Gait assessment and prognosis in low back pain

Executive assessment

A walking assessment in low-back pain is useful when its intended meaning is explicit. Short-distance walking measures capacity under specified instructions; sustained walking adds endurance and symptom tolerance; laboratory gait analysis describes movement strategy; and daily-life monitoring describes activity in a particular environment. None of these, alone, identifies the anatomical cause of pain, establishes that a movement is harmful, or provides a validated forecast of disability, work participation or falls.

For chronic nonspecific low-back pain, standardized walking tests can complement patient-reported function. The older measurement literature supports the 50-foot fast walk and 5-minute walk more consistently than it supports patient-important change thresholds. High reliability does not remove substantial day-to-day error. Moreover, some numerical summaries require original-source resolution before use: one influential review contains Simmonds reliability values that are difficult to reconcile with the original abstract, and the frequently quoted important-change estimate from a 400-m/200-m walking study was anchored to a disability questionnaire's detectable-change criterion. That is not equivalent to a direct patient judgment of important walking improvement. [1‑8]

People with persistent pain often walk more slowly and with shorter strides, but this is a group tendency with considerable overlap. Reviews disagree in their certainty descriptions because they use different eligibility rules, tasks and appraisal methods. Their pooled differences are neither diagnostic cut-offs nor within-person rehabilitation targets. Imposing the same speed may help isolate mechanics but removes a clinically relevant self-selected adaptation. [9, 10]

The technological opportunity is real but output-specific. A 2026 smartphone-IMU study showed strong internal gait-event detection in older adults including 18 participants with nonspecific low-back pain, while transfer to another dataset required substantial retraining. A separate 2026 pain-phenotyping model excluded the intermediate pain group and developed its models within a very small cohort; its own calibration results argue against individual probability estimates. These findings support research and transparent measurement, not automated pain diagnosis or prognostic scoring. [11, 12]

For rehabtools, the most defensible first application is a protocol-labelled walking result with pain, assistance and data-quality information, followed by a transparent comparison with the person's previous performance. A future-risk claim requires a distinct prospective validation programme. This report does not provide individual treatment advice.

Scope and search approach

The principal population is adults with nonspecific or primary low-back pain, especially chronic presentations. Acute, subacute and recurrent pain are not assumed to have the same measurement properties. Radicular symptoms, confirmed radiculopathy, spinal stenosis, postoperative cohorts and specific spinal disease are identified separately when they inform an evidence boundary. Studies of all spinal disorders are not pooled into nonspecific low-back-pain evidence.

This is a critical narrative synthesis, not a registered systematic review or exhaustive dual-screened review. Three dated PubMed queries covering measurement, prognosis, and protocol/mechanism/technology retrieved 247, 1,047 and 697 records respectively. Connector pagination did not preserve all unique identifiers: the corresponding streams contained 182, 637 and 474 unique PubMed IDs. Official NCBI ESearch/EFetch reconciliation recovered every record in each query, without missing IDs. Native Scopus queries returned 545, 3,004 and 1,831 unique records, with all pages archived and totals reconciled. These query totals overlap and must not be added to imply unique studies.

Population and clinical state define the interpretation

Chronic does not mean homogeneous

A duration threshold is not a complete clinical phenotype. Persistent daily pain, intermittent pain recurring over several months, an acute flare in a person with long-standing pain, and testing during a remission period can all be described as chronic or recurrent low-back pain. Record the study's actual definition and the symptom state at testing. The 2022 gait review included 97 studies, but only two concerned acute pain; its findings therefore predominantly concern persistent presentations. The 2019 capacity review and its subsequent author response similarly caution against generalizing chronic-pain findings to subacute pain. [1, 9, 13]

Nonspecific low-back pain is a clinical category, not a guarantee of identical mechanisms. Age, coexisting hip or knee symptoms, obesity, cardiovascular capacity, neurologic impairment, analgesic effects and prior surgery can influence walking. Narrow research samples are particularly important here. Bagheri's reliability study enrolled adults aged 30–40 years, excluded BMI above 25 kg/m², and excluded below-knee radiation and neurologic signs. Its kinematic error estimates are not population norms for older, heavier or more disabled patients. [14]

Leg pain needs its own descriptor. Referred pain without neurologic involvement, radicular pain, and confirmed radiculopathy should not be collapsed into one binary field. Simmonds and colleagues demonstrated that pain distribution was associated with walking performance and horizontal ground-reaction forces. This is a reason to preserve symptom distribution during assessment, not a reason to infer a nerve lesion from the gait trace. [15]

Spinal stenosis deserves a separate assessment pathway because neurogenic claudication, posture-dependent symptoms and stopping distance can dominate the task. Surgical selection and postoperative recovery create further differences. A highly reproducible stenosis walking test or a postoperative TUG change threshold is not automatically suitable for nonspecific low-back pain. Even a multidisciplinary secondary-care cohort may be mixed: the 2026 EL DORADO feasibility report contained diagnostic codes for nonspecific pain, stenosis and disc herniation. Its descriptive walking results cannot be treated as nonspecific-LBP reference values. [1, 16]

Pain fear and observed performance

Walking is an observable activity, but performance is not independent of pain or context. Anticipated pain, confidence, perceived threat, task instructions, observation and effort can affect pace. The relationship is not adequately described by saying that a slow result is either physical or psychological. Fear-related constructs show associations with maximal physical performance across a heterogeneous literature; they do not explain every individual's result and do not establish intentional underperformance. [17‑19]

Ask about the task rather than infer the reason for a strategy from appearance. A person may reduce speed to manage symptoms, preserve balance, accommodate another painful joint, or follow an interpretation of the instruction. The same measured time can arise from several combinations of capacity and strategy. Conversely, normal speed does not establish absence of pain or participation restriction.

Movement-evoked pain can add information beyond a resting pain rating. In older adults, pain recorded after a chair-rise, walking and stair battery was associated with contemporaneous self-reported and observed function after covariate adjustment. Another study of 39 older adults examined a purpose-built movement-evoked-pain task and usual gait speed. These are useful assessment studies, but their baseline associations are not proof that a walking test predicts later disability. [20, 21]

Clinical walking tests select the construct before the distance

Short walking capacity

Comfortable and fast walking are different challenges. Comfortable pace documents the selected strategy in a standardized setting. Fast walking, with running prohibited, tests a more demanding capacity. A person can have a relatively preserved comfortable pace but limited ability to increase speed. Reporting both can be useful if the added task addresses a clinical question, but the difference between them is not itself a validated diagnostic score.

The 50-foot test has direct historical support in low-back pain. It should not be silently relabelled a 10-m walk. Fifty feet is approximately 15.24 m, but published versions include different path shapes and turns. Smeets used an eight-shaped course, while contemporary straight-path testing may have a moving start and a separately timed central segment. Distance divided by time gives average speed only for the distance actually timed; it does not remove differences in acceleration or turning demands. [2, 3]

A 4-m, 10-m or other short straight walk may be feasible and clinically descriptive without having a validated LBP-specific minimal important change. Piva's large chronic-LBP feasibility study included a 4-m gait-speed test, but it was not designed to establish its test–retest error or patient-important threshold. That distinction matters when a practical programme adopts a convenient test: feasibility is necessary but does not complete the measurement argument. [22]

The Timed Up and Go combines chair rise, walking, turning and sitting. It is useful as a composite mobility measure, not a pure gait-speed measure. Total time cannot identify which component changed. A video-derived walking segment can be reported separately, but its validity and error then need evaluation as a new output. The same principle applies to the Dubousset Functional Test, whose components include walking, steps, floor-related transfers and dual tasking. [1, 23]

Sustained and externally paced walking

The 5-minute walk is a well-established LBP capacity task, often performed as fast and far as possible without running. Course length, shape, rests and aid rules are part of the test. The 6-minute walk is another sustained task, but a 5-minute result should not be multiplied by 6/5 to manufacture a 6-minute distance. Symptoms and speed need not evolve linearly.

The distinction is supported directly by the comparison published online in 2023 of the first two minutes of a 6-minute test with the complete test. In 124 people with LBP, the measures correlated strongly, yet tripling the early distance overestimated the observed 6-minute distance by 46.8 m on average, with substantial variability. Crucially, the shorter measure was embedded inside the longer test. That design does not fully establish equivalence to a separately instructed, standalone 2-minute walk. [8]

A 400-m comfortable walk and a 200-m fast walk can probe sustained performance at different intensities. They are distance-limited tests, so time and speed are the main outputs, whereas the 5- or 6-minute tests are time-limited and yield distance. The multimodal-rehabilitation study, published online in 2019, included 127 participants with or without radicular pain. It supports the feasibility of those particular distance-limited tasks but does not establish that they are interchangeable with shorter clinical tests. [7]

An externally paced shuttle test imposes increasing or prescribed pacing demands and repeated turns. Taylor's chronic-LBP work is relevant to reliability and responsiveness, but this is not a measure of spontaneous gait mechanics. Its value depends on whether paced endurance and symptom tolerance answer the clinical question. A symptom-limited treadmill test in stenosis has a different objective again. [6]

Table 1 Choose the walking assessment for its intended construct

Practical synthesis; choose the smallest set that answers the assessment question.

Reliability error and patient important change

Relative ranking and absolute agreement answer different questions

An intraclass correlation coefficient describes reproducibility relative to between-person variation. A heterogeneous sample can produce a high ICC despite sizeable individual error. The ICC model, agreement versus consistency definition, use of one or averaged performances, retest interval and rater design should accompany the value. Simultaneous stopwatch raters primarily evaluate scoring disagreement; measurements taken on different days include real short-term fluctuation, task execution and setup variation.

Simmonds's foundational study included 44 people with LBP and 48 controls and tested a battery twice on two days. The original abstract supports high intertester reliability and shows that day-to-day reliability was less impressive, with improvement when selected tasks were averaged. The 2019 review's table assigns a very low same-session value to chair rising even though the original abstract assigns that exception to repeated trunk flexion. Its displayed 5-minute-walk day-to-day value also sits uneasily with the abstract's overall range. Those secondary-table values are withheld from implementation pending the original tables. [1, 2]

Smeets studied 53 participants over 5–9 days. The sample was labelled nonspecific CLBP, but 50.9% reported below-knee radiation; this does not establish confirmed radiculopathy. The figure-eight courses were 30 m for the 5-minute walk and 50 feet for the fast walk. The 5-minute walk ICC was 0.89 and its limits-of-agreement half-width 82.7 m. For the 50-foot fast walk, analysed in 52 participants, ICC was 0.76; the reported 3.9-s error summary was back-converted from inverse time at the sample mean. These are protocol- and scale-dependent error estimates, not universal speed thresholds. [3]

A Brazilian study of 30 participants with LBP also found high reproducibility of several performance tasks, but small samples, a restricted clinical setting and differing protocols constrain transportability. The primary abstract does not supply all task-specific absolute-error estimates needed for an app. A reassuring ICC alone is therefore not enough to populate a change alert. [5]

For two independent measurements with the same standard error, the conventional individual MDC95 is 1.96 × square root of 2 × SEM. That formula is only appropriate when the SEM and error model match the comparison. Averaged trials, unequal errors, transformations and systematic learning effects require care. A group-mean confidence interval is not an individual MDC, and a limit-of-agreement interval may include nonzero bias. The app should preserve the actual estimate and its assumptions rather than rename every error statistic “MCID.”

Responsiveness and importance

Andersson followed 198 people with chronic nonspecific LBP through 10 weeks of treatment, using global perceived effect to classify improvement. Only chair rising and stair climbing achieved the prespecified responsiveness criterion; the walking tests did not. This does not mean walking is useless. It means that the observed changes were not sufficiently aligned with the selected global anchor in that cohort to support the same evaluative claim. [4]

A global recovery anchor can reflect pain, confidence, sleep and valued activities beyond walking. Its correlation with walking change, the anchor question and response categories, the definition of “improved,” and baseline dependence need inspection before interpreting a threshold. An intervention can improve average walking distance while failing to produce a useful individual important-change threshold.

The 2019 400-m/200-m study reported speed improvements and an important-change estimate linked to the Quebec disability scale. Its abstract specifies that the questionnaire threshold was a minimum detectable change. That anchor identifies change beyond an error criterion, rather than directly establishing patient-perceived importance. The reported 95% confidence interval of 0.14–0.22 m/s around the study's important-change estimate should therefore not become a universal LBP gait-speed responder rule. Original tables and anchor details remain an acquisition priority. [7]

The Dubousset paper also illustrates a terminology problem. It reports “MCID” values in a one-hour retest study, with accompanying SEMs. The abstract does not describe a longitudinal patient anchor. Those values cannot be treated as patient-important rehabilitation changes on the available evidence. [23]

The practical conclusion is to report observed change first. Add a protocol-matched error estimate if available. Label patient-important improvement only when a relevant anchor-based threshold has been justified, and preserve uncertainty when a change is important to the person but smaller than the measurement's individual error.

Table 2 Consequential measurement claims and implementation limits

MDC/SDC/LoA describe error; MIC describes importance. These are not interchangeable.

Instrumented gait what does an apparent abnormality mean

Group differences and speed dependence

Smith and colleagues found slower walking, shorter strides, more in-phase thorax–pelvis coordination and greater paraspinal activation in persistent LBP. Their review did not establish consistent reductions in the motion amplitude of every trunk or lower-limb segment. Dal Farra's narrower 2025 review of spontaneous walking reported lower speed, cadence and step length but rated that evidence very low. Different study sets and appraisal systems explain part of the apparent certainty contrast; neither establishes a single pathological gait signature. [9, 10]

Speed is both an outcome and a mechanical modifier. Comparisons at self-selected speeds include the person's pace adaptation. Comparisons at matched or imposed speeds answer a narrower mechanical question and can make the task relatively more demanding for the pain group. Report both speed conditions when available, rather than assume that adjustment removes all confounding. Treadmill and overground walking, steady-state and turning, and single- versus dual-task performance also deserve separate interpretation.

A rigid-looking trunk may be a protective strategy, a response to instruction, or one of several adaptations. Reduced variability is not always better, and increased variability is not invariably instability. Coordination depends on the segments, angle conventions, signal processing and phase calculation. It cannot be inferred reliably from one visually unusual stride, nor can a group difference establish the causal direction between pain and movement.

Reproducibility is variable specific

Fernandes's 2015 three-dimensional gait study directly addressed test–retest reliability and detectable change in CLBP. The 2016 corrigendum corrects the third author's name only. The separate 2015 original-table gap remains, so individual numerical thresholds are still withheld. [24, 25]

Bagheri's complete body provides a more inspectable example. Forty selected participants completed six barefoot self-paced 10-m trials, with the same assessor and a 6–11-day retest. Reliability differed between spinal regions, planes and sex strata; some lower-lumbar estimates had very wide confidence intervals. Between-session MDCs for all reported movements ranged into double-digit degrees, while selected lumbar-segment measures were more precise. The 2018 correction concerned affiliation, not a change to the measurement results. [14, 26]

These results do not authorize a generic “lumbar motion MDC.” A thorax-relative-to-pelvis angle, upper-lumbar angle and lower-lumbar angle are different quantities. Surface markers also estimate skin-defined segments rather than direct intervertebral bone motion. Error from marker replacement, soft tissue, event identification and differentiation should be considered for each output.

Moissenet's study assessed 72 candidate biomarkers in 30 symptomatic and 30 asymptomatic participants, including walking and chair-rise tasks. Only four met the authors' combined reliability and discrimination criteria; all four concerned forward bending. This independent assessment tempers the idea that any previously significant walking feature is ready for individual monitoring. Its MDC-based “interpretability” was not patient-anchored importance. [27]

Wearables video and practical technology

Measure an output before making a clinical claim

Three separate questions should be asked. Does the device agree with an appropriate reference? Is its output reproducible in the target population and setting? Does that output improve a clinical decision? A model may answer the first question while leaving the other two open. Correlation alone is inadequate for agreement; inspect bias, limits of agreement, missed events and clinically relevant error across the performance range.

Vivar's 2026 work used a shank-mounted smartphone IMU, not ordinary hand-held video. Its internal dataset comprised 28 healthy older adults and 18 people with nonspecific LBP. Participant-independent splitting reduced leakage between training and testing. Internal reference labels were algorithm-assisted, manually reviewed events derived from the same shank angular-velocity signal; Vicon data were not used for annotation or analysis. The reported event detection and stance/stride agreement therefore concern that labeling workflow, rather than an independent biomechanical reference. Zero-shot external transfer failed and full-network fine-tuning was required, using labelled data from half of a separate 22-person healthy young-adult dataset. Thus, changing the phone, mounting or acquisition conditions is not a minor implementation detail. The study does not validate free-living LBP prognosis. [11]

Koltermann's 2026 model provides a different caution. Of 36 matched participants, the principal analysis retained 25 after excluding intermediate pain. Many features and model configurations were explored. Internal discrimination of pain strata does not establish an objective pain meter; the reported calibration slope was extremely low, and there was no independent external cohort. A practical tool should not convert those outputs into personal pain-severity or future-disability probabilities. [12]

Markerless feasibility should be kept separate from criterion validity. The BACPAC protocol describes ambitious multimodal biomechanical measurement, including depth-camera chair rising, but a protocol is not evidence that every output is validated for gait in symptomatic patients. EL DORADO demonstrates real-world acquisition challenges: gait on a walkpad was substantially less usable than some other motion tasks, owing partly to balance-support requirements and tracking limitations. An application should retain failed or unanalysable recordings in its denominator. [16, 28]

Daily life walking is a distinct outcome

Steps, time walking, bout distribution and intensity reflect participation opportunity as well as capacity. Weather, work, caregiving, transport and device wear can alter them. A low count is not necessarily reduced physical capacity, and an increase is not necessarily a reduction in pain-related disability.

A small activity-tracker feasibility study followed 17 people with LBP. Average wear-and-sync adherence was approximately 70% over 26 weeks, and correlations with questionnaire-reported activity were weak. The questionnaire was not a criterion-standard measure of steps. Nonwear, account problems and lost devices complicate interpretation; excluding zero-step days can change the resulting average. Such findings support careful feasibility design rather than a blanket claim that consumer step data are either valid or invalid. [29]

Table 3 Technology evidence is specific to the output and setting

Training/test splits must separate people, and the full measurement pipeline must match deployment.

Prognosis separate later outcomes from concurrent labels

Persistent disability and functional recovery

The evidence appraised here does not justify a single gait-speed threshold for future disability in nonspecific LBP. A cross-sectional study found better walking and obstacle performance in people classified as low risk by the STarT Back Tool. The outcome in that analysis was the current risk category, not observed subsequent disability. Adding gait measures to a questionnaire requires prospective testing of incremental prediction, calibration and clinical utility. [30]

Treatment-response correlations require the same caution. In an older obese cohort, change in lumbar strength explained part of change in walking endurance during a resistance-exercise study. The between-group walking-endurance improvement was not statistically significant (P = .11), while gait speed favoured total-body training (P < .05). The strength–endurance result is a change–change relationship under an intervention, not an independently validated baseline gait predictor of natural recovery. Such analyses may illuminate mechanisms or candidate outcomes without supplying a forecast suitable for a new patient. [31]

Falls

Prospective evidence that LBP-related disability is associated with later falls exists, including the LOHAS cohort. However, its exposure was questionnaire-defined disability, not a gait test. Of 2,738 enrolled residents aged at least 60 years, 1,358 contributed to the analysis, and falls were recalled at one year. This supports attention to falls in older people with LBP, while leaving gait-specific incremental prediction and threshold transportability unresolved. [32]

Concurrent sway, slower walking, fear of falling, a retrospective history of falls and prospectively observed falls are not interchangeable endpoints. A gait app should not display a future-fall probability based on a group difference or on a cut-off imported from healthy older adults, stroke or knee OA.

Work and incident pain

Return to work depends on job demands, workplace conditions, expectations, compensation context and available accommodations as well as capacity. A rehabilitation cohort examining work-related outcomes did not identify baseline walking tolerance as a convincing individual predictor. Its endpoint required meeting the reported job-demand category and workplace reintegration at programme completion; its retrospective design and endpoint differ from prospective sustained employment. Passing a walking task therefore does not establish fitness for a particular job. [33]

The HUNT cohort provides genuine temporal evidence about walking and incident chronic LBP. In more than 11,000 adults without chronic LBP at baseline, device-measured walking was associated with subsequent chronic-LBP risk over approximately four years. This is prevention-oriented population evidence. It does not validate a clinical walking test as a prognostic instrument among patients who already have chronic pain, nor does it establish a treatment dose from observational associations. [34]

Table 4 A temporal test for claims described as prediction

A future-risk model also needs calibration, external validation and useful incremental prediction.

Protocol checklist and reporting for rehabtools

Record the clinical population and current state: pain duration and recurrence pattern; pain distribution; current and movement-evoked pain; relevant neurologic or claudication features; other limiting conditions; recent medication or symptom changes; and whether a walking aid is ordinarily used.

Record the task: exact path and timed distance; course shape and turn rules; comfortable, fast or externally imposed speed; standing or moving start; acceleration/deceleration allowance; footwear; aid and assistance; practice trials; number of scored trials; aggregation; rest; encouragement; stopping rules; and noncompletion.

Record the measurement: timing method; device and algorithm version; sensor placement or camera view; calibration; sampling frequency; filtering and event detection; trial exclusions; missingness; and the reference population for any interpretation.

Display the observation separately from interpretation. A suitable output is “comfortable 10-m speed under this protocol,” accompanied by pain and aid context. Change can be shown numerically even where no defensible responder threshold exists. If an error estimate is displayed, name its source population, retest design and confidence level. If those conditions differ materially, use a caution rather than a green/red decision rule.

A safe development sequence is protocol reproducibility, criterion agreement, between-day measurement error, responsiveness with a patient-relevant anchor, and then prospective outcome prediction. External validation should include symptomatic people with varying disability, body size, age and aid use, as well as the intended home environment. Testing a model on new frames from the same people is not external validation.

Limitations and research priorities

The strongest available conclusions concern measurement selection and limits of interpretation. Direct LBP-specific evidence remains uneven across common short walks, newer sensors and patient-important change. Several foundational original bodies and detailed supplements were inaccessible through the permitted routes, so this report deliberately withholds some frequently cited numerical thresholds. It does not substitute values from healthy cohorts or other diagnoses.

Priorities are repeated-measures studies with stable-state criteria and explicit pain context; patient-anchored responsiveness studies; output-specific validation of low-cost technology; prospective functional, participation and falls outcomes; and transparent treatment of inability and missing recordings. Future models should demonstrate improvement over routine clinical information, report calibration and validate beyond their development setting.

The central conclusion is practical: use walking to document what the person can do under clearly stated conditions, and use kinematics to describe how the task was accomplished. Greater measurement detail is valuable only when its error and clinical meaning are equally clear.

Primary study characteristics

The study profiles preserve the population, design, protocol, measurement findings, change interpretation, later outcomes and limitations for each appraised original study. A source can be useful for one question while remaining insufficient for another. Contextual and mixed-population studies are explicitly identified, and related publications are not assumed to represent independent cohorts.

Simmonds 1998

Study and population [2] Direct measurement study. 44 LBP; 48 controls; nonspecific/mechanical, variable duration

Protocol Nine-task battery, twice on two days; 50-ft preferred/fast, 5-min walk, 5 STS among tasks

Measurement and change Original abstract: intertester ICC(1,1)>0.95; day-to-day LBP ICC range 0.59–0.88; averaging selected tasks improved reliability. No patient-anchored MIC in original abstract

Later outcomes and interpretation No later clinical outcome. Original tables unavailable; review entries for same-day STS and 5-min day-to-day reliability conflict with original abstract; withhold contested numbers

Source examined PMID 9836355 abstract; Jakobsson Table 4 comparison. Original abstract or bibliographic record only; complete numerical tables unavailable

Smeets 2006

Study and population [3] Direct measurement study. 53 nonspecific CLBP; 50.9% below-knee radiation (not confirmation of radiculopathy); 50-ft analysis n=52

Protocol Same-observer 5–9-day retest; 5MW on 30-m figure-eight; fast 50-ft figure-eight; mean of two 5STS trials; inverse-time analysis for timed tasks

Measurement and change ICC: 5 MW .89; inverse 50-ft .76; inverse 5 STS .91. LoA half-width 82.7 m; time summaries 3.9/7.6 s back-converted at sample mean. Error, not MIC

Later outcomes and interpretation No clinical prognosis. Inverse scale and average-of-two matter; do not apply constant second thresholds across baselines

Source examined Official Maastricht thesis DOI 10.26481/dis.20061208rs, Chapter 8, printed pp 166–173, Table 3; original repository version linked below. Original author repository chapter methods and results inspected; complete journal article unavailable

Original repository version

Andersson 2010

Study and population [4] Longitudinal measurement responsiveness. 198 chronic nonspecific LBP; 10 weeks treatment

Protocol Six physical tasks; global perceived effect anchor

Measurement and change MDC-for-improvement and ROC methods are distinct. Only STS AUC .75 and stairs AUC .72 met .70 criterion; STS ROC improvement 4.1 s versus one-sided detectable improvement 9.8 s; GPE 3–5 unchanged and 6–7 improved

Later outcomes and interpretation Responsiveness, not baseline prognosis. Original author-posted full article available; GPE 3–5 unchanged and 6–7 improved; no universal 4.1 s rule

Source examined PMID 20634779; original author-posted full article now available

Teixeira da Cunha Filho 2010

Study and population [5] Direct measurement and construct validity. 30 Brazilian Portuguese-speaking LBP; 30 controls

Protocol 7 tasks including 50-ft/5-min walking and sit–stand

Measurement and change Abstract ICC range 0.92–.99 across tasks; RMDQ correlations with selected tasks .38–.47. No verified task-specific MIC

Later outcomes and interpretation Concurrent only. Small sample; task-specific error and protocol details not obtained

Source examined PMID 20067353 abstract. Original abstract or bibliographic record only; complete numerical tables unavailable

Taylor 2001

Study and population [6] Direct measurement and responsiveness. 44 mechanical CLBP with or without sciatica (review extraction)

Protocol Externally paced shuttle walking

Measurement and change Original abstract reports ICC 0.99, bias 2.5 m and LoA −47 to 52 m (half-width 49.5 m). Responsiveness used standardized effect sizes; this is not an individual patient-important threshold

Later outcomes and interpretation Not future disability prediction. Mixed symptoms; externally paced/turning task; original body unavailable

Source examined PMID 11725598 abstract; Jakobsson Tables 2/4. Original abstract or bibliographic record only; complete numerical tables unavailable

Trampe 2020

Study and population [7] Treatment-response measurement context. 127 CLBP, with or without radicular pain

Protocol 400-m comfortable and 200-m fast walk before/after multimodal programme

Measurement and change Mean speed changes .18±0.15 and .17±0.17 m/s. Reported MCIC 95% confidence interval .14–.22 m/s linked to Quebec scale detectable change, not a direct patient-importance anchor

Later outcomes and interpretation Change–change relationship, not prognostic validation. Online 2019/issue 2020; original tables unavailable; mixed radicular population

Source examined PMID 31561323 abstract. Original abstract or bibliographic record only; complete numerical tables unavailable

Hansen 2024

Study and population [8] Concurrent test comparison. 124 LBP

Protocol First two minutes of a 6 MWT versus full six minutes

Measurement and change r≈.83; 3 × early distance overestimated full distance by 46.8 m (SD67.0), about 9%. No patient-anchored MIC

Later outcomes and interpretation Concurrent estimation, not prognosis. Embedded shorter segment differs from standalone 2 MWT; original full text unavailable; online 2023/issue 2024

Source examined PMID 37026438 abstract. Original abstract or bibliographic record only; complete numerical tables unavailable

Bagheri 2018

Study and population [14] Direct instrumented reliability. 40 nonspecific CLBP; 20 male/20 female; age 30–40; BMI≤25

Protocol Six barefoot 10-m self-paced trials; 13 markers,100 Hz; 6–11-day same-assessor retest

Measurement and change Two-way random ICC; intersession generally high but lower-lumbar exceptions; sex/region-dependent MDCs up to 13.2°/14.7°. MDC, not MIC

Later outcomes and interpretation No prognosis. Highly selected group; wide/negative lower CI limits for some ICCs; segment-specific only

Source examined Full text Methods/Results; Tables 2/3 not used for new cut-offs; correction affiliation only. Complete article text

Fernandes 2015

Study and population [24] Direct instrumented reliability. CLBP; original detailed eligibility not inspected

Protocol Three-dimensional gait analysis

Measurement and change Original numerical table estimates withheld pending full body and corrigendum. MDC study; no verified MIC

Later outcomes and interpretation No clinical outcome prediction. 2015 original-table gap remains; 2016 corrigendum corrects the third author's name only; no second-hand thresholds implemented

Source examined PMID 26349524 and PMID 27318882; DOI10.1016/j.gaitpost.2016.05.009. Original abstract or bibliographic record only; complete numerical tables unavailable

Simmonds 2012

Study and population [15] Concurrent association. LBP pain-distribution groups and controls

Protocol Walking velocity and horizontal ground-reaction force

Measurement and change Distribution-related differences; not criterion-validation or change-threshold study. No MIC

Later outcomes and interpretation No later outcome. Preserve referred/radiating symptoms; distribution is not a gait-based anatomical diagnosis

Source examined Complete body, Methods/Results; PMC3325118. Complete article text

Moissenet 2023

Study and population [27] Direct reliability and concurrent discrimination. 30 nonspecific-CLBP participants and 30 controls; 24 patients returned

Protocol One-week retest; five usual 10-m walks; three height-adjusted chair rises

Measurement and change Hip chair-rise ROM: within-session ICC 0.90, between-session 0.57, MDC95 17.6 degrees. Lumbo-pelvic ROM: between-session ICC 0.60, MDC95 11.7 degrees. MDC concerns error, not patient importance

Later outcomes and interpretation No prognosis; four retained combined reliability/discrimination candidates concerned bending. Single centre, many features and study-specific rating bands

Source examined Supplement 3: Patient_BMo 3/Patient_BMo 33, reliability E97/E100 and MDC E97/E100; Supplement 4 protocols. Complete original article and original supplementary workbooks

Vivar 2026

Study and population [11] Device/algorithm agreement; internal development. 28 healthy older adults and 18 nonspecific-LBP participants; external dataset 22 healthy young adults, half used for adaptation

Protocol Shank smartphone IMU; 100 Hz; participant-independent 5-fold CV; internal CSAV-assisted manual labels from same angular-velocity signal; Vicon not used as event reference

Measurement and change NSLBP median IC/FO F1 .963/.988; event MAE≈10.6/11.2 ms; timing CCC .980–.994 across datasets. No responsiveness/MIC

Later outcomes and interpretation Gait-event prediction, not clinical prognosis. Agreement with internal signal-derived labeling workflow, not independent biomechanical reference; zero-shot and final-layer external adaptation failed; whole-network fine-tuning used half external cohort; repeated sessions and free living unvalidated

Source examined Complete body Methods 2.1–2.4 and Results 3.4–3.6; Table 3; Limitations. Complete article text

Koltermann 2026

Study and population [12] Exploratory current-state classifier. 36 matched participants; primary extreme-group subset 25/50 trials; 11 mild-pain participants excluded

Protocol Many gait/EMG variables reduced to 15-feature pipeline; subject-wise LOSO

Measurement and change Primary accuracy .780 (CI .620–.920); balanced accuracy .731; no external cohort. No change interpretation

Later outcomes and interpretation Current pain stratum; calibration slope .089; not an individual future-risk model. Configuration selection within cohort; extreme-group design; no diagnostic pain reference; poor calibration

Source examined Complete body 2.1/3.1/limitations; Table 2. Complete article text

Zhuo 2021

Study and population [29] Feasibility and convergent association. 17 LBP; 26-week tracker; 12-week questionnaire comparison

Protocol Garmin wearing/syncing; weekly activity questionnaire

Measurement and change Mean adherence 128/182 days≈70%; correlations did not meet prespecified .6 criterion. No device-specific MIC

Later outcomes and interpretation Not prognosis. Self-report not step-count gold standard; zero-step days excluded; device loss/login/nonwear matter

Source examined Complete body Results, Tables 1–3. Complete article text

Simon 2023

Study and population [20] Feasibility/concurrent association. 39 older adults with persistent LBP

Protocol MEPLO task; movement-evoked pain versus self-reported function/usual speed

Measurement and change Safety/feasibility and concurrent adjusted associations. Not a gait-speed MIC

Later outcomes and interpretation Baseline, not future decline. Small older sample; distinct task/scoring methods; no unrestricted remote safety inference

Source examined PMID 36943160; complete PMC BioC body Methods/Results. Complete article text

Knox 2021

Study and population [21] Concurrent association from prospective parent cohort. 230 older adults with CLBP

Protocol Sum of 0–10 pain ratings after repeated chair rise, 6 MWT, stairs; score 0–30

Measurement and change Adjusted associations with LLFDI b−.30 and TUG b.081. No task-specific MIC

Later outcomes and interpretation Baseline analysis; not longitudinal prediction. Aggregate cannot isolate STS/gait effect; parent cohort label does not change temporal order

Source examined Complete original article Methods/Results; Tables 3/4. Complete article text

Kimachi 2019

Study and population [32] Prospective association; adjacent exposure. 2738 enrolled residents≥60; 1358 analysed

Protocol Baseline RMDQ-defined LBP disability; one-year fall recall

Measurement and change Not a walking measurement-validation study. Not applicable

Later outcomes and interpretation Medium/high disability: adjusted RR1.53 any fall; 2.55 treated fall. Exposure is disability questionnaire, not gait; substantial analytic attrition; original abstract only

Source examined PMID 30856262 abstract. Original abstract or bibliographic record only; complete numerical tables unavailable

Haddadj 2025

Study and population [34] Prospective prevention association. 11194 adults without baseline chronic LBP; 1659 incident cases; mean 4.2 years

Protocol Device-derived walking minutes/intensity; follow-up self-report≥3 months pain

Measurement and change Not a clinical walking-test reliability study. No treatment MIC

Later outcomes and interpretation Higher walking volume associated with lower incident chronic LBP risk. Prevention population; observational confounding; not prediction among established CLBP patients

Source examined Complete HUNT body, Methods/Results. Complete article text

Butera 2022

Study and population [30] Concurrent risk-group comparison. 76 adults with LBP

Protocol STarT Back category versus speed, obstacles,TUG, other measures

Measurement and change Low-risk group performed better. No change threshold

Later outcomes and interpretation Current risk label; no observed future disability endpoint. Cannot claim gait adds prognostic value without longitudinal validation

Source examined PMID 35079824 abstract. Original abstract or bibliographic record only; complete numerical tables unavailable

Vincent 2014

Study and population [31] Treatment-response association. 49 obese older CLBP participants, 60–85 years; 4-month randomized exercise context

Protocol Lumbar/total resistance training; strength and walking at baseline/month 4

Measurement and change Strength change explained 10.6% of walking-endurance-change variance; between-group endurance P=.11; gait speed favoured total-body training P<.05. No gait MIC

Later outcomes and interpretation Change–change association, not baseline forecast. Small trial; treatment-specific; responder threshold is strength-based

Source examined PMID 24211698 abstract. Original abstract or bibliographic record only; complete numerical tables unavailable

Leung 2021

Study and population [33] Retrospective rehabilitation outcome model. 191 enrolled; 158 analysed; 91.1% injury-on-duty; recruitment 1996–2014

Protocol 14-week programme; walking tolerance; job-demand comparison and workplace reintegration at completion

Measurement and change No timed-gait validity or MDC evaluated. No walking MIC

Later outcomes and interpretation Walking tolerance was not a significant predictor; job demands and SFSS were associated with the work endpoint. Selected cohort, long recruitment period and imprecise odds ratios; not sustained long-term employment

Source examined Complete Methods: return-to-work definition; Tables 2–4. Complete article text

Piva 2025

Study and population [22] Large descriptive feasibility study. 1,006 participants with CLBP; 999 contributed some clinical examination data

Protocol Supervised comprehensive battery, including 4-m gait, five-rise chair test, 2-minute walk and balance

Measurement and change Tests not done: gait 1.9%, chair rise 10.7%, 2-minute walk 7.4%; timing included preparation and data entry. No reliability or MIC validation

Later outcomes and interpretation No clinical outcome prediction. Safety screening and task-specific denominators; sample means are not normality thresholds

Source examined Complete body, Table 3 and footnote a; Table 4. Complete article text

Hansen 2026

Study and population [16] Descriptive multidomain feasibility. 542 secondary-care patients; diagnosis codes: 69% nonspecific LBP, 13% stenosis, 11% disc herniation

Protocol 2-minute walk, five-rise chair test, markerless recordings, sensory tests, self-report and SMS follow-up

Measurement and change High functional-test completion; gait tracking less usable; calibration matured during recruitment. Descriptive medians are not thresholds

Later outcomes and interpretation No validated multidomain prognosis; follow-up demonstrates feasibility. Mixed diagnoses and no complete eligible-population denominator

Source examined Complete body, Methods and Results; Tables 1–3. Complete article text

Search methods and source access

Table 5 Search retrieval and reconciliation

Across the three overlapping slices: 1598 unique PubMed IDs and 4089 unique Scopus IDs. These are database records, not unique eligible or appraised studies.

Exact executed native queries

Measurement PubMed

("Low Back Pain"[MeSH Terms] OR "low back pain"[Title/Abstract] OR "low-back pain"[Title/Abstract] OR lumbago[Title/Abstract] OR "lumbar pain"[Title/Abstract]) AND ("Gait"[MeSH Terms] OR "Walking"[MeSH Terms] OR (gait[Title/Abstract] OR walk*[Title/Abstract] OR ambulat*[Title/Abstract] OR locomot*[Title/Abstract] OR "six minute"[Title/Abstract] OR "6 minute"[Title/Abstract] OR 6MWT[Title/Abstract] OR 10MWT[Title/Abstract] OR "40 m"[Title/Abstract] OR "40 meter"[Title/Abstract] OR "40 metre"[Title/Abstract])) AND (reliab*[Title/Abstract] OR valid*[Title/Abstract] OR reproducib*[Title/Abstract] OR psychometr*[Title/Abstract] OR clinimetr*[Title/Abstract] OR agreement[Title/Abstract] OR "measurement error"[Title/Abstract] OR "standard error"[Title/Abstract] OR "minimal detectable"[Title/Abstract] OR "minimum detectable"[Title/Abstract] OR "smallest detectable"[Title/Abstract] OR "minimal important"[Title/Abstract] OR "minimally important"[Title/Abstract] OR "minimum important"[Title/Abstract] OR responsiv*[Title/Abstract] OR interpretabil*[Title/Abstract] OR "floor effect"[Title/Abstract] OR "ceiling effect"[Title/Abstract]) AND ("1800/01/01"[Date - Publication] : "2026/10/02"[Date - Publication])

Measurement Scopus

TITLE-ABS-KEY(("low back pain" OR "low-back pain" OR lumbago OR "lumbar pain") AND (gait OR walk* OR ambulat* OR locomot* OR "six minute" OR "6 minute" OR 6MWT OR 10MWT OR "40 m" OR "40 meter" OR "40 metre") AND (reliab* OR valid* OR reproducib* OR psychometr* OR clinimetr* OR agreement OR "measurement error" OR "standard error" OR "minimal detectable" OR "minimum detectable" OR "smallest detectable" OR "minimal important" OR "minimally important" OR "minimum important" OR responsiv* OR interpretabil* OR "floor effect" OR "ceiling effect")) AND PUBYEAR BEF 2027

Prognosis PubMed

("Low Back Pain"[MeSH Terms] OR "low back pain"[Title/Abstract] OR "low-back pain"[Title/Abstract] OR lumbago[Title/Abstract] OR "lumbar pain"[Title/Abstract]) AND ("Gait"[MeSH Terms] OR "Walking"[MeSH Terms] OR (gait[Title/Abstract] OR walk*[Title/Abstract] OR ambulat*[Title/Abstract] OR locomot*[Title/Abstract] OR "six minute"[Title/Abstract] OR "6 minute"[Title/Abstract] OR 6MWT[Title/Abstract] OR 10MWT[Title/Abstract] OR "40 m"[Title/Abstract] OR "40 meter"[Title/Abstract] OR "40 metre"[Title/Abstract])) AND (prognos*[Title/Abstract] OR predict*[Title/Abstract] OR longitudinal[Title/Abstract] OR prospective[Title/Abstract] OR cohort[Title/Abstract] OR "follow up"[Title/Abstract] OR "follow-up"[Title/Abstract] OR recovery[Title/Abstract] OR deteriorat*[Title/Abstract] OR fall*[Title/Abstract] OR "natural history"[Title/Abstract] OR "return to work"[Title/Abstract] OR discharge[Title/Abstract] OR "risk factor"[Title/Abstract]) AND ("1800/01/01"[Date - Publication] : "2026/10/02"[Date - Publication])

Prognosis Scopus

TITLE-ABS-KEY(("low back pain" OR "low-back pain" OR lumbago OR "lumbar pain") AND (gait OR walk* OR ambulat* OR locomot* OR "six minute" OR "6 minute" OR 6MWT OR 10MWT OR "40 m" OR "40 meter" OR "40 metre") AND (prognos* OR predict* OR longitudinal OR prospective OR cohort OR "follow up" OR "follow-up" OR recovery OR deteriorat* OR fall* OR "natural history" OR "return to work" OR discharge OR "risk factor")) AND PUBYEAR BEF 2027

Protocol mechanism technology PubMed

("Low Back Pain"[MeSH Terms] OR "low back pain"[Title/Abstract] OR "low-back pain"[Title/Abstract] OR lumbago[Title/Abstract] OR "lumbar pain"[Title/Abstract]) AND ("Gait"[MeSH Terms] OR "Walking"[MeSH Terms] OR (gait[Title/Abstract] OR walk*[Title/Abstract] OR ambulat*[Title/Abstract] OR locomot*[Title/Abstract] OR "six minute"[Title/Abstract] OR "6 minute"[Title/Abstract] OR 6MWT[Title/Abstract] OR 10MWT[Title/Abstract] OR "40 m"[Title/Abstract] OR "40 meter"[Title/Abstract] OR "40 metre"[Title/Abstract])) AND (protocol[Title/Abstract] OR biomechan*[Title/Abstract] OR kinematic*[Title/Abstract] OR kinetic*[Title/Abstract] OR symmetr*[Title/Abstract] OR asymmetr*[Title/Abstract] OR "weight bearing"[Title/Abstract] OR "weight-bearing"[Title/Abstract] OR "ground reaction"[Title/Abstract] OR "force plate"[Title/Abstract] OR "force platform"[Title/Abstract] OR sensor*[Title/Abstract] OR wearable*[Title/Abstract] OR inertial[Title/Abstract] OR acceleromet*[Title/Abstract] OR markerless[Title/Abstract] OR "motion capture"[Title/Abstract] OR camera[Title/Abstract] OR video[Title/Abstract] OR algorithm[Title/Abstract] OR electromyogra*[Title/Abstract] OR activation[Title/Abstract] OR "sampling frequency"[Title/Abstract] OR "filter cutoff"[Title/Abstract]) AND ("1800/01/01"[Date - Publication] : "2026/10/02"[Date - Publication])

Protocol mechanism technology Scopus

TITLE-ABS-KEY(("low back pain" OR "low-back pain" OR lumbago OR "lumbar pain") AND (gait OR walk* OR ambulat* OR locomot* OR "six minute" OR "6 minute" OR 6MWT OR 10MWT OR "40 m" OR "40 meter" OR "40 metre") AND (protocol OR biomechan* OR kinematic* OR kinetic* OR symmetr* OR asymmetr* OR "weight bearing" OR "weight-bearing" OR "ground reaction" OR "force plate" OR "force platform" OR sensor* OR wearable* OR inertial OR acceleromet* OR markerless OR "motion capture" OR camera OR video OR algorithm OR electromyogra* OR activation OR "sampling frequency" OR "filter cutoff")) AND PUBYEAR BEF 2027

Coverage reconciliation and source selection

All returned database pages were retrieved through the terminal page. Scopus unique-ID counts matched each provider total. PubMed pages repeatedly returned some records while omitting others; exact-query official NCBI ESearch/EFetch recovered all IDs and records without missing fetches. PubMed BookArticle records were handled where present. Native Scopus queries used publication year before 2027; individual included-source dates were checked against the 2 October cutoff, since issue year alone is not exact date eligibility.

Search slices are overlapping discovery sets. No claim is made that all retrieved records were independently screened. Intensive appraisal prioritized original measurement properties, influential change thresholds, useful clinical protocols, technology validation and prospective clinical outcomes. The earlier low-back-pain landscape search, review references and bounded citation chasing supplemented native queries.

Forward citations of the gait review were retrieved from OpenAlex and Semantic Scholar (93 and 85 records respectively); backward references of the STS review returned 26 OpenAlex records, with a Semantic Scholar rate-limit failure. These were bounded discovery calls, not exhaustive citation networks. Forward discovery from the 2019 capacity review was also inspected (47 OpenAlex records; Semantic Scholar rate limit). Relevant new studies were verified at their primary sources.

Original full articles were sought through bibliographic services, publishers and lawful repositories. Some subscription originals remained abstract-only. Complete text, partial chapter inspection, original abstracts and corrections remain distinguished. One changed-content retrieval was resolved by obtaining the complete article again.

Key original supplements retrieved: Moissenet 2023 Supplementary Information 1 (file guide), Supplementary Information 3 (full numerical properties workbook) and Supplementary Information 4 (task protocols). Ozsoy/Uz 2024 publisher PDF Tables 2–3 were visually verified. The Fernandes 2016 corrigendum corrects the third author's name only; its separate 2015 original-table gap remains and thresholds are withheld.

Related source and validation needs

These source acquisition priorities include foundational studies shared across the walking and chair rise evidence base.

• Obtain Simmonds 1998 original tables before disputed reliability/SDC values are used

• Andersson’s original anchor and ROC/error tables are now available. Keep ROC improvement of 4.1 s distinct from one-sided detectable improvement of 9.8 s; no universal MIC is established.

• Obtain Kahraman’s original full tables for absolute-error appraisal. Özüdoğru’s article is now available, but the SD input and SEM/MDC derivation still require clarification.

• Gait: obtain Fernandes's original 2015 tables; the 2016 corrigendum corrects the third author's name only

• Kinematics: obtain exact Christe 2022 supplementary task/output tables before individual threshold use

References

References are numbered in first citation order. Source descriptions identify the material examined and are not study quality ratings. Links identify the original publication or the explicitly named original source version.

1. Jakobsson M, Gutke A, Mokkink LB, Smeets R, Lundberg M. Level of Evidence for Reliability, Validity, and Responsiveness of Physical Capacity Tasks Designed to Assess Functioning in Patients With Low Back Pain: A Systematic Review Using the COSMIN Standards. Physical therapy. 2019;99(4):457-477. DOI 10.1093/ptj/pzy159 Source examined: Complete article text.

Source note: SRC-418578aae366 Jakobsson M 2019

2. Simmonds MJ, Olson SL, Jones S, Hussein T, Lee CE, Novy D et al. Psychometric characteristics and clinical usefulness of physical performance tests in patients with low back pain. Spine. 1998;23(22):2412-21. DOI 10.1097/00007632-199811150-00011 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-e0690127ab0a Simmonds MJ 1998

3. Smeets RJ, Hijdra HJ, Kester AD, Hitters MW, Knottnerus JA. The usability of six physical performance tasks in a rehabilitation population with chronic low back pain. Clinical rehabilitation. 2006;20(11):989-97. DOI 10.1177/0269215506070698 Source examined: Original author repository chapter methods and results inspected; complete journal article unavailable.

Source note: SRC-606df2e53a98 Smeets RJ 2006

4. Andersson EI, Lin CC, Smeets RJ. Performance tests in people with chronic low back pain: responsiveness and minimal clinically important change. Spine. 2010;35(26):E1559-63. DOI 10.1097/brs.0b013e3181cea12e Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-9e89524f8f98 Andersson EI 2010

5. Teixeira da Cunha-Filho I, Lima FC, Guimarães FR, Leite HR. Use of physical performance tests in a group of Brazilian Portuguese-speaking individuals with low back pain. Physiotherapy theory and practice. 2010;26(1):49-55. DOI 10.3109/09593980802602844 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-875bbc250a64 Teixeira da Cunha-Filho I 2010

6. Taylor S, Frost H, Taylor A, Barker K. Reliability and responsiveness of the shuttle walking test in patients with chronic low back pain. Physiotherapy research international : the journal for researchers and clinicians in physical therapy. 2001;6(3):170-8. DOI 10.1002/pri.225 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-49266bf67111 Taylor S 2001

7. Trampe D, Gouteron A, Naaim A, Laroche D, Grelat M, Casillas JM. A tool to improve functional outcome assessment of a multimodal program for patients with chronic low back pain: A study on walk tests (at comfortable and fast speed). Journal of back and musculoskeletal rehabilitation. 2020;33(3):485-494. DOI 10.3233/bmr-171019 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-2b5068b99de8 Trampe D 2020

8. Hansen A, Nim CG, O'Sullivan K, O'Neill S. Testing walking performance in patients with low back pain: will two minutes do instead of six minutes? Disability and rehabilitation. 2024;46(6):1173-1177. DOI 10.1080/09638288.2023.2194683 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-3cbe35e20365 Hansen A 2024

9. Smith JA, Stabbert H, Bagwell JJ, Teng HL, Wade V, Lee SP. Do people with low back pain walk differently? A systematic review and meta-analysis. Journal of sport and health science. 2022;11(4):450-465. DOI 10.1016/j.jshs.2022.02.001 Source examined: Complete article text.

Source note: SRC-a18ea3bbb9ec Smith JA 2022

10. Dal Farra F, Lopomo NF, Fascia M, Scalona E, Cerfoglio S, Cimolin V. How non-specific low back pain affects gait kinematics: a systematic review and meta-analysis. Frontiers in pain research (Lausanne, Switzerland). 2025;6:1693068. DOI 10.3389/fpain.2025.1693068 Source examined: Complete article text.

Source note: SRC-f05e1f7c1db7 Dal Farra F 2025

11. Vivar G, Singh S, Bea T, Saal C, Munoz-Martel V, Schega L. Deep Learning-Based Temporal Gait Analysis Using a Smartphone IMU in Older Adults with and Without Non-Specific Low Back Pain. Bioengineering (Basel, Switzerland). 2026;13(8):924. DOI 10.3390/bioengineering13080924 Source examined: Complete article text.

Source note: SRC-0ca21bafa7f6 Vivar G 2026

12. Koltermann JJ, Floessel P, Wunderlich FC, Funke JJ, Kaplick H, Disch AC. AI-Supported Functional Pain Phenotyping in Low Back Pain Using Sensor-Based Gait and Neuromuscular Biomarkers. Sensors (Basel, Switzerland). 2026;26(16):5178. DOI 10.3390/s26165178 Source examined: Complete article text.

Source note: SRC-f5baeb3d6597 Koltermann JJ 2026

13. Jakobsson M, Gutke A, Mokkink LB, Smeets R, Lundberg M. Author Response to Denteneer et al. Physical therapy. 2020;100(6):1036-1037. DOI 10.1093/ptj/pzaa040 Source examined: Complete article text.

Source note: SRC-cb48b9ab5ffc Jakobsson M 2020

14. Bagheri R, Ebrahimi Takamjani I, Dadgoo M, Ahmadi A, Sarrafzadeh J, Pourahmadi MR et al. Gender-Related Differences in Reliability of Thorax, Lumbar, and Pelvis Kinematics During Gait in Patients With Non-specific Chronic Low Back Pain. Annals of rehabilitation medicine. 2018;42(2):239-249. DOI 10.5535/arm.2018.42.2.239 Source examined: Complete article text.

Source note: SRC-3c24069cd66b Bagheri R 2018

15. Simmonds MJ, Lee CE, Etnyre BR, Morris GS. The influence of pain distribution on walking velocity and horizontal ground reaction forces in patients with low back pain. Pain research and treatment. 2012;2012:214980. DOI 10.1155/2012/214980 Source examined: Complete article text.

Source note: SRC-47b527141a58 Simmonds MJ 2012

16. Hansen A, Harsted S, Nim C, O'Sullivan K, O'Neill S. Integrated disability evaluation in low back pain: feasibility and multidomain profiling from the EL DORADO cohort. Frontiers in rehabilitation sciences. 2026;7:1731260. DOI 10.3389/fresc.2026.1731260 Source examined: Complete article text.

Source note: SRC-ce6f19f13df1 Hansen A 2026

17. Al-Obaidi SM, Al-Zoabi B, Al-Shuwaie N, Al-Zaabie N, Nelson RM. The influence of pain and pain-related fear and disability beliefs on walking velocity in chronic low back pain. International journal of rehabilitation research. Internationale Zeitschrift fur Rehabilitationsforschung. Revue internationale de recherches de readaptation. 2003;26(2):101-8. DOI 10.1097/00004356-200306000-00004 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-a19e493c186e Al-Obaidi SM 2003

18. Vickers J, Reed A, Decker R, Conrad BP, Olegario-Nebel M, Vincent HK. Effect of investigator observation on gait parameters in individuals with and without chronic low back pain. Gait & posture. 2017;53:35-40. DOI 10.1016/j.gaitpost.2017.01.002 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-a39a58a81189 Vickers J 2017

19. Matheve T, Janssens L, Goossens N, Danneels L, Willems T, Van Oosterwijck J et al. The Relationship Between Pain-Related Psychological Factors and Maximal Physical Performance in Low Back Pain: A Systematic Review and Meta-Analysis. The journal of pain. 2022;23(12):2036-2051. DOI 10.1016/j.jpain.2022.08.001 Source examined: Complete article text.

Source note: SRC-4dcdcbbbd059 Matheve T 2022

20. Simon CB, Hicks GE, Pieper CF, Byers Kraus V, Keefe FJ, Colón-Emeric C. A Novel Movement-Evoked Pain Provocation Test for Older Adults With Persistent Low Back Pain: Safety, Feasibility, and Associations With Self-reported Physical Function and Usual Gait Speed. The Clinical journal of pain. 2023;39(4):166-174. DOI 10.1097/ajp.0000000000001101 Source examined: Complete article text.

Source note: SRC-c030d3e9807f Simon CB 2023

21. Knox PJ, Simon CB, Pohlig RT, Pugliese JM, Coyle PC, Sions JM et al. A Standardized Assessment of Movement-evoked Pain Ratings Is Associated With Functional Outcomes in Older Adults With Chronic Low Back Pain. The Clinical journal of pain. 2021;38(4):241-249. DOI 10.1097/ajp.0000000000001016 Source examined: Complete article text.

Source note: SRC-719f60899a73 Knox PJ 2021

22. Piva SR, Alfikri Z, Anderst W, Bell KM, Carlesso C, Darwin J et al. Feasibility of Physical Exam and Performance-Based Tests in Individuals With Chronic Low Back Pain: A Descriptive Study. JOR spine. 2025;8(3):e70096. DOI 10.1002/jsp2.70096 Source examined: Complete article text.

Source note: SRC-440cc699ea82 Piva SR 2025

23. Unver T, Unver B, Kacmaz KS. The test-retest reliability and minimal clinically important difference of the Dubousset Functional Test and its correlation with Rolland Morris disability questionnaire in chronic non-specific low back pain. European spine journal : official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society. 2023;32(6):2086-2092. DOI 10.1007/s00586-023-07720-6 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-0082399c402f Unver T 2023

24. Fernandes R, Armada-da-Silva P, Pool-Goudzwaard AL, Moniz-Pereira V, Veloso AP. Test-retest reliability and minimal detectable change of three-dimensional gait analysis in chronic low back pain patients. Gait & posture. 2015;42(4):491-7. DOI 10.1016/j.gaitpost.2015.08.002 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-d0e4de713936 Fernandes R 2015

25. Fernandes RS, Armada-da-Silva P, Pool-Goudzwaard AL, Moniz-Pereira V, Veloso ANP. Corrigendum to "Test-retest reliability and minimal detectable change of three-dimensional gait analysis in chronic low back pain patients" [Gait Posture 42 (2015) 491-497]. Gait & posture. 2016;48:268. DOI 10.1016/j.gaitpost.2016.05.009 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-e9c0eaeb2004 Fernandes RS 2015

26. Bagheri R, Takamjani IE, Dadgoo M, Ahmadi A, Sarrafzadeh J, Pourahmadi MR et al. Correction: Gender-Related Differences in Reliability of Thorax, Lumbar, and Pelvis Kinematics During Gait in Patients With Non-specific Chronic Low Back Pain. Annals of rehabilitation medicine. 2018;42(6):888. DOI 10.5535/arm.2018.42.6.888 Source examined: Complete article text.

Source note: SRC-699cb7f376b2 Bagheri R 2018

27. Moissenet F, Armand S, Genevay S. Measurement properties of 72 movement biomarkers aiming to discriminate non‑specific chronic low back pain patients from an asymptomatic population. Scientific reports. 2023;13(1):6483. DOI 10.1038/s41598-023-33504-5 Source examined: Complete original article and original supplementary workbooks.

Source note: SRC-de0028260239 Moissenet F 2023

28. Quirk DA, Johnson ME, Anderson DE, Smuck M, Sun R, Matthew R et al. Biomechanical Phenotyping of Chronic Low Back Pain: Protocol for BACPAC. Pain medicine (Malden, Mass.). 2023;24(Suppl 1):S48-S60. DOI 10.1093/pm/pnac163 Source examined: Complete article text.

Source note: SRC-e346cdccc99b Quirk DA 2023

29. Zhuo LX, Macedo LG. Feasibility and Convergent Validity of an Activity Tracker for Low Back Pain Within a Clinical Study: Cross-sectional Study. JMIR rehabilitation and assistive technologies. 2021;8(1):e18942. DOI 10.2196/18942 Source examined: Complete article text.

Source note: SRC-d9a1afbf51a0 Zhuo LX 2021

30. Butera KA, Fox EJ, Bishop MD, Coombes SA, Beneciuk JM, George SZ. Low Risk for Persistent Back Pain Disability Is Characterized by Lower Pain Sensitivity and Higher Physical Performance. Physical therapy. 2022;102(3):pzab283. DOI 10.1093/ptj/pzab283 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-caf311c20736 Butera KA 2022

31. Vincent HK, Vincent KR, Seay AN, Conrad BP, Hurley RW, George SZ. Back strength predicts walking improvement in obese, older adults with chronic low back pain. PM & R : the journal of injury, function, and rehabilitation. 2014;6(5):418-26. DOI 10.1016/j.pmrj.2013.11.002 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-3899c92ce557 Vincent HK 2014

32. Kimachi K, Kimachi M, Takegami M, Ono R, Yamazaki S, Goto Y et al. Level of Low Back Pain-Related Disability Is Associated with Risk of Subsequent Falls in an Older Population: Locomotive Syndrome and Health Outcomes in Aizu Cohort Study (LOHAS). Pain medicine (Malden, Mass.). 2019;20(12):2377-2384. DOI 10.1093/pm/pny313 Source examined: Original abstract or bibliographic record only; complete numerical tables unavailable.

Source note: SRC-850c6232ad04 Kimachi K 2019

33. Leung GCN, Cheung PWH, Lau G, Lau ST, Luk KDK, Wong YW et al. Multidisciplinary programme for rehabilitation of chronic low back pain - factors predicting successful return to work. BMC musculoskeletal disorders. 2021;22(1):251. DOI 10.1186/s12891-021-04122-x Source examined: Complete article text.

Source note: SRC-88dd0ea1f02f Leung GCN 2021

34. Haddadj R, Nordstoga AL, Nilsen TIL, Skarpsno ES, Kongsvold A, Flaaten M et al. Volume and Intensity of Walking and Risk of Chronic Low Back Pain. JAMA network open. 2025;8(6):e2515592. DOI 10.1001/jamanetworkopen.2025.15592 Source examined: Complete article text.

Source note: SRC-2a5a5c3f2bdc Haddadj R 2025