Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Study Protocol

Physical measures of physical functioning as prognostic factors in predicting outcomes for neck pain: Protocol for a prospective longitudinal cohort study

  • Rabea Begum ,

    Roles Conceptualization, Methodology, Writing – original draft, Writing – review & editing

    mbegum6@uwo.ca

    Affiliations School of Physical Therapy, Western University, London, Ontario, Canada, CANSpine Research Group, Western University, London, Ontario, Canada

    ⨯
  • Paul Parikh,

    Roles Conceptualization, Methodology, Writing – review & editing

    Affiliations School of Physical Therapy, Western University, London, Ontario, Canada, CANSpine Research Group, Western University, London, Ontario, Canada

    ⨯
  • David Walton,

    Roles Conceptualization, Methodology, Validation, Writing – review & editing

    Affiliations School of Physical Therapy, Western University, London, Ontario, Canada, CANSpine Research Group, Western University, London, Ontario, Canada

    ⨯
  • Alison Rushton

    Roles Conceptualization, Methodology, Validation, Writing – review & editing

    Affiliations School of Physical Therapy, Western University, London, Ontario, Canada, CANSpine Research Group, Western University, London, Ontario, Canada

    ⨯

Abstract

Introduction

Neck pain is a common musculoskeletal condition that contributes to significant global burden. Evidence on physical measures of functioning as prognostic factors for neck pain remains limited. The objective of this study is to investigate whether physical measures of physical functioning as prognostic factors predict outcomes for people with neck pain.

Methods

This protocol for a prospective longitudinal cohort study will recruit approximately 156 adults with neck pain of any duration. Physical measures of physical functioning including impairment, performance, and activity measures in a real environment, will be assessed as prognostic factors. Outcomes at 12-month follow-up will include pain, disability, and overall changes, assessed using the Numeric Pain Rating Scale (NPRS), Neck Disability Index (NDI), and Global Rating of Change (GROC), respectively. Hierarchical regression analyses will be conducted separately for each physical measure and outcome. Physical measures will first be entered alone to examine the initial unadjusted (bivariate) association with the outcome, followed by adjustment for covariates (age, sex and baseline NPRS/NDI). Additional contribution will be evaluated using changes in explained variance ΔR2 and F-change statistics. Physical measures demonstrating additional contribution will subsequently be entered into the final penalized multivariable regression model using least absolute shrinkage and selection operator (LASSO) regression to reduce overfitting and identify the most relevant prognostic factors. Sensitivity analyses comparing penalized and non-penalized multivariable regression models will also be conducted to examine the consistency and stability of the findings. Prognostic effects will be reported as beta coefficients (β) or odds ratios (OR) with 95% confidence intervals (CIs). Each model will be internally validated using bootstrapping, with model performance evaluated using calibration and discrimination.

Discussion

This study will provide new insights into how physical measures of physical functioning relate to outcomes for neck pain. The resulting prognostic models have the potential to strengthen clinical decision-making, support personalized care, and guide the development of targeted interventions for people with neck pain.

Registration number

The protocol has prospectively registered on the Open Science Framework (OSF). The link is https://doi.org/10.17605/OSF.IO/HG6PJ.

Citation: Begum R, Parikh P, Walton D, Rushton A (2026) Physical measures of physical functioning as prognostic factors in predicting outcomes for neck pain: Protocol for a prospective longitudinal cohort study. PLoS One 21(7): e0354102. https://doi.org/10.1371/journal.pone.0354102

Editor: Mario Lopes, Universidade de Aveiro Escola Superior de Saude de Aveiro, PORTUGAL

Received: October 14, 2025; Accepted: June 30, 2026; Published: July 17, 2026

Copyright: © 2026 Begum et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The research data will be made publicly available when the study is completed and published.

Funding: The author(s) received no specific funding for this work.

Competing interests: The author have declared that no competing interests exist.

Introduction

Neck pain is a highly prevalent and debilitating musculoskeletal condition, currently ranked as the fourth leading cause of global years lived with disability [1]. Neck pain is best characterized as multifactorial and episodic in nature, often with an unclear etiology [2]. More than one-third of people with neck pain will develop chronic symptoms lasting more than six months [3,4] representing a serious health concern [5,6]. The socioeconomic burden of neck pain is therefore substantial, including long-term healthcare utilization and lost productivity [7]. People with neck pain demonstrate reduced physical functioning [8] which is one of the core domains recommended by clinical research guidelines [9,10].

Physical functioning provides evaluation of meaningful aspects of a person’s life, including the ability to perform household chores, walking, work, and self-care, as well as strength, endurance, and flexibility [9], categorized within the domains of the International Classification of Functioning, Disability, and Health (ICF) [11]. Physical functioning can be assessed using patient-reported outcome measures (PROMs, self-perceived function) or physical measures (observed function) [12,13]. PROMs are useful for providing perceived physical functioning from the patient’s perspective [14] but are insufficient to measure the multiple domains of physical functioning [15,16]. Physical measures can be classified as impairment-based measures (e.g., limited range of motion measured by a cervical range of motion, CROM device), performance-based measures (e.g., endurance of deep neck flexors measured using the Craniocervical Flexion Test, CCFT) or activity measures in a real environment (e.g., physical activity measured by accelerometry) [9,11–13]. Knowledge of the prognostic value of these measures will enable clinicians to better assess prognosis, individualize management, and potentially improve outcomes [17].

Our recent systematic review [18] synthesized existing evidence for physical measures of physical functioning for predicting outcomes in neck pain. Using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach, the overall level of evidence was low to very low. Low level evidence supports that cervical movement velocity in the frontal plane was a potential predictor of short-term disability, while low to very low level evidence indicates that others impairment- and performance-based measures were not predictors of pain, disability or overall changes in the short-term (< 3-month) and medium-term (3–12-month) follow up for people with neck pain. For measures including cervical active range of motion (AROM), joint position error (JPE), and endurance of the neck flexors, the evidence was conflicting or inconclusive, underscoring the need for high-quality prospective research. The review also revealed key gaps: most studies focused mainly on ROM as impairment and endurance as performance measures, with none examined activity measures in real environments. Studies were disproportionately focused on acute whiplash populations, limiting generalizability, and few adjusted adequately for confounders.

This study seeks to address evidence gaps by investigating the predictive ability of physical measures of physical functioning for neck pain. The objective of this study is to investigate whether physical measures of physical functioning as prognostic factors predict outcomes for people with neck pain.

Methods

Study design

This prospective longitudinal inception cohort observational study will be conducted following the recommendations of the Prognosis Research Strategy framework (PROGRESS) [19]. Study design and procedures are informed by the Quality In Prognosis Studies (QUIPS) tool to address potential sources of bias in prognostic factor research [20]. Reporting will follow the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement [21] (S1 Table). Data will be obtained from the wider COMPrehensive 360° Assessment of Spinal pain Study (COMPASS). Ethical approval was obtained from Western University Research Ethics Board (application no. 122736). The study has been registered with the Open Science Framework (OSF) prior to the start of data collection.

Study settings

Data will be collected at two primary locations in London Ontario, Canada: one publicly funded hospital-based clinic and one non-profit community rehabilitation clinic in London, Ontario, Canada. Study recruitment will commence in October 13, 2025 and is expected to continue until mid-June, 2027.

Participants

Adults aged 18 years or older with a primary complaint of neck pain, with or without arm symptoms, of any duration, will be eligible for this study. Exclusion criteria will be: anyone unable or unwilling to provide informed consent, unable to communicate in English, or with medical conditions that could affect neck symptoms or function, including space-occupying lesion (e.g., tumors, cysts), serious neurological disorders (e.g., spinal cord injury, myelopathy), disc infection, inflammatory disorders (e.g., ankylosing spondylitis), fibromyalgia, metabolic bone diseases (e.g., osteoporosis), cervical spine dislocation or fracture, and pain of non-neuromusculoskeletal origin.

Study recruitment

Consecutive participants will be approached to provide verbal consent to be contacted. Participants who will provide consent to be contacted will be first contacted through email with a letter of information. Participants will be contacted by phone call if they do not reply within 48 hours if they have any questions and to check their interest in taking part in the study. Participants who will provide consent, will be scheduled for an in-person data collection for baseline physical measures assessment at the CanMOVE Research Laboratory. After entering participant into study through EmPOWER database, participants will create their account and receive a unique ID number to link study participants with personal identifiers. Participants will fill out demographic and clinical variables surveys electronically (≈20 minutes), and will then undergo physical measures (≈30 minutes). Participants will complete 12-month outcome questionnaires through their EmPOWER account within 12 months (±1 month) after baseline assessment. Participants will fill out NPRS, NDI, and GROC at 12-month follow-up electronically (≈10 minutes). The study recruitment process is presented in Fig 1.

thumbnail
Download:
Fig 1. Study flow diagram.

Overview of participants’ recruitment.

https://doi.org/10.1371/journal.pone.0354102.g001

Data management

All data will be entered into SPINA Spinal Rehabilitation Registry platform [22] through the EmPOWER Health Research Inc database, which has high standard of physical and virtual security for data management. Ethical approval for SPINA was obtained from Western University Research Ethics Board (application no. 122735) and Lawson Health Research Institute ReDA (application no. 13205). Only de-identified data will be used for analysis.

Variables

Demographic variables including age, sex, height and weight to calculate body mass index (BMI), and clinical variables, such as types of symptoms (pain, numbness or pins and needles), and pain duration, will be recorded to describe the sample characteristics for descriptive reporting.

Physical measures of physical functioning as prognostic factors.

Prognostic factors (PFs) are selected based on current evidence synthesis regarding physical measures of physical functioning in predicting outcomes for neck pain [18]. The PFs are also chosen for their feasibility of measurement, as recommended in clinical practice guideline for neck pain [23], and their common use in clinical practice [24,25].

Impairment-based measures: In this study, cervical AROM, JPE, and handgrip strength will be assessed as impairment-based measures as they evaluate physical limitations or dysfunctions at the body structure or function level, in line with World Health Organization (WHO) International Classification of Functioning, Disability and Health (ICF) [11].

Cervical AROM. Cervical AROM will be measured using the cervical range of motion (CROM) device (Performance Attainment Associates, Minnesota, USA) [26–28], which incorporates 3 embedded inclinometers to measure three-dimensional ROM with a single device attached to the participant’s head. The CROM device has demonstrated good–excellent intrarater reliability for measuring cervical AROM including all movements in individuals with neck pain (ICC3,1 = 0.87 to 0.94) and without neck pain (ICC3,1 = 0.88 to 0.96), [29], with good concurrent validity compared with the Fastrak (Polhemus, Colchester, Vermont, USA) system (r = 0.98 for extension and rotations, 0.93 for flexion [30]. Participants will be seated in a chair with feet flat on the floor and back supported. A strap will be used to gently restrict shoulders and trunk movement, isolating movements to the neck [30]. The CROM device will be placed on the participant’s head, and they will be instructed to move their head as far as possible comfortably in 4 directions: flexion, extension, right and left rotation [30]. Two trials will be performed for each movement direction, and the mean value of the two trials will be used for analysis. While some previous studies [31–34] have combined flexion and extension into a single sagittal plane motion value, in this study they will be analyzed separately as PFs, given their distinct biomechanical properties (i.e., involvement of the upper vs. lower cervical segments, respectively) [35], and their potential to reflect direction-specific functional impairments in individuals with neck pain [35,36], in accordance with previous studies [33,37–41]. For cervical rotation, left and right rotation will be combined to calculate total cervical rotation, which will be analyzed as a single predictor. This approach is considered appropriate because the study includes participants with both unilateral and bilateral neck pain, where direction-specific limitations may vary depending on the dominant or affected side. Combining left and right rotation therefore provides a more stable estimate of cervical AROM in rotation, consistent with previous studies [32,38,42–44].

Cervical Joint Position Error (JPE). Cervical JPE will be assessed using the CROM device (Performance Attainment Associates, Minnesota, USA) [26–28]. JPE for rotation measured with the CROM device shows good to excellent intra-rater (ICC3,1 = 0.78–0.92) and inter-rater (ICC2,1 = 0.75–0.88) reliability in people with neck pain, with convergent validity supported between symptomatic and healthy individuals [45,46]. Rotation is selected as the most frequently reported movement for JPE in the literature for neck pain [46,47]. It has also shown greater deficits in JPE during left and right rotation rather than flexion and extension in neck pain population, suggesting that rotation is more sensitive for detecting cervical proprioceptive impairment [35,48,49]. For testing, the CROM device will be placed on the participant’s head. Participants will move their neck to the mid-point of their available rotational range and hold for 3 seconds to memorize this target position. After returning to the starting position, they will reproduce the target position as accurately as possible without feedback, holding for 3 seconds while the ROM will be recorded. The error between target and reproduced position (in degrees) will be recorded. Two trials will be performed for left and right rotation, and mean values will be calculated. Left and right rotation will be combined to calculate total cervical rotation for analysis as a single predictor to reduce variability from side-to-side differences and provide a stable estimate of overall cervical proprioception. This approach was considered appropriate because the study includes participants with both unilateral and bilateral neck pain, where side-specific measures may vary depending on the dominant or affected side. Combining both directions therefore provides a more consistent estimate of cervical JPE, and is also consistent with previous methods used in proprioception research for neck pain [50].

Handgrip strength. Handgrip strength is a widely used physical measure of physical function and a useful indicator of general health status, and functional decline across various health conditions [51]. The handgrip strength for dominant hand will be measured by using a reliable and valid JTECH handgrip dynamometer (JTECH Medical, Midvale, Utah, USA) [52], in accordance with standardized protocol [53]. This device demonstrates good to excellent intra-rater reliability (ICC2,1 = 0.78–0.9) and inter-rater reliability (ICC2,1 = 0.76–0.95), with substantial concurrent validity for measuring handgrip strength in healthy adults [54]. For testing, the participant will be seated with feet flat, shoulder adducted, elbow flexed to 90°, forearm unsupported in neutral, writs positioned between 0–30° extension and 0–15° ulnar deviation. Participants will be instructed to squeeze as hard as they can. They will also receive encouragement while performing the test (e.g., “harder, harder, harder”) to facilitate them achieve maximum voluntary contraction. Two trials will be performed instead of three trials, to reduce to burden of assessment, with a 1 min rest interval. The handgrip strength, in kg, will be recorded and the mean value of 2 trials will be used for analysis, as this has previously been shown to be a more accurate reflection of true strength [55,56].

Performance-based measure: Performance-based measure assesses how well an individual perform specific tasks or actions in a standardized settings (e.g., hospital or clinic) [13]. Endurance of the deep neck flexors using the CCFT will be considered a performance-based measure as it requires participants to actively demonstrate graded activation and sustained control of the cervical flexor muscles in a standardized environment. This endurance test measures the activity level functioning (code: d220 in ICF coreset for musculoskeletal condition) defined by ICF.

Endurance of deep neck flexor muscles. Endurance of the deep neck flexor muscles will be assessed using the CCFT with a pressure biofeedback unit (Stabilizer, Chattanooga Group, Hixson, TN, USA), with following standardized protocol as previously described [57]. The CCFT demonstrates good intra-rater reliability (ICC 3,1 = 0.85–0.87) and inter-rater reliability (ICC2,1 = 0.78–0.81) in people with neck pain. Its construct validity is also supported by its ability to distinguish symptomatic from healthy individuals and by its correlation with deep cervical flexor thickness (r = 0.65–0.74) [58,59]. The test consists of two phases. In the activation phase, participants will gently nod to increase pressure in 2 mmHg at each level (22, 24, 26, 28, and 30 mmHg), holding each for 10 seconds while correct activation without superficial muscle substitution will be confirmed through observation or palpation. In the performance (endurance) phase, at the highest pressure achieved without substitution, participants will perform repeated 10-second holds with short rests between trials until they can no longer maintain the target pressure for 10 seconds without any compensatory movement or muscle activation. The highest pressure successfully held for 10 seconds will be recorded as the endurance measure.

Activity measures in a natural environment: ActiGraph activity monitor has been used in recent clinical research [60–63]. Physical activity and sleep will be objectively measured using the ActiGraph GT9X Link, a research-grade accelerometer [64–67]. Participants will wear the watch on their non-dominant wrist for 7 consecutive days and nights, ensuring comfort and compliance while maintaining accuracy and validity [68]. Only 5 continuous days with ≥10 hours (600 minutes) per day of wear time will be considered as valid measurement [69]. Participants will receive written instructions and a prepaid envelope for device return after the 7-day period. Data will be processed using the manufacturer’s validated algorithms, with mean values across valid days will be used for analysis. It is recommended to include physical activity measurements including both weekend and weekdays [70] as Gretebeck and Montoye [71] found no day-to-day difference when only weekdays were compared. Single-day ICCs for sleep can also be as low as 0.38 for weekdays and 0.27 for weekends; wearing the device for 7 days and nights achieves ICC of 0.7 for both [72], supporting include of both weekdays and weekend data.

Physical activity. Physical activity will be objectively measured using ActiGraph GT9X Link accelerometer (ActiGraph LLC, Pensacola, FL, USA), which shows strong criterion validity for estimating physical activity [73,74] and good test–retest reliability in healthy adults (ICC 2,1 = 0.806, 95% CI = 0.669–0.898) [74] for healthy adults. Physical activity level will be classified according to the Freedson Adult [75] cut-off points, a validated method developed for ActiGraph accelerometers. These cut-offs define sedentary activity as <100 counts per minute (cpm), light activity as 100–1,951 cpm, moderate activity as 1,952–5,724 cpm, vigorous activity as 5,725–9,498 cpm, and very vigorous activity as ≥9,499 cpm. Time in each category will be calculated in minutes per day, with moderate, vigorous, and very vigorous activity combined into a single moderate to vigorous physical activity (MVPA) category, following WHO physical activity guideline [76], consistent with previous literature [77–79]. These categories examined their independent associations with health outcomes in musculoskeletal pain and recovery [79]; however, the prognostic value of physical activity for neck pain does not exist due to insufficient evidence [18].

Sleep. The sleep quantity and sleep efficiency will be objectively measured using ActiGraph GT9X Link accelerometer (ActiGraph LLC, Pensacola, FL, USA). The ActiGraph demonstrates sufficient concurrent validity and moderate test–retest reliability in healthy adults (ICC2,1 = 0.45 to 0.50) for sleep efficiency and ICC 2,1 = 0.41 to 0.44 for total sleep time [16] for healthy adults [80]. Data will be processed with the manufacturer’s validated Cole–Kripke algorithm for adults, which has been validated against polysomnography among healthy adults [81]. Sleep quantity, expressed as TST (minutes/day) indicates how much sleep an individual obtains, and sleep quality as sleep efficiency (percentage of time in bed spent asleep), reflects how well an individual has slept Sleep quality is an important measure of sleep health [82], and provides additional information beyond sleep quantity [83], supporting its inclusion as a separate factor in prognostic modeling.

Outcomes and outcome measures

Outcomes including pain, disability, and overall changes, will be measured at 12 months using patient-reported outcome measures (PROMs): NPRS, NDI, and GROC, respectively. PROMs that have previously demonstrated adequate validity and reliability in adults with neck pain [84,85].These instruments are widely recommended in neck pain research [84,86], and implementing an electronic PROMs survey is relatively easy and well accepted by patients [87,88].

Pain.

Pain intensity will be measured using the 11-point NPRS [89], which demonstrates moderate test–retest reliability for neck pain (ICC2,1 = 0.67; 95% CI: 0.27–0.84) and acceptable construct validity [90]. Participants will be asked to rate their average pain over the past week on a scale from 0 (no pain) to 10 scale (worst pain imaginable). Continuous NPRS scores will be used in the analyses to preserve information and support adequately powered linear regression modelling.

Disability.

Disability will be measured by using the NDI [91]. The NDI consists of 10 items covering daily activities including personal care, lifting, reading, work, driving, sleeping and recreation. Each item has five ordinal response options from 0 (no disability) to 5 (complete disability), producing a maximum total score of 50, which can be expressed as a percentage (0% to 100%). The NDI has demonstrated excellent test–retest reliability (ICC2,1 = 0.88; 95% CI: 0.63–0.95) and acceptable construct validity for mechanical neck pain population [90]. Continuous NDI scores will be used in the analyses to preserve information and support adequately powered linear regression modelling.

Overall changes.

Overall changes will be measured using the GROC [92], a 15-point ordinal scale, which has excellent test-retest reliability (ICC2,1 = 0.84; 95% CI: 0.65–0.94) for spinal population [93] and acceptable construct validity for people with neck disorders [94]. Participants will be asked to rate their neck condition at 7 levels of improvement (positive scores: “A tiny bit better” to “A very great deal better”), no change (0 score), or 7 levels of worsening (negative scores: “A tiny bit worse” to “A very great deal worse”). Previous studies [92,95] reported meaningful improvement ranges from 2 to 5 points of GROC scale for spinal pain population. In this study, participants reporting an improvement of ≥2 points (positive scores) will be classified as improved, while those reporting no change or worsening (negative scores) will be classified as not improved.

The study variables measured at baseline and 12-month follow-up are presented in Table 1.

thumbnail
Download:
Table 1. Summary of study variables.

https://doi.org/10.1371/journal.pone.0354102.t001

Covariates

Age and sex will be included as covariates in the hierarchical regression model to account for factors that may influence both prognostic factors and outcomes, consistent with previous prognostic studies for neck pain [96–100]. Age and sex are associated with physical measures of physical functioning [101–104] and neck pain outcomes [105,106]. Increasing age is associated with reduced physical functioning (e.g., reduced strength), and higher pain and disability [107–109], while females demonstrate lower physical functioning (e.g., reduced strength), and higher pain intensity and disability [110–112]. Baseline pain intensity (NPRS) and disability (NDI) will also be included to examine whether physical measures provide additional prognostic information beyond baseline pain intensity and disability as individuals with higher baseline pain intensity and disability may demonstrate poorer physical functioning and different outcomes over time. Adjustment for these variables will help account for their shared variance with both physical functioning measures and outcomes after partialling out the variance explained by age, sex, and baseline outcome measures (NPRS, NDI).

The Directed Acyclic Graph [113] is constructed using DAGitty tool [114] (Fig 2) as a conceptual framework to illustrate the relationship between prognostic factors as exposures and outcomes (pain, disability and overall change).

thumbnail
Download:
Fig 2. Directed acyclic graph: Green nodes are exposures, blue nodes are outcomes, and pink nodes are covariates.

https://doi.org/10.1371/journal.pone.0354102.g002

Bias

Several strategies will be implemented to minimize potential sources of bias throughout the study. Selection bias will be minimized by recruiting participants using predefined eligibility criteria and standardized recruitment procedures across study sites. Measurement bias will be reduced through the use of standardized outcome measures and physical assessment procedures conducted by trained assessors. Assessors will follow standardized testing protocols to reduce measurement bias and improve consistency of data collection.

Sample size

This study will examine whether 15 variables (Table 1) predict 12-month pain, disability, and overall change outcomes using multivariable linear and logistic regression analyses. To support prognostic model development and reduce the risk of overfitting, sample size considerations were informed by both conventional and contemporary recommendations for prognostic modelling. Conventional recommendations of at least 10 participants per predictor variable are considered to support adequately powered linear regression analyses, resulting in an estimated minimum sample size of 150 participants for 15 candidate parameters. Because the shrinkage method using LASSO penalized regression produces models with fewer retained predictors [115], the final prognostic models are expected to include fewer predictors than the initial candidate set. In addition, applying the contemporary Riley et al. framework [116] for continuous outcomes, using 15 variables, a shrinkage factor of 0.90, and an anticipated model R2 of 0.63 [41] produced a comparable estimated sample size of approximately 125 participants for adequately powered linear regression. After accounting for an anticipated 20% attrition rate, the recruitment target was set at 156 participants.

For logistic regression analyses using the LASSO penalized regression to examine overall change (GROC) as a binary outcome, a minimum of 5 events per predictor variable was considered sufficient for model development [115]. Based on 15 variables, at least 75 events were required. Using an anticipated event rate of 62% [117], the estimated minimum sample size is 121 participants. After adjusting for an anticipated 20% attrition rate, the required sample size increases to 151 participants.

Statistical analyses

All statistical analyses will be performed by R software (Rstudio, version 4.5.0) [118]. A p value of <0.05 will be considered statistically significant for all analyses, except variable selection. Descriptive statistics will summarize sample characteristic: continuous variables as mean with standard deviation, and categorical variables as numbers with percentages. Loss to follow-up will be reported descriptively, including the number of participants lost to follow-up and reasons for non-participation at the 12-month follow-up where available. Baseline characteristics of completers and non-completers will also be compared using independent t-tests or Mann–Whitney U tests for continuous variables and Chi-squared (χ2) or Fisher’s exact tests for categorical variables to assess potential attrition bias.

Prior to model development, data screening and missing data patterns will be examined. Descriptive statistics will be used to summarize participant characteristics, prognostic factors, and outcomes. Separate regression models will be developed for the continuous outcomes of pain (NPRS) and disability, and the binary outcome of overall change (GROC), dichotomized as improved versus not improved.

Model assumptions will be assessed before regression analyses, including linearity, normality of residuals, homoscedasticity, independence of errors, multicollinearity, and linearity of continuous predictors with the logit where appropriate. Multicollinearity among candidate prognostic factors will be assessed using variance inflation factors (VIF), with VIF values >5 considered indicative of potential multicollinearity.

Variable selection and model development.

Hierarchical regression analyses will be conducted separately for each physical measure and each outcome. In Step 1, the physical measure will be entered alone to examine the initial unadjusted (bivariate) association with the outcome. In Step 2, age and sex will be added to examine the adjusted multivariable association between the physical measure and the outcome after adjusting for age and sex. InStep 3, baseline NPRS/NDI (depending on the outcome) will be added to determine whether the physical measure provides additional contribution beyond baseline NPRS/NDI. Steps 2 and 3 will examine whether the physical measure contributes to the outcome after partialling out the variance explained by age, sex, and baseline outcome measures. Additional contribution will be evaluated using changes in explained variance (ΔR2) [119] and F-change statistics [120].

Following this, candidate predictors demonstrating significant F-change statistics (p < 0.10) [120] and meaningful ΔR2 (≥2.5%) in the hierarchical regression analyses will be entered together with confounders and covariates into a final penalized multivariable regression model using least absolute shrinkage and selection operator (LASSO) regression [121]. LASSO regression applies coefficient shrinkage coefficients toward zero to reduce overfitting and improve model stability when multiple predictors are included in the same model. This approach is consistent with current recommendations for prognostic model development to reduce overfitting [122]. The optimal penalty parameter will be selected using the one-standard-error criterion, lamda.1se (λ.1SE) from cross-validation to promote a more parsimonious and stable model [123]. Separate penalized multivariable linear regression models will be developed for the NPRS and NDI outcomes, while a penalized multivariable logistic regression model will be developed for the GROC outcome.

Prognostic effects will be reported as beta coefficients (β) or odds ratios (OR) with 95% confidence intervals (CIs) and corresponding p values. β values equal to 0 indicate no change in the outcome, values >0 indicate increased outcome scores, and values <0 indicate decreased outcome scores. OR values equal to 1 indicate no association with the outcome, values >1 indicate increased odds of the outcome, and values <1 indicate decreased odds of the outcome.

Sensitivity analyses comparing penalized and non-penalized multivariable regression models will be conducted to examine the consistency and stability of the findings. The model development workflow is presented in Fig 3.

thumbnail
Download:
Fig 3. Statistical analyses workflow.

The figure illustrates the model development process.

https://doi.org/10.1371/journal.pone.0354102.g003

Model performance.

Predictive performance of the prognostic models [124] will be evaluated by examining overall performance, calibration, and discrimination. For the continuous NPRS and NDI outcomes, overall performance will be assessed using the coefficient of determination (R2) and root mean squared error (RMSE), where higher R2 values and lower RMSE values indicate better model performance. For the binary GROC outcome, overall performance will be assessed using the Nagelkerke R2 statistic [125], where higher values indicate better model performance.

Calibration reflects the agreement between predicted and observed outcomes. For the continuous NPRS and NDI outcomes, calibration will be evaluated using the calibration slope and intercept, with values close to 1 and 0 indicating good calibration, respectively. For the binary GROC outcome, calibration will be assessed using calibration slope, calibration intercept, and the Hosmer– Lemeshow statistic goodness-of-fit test [126], where a non-significant p value indicates good model fit.

Discrimination reflects the model’s ability to distinguish between participants with different outcomes. For the binary GROC outcome, discrimination will be assessed using the concordance index (C-statistic) to evaluate the model’s ability to distinguish between participants who do and do not experience improvement, where values of 0.5 indicate no discrimination, values between 0.7 and 0.8 indicate acceptable discrimination, and values >0.8 indicate good discrimination. To account for model optimism and assess stability, internal validation will be conducted using bootstrap resampling with 1,000 iterations [127].

Handling of missing data

Missing values will be imputed using the multiple imputation by chained equations (MICE) method [128]. Missing predictor and 12-month follow-up outcome data will be imputed under the assumption that data are missing at random (MAR). Imputation will be supported by Little’s test of data missing completely at random (p < 0.05) [129]. Sensitivity analyses [130] will be conducted by comparing estimates from complete-case analyses and imputed datasets to assess how missing data handling impacts model performance and results. Complete case analysis excludes participants with missing data, which can result in biased parameter estimates and poor predictive performance if the missingness is related to the outcome [131]. The comparison evaluates differences in discrimination and calibration, with results often showing imputed dataset provides more accurate and less biased predictions [132]. Comparison of the effect estimated (e.g., β, OR) between complete-case and imputed dataset will help us to understand differences in the identified predictors.

Subgroup or interaction analyses

No subgroup analyses are planned a priori. However, hierarchical regression analyses will examine whether physical measures provide additional prognostic contribution beyond age, sex.

Patient and public involvement

This study is developed in close collaboration with the Spinal Pain Patient Partner Advisory Group (PPAG), School of Physical Therapy, Western University. Patient partners were engaged through regular meetings with the CANSpine clinical research group, School of Physical Therapy, Western University, an established forum for discussion and co-design. Through these meetings, patient partners provided critical input in two key areas. First, they supported the inclusion of activity measures in a real environment as potential PFs, such as sleep quality and physical activity, highlighting their relevance to the lived experience of neck pain. Second, they emphasized the importance of using a validated research tool for activity measure (e.g., accelerometer), rather than relying on smartwatches. These insights directly influenced the selection of PFs in the study. The PPAG will continue to be engaged through ongoing CANSpine meetings, including participation in interpretating findings and contributing to public dissemination of results.

Discussion

This prospective longitudinal observational cohort study aims to investigate whether physical measures of physical functioning including AROM, JPE and handgrip strength as impairment-based measures, endurance test for deep neck flexor as performance-based measure, physical activity and sleep as activity measures in a real environment, predict 12-month pain, disability, and overall changes for people with neck pain. By focusing on physical functioning, the study will provide a more robust understanding of how physical measures contribute to future outcomes for neck pain population. The analysis will be guided by contemporary methodological recommendation for prognostic modeling. The prognostic model will be developed using hierarchical regression analyses, followed by penalized regression using LASSO for final model development. Traditional events-per-variable [115] and contemporary Riley’s framework [116] were used to estimate sample size, and model performance will be using discrimination and calibration metrics, enhancing the methodological rigor of the final model. The selected prognostic factors are included based on our current evidence syntheses [18], relevance to clinical practice, and patient partner feedback.

The findings of this study will add to the literature by addressing the limited evidence identified in our systematic review [18], clarifying the prognostic role of novel physical measures and outcomes in people with neck pain. The knowledge gain through this study may inform clinical decision-making by identifying which physical measures can help predict outcomes for people with neck pain. If the model demonstrates adequate performance and calibration, the findings may support prognostic assessment in people with neck pain using physical measures of physical functioning. This could also support early patient identification, personalized treatment planning, goal setting, and the use of simple clinical tools for the management of neck pain. Additionally, the integration of objective and measurable physical measures into prognostic modeling reflects a shift toward more functionally meaningful data-driven care in musculoskeletal rehabilitation. Future work will be needed to externally validate the model and assess its impact on clinical outcomes, healthcare resource utilization, and patient-centered care.

The study incorporates several methodological features anticipated to strengthen its internal validity and applicability, including its prospective nature, inclusion of participants with different types and duration of neck pain, and inclusion of a coherent candidate set of PFs informed by our recent systematic review. Additional features include the use of comprehensive physical measures of physical functioning, including impairment, performance, and activity measures in a real environment, rigorous prognostic model development using multivariate linear and logistic regression analyses with model performance and internal validation procedures, blinded data analysis and transparent reporting aligned with contemporary best-practice recommendations for prognostic research. Anticipated limitations relate to external validity. First, the lack of external validation within the current study will limit the immediate generalizability of the prognostic model; future research will be required to validate the model in independent cohorts. Second, the observational design of the study limits the ability to infer causal relationships between physical measures and outcomes.

Conclusion

This study will provide new evidence on the prognostic value of physical measures of physical functioning as PFs for long-term neck pain outcomes. By integrating impairment, performance-based, and activity measure in a real environment within a rigorous prognostic modelling and validation framework, the results may have the potential to inform personalized rehabilitation strategies and shape future research on prognostic tools in neck pain.

Supporting information

S1 Table. TRIPOD checklist.

Checklist of items reported according to the TRIPOD reporting guideline.

https://doi.org/10.1371/journal.pone.0354102.s001

(DOCX)

Acknowledgments

We thank the CANSpine Patient Partner Advisory Group for their valuable input while developing the protocol of this study.

References

  1. 1. Vos T, Barber RM, Bell B, Bertozzi-Villa A, Biryukov S, Bolliger I, et al. Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. The Lancet. 2015;386(9995):743–800.
  2. 2. Guzman J, Hurwitz E, Carroll L, Haldeman S, Côté P, Carragee E, et al. A new conceptual model of neck pain: linking onset, course, and care: the bone and joint decade 2000–2010 task force on neck pain and its associated disorders task force on neck pain and its associated disorderstask force on neck pain and its associated disorders. Spine. 2008;33(4 Suppl):S14–23.
  3. 3. Bovim G, Schrader H, Sand T. Neck pain in the general population. Spine (Phila Pa 1976). 1994;19(12):1307–9. pmid:8066508
  4. 4. Côté P, Cassidy DJ, Carroll LJ, Kristman V. The annual incidence and course of neck pain in the general population: a population-based cohort study. Pain. 2004;112(3):267–73. pmid:15561381
  5. 5. Côté P, Cassidy JD, Carroll LJ. The epidemiology of neck pain: what we have learned from our population-based studies. J Can Chiropr Assoc. 2003;47(4):284–90.
  6. 6. Palmer KT, Walker-Bone K, Griffin MJ, Syddall H, Pannett B, Coggon D, et al. Prevalence and occupational associations of neck pain in the British population. Scand J Work Environ Health. 2001;27(1):49–56. pmid:11266146
  7. 7. Dieleman JL, Cao J, Chapin A, Chen C, Li Z, Liu A, Horst C, Kaldjian A, Matyasz T, Scott KW. US health care spending by payer and health condition, 1996-2016. JAMA. 2020;323(9):863–84.
  8. 8. Aafreen A, Khan AR, Khan A, Ahmad A, Alzahrani AH, Alhusayni AI, et al. Neck Health metrics and quality of life: a comparative study in bike drivers with and without neck pain. J Multidiscip Healthc. 2023;16:3575–84. pmid:38024127
  9. 9. Turk DC, Dworkin RH, Allen RR, Bellamy N, Brandenburg N, Carr DB, et al. Core outcome domains for chronic pain clinical trials: IMMPACT recommendations. Pain. 2003;106(3):337–45. pmid:14659516
  10. 10. Mayer KP, Johnson AM, Smith DJ, Crandall CM, Johnson MD, Fresenko LE, et al. A core set of outcome measures to assess physical function for adults participating in physical therapist treatment in the hospital: a clinical practice guideline. Phys Ther. 2025;105(6):pzaf076. pmid:40403754
  11. 11. World Health Organization (WHO).The international classification of functioning, disability and health (ICF). 2001. [cited 2025 Oct 10]. Available from: https://www.who.int/standards/classifications/international-classification-of-functioning-disability-and-health
  12. 12. Reiman MP, Manske RC. The assessment of function: how is it measured? A clinical perspective. J Man Manip Ther. 2011;19(2):91–9. pmid:22547919
  13. 13. Taylor AM, Phillips K, Patel KV, Turk DC, Dworkin RH, Beaton D, et al. Assessment of physical function and participation in chronic pain clinical trials: IMMPACT/OMERACT recommendations. Pain. 2016;157(9):1836–50. pmid:27058676
  14. 14. Weldring T, Smith SMS. Patient-reported outcomes (PROs) and patient-reported outcome measures (PROMs). Health Serv Insights. 2013;6:61–8. pmid:25114561
  15. 15. Chiarotto A. Patient-reported outcome measures: best is the enemy of good (but what if good is not good enough?). J Orthop Sports Phys Ther. 2019;49(2):39–42. pmid:30704358
  16. 16. Cook CE, Wright A, Wittstein J, Barbero M, Tousignant-Laflamme Y. Five Recommendations to address the limitations of patient-reported outcome measures. J Orthop Sports Phys Ther. 2021;51(12):562–5. pmid:34783262
  17. 17. Luo X, Edwards CL, Richardson W, Hey L. Relationships of clinical, psychologic, and individual factors with the functional status of neck pain patients. Value Health. 2004;7(1):61–9. pmid:14720131
  18. 18. Begum R, Chamaa AE, Parikh P, Walton D, Rushton A. Physical measures of physical functioning as prognostic factors in predicting outcomes for neck and thoracic pain: a systematic review. BMC Musculoskelet Disord. 2025;27(1):3. pmid:41316109
  19. 19. Steyerberg EW, Moons KGM, van der Windt DA, Hayden JA, Perel P, Schroter S, et al. Prognosis Research Strategy (PROGRESS) 3: prognostic model research. PLoS Med. 2013;10(2):e1001381. pmid:23393430
  20. 20. Hayden JA, van der Windt DA, Cartwright JL, Côté P, Bombardier C. Assessing bias in studies of prognostic factors. Ann Intern Med. 2013;158(4):280–6. pmid:23420236
  21. 21. Collins GS, Reitsma JB, Altman DG, Moons KGM, TRIPOD Group. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. The TRIPOD Group. Circulation. 2015;131(2):211–9. pmid:25561516
  22. 22. CANSpine. SPINA Spinal Rehabilitation Registry. Western University, London, ON, Canada. [cited 2025 Sep 8]. Available from: https://www.uwo.ca/fhs/canspine/spina-registry/index.html
  23. 23. Blanpied PR, Gross AR, Elliott JM, Devaney LL, Clewley D, Walton DM, et al. Neck pain: revision 2017: clinical practice guidelines linked to the international classification of functioning, disability and health from the orthopaedic section of the American Physical Therapy Association. J Orthopaed Sports Phys Ther. 2017;47(7):A1–83.
  24. 24. Shahidi B, Johnson CL, Curran-Everett D, Maluf KS. Reliability and group differences in quantitative cervicothoracic measures among individuals with and without chronic neck pain. BMC Musculoskelet Disord. 2012;13:215. pmid:23114092
  25. 25. Strimpakos N. The assessment of the cervical spine. Part 2: strength and endurance/fatigue. J Bodyw Mov Ther. 2011;15(4):417–30. pmid:21943615
  26. 26. Capuano-Pucci D, Rheault W, Aukai J, Bracke M, Day R, Pastrick M. Intratester and intertester reliability of the cervical range of motion device. Arch Phys Med Rehabil. 1991;72(5):338–40. pmid:2009054
  27. 27. Youdas JW, Carey JR, Garrett TR. Reliability of measurements of cervical spine range of motion--comparison of three methods. Phys Ther. 1991;71(2):98–104; discussion 105-6. pmid:1989013
  28. 28. Youdas JW, Garrett TR, Suman VJ, Bogard CL, Hallman HO, Carey JR. Normal range of motion of the cervical spine: an initial goniometric study. Phys Ther. 1992;72(11):770–80. pmid:1409874
  29. 29. Fletcher JP, Bandy WD. Intrarater reliability of CROM measurement of cervical spine active range of motion in persons with and without neck pain. J Orthop Sports Phys Ther. 2008;38(10):640–5. pmid:18827326
  30. 30. Audette I, Dumas J-P, Côté JN, De Serres SJ. Validity and between-day reliability of the cervical range of motion (CROM) device. J Orthop Sports Phys Ther. 2010;40(5):318–23. pmid:20436238
  31. 31. Malik AA, Robinson S, Khan WS, Dillon B, Lovell ME. Assessment of range of movement, pain and disability following a whiplash injury. Open Orthop J. 2017;11:541–5. pmid:28839498
  32. 32. Alsultan F, Cescon C, De Nunzio AM, Barbero M, Heneghan NR, Rushton A, et al. Variability of the helical axis during active cervical movements in people with chronic neck pain. Clin Biomech (Bristol). 2019;62:50–7. pmid:30690409
  33. 33. De Rosario H, Vivas MJ, Sinovas MI, Page Á. Relationship between neck motion and self-reported pain in patients with whiplash associated disorders during the acute phase. Musculoskelet Sci Pract. 2018;38:23–9. pmid:30218881
  34. 34. Lindenmann S, Tsagkaris C, Farshad M, Widmer J. Kinematics of the cervical spine under healthy and degenerative conditions: a systematic review. Ann Biomed Eng. 2022;50(12):1705–33. pmid:36496482
  35. 35. Rudolfsson T, Björklund M, Djupsjöbacka M. Range of motion in the upper and lower cervical spine in people with chronic neck pain. Man Ther. 2012;17(1):53–9. pmid:21945295
  36. 36. Bogduk N, Mercer S. Biomechanics of the cervical spine. I: normal kinematics. Clin Biomech (Bristol). 2000;15(9):633–48. pmid:10946096
  37. 37. Fernández-Pérez AM, Villaverde-Gutiérrez C, Mora-Sánchez A, Alonso-Blanco C, Sterling M, Fernández-de-Las-Peñas C. Muscle trigger points, pressure pain threshold, and cervical range of motion in patients with high level of disability related to acute whiplash injury. J Orthop Sports Phys Ther. 2012;42(7):634–41. pmid:22677576
  38. 38. Stenneberg MS, Rood M, de Bie R, Schmitt MA, Cattrysse E, Scholten-Peeters GG. To what degree does active cervical range of motion differ between patients with neck pain, patients with whiplash, and those without neck pain? A systematic review and meta-analysis. Arch Phys Med Rehab. 2017;98(7):1407–34.
  39. 39. Sterling M, Jull G, Kenardy J. Physical and psychological factors maintain long-term predictive capacity post-whiplash injury. Pain. 2006;122(1–2):102–8. pmid:16527397
  40. 40. Sterling M, Jull G, Vicenzino B, Kenardy J, Darnell R. Development of motor system dysfunction following whiplash injury. Pain. 2003;103(1–2):65–73. pmid:12749960
  41. 41. Sterling M, Jull G, Vicenzino B, Kenardy J, Darnell R. Physical and psychological factors predict outcome following whiplash injury. Pain. 2005;114(1–2):141–8. pmid:15733639
  42. 42. Smith K, Hall T, Robinson K. The influence of age, gender, lifestyle factors and sub-clinical neck pain on the cervical flexion-rotation test and cervical range of motion. Man Ther. 2008;13(6):552–9. pmid:17945529
  43. 43. Yoshida R, Yasuda T, Kuruma H. Analysis of cervical and upper thoracic spinal segmental rotation angles during end-range neck rotation: comparison with and without neck pain. J Man Manip Ther. 2022;30(6):328–33. pmid:35384789
  44. 44. Kasch H, Stengaard-Pedersen K, Arendt-Nielsen L, Staehelin Jensen T. Headache, neck pain, and neck mobility after acute whiplash injury: a prospective study. Spine (Phila Pa 1976). 2001;26(11):1246–51. pmid:11389391
  45. 45. AlDahas A, Devecchi V, Deane JA, Falla D. Measurement properties of cervical joint position error in people with and without chronic neck pain. PLoS One. 2023;18(10):e0292798. pmid:37824590
  46. 46. AlDahas A, Heneghan NR, Althobaiti S, Deane JA, Rushton A, Falla D. Measurement properties of cervical joint position error in people with and without neck pain: a systematic review and narrative synthesis. BMC Musculoskelet Disord. 2024;25(1):44. pmid:38200520
  47. 47. Nagai T, Clark NC, Abt JP, Sell TC, Heebner NR, Smalley BW, et al. The effect of target position on the accuracy of cervical-spine-rotation active joint-position sense. J Sport Rehabil. 2016;25(1):58–63. pmid:25803657
  48. 48. Treleaven J, Jull G, Sterling M. Dizziness and unsteadiness following whiplash injury: characteristic features and relationship with cervical joint position error. J Rehabil Med. 2003;35(1):36–43. pmid:12610847
  49. 49. Micarelli A, Viziano A, Carlino P, Granito I, Micarelli RX, Alessandrini M. Reciprocal roles of joint position error, visual dependency and subjective perception in cervicogenic dizziness. Somatosens Mot Res. 2020;37(4):262–70. pmid:32772608
  50. 50. Woodhouse A, Vasseljen O. Altered motor control patterns in whiplash and chronic neck pain. BMC Musculoskelet Disord. 2008;9:90. pmid:18570647
  51. 51. Soysal P, Hurst C, Demurtas J, Firth J, Howden R, Yang L, et al. Handgrip strength and health outcomes: Umbrella review of systematic reviews with meta-analyses of observational studies. J Sport Health Sci. 2021;10(3):290–5. pmid:32565244
  52. 52. Bobos P, Nazari G, Lu Z, MacDermid JC. Measurement properties of the hand grip strength assessment: a systematic review with meta-analysis. Arch Phys Med Rehabil. 2020;101(3):553–65. pmid:31730754
  53. 53. Wilkinson TJ, Gabrys I, Lightfoot CJ, Lambert K, Baker LA, Billany RE, et al. A Systematic review of handgrip strength measurement in clinical and epidemiological studies of kidney disease: toward a standardized approach. J Ren Nutr. 2022;32(4):371–81. pmid:34294555
  54. 54. Biasini NR, Bannon B, Pellegrino M, Qaderi A, Trinh W, Switzer-McIntyre S, et al. Reliability and validity of shoulder and handgrip strength testing. Physiother Can. 2023;75(1):65–71. pmid:37250730
  55. 55. Mathiowetz V, Weber K, Volland G, Kashman N. Reliability and validity of grip and pinch strength evaluations. J Hand Surg Am. 1984;9(2):222–6. pmid:6715829
  56. 56. Núñez-Cortés R, Cruz BDP, Gallardo-Gómez D, Calatayud J, Cruz-Montecinos C, López-Gil JF, et al. Handgrip strength measurement protocols for all-cause and cause-specific mortality outcomes in more than 3 million participants: a systematic review and meta-regression analysis. Clin Nutr. 2022;41(11):2473–89. pmid:36215867
  57. 57. Jull GA, O’Leary SP, Falla DL. Clinical assessment of the deep cervical flexor muscles: the craniocervical flexion test. J Manipulative Physiol Ther. 2008;31(7):525–33. pmid:18804003
  58. 58. Selistre LFA, Melo CDS, de Noronha MA. Reliability and validity of clinical tests for measuring strength or endurance of cervical muscles: a systematic review and meta-analysis. Arch Phys Med Rehabil. 2021;102(6):1210–27. pmid:33383030
  59. 59. Jørgensen R, Ris I, Falla D, Juul-Kristensen B. Reliability, construct and discriminative validity of clinical testing in subjects with and without chronic neck pain. BMC Musculoskelet Disord. 2014;15:408. pmid:25477032
  60. 60. O’Brien JR, Stoyles S, Battison EAJ, Heierle J, Rabbitts JA, Palermo TM, et al. Daily and weekly associations among pain intensity, self-reported activity limitations, and objectively assessed physical activity in youth with acute musculoskeletal pain. J Pain. 2024;25(12):104680. pmid:39306060
  61. 61. Hoydick JF, Johnson ME, Cook HA, Alfikri ZF, Jakicic JM, Piva SR, et al. Algorithm validation for quantifying ActiGraph™ physical activity metrics in individuals with chronic low back pain and healthy controls. Sensors (Basel). 2024;24(16):5323. pmid:39205017
  62. 62. Abeler K, Bergvik S, Sand T, Friborg O. Daily associations between sleep and pain in patients with chronic musculoskeletal pain. J Sleep Res. 2021;30(4):e13237. pmid:33529464
  63. 63. Goossens Z, Van Stallen A, Vermuyten J, De Deyne M, Rice D, Runge N, et al. Day-to-day associations between pain intensity and sleep outcomes in an adult chronic musculoskeletal pain population: a systematic review. Sleep Med Rev. 2025;79:102013. pmid:39467486
  64. 64. Rhudy MB, Dreisbach SB, Moran MD, Ruggiero MJ, Veerabhadrappa P. Cut points of the Actigraph GT9X for moderate and vigorous intensity physical activity at four different wear locations. J Sports Sci. 2020;38(5):503–10. pmid:31865845
  65. 65. Valkenet K, Veenhof C. Validity of three accelerometers to investigate lying, sitting, standing and walking. PLoS One. 2019;14(5):e0217545. pmid:31120994
  66. 66. Sasaki JE, John D, Freedson PS. Validation and comparison of ActiGraph activity monitors. J Sci Med Sport. 2011;14(5):411–6. pmid:21616714
  67. 67. Negrini F, de Sire A, Lazzarini SG, Pennestrì F, Sorce S, Arienti C, et al. Reliability of activity monitors for physical activity assessment in patients with musculoskeletal disorders: a systematic review. J Back Musculoskelet Rehabil. 2021;34(6):915–23. pmid:33935067
  68. 68. Migueles JH, Cadenas-Sanchez C, Ekelund U, Delisle Nyström C, Mora-Gonzalez J, Löf M, et al. Accelerometer data collection and processing criteria to assess physical activity and other outcomes: a systematic review and practical considerations. Sports Med. 2017;47(9):1821–45. pmid:28303543
  69. 69. Quante M, Kaplan ER, Rueschman M, Cailler M, Buxton OM, Redline S. Practical considerations in using accelerometers to assess physical activity, sedentary behavior, and sleep. Sleep Health. 2015;1(4):275–84. pmid:29073403
  70. 70. Trost SG, Pate RR, Freedson PS, Sallis JF, Taylor WC. Using objective physical activity measures with youth: how many days of monitoring are needed? Med Sci Sports Exerc. 2000;32(2):426–31. pmid:10694127
  71. 71. Gretebeck RJ, Montoye HJ. Variability of some objective measures of physical activity. Med Sci Sports Exerc. 1992;24(10):1167–72. pmid:1435166
  72. 72. Lee PH. Number of days required to reliably measure weekday and weekend total sleeping time with accelerometer: a secondary data analysis with National Health and Nutritional Survey (NHANES) 2011-2014 data. Sleep Med. 2024;114:178–81. pmid:38211376
  73. 73. Chinapaw MJM, Slootmaker SM, Schuit AJ, van Zuidam M, van Mechelen W. Reliability and validity of the Activity Questionnaire for Adults and Adolescents (AQuAA). BMC Med Res Methodol. 2009;9:58. pmid:19664254
  74. 74. Yang Y, Schumann M, Le S, Cheng S. Reliability and validity of a new accelerometer-based device for detecting physical activities and energy expenditure. PeerJ. 2018;6:e5775. pmid:30324032
  75. 75. Freedson PS, Melanson E, Sirard J. Calibration of the computer science and applications, Inc. accelerometer. Med Sci Sports Exerc. 1998;30(5):777–81. pmid:9588623
  76. 76. World Health Organisation (WHO). Global recommendations on physical activity for health. 2010. [cited 2025 Sep 7]. Available from: https://www.who.int/publications/i/item/9789241599979
  77. 77. Dupré C, Brégère M, Berger M, Pichot V, Celle S, Garet M, et al. Relationship between moderate-to-vigorous, light intensity physical activity and sedentary behavior in a prospective cohort of older French adults: a 18-year follow-up of mortality and cardiovascular events ─ the PROOF cohort study. Front Public Health. 2023;11:1182552. pmid:37351092
  78. 78. Oliveira CB, Pinheiro MB, Teixeira RJ, Franco MR, Silva FG, Hisamatsu TM, et al. Physical activity as a prognostic factor of pain intensity and disability in patients with low back pain: A systematic review. Eur J Pain. 2019;23(7):1251–63. pmid:30920074
  79. 79. Freire APCF, de Oliveira CBS, Christofaro DGD, Shumate S, Marques LBF, Pinto RZ. Isotemporal substitution of sedentary behavior by different physical activity intensities on pain and disability of patients with chronic low back pain: a cross-sectional study. Arch Phys Med Rehabil. 2022;103(10):1944–50. pmid:35430240
  80. 80. Aili K, Åström-Paulsson S, Stoetzer U, Svartengren M, Hillert L. Reliability of actigraphy and subjective sleep measurements in adults: the design of sleep assessments. J Clin Sleep Med. 2017;13(1):39–47. pmid:27707448
  81. 81. Gao C, Li P, Morris CJ, Zheng X, Ulsa MC, Gao L, et al. Actigraphy-based sleep detection: validation with polysomnography and comparison of performance for nighttime and daytime sleep during simulated shift work. Nat Sci Sleep. 2022;14:1801–16. pmid:36275180
  82. 82. Kohyama J. Which is more important for health: sleep quantity or sleep quality? Children (Basel). 2021;8(7):542. pmid:34202755
  83. 83. Pilcher JJ, Rummel EG, Lamm C. Sleep quality, sleep quantity, and sleep timing: contrasts in Austrian and U.S. college students. Front Sleep. 2024;3:1487739. pmid:41424489
  84. 84. Ramirez MM, Shepherd MH, Melnick SJ, Hanebuth C, Bazemore C, Couce L, et al. Patient-reported outcome measures in physical therapy practice for neck pain: an overview of reviews. J Patient Rep Outcomes. 2023;7(1):97. pmid:37782344
  85. 85. Bobos P, MacDermid JC, Walton DM, Gross A, Santaguida PL. Patient-reported outcome measures used for neck disorders: an overview of systematic reviews. J Orthop Sports Phys Ther. 2018;48(10):775–88. pmid:29932874
  86. 86. van Dulmen SA, van der Wees PJ, Bart Staal J, Braspenning JCC, Nijhuis-van der Sanden MWG. Patient reported outcome measures (PROMs) for goalsetting and outcome measurement in primary care physiotherapy, an explorative field study. Physiotherapy. 2017;103(1):66–72. pmid:27033783
  87. 87. Brusco NK, Atkinson V, Woods J, Myles PS, Hodge A, Jones C, et al. Implementing PROMS for elective surgery patients: feasibility, response rate, degree of recovery and patient acceptability. J Patient Rep Outcomes. 2022;6(1):73. pmid:35798915
  88. 88. Thoomes E, de Graaf M, Gallina A, Falla D, Stathi A. Comparison between two patient-reported outcome measures for patients with cervical radiculopathy: a think-aloud study. Musculoskelet Sci Pract. 2023;65:102764. pmid:37094507
  89. 89. Jensen MP, Karoly P, Braver S. The measurement of clinical pain intensity: a comparison of six methods. Pain. 1986;27(1):117–26. pmid:3785962
  90. 90. Young IA, Dunning J, Butts R, Mourad F, Cleland JA. Reliability, construct validity, and responsiveness of the neck disability index and numeric pain rating scale in patients with mechanical neck pain without upper extremity symptoms. Physiother Theory Pract. 2019;35(12):1328–35. pmid:29856244
  91. 91. Vernon H, Mior S. The neck disability index: a study of reliability and validity. J Manipulative Physiol Ther. 1991;14(7):409–15. pmid:1834753
  92. 92. Jaeschke R, Singer J, Guyatt GH. Measurement of health status. Ascertaining the minimal clinically important difference. Control Clin Trials. 1989;10(4):407–15. pmid:2691207
  93. 93. Bobos P, Ziebart C, Furtado R, Lu Z, MacDermid JC. Psychometric properties of the global rating of change scales in patients with low back pain, upper and lower extremity disorders. A systematic review with meta-analysis. J Orthop. 2020;21:40–8. pmid:32082038
  94. 94. Bobos P, MacDermid J, Nazari G, Furtado R. Psychometric properties of the global rating of change scales in patients with neck disorders: a systematic review with meta-analysis and meta-regression. BMJ Open. 2019;9(11):e033909. pmid:31772112
  95. 95. Kamper SJ, Maher CG, Mackay G. Global rating of change scales: a review of strengths and weaknesses and considerations for design. J Man Manip Ther. 2009;17(3):163–70. pmid:20046623
  96. 96. Grooten WJA, MUSIC Norrtälje Study Group. Predictors for persistent neck/shoulder pain, medical care-seeking due to neck/shoulder pain and sickness absence. Clin Rehabil. 2007;21(7):648–59. pmid:17702707
  97. 97. Kasch H, Bach FW, Jensen TS. Handicap after acute whiplash injury: a 1-year prospective study of risk factors. Neurology. 2001;56(12):1637–43. pmid:11425927
  98. 98. Meisingset I, Stensdotter A-K, Woodhouse A, Vasseljen O. Neck motion, motor control, pain and disability: a longitudinal study of associations in neck pain patients in physiotherapy treatment. Man Ther. 2016;22:94–100. pmid:26586133
  99. 99. Trippolini MA, Dijkstra PU, Côté P, Scholz-Odermatt SM, Geertzen JH, Reneman MF. Can functional capacity tests predict future work capacity in patients with whiplash-associated disorders? Arch Phys Med Rehab. 2014;95(12):2357–66.
  100. 100. Tsang SMH, Szeto GPY, So BCL, Lau RWL, Tai JJ. Using cervical movement velocity to assist the prediction of pain and functional recovery for people with chronic mechanical neck pain. Clin Biomech (Bristol). 2022;93:105607. pmid:35245780
  101. 101. Pan F, Arshad R, Zander T, Reitmaier S, Schroll A, Schmidt H. The effect of age and sex on the cervical range of motion - a systematic review and meta-analysis. J Biomech. 2018;75:13–27. pmid:29776822
  102. 102. Milanović Z, Pantelić S, Trajković N, Sporiš G, Kostić R, James N. Age-related decrease in physical activity and functional fitness among elderly men and women. Clin Interv Aging. 2013;8:549–56. pmid:23723694
  103. 103. Haynes EMK, Neubauer NA, Cornett KMD, O’Connor BP, Jones GR, Jakobi JM. Age and sex-related decline of muscle strength across the adult lifespan: a scoping review of aggregated data. Appl Physiol Nutr Metab. 2020;45(11):1185–96. pmid:32598857
  104. 104. Alahmari KA, Reddy RS, Silvian PS, Ahmad I, Kakaraparthi VN, Alam MM. Association of age on cervical joint position error. J Adv Res. 2017;8(3):201–7. pmid:28203459
  105. 105. Cook AJ, Chastain DC. The classification of patients with chronic pain: age and sex differences. Pain Res Manag. 2001;6(3):142–51. pmid:11854777
  106. 106. Lipson P, Bansal A, Leveque J-C, Kumar R, Fujii T, Nemani VM, et al. Age-adjusted baseline neck disability index values: an analysis of neck pain disability in the General American Population. Spine (Phila Pa 1976). 2025;50(2):137–44. pmid:39318202
  107. 107. Janssen TAH, Lowisz CV, Phillips S. From molecular to physical function: the aging trajectory. Curr Res Physiol. 2024;8:100138. pmid:39811024
  108. 108. Nascimento MDM, Gouveia ÉR, Marques A, Gouveia BR, Marconcin P, França C, et al. The role of physical function in the association between physical activity and gait speed in older adults: a mediation analysis. Int J Environ Res Public Health. 2022;19(19):12581. pmid:36231881
  109. 109. Wettstein M, Eich W, Bieber C, Tesarz J. Pain intensity, disability, and quality of life in patients with chronic low back pain: does age matter? Pain Med. 2019;20(3):464–75. pmid:29701812
  110. 110. Okabe T, Suzuki M, Goto H, Iso N, Cho K, Hirata K, et al. Sex differences in age-related physical changes among community-dwelling adults. J Clin Med. 2021;10(20):4800. pmid:34682922
  111. 111. Sialino LD, Picavet HSJ, Wijnhoven HAH, Loyen A, Verschuren WMM, Visser M, et al. Exploring the difference between men and women in physical functioning: How do sociodemographic, lifestyle- and health-related determinants contribute? BMC Geriatr. 2022;22(1):610. pmid:35864451
  112. 112. Stubbs D, Krebs E, Bair M, Damush T, Wu J, Sutherland J, et al. Sex differences in pain and pain-related disability among primary care patients with chronic musculoskeletal pain. Pain Med. 2010;11(2):232–9. pmid:20002591
  113. 113. Yoo JS, Ahn J, Mayo BC, Bohl DD, Ahn J, Hrynewycz NM, et al. Improvements in grip and pinch strength and patient-reported outcomes after anterior cervical discectomy and fusion. Clin Spine Surg. 2019;32(9):403–8. pmid:31567533
  114. 114. Tennant PWG, Murray EJ, Arnold KF, Berrie L, Fox MP, Gadd SC, et al. Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: review and recommendations. Int J Epidemiol. 2021;50(2):620–32. pmid:33330936
  115. 115. Pavlou M, Ambler G, Seaman SR, Guttmann O, Elliott P, King M, et al. How to develop a more accurate risk prediction model when there are few events. BMJ. 2015;351:h3868. pmid:26264962
  116. 116. Riley RD, Snell KI, Ensor J, Burke DL, Harrell Jr FE, Moons KG, Collins GS. Minimum sample size for developing a multivariable prediction model: Part I–Continuous outcomes. Stat Med. 2019;38(7):1262–75.
  117. 117. Bishop MD, Mintken PE, Bialosky JE, Cleland JA. Patient expectations of benefit from interventions for neck pain and resulting influence on outcomes. J Orthop Sports Phys Ther. 2013;43(7):457–65. pmid:23508341
  118. 118. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2026.
    • 119. Rights JD, Sterba SK. New Recommendations on the Use of R-Squared Differences in Multilevel Model Comparisons. Multivariate Behav Res. 2020;55(4):568–99. pmid:31559890
    • 120. Pope PT, Webster JT. The use of an F-statistic in stepwise regression procedures. Technometrics. 1972;14(2):327.
    • 121. Tibshirani R. Regression Shrinkage and selection via the Lasso. J R Stat Soc Series B: Stat Methodol. 1996;58(1):267–88.
    • 122. Efthimiou O, Seo M, Chalkou K, Debray T, Egger M, Salanti G. Developing clinical prediction models: a step-by-step guide. BMJ. 2024;386:e078276. pmid:39227063
    • 123. Melkumova LE, Shatskikh SY. Comparing Ridge and LASSO estimators for data analysis. Procedia Eng. 2017;201:746–55.
    • 124. Steyerberg EW, Vickers AJ, Cook NR, Gerds T, Gonen M, Obuchowski N, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128–38. pmid:20010215
    • 125. Nagelkerke NJD. A note on a general definition of the coefficient of determination. Biometrika. 1991;78(3):691–2.
    • 126. Hosner DW, Lemeshow S. Applied logistic regression. New York: Jhon Wiley & Sons; 1989.
      • 127. Fernandez-Felix BM, García-Esquinas E, Muriel A, Royuela A, Zamora J. Bootstrap internal validation command for predictive logistic regression models. The Stata J: Promot Commun Stat Stata. 2021;21(2):498–509.
      • 128. Azur MJ, Stuart EA, Frangakis C, Leaf PJ. Multiple imputation by chained equations: what is it and how does it work? Int J Methods Psychiatric Res. 2011;20(1):40–9.
      • 129. Chen H. A test of missing completely at random for generalised estimating equations with missing data. Biometrika. 1999;86(1):1–13.
      • 130. Drechsler M. Sensitivity analysis of complex models. Biol Conserv. 1998;86(3):401–12.
      • 131. Nijman S, Leeuwenberg AM, Beekers I, Verkouter I, Jacobs J, Bots ML, et al. Missing data is poorly handled and reported in prediction model studies using machine learning: a literature review. J Clin Epidemiol. 2022;142:218–29. pmid:34798287
      • 132. Li J, Guo S, Ma R, He J, Zhang X, Rui D, et al. Comparison of the effects of imputation methods for missing data in predictive modelling of cohort study datasets. BMC Med Res Methodol. 2024;24(1):41. pmid:38365610