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

Peer-reviewed

Research Article

Pain phenotyping and investigation of outcomes in physical therapy: An exploratory study in patients with low back pain

  • Abigail T. Wilson ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing

    Abigail.Wilson@ucf.edu

    Affiliation School of Kinesiology and Rehabilitation Sciences, College of Health Professions and Sciences, University of Central Florida, Orlando, Florida, United States of America

    ⨯
  • Joseph L. Riley III,

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

    Affiliations Department of Community Dentistry and Behavioral Science, University of Florida, Gainesville, Florida, United States of America, Pain Research & Intervention Center of Excellence, University of Florida, Gainesville, Florida, United States of America

    ⨯
  • Mark D. Bishop,

    Roles Conceptualization, Formal analysis, Methodology, Supervision, Writing – original draft, Writing – review & editing

    Affiliation University of Florida Department of Physical Therapy, Gainesville, Florida, United States of America

    ⨯
  • Jason M. Beneciuk,

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

    Affiliations University of Florida Department of Physical Therapy, Gainesville, Florida, United States of America, Clinical Research Center, Brooks Rehabilitation, Jacksonville, Florida, United States of America

    ⨯
  • Yenisel Cruz-Almeida,

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

    Affiliations Department of Community Dentistry and Behavioral Science, University of Florida, Gainesville, Florida, United States of America, Pain Research & Intervention Center of Excellence, University of Florida, Gainesville, Florida, United States of America

    ⨯
  • Keri Markut,

    Roles Data curation, Resources, Writing – review & editing

    Affiliation University of Florida Health Rehab Center-Orthopedic and Sports Medicine Institute, Gainesville, Florida, United States of America

    ⨯
  • Charlotte Redd,

    Roles Data curation, Resources, Writing – original draft, Writing – review & editing

    Affiliation University of Florida Health Rehab Center-Orthopedic and Sports Medicine Institute, Gainesville, Florida, United States of America

    ⨯
  • Nicholas LeBlond,

    Roles Data curation, Resources, Writing – original draft, Writing – review & editing

    Affiliation Duke University Health System Durham, North Carolina, United States of America

    ⨯
  • Patrick H. Pham,

    Roles Data curation, Resources, Writing – original draft, Writing – review & editing

    Affiliation Brooks Rehabilitation, Jacksonville, Florida, United States of America

    ⨯
  • David Shirey,

    Roles Data curation, Resources, Writing – original draft, Writing – review & editing

    Affiliation Brooks Rehabilitation, Jacksonville, Florida, United States of America

    ⨯
  • Joel E. Bialosky

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing

    Affiliations University of Florida Department of Physical Therapy, Gainesville, Florida, United States of America, Clinical Research Center, Brooks Rehabilitation, Jacksonville, Florida, United States of America

    ⨯

Abstract

Phenotypes have been proposed as a method of characterizing subgroups based on biopsychosocial factors to identify responders to analgesic treatments. This study aimed to, first, confirm phenotypes in patients with low back pain receiving physical therapy based on an a priori set of factors used to derive subgroups in other pain populations. Second, an exploratory analysis examined if phenotypes differentiated pain and disability outcomes at four weeks of physical therapy. Fifty-five participants completed psychological questionnaires and pressure pain threshold (PPT). Somatization, anxiety, and depression domains of the Symptom-Checklist-90-Revised, and PPT, were entered into a hierarchical agglomerative cluster analysis with Ward’s method to identify phenotypes. Repeated measures ANOVAs assessed pain ratings and disability by phenotype at four weeks. Three clusters emerged: 1) high emotional distress and pain sensitivity (n = 10), 2) low emotional distress (n = 34), 3) low pain sensitivity (n = 11). As an exploratory study, clusters did not differentiate pain ratings or disability after four weeks of physical therapy (p’s>0.05). However, trends were observed as magnitude of change for pain varied by phenotype. This supports the characterization of homogenous subgroups based on a protocol conducted in the clinical setting with varying effect sizes noted by phenotype for short-term changes in pain. As an exploratory study, future studies should aim to repeat this trial in a larger sample of patients.

Citation: Wilson AT, Riley JL III, Bishop MD, Beneciuk JM, Cruz-Almeida Y, Markut K, et al. (2023) Pain phenotyping and investigation of outcomes in physical therapy: An exploratory study in patients with low back pain. PLoS ONE 18(2): e0281517. https://doi.org/10.1371/journal.pone.0281517

Editor: Fatih Özden, Mugla Sitki Kocman Universitesi, TURKEY

Received: November 10, 2022; Accepted: January 25, 2023; Published: February 14, 2023

Copyright: © 2023 Wilson 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: All relevant data are within the manuscript and its Supporting Information files.

Funding: This project was funded by Brooks-PHHP Research Collaboration Funds. Abigail Wilson would also like to acknowledge this project was supported in part by a Promotion of Doctoral Studies I Scholarship in 2018 from the Foundation for Physical Therapy Research. REDCap at the University of Florida was funded by NCATS grant UL1 TR000064. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

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

Introduction

Globally, the need for rehabilitation services is large with 2.41 billion individuals having a neuromusculoskeletal condition that would benefit from a non-pharmacologic treatment approach [1]. Low back pain (LBP) is the most prevalent condition contributing to this need[1] and the leading cause of years lived with disability [2]. A current challenge in the management of LBP is the large variation in rehabilitation outcomes [3,4], which may stem from a discrepancy between the heterogenous nature of LBP and a traditional ‘one size fits all’ approach that is commonly applied. Given the large need for rehabilitation of LBP [1] and inconsistent outcomes, a more precise approach to inform clinical decision making is needed.

Treatment stratification identifies homogenous subgroups to optimize clinical outcome prediction and has been endorsed as a step toward precision medicine. However, current approaches have limitations. For example, clinical prediction rules fail to be replicated in validation studies [5,6]. Prognostic based approaches that incorporate the STaRT Back Tool show promising results [7–9]; however, recent results found no effect on the transition to chronic pain [10]. Additionally, these approaches focus on negative pain related psychological factors and symptom location yet do not account for other potentially relevant prognostic factors, such as pain sensitivity.

Phenotyping is a more comprehensive method of identifying homogenous subgroups based on the interaction between prognostic indicators [11], psychological [12], or pain sensitivity factors [13]. Phenotypes have the potential to differentiate pain trajectories and clinical outcomes [11]. In a study of patients receiving physical therapy for musculoskeletal pain, five phenotypes were identified based on eleven pain-related prognostic factors [14]. Greater variability in pain, function, and recovery [15] were observed at fifty-two weeks across phenotypes compared to the anatomical location of pain, suggesting phenotypes more precisely distinguished long-term trajectories [14]. Numerous studies have identified phenotypes; however, few have prospectively examined their utility in distinguishing short-term clinical outcomes in rehabilitation despite this being a research priority [11]. An additional limitation of the prior work is phenotypes were derived based on a number of clinical factors that may not be feasible to implement in the clinical setting. Phenotyping can also be challenging to implement in the clinical setting due to extensive clinician and patient burden.

In the Orofacial Pain Prospective Evaluation and Risk Assessment (OPPERA) study, reliable phenotypes of patients with temporomandibular dysfunction and healthy controls were derived based on somatization, depression, anxiety dimensions of the Symptom Checklist-90-Revised (SCL-90-R) as well as Pressure Pain Threshold (PPT) applied to the upper trapezius [16,17]. Three phenotypes were identified that distinguished individuals who developed chronic pain [16] and were further characterized by differences in sex, age, pain ratings, psychological factors, and quantitative sensory testing [17]. In addition to being validated in a large sample of individuals with mixed chronic pain conditions [17], this protocol could be feasibly implemented in a clinical setting due to the relatively low clinician and patient burden (one questionnaire and PPT). The prior studies examined this set of variables in patients receiving treatment at a multidisciplinary chronic pain clinic [17]; however, this had never been specifically applied to patients receiving physical therapy for LBP.

Therefore, this current research aimed to, first, identify if these phenotypes are present in patients receiving physical therapy for LBP based on the set of established [16] and validated [17] variables using clinically feasible measures. Second, as an exploratory analysis, this study prospectively examined if identified phenotypes were associated with differences in clinical outcomes of pain intensity and disability after 4 weeks of physical therapy. Four weeks were selected due to prior research selecting this time point for short-term outcomes [18,19]. We add to the existing body of literature by applying this approach to a new sample, patients receiving physical therapy, and examining phenotypes’ prognostic utility for rehabilitation outcomes.

Materials and methods

Data for this observational, prospective cohort study was collected between December 2020-August 2021 from six outpatient physical therapy clinics within Brooks Rehabilitation in Jacksonville, FL, two outpatient physical therapy clinics within University of Florida Health in Gainesville, FL, and one outpatient physical therapy clinic within University of Florida Health in Jacksonville, FL. Thirteen physical therapists were recruited by word of mouth and underwent IRB-01 training and a 30-minute training session by the study coordinator to standardize data collection and recruitment methods. This study was approved by the University of Florida Institutional Review Board for Human Subjects Research. All participants provided written informed consent to enroll in the study.

Participants

Consecutive participants were recruited by the patient’s physical therapist at the initial evaluation or first follow-up appointment using a standard script and study flyer. Participants between 18–75 years old who were currently receiving outpatient physical therapy for LBP at one of the approved study sites were eligible to participate. LBP was defined as pain between the inferior posterior margin of the ribs and the horizontal gluteal fold [20]. Participants who did not speak English, had a systemic medical condition known to affect sensation, or low back surgery or fracture within the past 6 months were excluded.

Study overview

At baseline, enrolled participants completed self-reported demographic and psychological measures online using Research Electronic Data Capture (REDCap) tools, a secure online software designed for collecting data in research studies, hosted at the University of Florida. PPT was assessed at baseline and once weekly for four weeks during scheduled physical therapy sessions by physical therapists trained by the study coordinator. Baseline measurement was within 2 weeks of the initial evaluation and defined as the appointment after the patient enrolled in the study. Clinical pain ratings and the Oswestry Disability Index were collected at baseline and 4 weeks, allowing for assessment of short-term clinical outcomes. Four weeks was selected due to prior literature selecting this as a measurement time point for short term outcomes.(7) Patients received treatment at the discretion of their physical therapist.

Measures

Measures to build phenotypes.

Based on results of the original phenotyping trial16 as well as a follow-up validation study [17], somatization, anxiety, depression, and PPT applied to the upper trapezius and lower back were selected as a priori clustering variables.

Symptom Checklist-90-Revised (SCL-90-R).

The SCL-90-R [21,22] is a valid 90-item questionnaire in which individual items are scored from 0 (not at all) to 4 (extremely) and higher scores indicate greater psychological distress. The SCL-90-R measures 9 psychological domains: somatization, anxiety, depression, obsessive-compulsive, interpersonal sensitivity, hostility, phobic anxiety, paranoid ideation, psychoticism. Only the somatization, anxiety, and depression subscales were included in the cluster analysis. The SCL-90R has been previously applied to patients with low back pain [23–25].

Pressure-Pain Threshold (PPT).

A digital pressure algometer (Wagner Instruments FPX 25, Greenwich, CT) with a 1 cm diameter rubber tip was applied at 1 kgf/s to two locations: 1) locally at the low back medial to the posterior superior iliac spine on the most painful side and 2) remotely to the upper trapezius ipsilateral to the most painful side of LBP. Participants positioned in prone for the low back site and in sitting for the upper trapezius site. Participants were instructed to indicate when the sensation first changed from pressure to pain (pain threshold). This procedure was repeated two times [26] and the average PPT analyzed. PPT demonstrates excellent intra-rater reliability and good to excellent inter-rater reliability in patients with low back pain [27,28].

Measures used to characterize phenotypes.

Study participants completed a demographic and clinical form including: gender, age, race, ethnicity, employment status, marital status, educational level, pain duration.

Pain Catastrophizing Scale (PCS).

The PCS is a 13-item valid [29–31] questionnaire in which individuals respond to a statement on a scale from 0 to 4. Scores range from 0–52 with higher scores indicating higher catastrophizing levels.

Fear-Avoidance Beliefs Questionnaire (FABQ).

The FABQ is a valid [32–34] measure of fear avoidance in patients with LBP. Items are scored from 0–6 with higher scores indicating greater fear avoidance. The FABQ includes a 7-item work subscale (scores range from 0–42 points) and a 4-item physical activity subscale (scores range from 0–24 points).

Tampa Scale of Kinesiophobia (TSK).

The TSK is an 11-item questionnaire with acceptable psychometric properties [35] that quantifies the fear of movement and injury/re-injury. Individual items are scored from 1 to 4 with totals ranging from 11–44. Higher TSK scores indicate greater kinesiophobia.

Pain Self-Efficacy Questionnaire (PSEQ).

The PSEQ measures the degree of pain related self-efficacy. The PSEQ consists of 10 items scored from 0–6 with totals ranging from 0–60. Higher scores indicate elevated levels of pain-related self-efficacy [36].

Pittsburgh Sleep Quality Index (PSQI).

The PSQI is a valid [37,38] self-report measure of sleep quality and disturbance in the past month consisting of 19 items with 7 components totaling a score ranging from 0–21 [39]. A higher score indicates a poorer sleep quality.

Selection of measures used to determine short-term outcomes

Clinical pain ratings.

Current, best, and worst clinical pain ratings over the past 24 hours were reported using a 101-point numeric pain rating scale (NRS) where 0 = no pain and 100 = worst pain imaginable [40–44]. Mean pain ratings were calculated by averaging the participant’s current, best, and worst LBP ratings.

Oswestry Disability Index (ODI).

The ODI is a 10-item valid [45] questionnaire that examines perceived disability specific to LBP. Participants answered questions on a 6-point scale with total scores ranging from 0–50 points (0–100%). Higher scores indicate greater perceived disability.

Statistical analysis

We acknowledge that recommendations vary with some recent sources recommending a minimum sample of 20–30 participants per cluster [46] and others recommending using a ratio of participants to items of at least 10:1, suggesting a minimum sample of 50 participants [47]. Pertinent to the current study, other pain-related work has used samples based on 2m where m = number of clustering variables [48,49]. We selected 5 variables a priori and, therefore, a minimum of 32 participants were required using this criterion. We acknowledge this is lower than other recommendations; therefore, as an exploratory study, we will focus on reporting effect sizes as future studies may aim to replicate these methods in a larger sample of individuals receiving rehabilitation.

Identification of phenotypes

SPSS v. 25 (IBM, Armonk, NY) was used for all data analysis. PPT, somatization, depression, and anxiety subscales of the SCL-90R were Z-transformed and entered into a Principal Components Analysis with Varimax Rotation. The Principal Components Analysis was conducted prior to the hierarchical cluster analysis for the purpose of reducing dimensionality of the data. Components with eigenvalues greater than 1 were retained per Kaiser’s rule [50]. Saved regression factors were entered into a hierarchical cluster analysis with Ward’s clustering method and squared Euclidean distances. Cluster determination was based on the largest change in agglomeration coefficients between two adjacent steps.

Descriptive statistics were calculated for the total sample and derived clusters for all psychological, pain sensitivity, and outcome variables. A one-way ANOVA determined if derived subgroups significantly differed by somatization, depression, anxiety, and PPT. Next, either a Chi-Square Analysis for categorical variables or a one-way ANOVA for continuous variables determined differences in demographic, psychological (PCS, FABQ-PA, FABQ-W, TSK, PSEQ, PSQI), and clinical factors (pain duration) between the derived subgroups. Our a priori hypothesis was that individuals who reported higher levels of somatization, depression, and anxiety would also report higher levels of other negative pain-related psychological factors, such as catastrophizing. Cohen’s d effect sizes were calculated to determine the magnitude of difference between clusters for significant demographic and psychological variables with the following formula: (Mean of first cluster–Mean of second cluster/pooled standard deviation) and interpreted based on the following thresholds: 0.2 = small, 0.5 = medium, and 0.8 = large [51].

Differences in short-term clinical outcomes between phenotypes.

Dependent variables were normally distributed (Kolmogorov-Smirnov/Shapiro-Wilk p>0.05). In two repeated measures ANOVAs, baseline cluster membership was included as the between subject factor and clinical pain intensity or ODI scores at baseline and four weeks as the within subject factor. A cluster x time interaction effect was examined with Bonferroni simple effects decomposition. Cohen’s d effect sizes were calculated to determine the magnitude of difference in short-term clinical outcomes of pain and disability within each derived subgroup. Cohen’s d was calculated with the following formula: (Mean of the Outcome at Baseline–Mean of the Outcome at 4-weeks/pooled standard deviation) and interpreted with the above effect size thresholds [51]. The p-value for significance was set at p<0.05 and did not adjust for multiple comparisons due the exploratory nature of this analysis.

Results

Physical therapists had an average of 6.8 years of clinical practice, 62% were Orthopedic Clinical Specialists, and 100% reported they were completely aware of the American Physical Therapy Association’s Orthopedic Clinical Practice Guidelines for LBP and had a positive viewpoint. 100% reported collecting at least one measurement of psychological factors. The most provided interventions included: core stabilization, motor control exercises, education, and manual therapy.

Identification and validation of three phenotypes

Fifty-five participants completed all baseline outcomes with 90% follow-up rate at 4 weeks (Fig 1). Our sample size exceeded the recommended number of participants to conduct a cluster analysis [48,49]. Demographic and clinical data is presented in Table 1. Two principal components emerged with an eigenvalue greater than 1.0 [19,34] and a total variance explained = 89.01%. Component 1 consisted of somatization, depression, and anxiety. Component 2 consisted of PPT (Table 2). Factor loadings were sufficient without cross-loading.

thumbnail
Download:
Fig 1. Flow chart of recruitment and enrollment.

https://doi.org/10.1371/journal.pone.0281517.g001

thumbnail
Download:
Table 1. Demographic and clinical factors.

https://doi.org/10.1371/journal.pone.0281517.t001

thumbnail
Download:
Table 2. Principal components analysis.

https://doi.org/10.1371/journal.pone.0281517.t002

Ward’s Hierarchical Cluster Analysis revealed 3 distinct clusters (48% change in agglomeration coefficient between adjacent steps) (Fig 2). Cluster 1 (n = 10) was the smallest and included individuals with high somatization, depression, anxiety, and pain sensitivity (“high emotional distress and pain sensitivity”). Cluster 2 was the largest (n = 34) and included individuals with low somatization, depression, and anxiety (“low emotional distress”). Cluster 2 was also characterized by average pain sensitivity. Cluster 3 (n = 11) was characterized by “low pain sensitivity.” This Cluster was also characterized by average distress. Somatization, depression, anxiety, and PPT significantly differed by cluster (p<0.01).

thumbnail
Download:
Fig 2. Z-scores of variables by phenotype.

https://doi.org/10.1371/journal.pone.0281517.g002

The numerical data is presented as standardized Z-score ± SD. Negative Z-scores reflect a lower somatization, depression, anxiety, and pain sensitivity. Positive Z-scores reflect higher somatization, depression, anxiety, and greater pain sensitivity.

Clusters significantly differed by age (p = 0.04) and ethnicity (p = 0.04) (Table 1). Individuals in the low pain sensitivity subgroup (Cluster 3) were younger than individuals in Cluster 1 (Cohen’s d = 0.69) and Cluster 2 (Cohen’s d = 0.87). In addition to having a low pain sensitivity, Cluster 3 was characterized by a younger age and being college educated. None of the individuals in the low emotional distress subgroup (Cluster 2) were Hispanic. Study site, duration of LBP, and previous number of episodes of LBP did not differ by phenotype (p>0.05).

As demonstrated in Table 3, Clusters significantly differed on all dimensions of the SCL-90R (p<0.05), PCS (p = 0.01), PSEQ (p = 0.03), and PSQI (p<0.01). In individuals with low emotional distress (Cluster 2), catastrophizing was significantly lower compared to Cluster 1 (Cohen’s d = 1.01) and Cluster 3 (Cohen’s d = 0.71) and self-efficacy was significantly higher compared to Cluster 1 (Cohen’s d = 0.96). Individuals with high emotional distress (Cluster 1) reported significantly poorer sleep patterns compared to Cluster 2 (Cohen’s d = 1.07). Clusters did not significantly differ by FABQ-PA (p = 0.67), FABQ-W (p = 0.15), nor TSK (p = 0.18).

thumbnail
Download:
Table 3. Total Scores of Pain-related psychological factors by cluster.

https://doi.org/10.1371/journal.pone.0281517.t003

Short-term outcomes

As demonstrated in Fig 3, a cluster x time interaction effect was not observed (F(2,47) = 2.52, p = 0.09, partial eta2 = 0.10) for pain ratings, suggesting pain ratings did not uniquely change by cluster after 4 weeks of physical therapy. A main effect of cluster membership (F(2,47) = 0.61, p = 0.55, partial eta2 = 0.03) was also not observed. A main effect of time (F(1,47) = 14.12, p<0.01, partial eta2 = 0.23) was observed, indicating pain ratings reduced after 4 weeks of physical therapy with a mean ± SE reduction in pain for the total sample = 11.52 ± 3.07. The cluster x time interaction effect did not reach our a priori threshold for significance (F(2,47) = 2.52, p = 0.09, partial eta2 = 0.10) for pain ratings due to the exploratory nature of this study.

thumbnail
Download:
Fig 3. Changes in pain and disability ratings at baseline and 4-weeks for the total sample and by phenotype.

Error bars represent standard deviation.

https://doi.org/10.1371/journal.pone.0281517.g003

Given the probability of chance was under 10%, we made a post-hoc decision to examine effect sizes to identify trends. Cluster 1 displayed a large effect size for reduction in pain ratings by 4 weeks (Cohen’s d = 0.95), Cluster 2 displayed a small to medium effect size (Cohen’s d = 0.32), and Cluster 3 displayed a small effect size (Cohen’s d = 0.21). These trends suggest individuals with high emotional distress and pain sensitivity (Cluster 1) have the largest reductions in pain in response to interventions provided by physical therapists.

For changes in disability, a cluster x time effect was not observed (F(2,47) = 0.03, p = 0.97, partial eta2 = 0.001). A main effect of cluster membership was not observed (F(2,47) = 1.07, p = 0.35, partial eta2 = 0.04) but a main effect of time was observed (F(1,47) = 6.57, p = 0.01, partial eta2 = 0.12) suggesting similar changes in disability over time. Results for disability and pain outcomes remained the same when controlling for age, ethnicity, study site, sex, and race. Given the probability of chance was under 10%, we made a post-hoc decision to examine effect sizes to identify trends. Cluster 1 displayed a small to medium effect size (Cohen’s d = 0.29), Cluster displayed a small to medium effect size (Cohen’s d = 0.26), and Cluster 3 displayed a small to medium effect size (Cohen’s d = 0.25). Collectively, the preliminary trends suggest small to medium reductions in disability across phenotypes.

Discussion

We identified three clusters in patients with LBP using clinically relevant methods to collect psychological and pain sensitivity variables previously validated in a sample of individuals with varying pain conditions [16,17]. In our trial, three phenotypes were derived: 1) high emotional distress and pain sensitivity (n = 10), 2) low emotional distress (n = 34), 3) low pain sensitivity (n = 11).

Prior research [16,17] using the same clustering variables also identified three phenotypes. Our first phenotype paralleled the global symptoms cluster (higher pain sensitivity and psychological distress) [16,17] and our second phenotype paralleled the previously published adaptive cluster (lower pain sensitivity and psychological distress) [16,17]. In contrast, we identified a third phenotype characterized by individuals with low pain sensitivity (high PPT) while prior studies [16,17] identified a third phenotype characterized by high pain sensitivity (low PPT). These variations may be due to sample characteristics. An important difference between our work and that of others was that this trial was conducted in patients receiving rehabilitation for LBP while the others were conducted in healthy individuals and patients with temporomandibular dysfunction [16,17], individuals with complex persistent pain [17], and those seeking treatment at a multidisciplinary pain clinic [17]. Individuals with LBP receiving physical therapy may differ from individuals with complex persistent pain receiving care at a multidisciplinary pain clinic by key factors known to influence pain sensitivity, such as widespread chronic pain or activity level [52–54]. Furthermore, we acknowledge that this is an exploratory analysis that should be replicated in a larger sample size.

Phenotypes in the present study were further characterized by differences in age, ethnicity, and pain-related psychological factors. First, individuals demonstrating low pain sensitivity (Cluster 3) were significantly younger than the other clusters which is consistent with pain and aging literature demonstrating lower pain sensitivity in younger individuals compared to older adults [55–58]. Second, individuals in the high emotional distress and pain sensitivity phenotype (Cluster 1) had the largest proportion of individuals who were Hispanic. Consistent with our results, individuals who are Hispanic demonstrate elevated pain sensitivity in other studies [59]. Third, individuals in the high emotional distress and pain sensitivity phenotype (Cluster 1) also demonstrated elevated pain catastrophizing, lower pain-related self-efficacy, and poorer sleep patterns. Variability in psychological profiles exist in patients with LBP [12] and, consistent with previous literature, a subgroup of patients have higher depression, anxiety, pain catastrophizing [60], and poorer sleep quality [61].

We did not see significant interaction effects in our statistical models examining the clinical relevance of the phenotypes, possibly due to the sample size. However, exploratory analysis of effect sizes suggests these phenotypes may have different magnitudes of reduction in pain ratings after four weeks of physical therapy. A large effect size for reduction in pain was observed for individuals in the high emotional distress and pain sensitivity phenotype (Cluster 1). Thus, the phenotypes might include treatment moderators. Individuals with high fear and pain catastrophizing demonstrated the largest change in pain at 4 weeks when receiving physical therapy compared to those with low psychological factors [62]. While treatment was not prescribed across participating clinical sites in the current study, 100% of physical therapists in this study reported modifying treatment for those with elevated pain related psychological factors using psychologically informed principles [63–65], such as motivational interviewing and graded exposure. While speculative, this suggests physical therapists may be equipped, or at least knowledgeable, to successfully managing patients with elevated pain-related psychological factors. Although disability significantly reduced for the total sample, phenotypes did not differentiate short-term change in disability. While depression and somatization are predictors of disability in patients with LBP [66], improvement may vary according to baseline disability or STarT Back Tool Risk Category (which was not assessed in this study) and may help account for this result [67]. Future trials are needed in larger samples to validate the trends observed in this study and determine prognostic over treatment direction value of the phenotypes. Future research is needed to determine the prognostic capabilities of this subgroups. However, patients with high pain intensity at baseline are likely to be similar to other groups within a month. This may reflect a regression to the mean for this group; however, future studies would need to determine if this is occurring.

Nonetheless, this study advances phenotyping research. While previous studies have been conducted in experimental settings or using multiple questionnaires not feasible to administer in a clinical setting, this study took a clinically applicable approach based on published questionnaires/variables that were applied it to a novel setting, rehabilitation. We used one questionnaire, the SCL-90-R, that was collected online and could be efficiently applied in the clinical setting to form phenotypes. PPT data was quick to administer by a trained physical therapist during the patient’s appointment, providing ecological validity. Prospectively collecting data limits bias and is consistent with current phenotyping research priorities [11]. This protocol represents an advancement in implementing phenotyping research in the clinical setting.

There are limitations worth considering when interpreting the results of this study, specifically the sample size and lack of standardization of physical therapist interventions. As noted throughout this manuscript, we acknowledge that these findings are preliminary due to the small sample size. Furthermore, thresholds for classifying patients in each cluster based on the variable are unknown. Effect sizes might be influenced by the fairly small cluster samples sizes. Furthermore, as an observational study, this study lacks a comparator group. Future studies may aim to include a comparator group examining alternative methods of prognostic classification, such as the STaRT Back Tool. Although physical therapists were aware of Clinical Practice Guidelines (CPG) and reported routinely screening for negative pain-related psychological factors, it is unknown if clinicians were providing CPG informed treatment or interventions to target psychological factors. Additionally, inter-rater reliability of the PPT measurements were not collected. Data regarding interventions provided by physical therapists, frequency of treatment, and compliance may have provided greater context informing the lack of difference in short-term clinical outcomes.

The study represents an important first phase of future phenotyping trials conducted in the rehabilitation setting. Future trials with larger sample sizes are needed to validate the trends reported in the results for short-term pain and disability outcomes. Next, identifying key intervention targets related phenotype membership and matching interventions to these targets is an essential next step to validate and implement these approaches into clinical care.

Conclusion

Similar to previously published studies, three exploratory phenotypes in a small sample of patients receiving physical therapy for LBP were identified when clustering based on somatization, depression, anxiety, and PPT characteristics. Similar phenotypes were observed in prior studies [16,17] with the exception of the phenotype based on pain sensitivity. As an exploratory analysis, we also investigated the relevance of phenotypes to short-term clinical outcomes of pain and disability. Effect sizes for reductions in pain suggest different magnitudes of change by phenotype. Specifically, a large effect size was observed for reductions in pain for individuals with high emotional distress and pain sensitivity. While future research is needed to validate these results in a larger sample, the clinically applicable design provides a novel approach for future clinical trials and suggests phenotyping is feasible in clinical practice.

References

  1. 1. Cieza A, Causey K, Kamenov K, Hanson SW, Chatterji S, Vos T. Global estimates of the need for rehabilitation based on the Global Burden of Disease study 2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet (London, England). 2021;396(10267):2006–17. pmid:33275908
  2. 2. Wu A, March L, Zheng X, Huang J, Wang X, Zhao J, et al. Global low back pain prevalence and years lived with disability from 1990 to 2017: estimates from the Global Burden of Disease Study 2017. Annals of translational medicine. 2020;8(6):299–. pmid:32355743
  3. 3. Hancock MJ, Hill JC. Are Small Effects for Back Pain Interventions Really Surprising? The Journal of orthopaedic and sports physical therapy. 2016;46(5):317–9. pmid:27133941
  4. 4. Keller A, Hayden J, Bombardier C, van Tulder M. Effect sizes of non-surgical treatments of non-specific low-back pain. Eur Spine J. 2007;16(11):1776–88. pmid:17619914
  5. 5. Fernandez-de-Las-Penas C, Cleland JA, Salom-Moreno J, Palacios-Cena M, Martinez-Perez A, Pareja JA, et al. Prediction of Outcome in Women With Carpal Tunnel Syndrome Who Receive Manual Physical Therapy Interventions: A Validation Study. J Orthop Sports Phys Ther. 2016;46(6):443–51. pmid:27011304
  6. 6. Patel S, Friede T, Froud R, Evans DW, Underwood M. Systematic review of randomized controlled trials of clinical prediction rules for physical therapy in low back pain. Spine (Phila Pa 1976). 2013;38(9):762–9. pmid:23132535
  7. 7. Beneciuk JM, Bishop MD, Fritz JM, Robinson ME, Asal NR, Nisenzon AN, et al. The STarT back screening tool and individual psychological measures: evaluation of prognostic capabilities for low back pain clinical outcomes in outpatient physical therapy settings. Phys Ther. 2013;93(3):321–33. pmid:23125279
  8. 8. Fritz JM, Beneciuk JM, George SZ. Relationship between categorization with the STarT Back Screening Tool and prognosis for people receiving physical therapy for low back pain. Phys Ther. 2011;91(5):722–32. pmid:21451094
  9. 9. Toh I, Chong HC, Suet-Ching Liaw J, Pua YH. Evaluation of the STarT Back Screening Tool for Prediction of Low Back Pain Intensity in an Outpatient Physical Therapy Setting. J Orthop Sports Phys Ther. 2017;47(4):261–7. pmid:28257616
  10. 10. Delitto A, Patterson CG, Stevans JM, Freburger JK, Khoja SS, Schneider MJ, et al. Stratified care to prevent chronic low back pain in high-risk patients: The TARGET trial. A multi-site pragmatic cluster randomized trial. EClinicalMedicine. 2021;34:100795–. pmid:33870150
  11. 11. Edwards RR, Dworkin RH, Turk DC, Angst MS, Dionne R, Freeman R, et al. Patient phenotyping in clinical trials of chronic pain treatments: IMMPACT recommendations. Pain. 2016;157(9):1851–71. pmid:27152687
  12. 12. Rabey M, Smith A, Beales D, Slater H, O’Sullivan P. Differing Psychologically Derived Clusters in People With Chronic Low Back Pain are Associated With Different Multidimensional Profiles. Clin J Pain. 2016;32(12):1015–27. pmid:26889613
  13. 13. Frey-Law LA, Bohr NL, Sluka KA, Herr K, Clark CR, Noiseux NO, et al. Pain sensitivity profiles in patients with advanced knee osteoarthritis. Pain. 2016;157(9):1988–99. pmid:27152688
  14. 14. Meisingset I, Vasseljen O, Vollestad NK, Robinson HS, Woodhouse A, Engebretsen KB, et al. Novel approach towards musculoskeletal phenotypes. European journal of pain (London, England). 2020.
  15. 15. Aasdahl L, Granviken F, Meisingset I, Woodhouse A, Evensen KAI, Vasseljen O. Recovery trajectories in common musculoskeletal complaints by diagnosis contra prognostic phenotypes. BMC musculoskeletal disorders. 2021;22(1):455–. pmid:34011349
  16. 16. Bair E, Gaynor S, Slade GD, Ohrbach R, Fillingim RB, Greenspan JD, et al. Identification of clusters of individuals relevant to temporomandibular disorders and other chronic pain conditions: the OPPERA study. Pain. 2016;157(6):1266–78. pmid:26928952
  17. 17. Gaynor SM, Bortsov A, Bair E, Fillingim RB, Greenspan JD, Ohrbach R, et al. Phenotypic profile clustering pragmatically identifies diagnostically and mechanistically informative subgroups of chronic pain patients. Pain. 2021;162(5):1528–38. pmid:33259458
  18. 18. Artus M, van der Windt D, Jordan KP, Croft PR. The clinical course of low back pain: a meta-analysis comparing outcomes in randomised clinical trials (RCTs) and observational studies. BMC Musculoskelet Disord. 2014;15:68–. pmid:24607083
  19. 19. Artus M, van der Windt DA, Jordan KP, Hay EM. Low back pain symptoms show a similar pattern of improvement following a wide range of primary care treatments: a systematic review of randomized clinical trials. Rheumatology (Oxford, England). 2010;49(12):2346–56. pmid:20713495
  20. 20. Deyo RA, Dworkin SF, Amtmann D, Andersson G, Borenstein D, Carragee E, et al. Report of the NIH Task Force on Research Standards for Chronic Low Back Pain. Int J Ther Massage Bodywork. 2015;8(3):16–33. pmid:26388962
  21. 21. Derogatis LR, Lipman RS, Covi L. SCL-90: an outpatient psychiatric rating scale—preliminary report. Psychopharmacol Bull. 1973;9(1):13–28. pmid:4682398
  22. 22. Derogatis LR, Rickels K, Rock AF. The SCL-90 and the MMPI: a step in the validation of a new self-report scale. Br J Psychiatry. 1976;128:280–9. pmid:1252693
  23. 23. Williams DA, Urban B, Keefe FJ, Shutty MS, France R. Cluster analyses of pain patients’ responses to the SCL-90R. Pain. 1995;61(1):81–91. pmid:7644252
  24. 24. Kinney RK, Gatchel RJ, Mayer TG. The SCL-90R evaluated as an alternative to the MMPI for psychological screening of chronic low-back pain patients. Spine (Phila Pa 1976). 1991;16(8):940–2. pmid:1835155
  25. 25. Keefe FJ, Crisson J, Urban BJ, Williams DA. Analyzing chronic low back pain: the relative contribution of pain coping strategies. Pain. 1990;40(3):293–301. pmid:2139204
  26. 26. Balaguier R, Madeleine P, Vuillerme N. Is One Trial Sufficient to Obtain Excellent Pressure Pain Threshold Reliability in the Low Back of Asymptomatic Individuals? A Test-Retest Study. PLoS One. 2016;11(8):e0160866. pmid:27513474
  27. 27. Paungmali A, Sitilertpisan P, Taneyhill K, Pirunsan U, Uthaikhup S. Intrarater reliability of pain intensity, tissue blood flow, thermal pain threshold, pressure pain threshold and lumbo-pelvic stability tests in subjects with low back pain. Asian J Sports Med. 2012;3(1):8–14. pmid:22461960
  28. 28. Tabatabaiee A, Takamjani IE, Sarrafzadeh J, Salehi R, Ahmadi M. Pressure Pain Threshold in Subjects With Piriformis Syndrome: Test-Retest, Intrarater, and Interrater Reliability, and Minimal Detectible Changes. Archives of physical medicine and rehabilitation. 2020;101(5):781–8. pmid:31821801
  29. 29. Van Damme S, Crombez G, Bijttebier P, Goubert L, Van Houdenhove B. A confirmatory factor analysis of the Pain Catastrophizing Scale: invariant factor structure across clinical and non-clinical populations. Pain. 2002;96(3):319–24. pmid:11973004
  30. 30. George SZ, Robinson ME. Preference, expectation, and satisfaction in a clinical trial of behavioral interventions for acute and sub-acute low back pain. J Pain. 2010;11(11):1074–82. pmid:20466596
  31. 31. Osman A, Barrios FX, Gutierrez PM, Kopper BA, Merrifield T, Grittmann L. The Pain Catastrophizing Scale: further psychometric evaluation with adult samples. J Behav Med. 2000;23(4):351–65. pmid:10984864
  32. 32. George SZ, Valencia C, Beneciuk JM. A psychometric investigation of fear-avoidance model measures in patients with chronic low back pain. J Orthop Sports Phys Ther. 2010;40(4):197–205. pmid:20357418
  33. 33. Swinkels-Meewisse EJ, Swinkels RA, Verbeek AL, Vlaeyen JW, Oostendorp RA. Psychometric properties of the Tampa Scale for kinesiophobia and the fear-avoidance beliefs questionnaire in acute low back pain. Man Ther. 2003;8(1):29–36. pmid:12586559
  34. 34. Waddell G, Newton M, Henderson I, Somerville D, Main CJ. A Fear-Avoidance Beliefs Questionnaire (FABQ) and the role of fear-avoidance beliefs in chronic low back pain and disability. Pain. 1993;52(2):157–68. pmid:8455963
  35. 35. Woby SR, Roach NK, Urmston M, Watson PJ. Psychometric properties of the TSK-11: a shortened version of the Tampa Scale for Kinesiophobia. Pain. 2005;117(1–2):137–44. pmid:16055269
  36. 36. Nicholas MK. The pain self-efficacy questionnaire: Taking pain into account. European journal of pain (London, England). 2007;11(2):153–63. pmid:16446108
  37. 37. Fontes F, Goncalves M, Maia S, Pereira S, Severo M, Lunet N. Reliability and validity of the Pittsburgh Sleep Quality Index in breast cancer patients. Supportive care in cancer: official journal of the Multinational Association of Supportive Care in Cancer. 2017;25(10):3059–66. pmid:28455545
  38. 38. Mollayeva T, Thurairajah P, Burton K, Mollayeva S, Shapiro CM, Colantonio A. The Pittsburgh sleep quality index as a screening tool for sleep dysfunction in clinical and non-clinical samples: A systematic review and meta-analysis. Sleep medicine reviews. 2016;25:52–73. pmid:26163057
  39. 39. Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry research. 1989;28(2):193–213. pmid:2748771
  40. 40. Bolton JE, Wilkinson RC. Responsiveness of pain scales: a comparison of three pain intensity measures in chiropractic patients. J Manipulative Physiol Ther. 1998;21(1):1–7. pmid:9467094
  41. 41. DeLoach LJ, Higgins MS, Caplan AB, Stiff JL. The visual analog scale in the immediate postoperative period: intrasubject variability and correlation with a numeric scale. Anesth Analg. 1998;86(1):102–6. pmid:9428860
  42. 42. Hartrick CT, Kovan JP, Shapiro S. The numeric rating scale for clinical pain measurement: a ratio measure? Pain Pract. 2003;3(4):310–6. pmid:17166126
  43. 43. Jensen MP. The validity and reliability of pain measures in adults with cancer. J Pain. 2003;4(1):2–21. pmid:14622723
  44. 44. Jensen MP, Turner JA, Romano JM, Fisher LD. Comparative reliability and validity of chronic pain intensity measures. Pain. 1999;83(2):157–62. pmid:10534586
  45. 45. Fairbank JC, Couper J, Davies JB, O’Brien JP. The Oswestry low back pain disability questionnaire. Physiotherapy. 1980;66(8):271–3. pmid:6450426
  46. 46. Dalmaijer ES, Nord CL, Astle DE. Statistical power for cluster analysis. BMC Bioinformatics. 2022;23(1):205. pmid:35641905
  47. 47. Osborne JWC, Anna B. Sample size and subject to item ratio in principal components analysis. Practical Assessment, Research & Evaluation. 2004;9(11).
  48. 48. Ka S. Heuristics for sample size determination in multivariate statistical techniques. Applied Sciences Journal. 2013;27(2):285–7.
  49. 49. S D. Using cluster analysis for market segmentation-typical misconceptions, established methodological weaknesses and some recommendations for improvement. Australian Journal of Market Research. 2003;11(2):5–12.
  50. 50. Kaiser HF. The application of electronic computers to factor analysis. Education & Psychological Measurement. 1960;20:141–51.
  51. 51. J C. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale: Erlbaum; 1988.
    • 52. Arendt-Nielsen L, Morlion B, Perrot S, Dahan A, Dickenson A, Kress HG, et al. Assessment and manifestation of central sensitisation across different chronic pain conditions. Eur J Pain. 2018;22(2):216–41. pmid:29105941
    • 53. Belavy DL, Van Oosterwijck J, Clarkson M, Dhondt E, Mundell NL, Miller CT, et al. Pain sensitivity is reduced by exercise training: Evidence from a systematic review and meta-analysis. Neuroscience and biobehavioral reviews. 2021;120:100–8. pmid:33253748
    • 54. Naugle KM, Fillingim RB, Riley J. L 3rd. A meta-analytic review of the hypoalgesic effects of exercise. J Pain. 2012;13(12):1139–50. pmid:23141188
    • 55. Huang H-W, Wang W-C, Lin C-CK. Influence of age on thermal thresholds, thermal pain thresholds, and reaction time. Journal of clinical neuroscience: official journal of the Neurosurgical Society of Australasia. 2010;17(6):722–6. pmid:20356747
    • 56. Lautenbacher S, Kunz M, Strate P, Nielsen J, Arendt-Nielsen L. Age effects on pain thresholds, temporal summation and spatial summation of heat and pressure pain. Pain. 2005;115(3):410–8. pmid:15876494
    • 57. Rolke R, Baron R, Maier C, Tolle TR, Treede RD, Beyer A, et al. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values. Pain. 2006;123(3):231–43. pmid:16697110
    • 58. El Tumi H, Johnson MI, Dantas PBF, Maynard MJ, Tashani OA. Age-related changes in pain sensitivity in healthy humans: A systematic review with meta-analysis. European journal of pain (London, England). 2017;21(6):955–64. pmid:28230292
    • 59. Ostrom C, Bair E, Maixner W, Dubner R, Fillingim RB, Ohrbach R, et al. Demographic Predictors of Pain Sensitivity: Results From the OPPERA Study. The journal of pain. 2017;18(3):295–307. pmid:27884689
    • 60. Aoyagi K, He J, Nicol AL, Clauw DJ, Kluding PM, Jernigan S, et al. A Subgroup of Chronic Low Back Pain Patients With Central Sensitization. The Clinical journal of pain. 2019;35(11):869–79. pmid:31408011
    • 61. Gerhart JI, Burns JW, Post KM, Smith DA, Porter LS, Burgess HJ, et al. Relationships Between Sleep Quality and Pain-Related Factors for People with Chronic Low Back Pain: Tests of Reciprocal and Time of Day Effects. Annals of behavioral medicine: a publication of the Society of Behavioral Medicine. 2017;51(3):365–75. pmid:27844327
    • 62. Beneciuk JM, Robinson ME, George SZ. Low back pain subgroups using fear-avoidance model measures: results of a cluster analysis. The Clinical journal of pain. 2012;28(8):658–66. pmid:22510537
    • 63. Beneciuk JM, Ballengee LA, George SZ. Treatment monitoring as a component of psychologically informed physical therapy: A case series of patients at high risk for persistent low back pain related disability. Musculoskeletal science & practice. 2019;41:36–42. pmid:30909109
    • 64. Coronado RA, Brintz CE, McKernan LC, Master H, Motzny N, Silva FM, et al. Psychologically informed physical therapy for musculoskeletal pain: current approaches, implications, and future directions from recent randomized trials. Pain reports. 2020;5(5):e847–e. pmid:33490842
    • 65. Keefe FJ, Main CJ, George SZ. Advancing Psychologically Informed Practice for Patients With Persistent Musculoskeletal Pain: Promise, Pitfalls, and Solutions. Phys Ther. 2018;98(5):398–407. pmid:29669084
    • 66. Pincus T, Burton AK, Vogel S, Field AP. A systematic review of psychological factors as predictors of chronicity/disability in prospective cohorts of low back pain. Spine. 2002;27(5):E109–20. pmid:11880847
    • 67. Katzan IL, Thompson NR, George SZ, Passek S, Frost F, Stilphen M. The use of STarT back screening tool to predict functional disability outcomes in patients receiving physical therapy for low back pain. The spine journal: official journal of the North American Spine Society. 2019;19(4):645–54. pmid:30308254