined checkpoint and triggers escalation or de-escalation using measurable criteria.

Illustrative, non-prescriptive thresholds improve implementability. A response threshold may be defined as at least 30% improvement in pain interference, a 10-point Oswestry Disability Index (ODI) improvement, meaningful improvement in a patient-specific functional goal, or clear progress in walking/activity tolerance. Failure should require both insufficient improvement and adequate pathway exposure, including attendance, adherence, and appropriate dosing. These thresholds are examples for implementation testing and should be adapted to local outcome measures and patient goals.

Clinical examples

Two brief vignettes illustrate how the method changes routing. Example 1: a patient with CLBP, ODI 44%, high fear-avoidance, poor sleep, and no red flags enters psychologically informed PT rather than generic exercise. At 4 weeks, ODI has improved by 4 points, and walking tolerance remains unchanged despite adequate attendance. The response phenotype is nonresponse with high complexity, so the patient is escalated to multidisciplinary pain rehabilitation while imaging is deferred unless new neurologic signs or concordant radicular features emerge.

Example 2: a patient has leg-dominant pain, dermatomal symptoms, objective weakness, and MRI showing concordant nerve-root compression. If symptoms are severe or neurologic deficit progresses, the safety override triggers expedited surgical evaluation. If symptoms are stable, the patient may complete a defined conservative trial; subsequent surgical consultation is framed as shared decision-making about expected benefit, risk, preferences, and alternatives rather than an automatic indication for surgery.

Program catalog

The program catalog uses the IDPO template shown in Figure 4. Each program specifies minimal inputs, decision logic, the downstream pathway, and measurable outcomes. The purpose is not to mandate one pathway for all systems, but to provide a reproducible template that programs can adapt and audit. The full dashboard is provided in Table 3, and illustrative thresholds are summarized in Table 4.

Figure 4 IDPO program template. Each phenotype program is expressed as Inputs, Decision logic, Pathway, and Outcomes, with implementation notes and evidence tier. This template converts phenotyping from descriptive labeling into an auditable operational unit. IDPO, Inputs-Decision-Pathway-Outcome.

First-contact triage phenotype

The first-contact program uses red flags, neurologic screen, pain duration, prior episodes, disability, psychosocial risk, work constraints, and patient goals. The decision is whether the patient needs urgent evaluation, expedited specialty referral, structured conservative care, or supported self-management. Outcomes include imaging rate and timing, opioid initiation, time to appropriate care level, disability at 6–12 weeks, emergency visits, and re-consultations (18-22,27,33-36,43-48,50,63-73).

Rehabilitation movement phenotype

The rehabilitation program uses irritability, hip and lumbar motion tolerance, functional tasks, and activity tolerance to place patients into simple treatable profiles such as mobility-limited, motor-control impairment, deconditioning/low tolerance, or mixed. The pathway is a matched PT bundle with a 2–4 week response checkpoint. Nonresponse should trigger reassessment for psychosocial, work, neurologic, or structural modifiers rather than indefinite extension of unchanged therapy (10,27,37,38,63,74-76).

Imaging-informed phenotype

Imaging is treated as a modifier rather than an automatic escalation trigger. The key decision is symptom-exam-structure congruence. When imaging is discordant or nonspecific, the pathway emphasizes education, rehabilitation, and avoidance of imaging-driven escalation. When symptoms, examination, and imaging are concordant and function remains impaired after adequate pathway exposure, interventional or surgical evaluation may be considered. The term low-value is used to describe interventions whose expected benefit is low in relation to burden, risk, cost, and patient goals; it is not intended to imply that every procedure has been definitively classified by randomized controlled trial (RCT) evidence (28-30,49,51-59,67,77).

Psychosocial and occupational complexity phenotypes

Psychosocial complexity and occupational load are pathway multipliers. Brief measures of fear, distress, sleep disruption, pain interference, work demands, work instability, and prior utilization identify patients who may need psychologically informed PT, cognitive behavioral therapy (CBT)/acceptance and commitment therapy (ACT) referral, occupational-health coordination, work conditioning, or multidisciplinary care. Outcomes should include disability trajectory, work participation, repeat visits, imaging, procedures, opioid exposure, flare frequency, and durable return to work (2,10,13,25,26,60,78-87).

Digital activity phenotype

Digital activity phenotypes are best treated as targeted add-ons rather than universal requirements. Step counts, activity variability, sleep proxies, and symptom diaries can identify guarded low-variability behavior, boom-bust cycling, or stable activity with persistent symptoms. Programs must include no-device alternatives to avoid equity gaps, and digital outputs should change coaching intensity, pacing, graded exposure, or relapse-prevention strategy rather than merely produce additional labels (10,37-42,88-90).

Surgical evaluation pathway

Surgical referral should be framed as entry into a surgical evaluation pathway, not as a directive selection decision or automatic indication for surgery. The minimal input set includes diagnosis, objective impairment, baseline pain and disability, psychosocial and work context, comorbidity/health risk, prior treatment response, patient goals, and selective imaging when it establishes remediable pathology. Validated prediction models may support expectation-setting and shared decision-making, especially for well-defined radicular compression, but should not replace clinical judgment or patient preference (28-32,51-62).

A surgical gate is therefore illustrative rather than prescriptive: remediable pathology, symptom-exam-imaging concordance, persistent disabling symptoms or objective deficit, adequate conservative-care dose unless a safety override applies, and a shared decision that expected benefits justify burdens and risks. These elements are documented so that conversion yield, durable responder rate, complications, reinterventions, and cost per responder can be audited.

Implementation and governance

Implementation should begin with the pathways that a local system can deliver. A practical inventory includes supported self-management, PT-first rehabilitation, psychologically informed rehabilitation, multidisciplinary pain care, occupational/work-conditioning support, diagnostic or therapeutic interventional pathways, and surgical evaluation. Once the inventory is explicit, the program chooses the minimal inputs needed to route patients into those pathways and defines the checkpoints that determine escalation or de-escalation.

Decision support should start with transparent rules rather than opaque models. Phase 1 should define safety overrides, imaging gates, diagnostic and therapeutic injection gates, conservative-care dosing, and early-response thresholds. Phase 2 may add predictive models only after workflow fidelity and outcome collection are stable. Exceptions, missing data, patient preference, and access limitations should be recorded and reviewed as part of governance rather than treated as undocumented deviations.

Governance requires dashboards that track clinical outcomes, utilization outcomes, work outcomes, value, and equity. Minimum measures include pain interference or Numeric Rating Scale (NRS), disability/function, global improvement, imaging timing, injections, surgical referral conversion, durable responder rate, return to work when relevant, pathway adherence, dropout, and differential benefit across patient groups. These measures make pathway performance visible and allow recalibration when local outcomes drift.

Discussion

This revision deliberately reframes the manuscript from a finished field guide to a methodology for developing a field guide. The distinction matters because the evidence base does not yet support rigid universal pathways for every CLBP phenotype. Instead, the evidence supports a disciplined implementation structure: define deliverable pathways, use minimal feasible inputs, add targeted assessments only when management would change, reassess early response, and audit outcomes and utilization.

A further implication is that interventional care should not be treated as a single category. Diagnostic procedures may define a phenotype; therapeutic procedures may test whether a concordant target produces meaningful functional improvement; and neither should be repeated without a prespecified outcome standard. Similarly, surgical evaluation should be a shared decision-making process informed by diagnosis, objective impairment, response to prior care, patient goals, and expected probability of benefit rather than a directive output of a model.

The main strength of the framework is its operational discipline. It converts the broad phenotyping literature into a repeatable IDPO template and links phenotype assignment to outcomes and utilization metrics. The main limitation is that many proposed pathways remain partly evidence-informed rather than evidence-proven. Conservative care is itself heterogeneous, and local resources differ substantially. Therefore, pathway definitions must often remain general, with explicit dosing, reassessment, and governance rather than overly detailed universal prescriptions.

Conclusions

Operational phenotyping for CLBP is a methodology for developing routable, measurable care pathways. Its core requirement is that every phenotype link feasible inputs to a decision, a deliverable pathway, and outcomes that can be audited. A minimal implementation can begin with safety screening, neurologic screening, pain duration/prior care, disability/function, psychosocial complexity, and an early response checkpoint. More advanced programs can add imaging, occupational assessment, digital activity monitoring, diagnostic procedures, or calibrated prediction models when those inputs change management. Used this way, phenotyping can support safer routing, shared decision-making, lower-value variation reduction, and more transparent escalation across spine care.

Acknowledgments

The authors thank the reviewers and editorial team for comments that improved the clarity, organization, and clinical applicability of the manuscript.

AI-Assisted Technology Disclosure: OpenAI ChatGPT, GPT-5.5 Thinking, was used on 2 May 2026 and 3 May 2026 for editorial assistance, including language refinement, structural organization, journal-instruction compliance review, and preparation of revision-response language. Representative prompts asked the tool to revise the manuscript for Annals of Palliative Medicine (APM) formatting, clarify reviewer responses, and improve organization without changing scientific conclusions. The AI tool was not used to generate primary data, perform independent literature screening, conduct statistical analysis, or make scientific conclusions. All scientific content, literature selection, interpretation, clinical recommendations, and final wording were reviewed, edited, and approved by the authors, who take full responsibility for the accuracy, integrity, and originality of the work.

Footnote

Reporting Checklist: The authors have completed the Narrative Review reporting checklist. Available at https://apm.amegroups.com/article/view/10.21037/apm-2026-1-0026/rc

Peer Review File: Available at https://apm.amegroups.com/article/view/10.21037/apm-2026-1-0026/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://apm.amegroups.com/article/view/10.21037/apm-2026-1-0026/coif). B.S.H. reports consulting fees and/or honoraria from MaxMore for surgeon teaching courses. A.R.V. receives royalty payments from Alphatec (Atec), Atlas Spine, Curiteva, Elsevier, Globus, Jaypee, Medtronic, Spinal Elements, SpineWave, Stryker Spine, Taylor & Francis/Hodder and Stoughton, Thieme, and Wheel House Medical. A.R.V. also reports consulting or independent-contractor relationships with Accellus, Curiteva, Ferring Pharmaceutical, Globus, Medcura, Spinal Elements, Stryker Spine, Wheel House Medical, PBC Biomedical, and Johnson & Johnson DePuy; expert testimony/AO Spine-related activity; leadership, fiduciary, scientific advisory board, board of directors, or committee roles with Accellus, the National Spine Health Foundation (NSHF), and Sentryx; and stock or stock-option interests in Accellus, Advanced Spinal Intellectual Properties, Atlas Spine, AVKN Patient Driven Care, Avaz Surgical, Cytonics, Deep Health, Dimension Orthotics LLC, Electrocore, Flagship Surgical, FlowPharma, Globus, Harvard Medtech, Innovative Surgical Design, Jushi, Orthobullets, Parvizi Surgical Innovation, Progressive Spinal Technologies, Rothman Institute and Related Properties, See All AI, Sentryx, Stout Medical, ViewFi Health, and Oxion Dental LLC. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Hassan S, Nesovic K, Babineau J, et al. Identifying chronic low back pain phenotypic domains and characteristics accounting for individual variation: a systematic review. Pain 2023;164:2148-90. [Crossref] [PubMed]
2. Kamper SJ, Apeldoorn AT, Chiarotto A, et al. Multidisciplinary biopsychosocial rehabilitation for chronic low back pain: Cochrane systematic review and meta-analysis. BMJ 2015;350:h444. [Crossref] [PubMed]
3. Jin MC, Azad TD, Fatemi P, et al. Defining and describing treatment heterogeneity in new-onset idiopathic lower back and extremity pain through reconstruction of longitudinal care sequences. Spine J 2021;21:1993-2002. [Crossref] [PubMed]
4. Daluiso-King G, Hebron C. Is the biopsychosocial model in musculoskeletal physiotherapy adequate? An evolutionary concept analysis. Physiother Theory Pract 2022;38:373-89. [Crossref] [PubMed]
5. Cormack B, Stilwell P, Coninx S, et al. The biopsychosocial model is lost in translation: from misrepresentation to an enactive modernization. Physiother Theory Pract 2023;39:2273-88. [Crossref] [PubMed]
6. Iriondo C. Characterizing Phenotypes of Musculoskeletal Degeneration Using Medical Imaging and Deep Learning. San Francisco: University of California; 2021.
7. Bravo A, Medina R. An unsupervised clustering framework for automatic segmentation of left ventricle cavity in human heart angiograms. Comput Med Imaging Graph 2008;32:396-408. [Crossref] [PubMed]
8. Nijs J, Kosek E, Chiarotto A, et al. Nociceptive, neuropathic, or nociplastic low back pain? The low back pain phenotyping (BACPAP) consortium's international and multidisciplinary consensus recommendations. Lancet Rheumatol 2024;6:e178-88. [Crossref] [PubMed]
9. Riley SP, Swanson BT, Cleland JA. The why, where, and how clinical reasoning model for the evaluation and treatment of patients with low back pain. Braz J Phys Ther 2021;25:407-14. [Crossref] [PubMed]
10. Quirk DA, Johnson ME, Anderson DE, et al. Biomechanical Phenotyping of Chronic Low Back Pain: Protocol for BACPAC. Pain Med 2023;24:S48-60. [Crossref] [PubMed]
11. Dolman AJ, Loggia ML, Edwards RR, et al. Phenotype matters: the absence of a positive association between cortical thinning and chronic low back pain when controlling for salient clinical variables. Clin J Pain 2014;30:839-45. [Crossref] [PubMed]
12. Vo NV, Piva SR, Patterson CG, et al. Toward the Identification of Distinct Phenotypes: Research Protocol for the Low Back Pain Biological, Biomechanical, and Behavioral (LB3P) Cohort Study and the BACPAC Mechanistic Research Center at the University of Pittsburgh. Pain Med 2023;24:S36-47. [Crossref] [PubMed]
13. Gholi Zadeh Kharrat F, Amar Kumar P, Mehling W, et al. Biobehavioral phenotypes of chronic low back pain: Psychosocial subgroup identification using latent profile analysis. Pain Med 2026;27:15-25. [Crossref] [PubMed]
14. Lötsch J, Ultsch A. A machine-learned knowledge discovery method for associating complex phenotypes with complex genotypes. Application to pain. J Biomed Inform 2013;46:921-8. [Crossref] [PubMed]
15. Vella S, Youssef P, Maher C, et al. Unwarranted escalation of care for back pain: a dilemma for emergency health services. Lancet Rheumatol 2025;7:e229-31. [Crossref] [PubMed]
16. Hill A, Joyner CH, Keith-Jopp C, et al. Assessing Serious Spinal Pathology Using Bayesian Network Decision Support: Development and Validation Study. JMIR Form Res 2023;7:e44187. [Crossref] [PubMed]
17. Sarikonda A, Leibold A, Sivaganesan A. When Does Intervention End and Surgery Begin? The Role of Interventional Pain Management in the Treatment of Spine Pathology. Curr Pain Headache Rep 2023;27:707-17. [Crossref] [PubMed]
18. Fu Y, Jenks AD, Rubinstein SM, et al. Prognostic ability of the sTarT back screening tool for disability and pain intensity outcomes in older adults with low back pain seeking chiropractic care: a multi-national external validation study. Chiropr Man Therap 2025;33:30. [Crossref] [PubMed]
19. Croft P, Hill JC, Foster NE, et al. Stratified health care for low back pain using the STarT Back approach: holy grail or doomed to fail? Pain 2024;165:2679-92. [Crossref] [PubMed]
20. Hill J, Try F, Agnew G, et al. Perspectives and experiences of physiotherapists and general practitioners in the use of the STarT Back Tool: a review and meta-synthesis. J Prim Health Care 2022;14:164-72. [Crossref] [PubMed]
21. Robarts S, Razmjou H, Yee A, et al. Validation of the Subgroups for Targeted Treatment for Back (STarT Back) screening tool at a tertiary care centre. Can J Surg 2022;65:E352-8. [Crossref] [PubMed]
22. Middleton A, Fitzgerald GK, Delitto A, et al. Implementing stratified care for acute low back pain in primary care using the STarT Back instrument: a process evaluation within the context of a large pragmatic cluster randomized trial. BMC Musculoskelet Disord 2020;21:776. [Crossref] [PubMed]
23. Vickery S, Junker F, Döding R, et al. Integrating multidimensional data analytics for precision diagnosis of chronic low back pain. Sci Rep 2025;15:9675. [Crossref] [PubMed]
24. Hoehl BU, Taheri N, Becker LA, et al. Detailed Analysis of Back Pain: A Cross-Sectional Study of Epidemiology and Previous Therapies in a German Population. Pain Pract 2025;25:e70072. [Crossref] [PubMed]
25. Kolodziejczak-Krupp K, Zipper V, Wilhelm LO, et al. Physical activity and psychosocial characteristics of individuals with and without chronic low back pain in daily life: protocol for the PRIA intensive longitudinal study. BMJ Open 2025;15:e109887. [Crossref] [PubMed]
26. Cruz-Almeida Y, Mehta B, Haelterman NA, et al. Clinical and biobehavioral phenotypic assessments and data harmonization for the RE-JOIN research consortium: Recommendations for common data element selection. Neurobiol Pain 2024;16:100163. [Crossref] [PubMed]
27. Verma D, Jansen D, Bach K, et al. Exploratory application of machine learning methods on patient reported data in the development of supervised models for predicting outcomes. BMC Med Inform Decis Mak 2022;22:227. [Crossref] [PubMed]
28. Hiyama A, Sakai D, Katoh H, et al. Machine learning insights into patient satisfaction following lateral lumbar interbody fusion. Eur Spine J 2025;34:3670-82. [Crossref] [PubMed]
29. Berg B, Gorosito MA, Fjeld O, et al. Machine Learning Models for Predicting Disability and Pain Following Lumbar Disc Herniation Surgery. JAMA Netw Open 2024;7:e2355024. [Crossref] [PubMed]
30. Staartjes VE, Stumpo V, Ricciardi L, et al. FUSE-ML: development and external validation of a clinical prediction model for mid-term outcomes after lumbar spinal fusion for degenerative disease. Eur Spine J 2022;31:2629-38. [Crossref] [PubMed]
31. Khor S, Lavallee D, Cizik AM, et al. Development and Validation of a Prediction Model for Pain and Functional Outcomes After Lumbar Spine Surgery. JAMA Surg 2018;153:634-42. [Crossref] [PubMed]
32. Halicka M, Wilby M, Duarte R, et al. Predicting patient-reported outcomes following lumbar spine surgery: development and external validation of multivariable prediction models. BMC Musculoskelet Disord 2023;24:333. [Crossref] [PubMed]
33. French SD, O'Connor DA, Green SE, et al. Improving adherence to acute low back pain guideline recommendations with chiropractors and physiotherapists: the ALIGN cluster randomised controlled trial. Trials 2022;23:142. [Crossref] [PubMed]
34. Killingmo RM, Storheim K, van der Windt D, et al. Healthcare utilization and related costs among older people seeking primary care due to back pain: findings from the BACE-N cohort study. BMJ Open 2022;12:e057778. [Crossref] [PubMed]
35. Stevans JM, Delitto A, Khoja SS, et al. Risk Factors Associated With Transition From Acute to Chronic Low Back Pain in US Patients Seeking Primary Care. JAMA Netw Open 2021;4:e2037371. [Crossref] [PubMed]
36. Foster NE, Konstantinou K, Lewis M, et al. Stratified versus usual care for the management of primary care patients with sciatica: the SCOPiC RCT. Health Technol Assess 2020;24:1-130. [Crossref] [PubMed]
37. Baker SA, Billmire DA, Bilodeau RA, et al. Wearable Nanocomposite Sensor System for Motion Phenotyping Chronic Low Back Pain: A BACPAC Technology Research Site. Pain Med 2023;24:S160-74. [Crossref] [PubMed]
38. Mageswaran P, Dufour J, Aurand A, et al. Wearable sensor-based spinal motion assessments for identifying phenotypic clusters in chronic low back pain. Eur Spine J 2026; Epub ahead of print. [Crossref] [PubMed]
39. Ryan JJ, Bowden E, Hancock MM, et al. Wearable sensor technologies for individuals with back pain: a scoping review. Pain Manag 2025;15:855-76. [Crossref] [PubMed]
40. Hassett AL, Williams DA, Harris RE, et al. An Interventional Response Phenotyping Study in Chronic Low Back Pain: Protocol for a Mechanistic Randomized Controlled Trial. Pain Med 2023;24:S126-38. [Crossref] [PubMed]
41. Gombatto SP, Bailey B, Bari M, et al. Identifying Clinical Phenotypes in People Who Are Hispanic/Latino With Chronic Low Back Pain: Use of Sensor-Based Measures of Posture and Movement, Pain, and Psychological Factors. Phys Ther 2024;104:pzad185. [Crossref] [PubMed]
42. Herrero P, Ríos-Asín I, Lapuente-Hernández D, et al. The Use of Sensors to Prevent, Predict Transition to Chronic and Personalize Treatment of Low Back Pain: A Systematic Review. Sensors (Basel) 2023;23:7695. [Crossref] [PubMed]
43. Corp N, Mansell G, Stynes S, et al. Evidence-based treatment recommendations for neck and low back pain across Europe: A systematic review of guidelines. Eur J Pain 2021;25:275-95. [Crossref] [PubMed]
44. Tahvonen P, Oikarinen H, Tervonen O. The effect of interventions on appropriate use of lumbar spine radiograph and CT examinations in young adults and children: a three-year follow-up. Acta Radiol 2020;61:1042-9. [Crossref] [PubMed]
45. Johnson SM, Hutchins T, Peckham M, et al. Effects of implementing evidence-based appropriateness guidelines for epidural steroid injection in chronic low back pain: the EAGER (Esi Appropriateness GuidElines pRotocol) study. BMJ Open Qual 2019;8:e000772. [Crossref] [PubMed]
46. Morgan T, Wu J, Ovchinikova L, et al. A national intervention to reduce imaging for low back pain by general practitioners: a retrospective economic program evaluation using Medicare Benefits Schedule data. BMC Health Serv Res 2019;19:983. [Crossref] [PubMed]
47. Coyle MJ, Roffey DM, Phan P, et al. The Use of a Self-Administered Questionnaire to Reduce Consultation Wait Times for Potential Elective Lumbar Spinal Surgical Candidates: A Prospective, Pragmatic, Blinded, Randomized Controlled Quality Improvement Study. J Bone Joint Surg Am 2018;100:2125-31. [Crossref] [PubMed]
48. Suman A, Schaafsma FG, van de Ven PM, et al. Effectiveness of a multifaceted implementation strategy compared to usual care on low back pain guideline adherence among general practitioners. BMC Health Serv Res 2018;18:358. [Crossref] [PubMed]
49. Chen D, Bhambhvani HP, Hom J, et al. Effect of Electronic Clinical Decision Support on Imaging for the Evaluation of Acute Low Back Pain in the Ambulatory Care Setting. World Neurosurg 2020;134:e874-7. [Crossref] [PubMed]
50. Kim EJ, Chotai S, Stonko DP, et al. Patient-reported outcomes after lumbar epidural steroid injection for degenerative spine disease in depressed versus non-depressed patients. Spine J 2017;17:511-7. [Crossref] [PubMed]
51. Tanyel Saraçoğlu T, Erken B. Impact of transforaminal epidural steroid injection on pain and disability outcomes by lumbar intervertebral disc herniation class: a prospective study. Pain Med 2025;26:440-50. [Crossref] [PubMed]
52. Verheijen EJA, van Haagen OBHAM, Bartels EC, et al. Prediction of transforaminal epidural injection success in sciatica (POTEISS): a protocol for the development of a multivariable prediction model for outcome after transforaminal epidural steroid injection in patients with lumbar radicular pain due to disc herniation or stenosis. BMC Neurol 2024;24:290. [Crossref] [PubMed]
53. Riegger M, Le H, van Kuijk SMJ, et al. Intradiscal Glucocorticoid Injection in Discogenic Back Pain and Influence on Modic Changes: A Systematic Review and Meta-analysis of RCTs. Pain Physician 2023;26:E449-65. [Crossref] [PubMed]
54. Abbott A, Pedersen CF, Hedevik H, et al. External validation of a prediction model for disability and pain after lumbar disc herniation surgery: a prospective international registry-based cohort study. Acta Orthop 2025;96:512-20. [Crossref] [PubMed]
55. Hébert JJ, Bigney EE, Nowell S, et al. Outcome prediction following lumbar disc surgery: a longitudinal study of outcome trajectories, prognostic factors, and risk models. J Neurosurg Spine 2025;42:33-42. [Crossref] [PubMed]
56. Harvie DS, Keating C, Fulia N, et al. Predictors of willingness to undergo back surgery: a survey of Australian privately insured adults with chronic back pain. Aust Health Rev 2025;49:AH25242. [Crossref] [PubMed]
57. Rundell SD, Saito A, Meier EN, et al. The Lumbar Stenosis Prognostic Subgroups for Personalizing Care and Treatment (PROSPECTS) study: protocol for an inception cohort study. BMC Musculoskelet Disord 2022;23:692. [Crossref] [PubMed]
58. Moojen WA, Schenck CD, Lycklama À. Preoperative MRI in Patients With Intermittent Neurogenic Claudication: Relevance for Diagnosis and Prognosis. Spine (Phila Pa 1976) 2018;43:348-55. [Crossref] [PubMed]
59. Roszell K, Sandella D, Haig AJ, et al. Spinal Stenosis: Factors That Influence Patients' Decision to Undergo Surgery. Clin Spine Surg 2016;29:E509-13. [Crossref] [PubMed]
60. Willems P. Decision making in surgical treatment of chronic low back pain: the performance of prognostic tests to select patients for lumbar spinal fusion. Acta Orthop Suppl 2013;84:1-35. [Crossref] [PubMed]
61. Hägg O, Fritzell P, Ekselius L, et al. Predictors of outcome in fusion surgery for chronic low back pain. A report from the Swedish Lumbar Spine Study. Eur Spine J 2003;12:22-33. [Crossref] [PubMed]
62. Perkins FM, Kehlet H. Chronic pain as an outcome of surgery. A review of predictive factors. Anesthesiology 2000;93:1123-33. [Crossref] [PubMed]
63. Evans K, Ko J, Ceprnja D, et al. Development and Implementation of MyPainHub, a Web-Based Resource for People With Musculoskeletal Conditions and Their Health Care Professionals: Mixed Methods Study. JMIR Form Res 2025;9:e63780. [Crossref] [PubMed]
64. Geraghty AWA, Becque T, Roberts LC, et al. Supporting self-management with an internet intervention for low back pain in primary care: a RCT (SupportBack 2). Health Technol Assess 2025;29:1-90. [Crossref] [PubMed]
65. Greenlee TA, George SZ, Pickens B, et al. Risk-stratified Care Improves Pain-related Knowledge and Reduces Psychological Distress for Low Back Pain: A Secondary Analysis of a Randomized Trial. Clin Orthop Relat Res 2025;483:607-20. [Crossref] [PubMed]
66. Rashed R, Kowalski K, Walton D, et al. Physical measures of physical functioning as prognostic factors to predict outcomes in low back pain: Protocol for a systematic review. PLoS One 2023;18:e0295761. [Crossref] [PubMed]
67. Rothman R, Sokunbi G, Malik GR, et al. Comparison of baseline PROMIS-10 scores of mental and physical health of patients with low back pain evaluated by physiatrists and orthopedic spine surgeons. PM R 2025;17:1069-79. [Crossref] [PubMed]
68. Bernaers L, Cnockaert E, Braeckman L, et al. Disability and return to work after a multidisciplinary intervention for (sub)acute low back pain: A systematic review. Clin Rehabil 2023;37:964-74. [Crossref] [PubMed]
69. Roseen EJ, Gerlovin H, Felson DT, et al. Which Chronic Low Back Pain Patients Respond Favorably to Yoga, Physical Therapy, and a Self-care Book? Responder Analyses from a Randomized Controlled Trial. Pain Med 2021;22:165-80. [Crossref] [PubMed]
70. Simula AS, Malmivaara A, Booth N, et al. A classification-based approach to low back pain in primary care - protocol for a benchmarking controlled trial. BMC Fam Pract 2020;21:61. [Crossref] [PubMed]
71. Matsudaira K, Oka H, Oshima Y, et al. Development of the Japanese Core Outcome Measures Index (COMI): cross-cultural adaptation and psychometric validation. BMC Musculoskelet Disord 2018;19:71. [Crossref] [PubMed]
72. Storm L, Rousing R, Andersen MO, et al. Usefulness of the STarT Back Screening Tool to predict pain problems after lumbar spine surgery. Dan Med J 2018;65:A5517. [PubMed]
73. Abdel Shaheed C, McFarlane B, Maher CG, et al. Investigating the Primary Care Management of Low Back Pain: A Simulated Patient Study. J Pain 2016;17:27-35. [Crossref] [PubMed]
74. Rushton A, Heneghan NR, Calvert M, et al. Physiotherapy Post Lumbar Discectomy: Prospective Feasibility and Pilot Randomised Controlled Trial. PLoS One 2015;10:e0142013. [Crossref] [PubMed]
75. Li C, Hu J, He C. Analgesic efficacy of transcranial combined peripheral magnetic stimulation in chronic nonspecific low back pain: a fNIRS study. Eur J Phys Rehabil Med 2025;61:93-101. [Crossref] [PubMed]
76. Lyons KM, Stout JR, Stock MS, et al. The Identification of Pain Phenotypes in Individuals With Low Back Pain in Response to Dynamic Resistance Exercise. Eur J Pain 2025;29:e70136. [Crossref] [PubMed]
77. Beall DP, Bae HW, DePalma MJ, et al. Efficacy and safety of allogeneic mesenchymal precursor cells with and without hyaluronic acid for treatment of chronic low back pain: a prospective, randomized, double blind, concurrent-controlled 36-month study. Spine J 2025;25:1997-2013. [Crossref] [PubMed]
78. Foster NE, Anema JR, Cherkin D, et al. Prevention and treatment of low back pain: evidence, challenges, and promising directions. Lancet 2018;391:2368-83. [Crossref] [PubMed]
79. Islam S, Frey N. Psychotherapy for the Treatment of Acute Musculoskeletal Pain: A Review of Clinical Effectiveness and Guidelines. Ottawa: Canadian Agency for Drugs and Technologies in Health; 2020.
80. Lin D, Koenig C, Kaplan S, et al. Investigating the impact of context on zero-shot entity resolution: applications in chronic lower back pain. In: 2025 8th International Conference on Information and Computer Technologies (ICICT). IEEE; 2025:390-6.
81. Nachemson AL. Newest knowledge of low back pain. A critical look. Clin Orthop Relat Res 1992;8-20. [Crossref] [PubMed]
82. Violante FS, Mattioli S, Bonfiglioli R. Low-back pain. Handb Clin Neurol 2015;131:397-410. [Crossref] [PubMed]
83. Wong AYL, Karppinen J, Samartzis D. Low back pain in older adults: risk factors, management options and future directions. Scoliosis Spinal Disord 2017;12:14. [Crossref] [PubMed]
84. Giaccone P, de Rinaldis E, D'Antoni F, et al. Exploring the Biopsychosocial Impact of Chronic Low Back Pain in Workers Through Artificial Intelligence-Driven Phenotyping. JOR Spine 2025;8:e70110. [Crossref] [PubMed]
85. Carlesso LC, Tousignant-Laflamme Y, Shaw W, et al. Exploring pain phenotypes in workers with chronic low back pain: Application of IMMPACT recommendations. Can J Pain 2021;5:43-55. [Crossref] [PubMed]
86. Wang Z, Feng W, Jin Q. Occupational factors and low back pain: a Mendelian randomization study. Front Public Health 2023;11:1236331. [Crossref] [PubMed]
87. McCarthy J. Phenotype. In: Gellman MD, editor. Encyclopedia of Behavioral Medicine. Cham: Springer International Publishing; 2020:1664.
88. Huckvale K, Venkatesh S, Christensen H. Toward clinical digital phenotyping: a timely opportunity to consider purpose, quality, and safety. NPJ Digit Med 2019;2:88. [Crossref] [PubMed]
89. Liu F, Panagiotakos D. Real-world data: a brief review of the methods, applications, challenges and opportunities. BMC Med Res Methodol 2022;22:287. [Crossref] [PubMed]
90. Onnela JP. Opportunities and challenges in the collection and analysis of digital phenotyping data. Neuropsychopharmacology 2021;46:45-54. [Crossref] [PubMed]