Abstract
Background The effectiveness of risk stratification for low back pain (LBP) management has not been demonstrated in outpatient physical therapy settings.
Objective The purposes of this study were: (1) to assess implementation of a stratified care approach for LBP management by evaluating short-term treatment effects and (2) to determine feasibility of conducting a larger-scale study.
Design This was a 2-phase, preliminary study.
Methods In phase 1, clinicians were randomly assigned to receive standard (n=6) or stratified care (n=6) training. Stratified care training included 8 hours of content focusing on psychologically informed practice. Changes in LBP attitudes and beliefs were assessed using the Pain Attitudes and Beliefs Scale for Physiotherapists (PABS-PT) and the Health Care Providers Pain and Impairment Relationship Scale (HC-PAIRS). In phase 2, clinicians receiving the stratified care training were instructed to incorporate those strategies in their practice and 4-week patient outcomes were collected using a numerical pain rating scale (NPRS), and the Oswestry Disability Index (ODI). Study feasibility was assessed to identify potential barriers for completion of a larger-scale study.
Results In phase 1, minimal changes were observed for PABS-PT and HC-PAIRS scores for standard care clinicians (Cohen d=0.00–0.28). Decreased biomedical (−4.5±2.5 points, d=1.08) and increased biopsychosocial (+5.5±2.0 points, d=2.86) treatment orientations were observed for stratified care clinicians, with these changes sustained 6 months later on the PABS-PT. In phase 2, patients receiving stratified care (n=67) had greater between-group improvements in NPRS (0.8 points; 95% confidence interval=0.1, 1.5; d=0.40) and ODI (8.9% points; 95% confidence interval=4.1, 13.6; d=0.76) scores compared with patients receiving standard physical therapy care (n=33).
Limitations In phase 2, treatment was not randomly assigned, and therapist adherence to treatment recommendations was not monitored. This study was not adequately powered to conduct subgroup analyses.
Conclusions In physical therapy settings, biomedical orientation can be modified, and risk-stratified care for LBP can be effectively implemented. Findings from this study can be used for planning of larger studies.
Stratified care approaches for low back pain (LBP) management are appealing because they involve targeting treatment to subgroups of patients based on key characteristics (eg, psychological factors).1 Recently, Hill and colleagues2 reported that a risk-stratified primary care model (ie, STarT Back approach) resulted in greater improvements in clinical outcomes and cost savings when compared to current best practice for patients with LBP. The effectiveness of prognostic risk stratification for LBP management has not been evaluated when initiated outside of primary care. Therefore, preliminary studies are warranted to determine feasibility of stratified care in secondary care settings and to provide estimates of effect sizes for a larger clinical trial.
Considering that not all patients will enter the health care system through primary care, there is a need to have physical therapists who are adept in implementing risk stratification approaches for LBP management. Psychologically informed physical therapist practice is a secondary prevention approach for chronic LBP that merges impairment-focused physical therapy with cognitive-behavioral principles.3 The primary goal of psychologically informed physical therapist practice is prevention of future LBP-associated disability by emphasizing: (1) identification of individuals who are at high risk for developing chronic LBP based on the presence of psychological distress and (2) targeted treatment aimed at psychological factors in conjunction with traditional, impairment-based physical therapy.3 Biomedical or impairment-based perspectives are predominantly emphasized during the education and clinical practice of many physical therapists, with little, if any, content being provided from a biopsychosocial perspective.3–8 Therefore, sustained clinician attitudes and beliefs toward biomedical treatment orientations may serve as a barrier for clinical practice approaches that rely heavily on psychologically informed principles.4,9 Consequently, there is a need to evaluate whether implementation strategies for shifting clinicians' attitudes and beliefs to a biopsychosocial treatment orientation9 are effective for alignment with current conceptualization of pain experiences.10
Attempts at changing physical therapists' attitudes and beliefs toward a more biopsychosocial treatment orientation have provided mixed results,11–16 with few studies evaluating patient outcomes.12,14 Embedding psychological elements into physical therapist management strategies for LBP is associated with several challenges, including uncertainty about education, implementation, and clinician culture.4 Prior to embracing a stratified care approach that consists of psychologically informed practice principles, there is a need to further investigate implementation strategies in physical therapy settings. First, there is a need to evaluate training and education strategies geared toward increasing a biopsychosocial treatment orientation for physical therapists. Second, there is a need to evaluate if improved clinical outcomes are associated with therapists who received stratified care training following the training for a biopsychosocial treatment orientation.
Therefore, the primary purpose of this study was to assess the feasibility of biopsychosocially oriented stratified care in a physical therapy setting and provide an estimate of the subsequent treatment effect on commonly used clinical outcomes for LBP. For the first phase of this study, our feasibility objectives were to assess: (1) clinician recruitment and sampling methods, (2) implementation of pragmatic clinician training and education strategies, and (3) estimates of training and education effect on clinicians' attitudes and beliefs. For the second phase of this study, our feasibility objectives were to assess: (1) patient recruitment, sampling methods, and follow-up rates and (2) sample size estimations that would allow for future subgroup analyses. Providing estimates of stratified care treatment effect on short-term clinical outcomes also was an objective for the second phase of the study. For exploratory purposes, we compared stratified care and standard care groups by risk categorization for 4-week clinical outcomes to provide preliminary subgroup estimates. The results of this study will be used for planning of future studies that will determine if primary care–based risk stratification strategies can be successfully translated to secondary care settings.
Method
Design Overview
This was a 2-phase, sequential study that evaluated feasibility and generated preliminary treatment effects for 4-week clinical outcomes (Fig. 1). The first phase consisted of clinician training and education (February 2013–April 2013). The second phase consisted of patient outcome collection from the previously trained clinicians (May 2013–February 2014). All clinicians and patients were selected from 7 outpatient physical therapy clinics of Brooks Rehabilitation (Jacksonville, Florida).
Flow diagram for 2-phase study. PABS-PT=Pain Attitudes and Beliefs Scale for Physical Therapists, HC-PAIRS=Health Care Providers' Pain and Impairment Relationship Scale, LBP=low back pain.
Phase 1.
Physical therapists (n=63) employed by a large rehabilitation health system who practiced in an outpatient setting were recruited by email to participate in this study. The only criterion for participation was that therapists commonly evaluate and treat patients with LBP. Twelve clinicians (19%) responded and were randomly assigned to 1 of 2 groups (standard care or stratified care) using a random number generator (http://www.randomization.com) with blocked design to ensure an equal number of clinicians in each group. This randomization process resulted in 6 clinicians in each group (eTable).
Standard care group clinicians attended 3 formal group meetings over the course of 4 weeks, each lasting approximately 60 minutes, where they were provided with a description of the study and received protocol training (eg, informed consent process, study packet review, screening and outcome measure administration). Stratified care group clinicians received similar study descriptions and protocol training; however, they received an additional 8 hours of stratified care education.
Clinicians randomized to receive stratified care education attended 3 sessions over the course of 4 weeks. Each session was provided by the study authors and lasted 2 to 4 hours in duration. Educational content (Appendix) focused on psychologically informed physical therapist practice principles3 and was developed to reflect protocols that have been used in previous studies,17–19 including both theoretical and clinical application components. The overall objective of this multifactorial education approach was for clinicians to embrace the biopsychosocial model of pain and disability.20
As a component of stratified care education, physical therapists were trained on implementation of the STarT Back approach for LBP management.2,21 The STarT Back approach provides an example of stratification based on prognostic risk for persistent LBP-related disability that consists of 2 corresponding components. First, patients are categorized into 1 of 3 subgroups (low, medium, or high risk) for persistent LBP-related disability using the 9-item STarT Back Screening Tool (SBT), which consists of statements about physical or psychological prognostic factors that are considered modifiable with treatment.22 Second, suggested targeted treatment pathways are matched to each SBT subgroup such that patients categorized as being at low risk receive minimal care, primarily consisting of reassurance and education. Patients categorized as being at medium risk also receive reassurance and education, but their treatment is supplemented with physical therapy focused on restoring function and targeting physical symptoms. For patients categorized as being at high risk, physical therapy is focused on restoring function using a combination of physical and psychological approaches. In addition, physical therapists in the stratified care group were educated on utilization of the American Physical Therapy Association (APTA) Orthopaedic Section LBP clinical practice guidelines (CPGs),23 specifically for targeting physical symptoms and impairments.
Validated questionnaires for attitudes and beliefs about LBP were completed by all participating clinicians and administered before training, upon completion of training, and 6 months later. Physical therapists' attitudes and beliefs were measured using the Pain Attitudes and Beliefs Scale for Physical Therapists (PABS-PT)24 and the Health Care Providers' Pain and Impairment Relationship Scale (HC-PAIRS).25
The PABS-PT consists of 19 items about treatment orientation that are rated using a 6-point Likert scale ranging from “totally disagree” to “totally agree.”24 The PABS-PT biomedical scale (10 items) has a potential score range from 10 to 60, and the PABS-PT biopsychosocial scale (9 items) has a potential score range from 9 to 54, with higher scores indicating increased treatment orientation for the respective scale. The PABS-PT has been reported to have fair-to-excellent levels of internal consistency (Cronbach α [biomedical scale: .77–.84, psychosocial scale: .58–.68]) and is responsive to educational interventions.26
The original HC-PAIRS consisted of 15 items and was developed to assess health care providers' attitudes and beliefs about the relationship between LBP and physical function.27 We used a revised 13-item version based on previous recommendations25 and removed the term “chronic” from statements describing LBP.28,29 Responses to items are rated using a 6-point Likert scale ranging from “completely disagree” to “completely agree.” This version of the HC-PAIRS has a potential score range from 13 to 78, with higher scores indicating beliefs in a strong relationship between pain and impairment and attitudes that LBP justifies disability and limitation of activities. The HC-PAIRS has been reported to have adequate levels of internal consistency (Cronbach α=.78–.84); however, its responsiveness to educational interventions has not been adequately evaluated.30
Phase 2.
The same group of physical therapists (n=12) who completed phase 1 of this study provided treatment for patients with LBP that was consistent with the type of training and education received.
The participants in this phase of the study were consecutive patients across 7 outpatient clinic locations who were referred by a physician for physical therapy for LBP and evaluated by a physical therapist and who had completed phase 1 of the study. The outpatient clinics were part of a large rehabilitation system located in the southeastern region of the United States where physical therapy treatment is provided for a wide spectrum of musculoskeletal conditions. Physical therapists recruited and screened all patients with LBP for study eligibility prior to enrollment. Potential study participants were consecutively recruited to determine if the following criteria were met before being enrolled in the study: (1) between the ages of 18 and 65 years and seeking physical therapy for LBP (defined as having symptoms at T12 or lower, including radiating pain into the buttocks and lower extremity) and (2) ability to read and speak the English language. Potential study participants were ineligible to participate in this study for any of the following: (1) the presence of systemic involvement related to metastatic or visceral disease, (2) recent spinal fracture, (3) osteoporosis, or (4) pregnancy. All eligible patients were briefed on the study, and informed consent was obtained in compliance with the University of Florida's Institutional Review Board.
Study participants were asked to complete a standardized self-report questionnaire consisting of questions related to age, sex, race, ethnicity, education, household income, marital status, and employment status. Information involving LBP clinical characteristics (ie, prior surgery, symptom duration, symptom onset, symptom location, work-related LBP) also was obtained.
The STarT Back Tool (SBT) consists of 9 items related to physical and psychosocial factors used to categorize patients with LBP in primary care settings based on their risk for poor future disability outcomes.22 The SBT overall scores (ranging from 0 to 9) are determined by summing all positive responses, and SBT psychosocial subscale scores (ranging from 0 to 5) are determined by summing items related to bothersomeness, fear, catastrophizing, anxiety, and depression. Based on overall and psychosocial subscale scoring, patients are categorized as “high risk” (psychosocial subscale scores ≥4), in which high levels of psychosocial prognostic factors are present, with or without physical factors present; “medium risk” (overall score >3; psychosocial subscale score <4), in which physical and psychosocial factors are present but not a high level of psychosocial factors; or “low risk” (overall score: 0–3), in which few prognostic factors are present. The SBT has demonstrated adequate-to-high levels of test-retest reliability,22 concurrent validity,31 and predictive validity.32,33 For the current study, the SBT was self-administered by all patients at intake and 4 weeks later; however, only physical therapists in the stratified care group were educated on SBT scoring methods. Prior to beginning this study, the SBT was not used in any of the clinics to assist with initial treatment decision making.
Patients were not randomized to different treatment groups in this study. Physical therapists in the standard care group were instructed to provide treatment for patients with LBP as they normally would have administered if not participating in this study. Physical therapists in the stratified care group received an additional 8 hours of training and education consisting of content related to psychologically informed physical therapist practice principles3,17 and APTA Orthopaedic Section LBP CPGs.23 As a component of these training experiences, clinicians were directed to incorporate the knowledge and skills learned into subsequent management strategies for their patients with LBP. Specifically, we asked clinicians to utilize SBT categorization to guide initial treatment decision making. For patients categorized as SBT low risk, we suggested a minimal physical therapist intervention approach (1–2 sessions per week) and adherence to the APTA Orthopaedic Section CPGs. For patients categorized as SBT medium risk, we suggested an increased physical therapist intervention approach (2–3 sessions per week) with adherence to the APTA Orthopaedic Section CPGs. For patients categorized as SBT high risk, we suggested an increased physical therapist intervention approach (2–3 sessions per week) with adherence to the APTA Orthopaedic Section CPGs and psychologically informed practice principles previously learned as a component of prestudy training (Appendix).
Self-report outcome measures were collected at intake and 4 weeks later and are described in more detail below.
Pain intensity was rated using a numerical pain rating scale (NPRS), ranging from 0 (“no pain”) to 10 (“worst pain imaginable”). Participants were asked to rate their current pain intensity as well as their best and worst levels of pain intensity over the previous 24 hours. These 3 pain ratings were averaged and used as the NPRS variable in this study.34 The NPRS has been found to have sound psychometric properties,34–36 with a minimal clinically important difference reported to be 2 points.37 A ≥30% improvement in NPRS scores from baseline has been recommended as a useful threshold for identifying clinically meaningful improvement for patients with LBP.38 Therefore, reporting the proportion of patients achieving this magnitude has been suggested.39
Low back pain–related disability was assessed with a modified version of the ODI, which has 10 items that assess how LBP affects common daily activities.40,41 The ODI has a range of 0% (“no disability due to LBP”) to 100% (“completely disabled due to LBP”), with higher scores indicating higher disability from LBP. The ODI has been found to have sound psychometric properties,40,42,43 with a reported minimal clinically important difference of 10 percentage points.38 A ≥30% improvement in ODI scores from baseline has been recommended as a useful threshold for identifying clinically meaningful improvement for patients with LBP.38 Therefore, reporting the proportion of patients achieving this magnitude has been suggested.39
We analyzed and reported NPRS and ODI data on a continuous scale (primary outcome) and on a binary scale (secondary outcome) to generate treatment effect estimates for future study planning.
Data Analysis
All statistical analyses were performed using IBM SPSS version 21.0 (IBM Corp, Armonk, New York). Descriptive statistics were used to provide a summary of data. Point estimates and 95% confidence intervals (95% CIs) were reported for selected analyses based on previous recommendations for evaluating pilot study findings.44 We also incorporated hypothesis testing and reported associated P values and effect sizes for selected analyses to provide preliminary estimates for treatment effectiveness.45
Phase 1.
Differences between clinician groups for demographic characteristics, PABS-PT (biomedical and biopsychosocial scale), and HC-PAIRS scores before training were tested using independent-samples t tests (for continuous data) and chi-square tests (for categorical data). We used separate 2 × 3 repeated-measures analysis of variance (ANOVA) models to test for relationships between training and attitude and belief scores, with clinician group (standard or stratified education) as the between-subjects variable and time (before training, upon completion of training, and 6 months later) as the within-subject variable. For this study phase, we were particularly interested in detection of a 2-way interaction between time and clinician group to determine if clinicians who received stratified care education were associated with changes in attitudes and beliefs favoring a biopsychosocial treatment approach. We also were interested in determining if within-group changes were observed immediately following education intervention and if those changes were maintained 6 months later. Three separate repeated-measures ANOVA models were used, with either PABS-PT (biomedical or biopsychosocial) or HC-PAIRS scores serving as the dependent variables.
Phase 2.
Differences between patient groups for demographic and clinical characteristics were tested using independent-samples t tests (for continuous data) and chi-square tests (for categorical data). We used separate 2 × 2 repeated-measures ANOVA models to test for relationships between treatment approaches and clinical outcome scores, with clinician group (standard or stratified care) as the between-subjects variable and time (intake and 4 weeks) as the within-subject variable. For this study phase, we were particularly interested in detection of a 2-way interaction between time and clinician group to determine if patients who received treatment from clinicians who received stratified care education were associated with superior outcomes. Two separate repeated-measures ANOVA models were used, with either NPRS or ODI scores serving as the dependent variables. We calculated effect sizes using the Cohen d coefficient to provide an estimate of magnitude between groups using the following formula: [(stratified care group change score − standard care group change score)/pooled standard deviation], with an effect size of 0.2 considered small, 0.5 moderate, and 0.8 large.46
A proportional responder analysis was performed to provide an indication of the percentage of patients achieving clinically meaningful improvement. A ≥30% improvement in NPRS and ODI scores from baseline has been recommended as a useful threshold for identifying clinically meaningful improvement for patients with LBP.38 Therefore, reporting the proportion of patients achieving this magnitude has been suggested.39 We also provided a full range of cutoff scores to provide data on the proportion of participants achieving above and below 30% improvement. Percent change scores were calculated for both the NPRS and ODI using the following formula: [(initial score − 4-week score)/initial score × 100]. Each patient was then coded as having “clinically meaningful improvement” (percent change ≥30%) or “nonclinically meaningful improvement” (percent change <30%) for NPRS and ODI outcomes at 4 weeks. Relative risk (RR) estimates were calculated based on previous recommendations47 to compare clinically and nonclinically meaningful improvement status between stratified and standard care. A cumulative proportion responder analysis48,49 figure was generated to describe the proportion of participants who experienced 4-week changes at each NPRS and ODI threshold level or higher.
For planned exploratory analyses, we compared stratified care and standard care groups by SBT risk categorization for visit frequency and 4-week changes in NPRS and ODI scores using independent-samples t testing. Our rationale for these analyses was to provide preliminary findings related to: (1) SBT low-risk participant visit frequency based on previous suggestions that overtreatment for patients categorized as SBT low risk should be avoided1,2,21 and (2) SBT high-risk participant outcomes because our stratified education content placed a strong emphasis on psychologically informed practice.
Results
Phase 1
Physical therapists.
A total of 12 physical therapists participated in this study. There were no differences between clinician groups for most demographic characteristics and PABS-PT and HC-PAIRS scores before training (eTable). The only exception was that standard care group clinicians were associated with a higher number of years in practice (X̅=10.1, SD=5.3) compared with stratified care group clinicians (X̅=4.4, SD=2.9), which may have been influenced by 2 clinicians who had more than 15 years of practice.
Physical therapists' attitudes and beliefs.
Results from the repeated-measures ANOVA indicated statistically significant group × time interactions for PABS-PT biomedical scale scores (F2,20=4.91, P=.018) and PABS-PT biopsychosocial scale scores (F2,20=4.95, P=.018). Compared with pretraining PABS-PT biomedical and biopsychosocial scores, minimal changes were observed during posttraining assessment (biomedical: mean change=+2.0, SD=3.0, Cohen d=0.28; biopsychosocial: mean change=+0.7, SD=3.3, Cohen d=0.19), with no changes observed 6 months later (biomedical: mean change=0.0, SD=4.5, Cohen d=0.0; biopsychosocial: mean change=+0.1, SD=2.7, Cohen d=0.02) for the clinician group that did not receive stratified care training (Tab. 1). For the stratified care–trained clinicians, decreased PABS-PT biomedical scale scores (mean change=−4.5, SD=2.5, Cohen d=1.08) and increased biopsychosocial scale scores (mean change=+5.5, SD=2.0, Cohen d=2.86) were observed during posttraining assessment, with these findings also maintained 6 months later (Tab. 1).
PABS-PT and HC-PAIRS Scoresa
Results from the repeated-measures ANOVA indicated no statistically significant group × time interactions for HC-PAIRS scores (F2,20=2.62, P=.097) or main effects (F2,20=2.06, P=.154). Compared with pretraining HC-PAIRS scores, minimal changes were observed during posttraining assessment (mean change=+0.4, SD=4.4, Cohen d=0.05) for the clinician group that did not receive stratified care training, with these findings also maintained 6 months later (Tab. 1). For the stratified care clinicians, decreased HC-PAIRS scores (mean change=−5.3, SD=6.4, Cohen d=0.74) were observed during posttraining assessment, with these findings also maintained 6 months later (Tab. 1).
Phase 2
During the study period, a total of 254 consecutive patients were screened for eligibility to participate in this study (Fig. 1). Of these patients, 145 were excluded from study participation, with the most common reason being that they were older than 65 years of age (n=91). Two patients were consulting with a lawyer due to work-related LBP and refused to participate. The remaining 109 patients provided informed consent and were enrolled in the study. Of these patients, 9 were not able to complete the study and provide 4-week follow-up data due to personal reasons. Therefore, intake data were obtained from 109 patients, and 4-week follow-up data were obtained from 100 patients.
Baseline demographic and clinical data for the entire study sample are presented in Table 2. Baseline characteristics were similar across treatment groups for all variables, with the exception of the frequency of participants reporting work-related LBP (standard care: n=6, 15.3%; stratified care: n=3, 14.3%), with 6 fully employed, 2 part-time employed, and 1 unemployed. An approximately normal distribution for initial pain intensity (NPRS scores) and LBP-related disability (ODI scores) was suggested based on visual inspection of histograms and normality plots.
Descriptive Characteristics of Study Samplea
Overall, the 4-week follow-up rate was 91.7%, with differences observed between groups (84.6% standard care, 95.7% stratified care) (P=.044). For participants who provided follow-up data, the average number of physical therapy visits at 4 weeks was similar among groups (X̅=6.3, SD=2.3, minimum=1, maximum=12) (P=.174). Four-week physical therapy status also was similar at 4 weeks, with participants still currently receiving physical therapy (73.0%), having completed and were discharged from physical therapy (11.0%), or electing to not continue with physical therapy (16.0%) (P=.559).
Results from the repeated-measures ANOVA indicated statistically significant group × time interactions for 4-week NPRS scores (F1,98=4.08, P=.046, Cohen d=0.40) and ODI scores (F1,98=13.6, P<.001, Cohen d=0.76) (Figs. 2A and 2B). From a group perspective, participants who received stratified care were associated with greater improvements in NPRS scores (between-group difference=0.8 points; 95% CI=0.1, 1.5) and ODI scores (between-group difference=9.0 percentage points; 95% CI=4.1, 13.6) compared with those who received standard care.
Changes in (A) pain intensity scores and (B) ODI scores. ODI=Oswestry Disability Index, NPRS=numerical pain rating scale. Error bars represent 1 standard error of the mean.
Achievement of ≥30% improvement rates for NPRS and ODI scores following 4 weeks of physical therapy is reported in Table 3. Thirty-nine percent of the participants achieved ≥30% improvement in NPRS scores, with a greater proportion observed for stratified care (47.8%) compared with standard care (21.2%) (RR=2.25; 95% CI=1.11, 4.55). Fifty-two percent of the participants achieved ≥30% improvement in ODI scores, with a greater proportion observed for stratified care (61.2%) compared with standard care (33.3%) (RR=1.84; 95% CI=1.10, 3.08). Interpretation of the cumulative distribution of responders based on NPRS scores (Fig. 3A) and ODI scores (Fig. 3B) indicates a similar trend, with a greater proportion of participants who received stratified care compared with those who received standard care achieving improvement for a range of threshold change scores.
Changes in NPRS and ODI Scores Following 4 Weeks of Physical Therapya
(A) Cumulative proportion of responders for 4-week pain intensity outcomes highlighting 47.8% of stratified care group (21.2% of standard care group) achieved at least 30% improvement in numerical pain rating scale (NPRS) (potential range: 0–10). Nonpositive responders for NPRS changes (ie, less than 0%) are not indicated in the figure. (B) Cumulative proportion of responders for 4-week pain disability outcomes highlighting 61.2% of stratified care group (33.3% of standard care group) achieved at least 30% improvement in Oswestry Disability Index (ODI) scores (potential range: 0%–100%). Nonpositive responders for ODI changes (ie, less than 0%) are not indicated in the figure.
For exploratory analyses, we used independent-samples t tests to compare frequency of visits and changes in NPRS and ODI scores following 4 weeks of physical therapy between treatment groups by SBT risk categorization (Tab. 3). For participants categorized as SBT low risk, there were no differences between groups in the mean number of treatment sessions (stratified care group: X̅=6.5, SD=1.7; standard care group: X̅=6.6, SD=2.2). There were no between-group differences for changes in NPRS scores (0.1 points; 95% CI=−1.3, 1.2) or ODI scores (4.2 percentage points; 95% CI=−3.0, 11.4); however, changes in ODI scores alone (10.4 percentage points; 95% CI=5.0, 15.9) favored stratified care at 4 weeks. For participants categorized as SBT medium risk, there were no differences between groups in the number of treatment sessions (stratified care group: X̅=6.3, SD=2.5; standard care group: X̅=5.9, SD=2.3). Between-group differences for changes in NPRS scores (1.4 points; 95% CI=0.2, 2.6) and ODI scores (11.5 percentage points; 95% CI=5.1, 17.8) favored stratified care at 4 weeks. For participants categorized as SBT high risk, those who received stratified care had more visits (X̅=6.9, SD=2.1) compared with those who received standard care (X̅=4.6, SD=2.2). There were no between-group differences for changes in NPRS scores (0.6 points; 95% CI=−1.1, 2.2) or ODI scores (9.8 percentage points; 95% CI=−3.8, 23.4); however, changes in ODI scores alone (16.7 percentage points; 95% CI=10.2, 23.2) favored stratified care at 4 weeks.
Discussion
This study investigated the feasibility of a risk stratification approach for LBP management that was initially implemented outside of primary care, and the results were favorable for the translation to physical therapy settings. In addition, we generated estimates of treatment effects based on 4-week clinical outcomes that may be helpful for planning of future studies. Specifically, our findings demonstrated that clinician biomedical treatment orientations can be modified and that patients who received stratified physical therapy care were associated with greater improvement compared with those who received standard physical therapy care. Furthermore, there was a greater cumulative proportion of positive pain intensity and disability responders among patients who received stratified care compared with those who received standard care. Collectively, these findings suggest that physical therapists' attitudes and beliefs can be changed and maintained toward a biopsychosocially oriented treatment approach and that those changes may positively influence patient clinical outcomes when combined with a stratified care approach. To our knowledge, this is the first study conducted in the United States that has shown favorable clinical outcomes using risk-stratified care in combination with guideline recommendations in settings outside of primary care.
Feasibility Issues–Phase 1
Several feasibility issues may aid in planning future studies. First, alternative clinician recruitment methods may be needed to generate a more diverse pool of physical therapists. For example, we used email to recruit potentially interested physical therapists, and a majority of the physical therapists who participated were residency or fellowship trained. In the future, we plan to provide brief in-service presentations at several clinical sites to describe potential research opportunities with the intent to improve clinician and researcher engagement and recruit a more heterogeneous group of physical therapists. Second, using the PABS-PT and HC-PAIRS as measures for educational intervention responsiveness appears to be appropriate based on the direction of changes observed; however, interpreting the magnitudes of change still warrants further investigation. For future study planning, we will consider analyses that evaluate relationships between magnitude of changes in PABS-PT and HC-PAIRS scores and clinical outcome scores. Finally, in retrospect, our educational intervention for clinicians may be improved by including more in-depth content related to cognitive-behavioral principles that is provided by a clinical psychologist and including follow-up mentoring sessions.9,50
Physical Therapists' Attitudes and Beliefs
We expected that adaptions toward less biomedical and more biopsychosocial treatment orientation would be observed for clinicians who received 8 hours of training and education focusing on psychologically informed physical therapist practice principles. Pretraining PABS-PT biomedical (X̅=28.1, SD=4.9) and biopsychosocial (X̅=36.2, SD=3.2) scores observed in this study were similar to those of previous studies11,15; however, HC-PAIRS scores (X̅=30.5, SD=6.2) were lower compared with findings of previous studies that involved physical therapists.11,29 Although the responsiveness of the PABS-PT and HC-PAIRS has not been extensively evaluated,26,30 within-group changes in PABS-PT biomedical (−4.5 points) and biopsychosocial (+5.5 points) scores and HC-PAIRS scores (−5.3 points) observed in this study for stratified care clinicians are consistent with the findings of previous studies that incorporated similar biopsychosocially oriented training programs.11,15 For example, a previous study11 investigating the effect of an 8-day biopsychosocially oriented course netted similar improvements and maintenance of beliefs that were measured using the PABS-PT and HC-PAIRS, although the findings varied slightly in magnitude from those of the current study. Our findings indicate that training periods for belief changing can be shortened from what was previously reported in the literature. The clinical meaningfulness of changes in PABS-PT and HC-PAIRS scores remains unclear and warrants further investigation.11,26,30 The findings from phase 2 of this study, however, provide a preliminary indication that these changes in beliefs and attitudes were potentially sufficient to affect clinical outcomes for patients who received stratified care education.
Feasibility Issues–Phase 2
The findings of phase 2 of this study also will inform planning of future studies. First, to allow for appropriately powered subgroup analyses, the number of patients categorized as SBT high risk needs to be considered. For example, 29% of the patients in this study were initially categorized as SBT high risk, and previous studies32,33 conducted in similar settings have indicated between 19% and 26%. For future study analyses, we will need to consider determining sample size calculations based on multiple hypothesis tests comparing stratified care and standard care by SBT categorization, similar to previous study protocols.21,51 We observed differences in clinical outcomes between stratified and standard care that were driven by SBT medium-risk categorization in our cohort (Tab. 3), which is not consistent with previous study findings of higher benefit for patients at high risk.2,52 One potential strategy to consider for future studies is to power differences in clinical outcome scores between medium and low risk or high and low risk, as the ability to detect differences between medium and high risk may be more difficult.
Second, the influence of treatment visits on clinical outcomes will need to be accounted for in future analyses, particularly if we are suggesting a minimal treatment approach for patients categorized as SBT low risk.
Finally, implementation of our stratified care approach could be improved on several levels. For example, we recommended a minimal physical therapist intervention approach consisting of 1 or 2 treatment visits per week for patients at low risk who are receiving stratified care; however, similar rates were observed for stratified and standard care groups at 4 weeks. During future study planning, we will better structure clinician adherence for treatment of patients at low risk. We also will incorporate consistent mentoring sessions between clinicians and researchers during training and active study time periods that will provide opportunities to address potential barriers to implementation.
Estimates of Treatment Effects for Clinical Outcomes
Greater improvements in LBP clinical outcomes at 4 weeks were observed in patients who received physical therapy following a stratified care management approach. Further inspection of pain and disability outcomes indicated between-group differences favoring stratified care, with associated small (Cohen d=0.40) and moderate (Cohen d=0.76) effect sizes, respectively. These results are similar to 4-month findings of the larger STarT Back trial2 that compared stratified care (n=568) with current best practice (n=283) in UK primary care settings. In that study, Hill et al2 reported between-group differences for NPRS and Roland-Morris Disability Questionnaire scores favoring stratified care, with smaller respective (Cohen d=0.24 and 0.29) effect sizes. Specifically related to patients categorized as SBT high risk, we observed a large effect size (0.61) for improvements in ODI scores favoring stratified care; however, we acknowledge that this study was not powered for subgroup analyses. Patients at SBT high risk who received stratified care had more visits (6.9) compared with those who received standard care (4.6), suggesting the beneficial effects may have been due to extended treatment and increased attention as designed by psychologically informed practice. Collectively, these findings provide continuing evidence that stratified care with SBT categorization has the potential to improve LBP outcomes when combined with psychologically informed practice.
We also incorporated a proportional responder analysis to determine if different statistical approaches would provide variations in the appraisal of our clinical outcomes data. The analysis of treatment response using previously suggested thresholds (ie, ≥30% improvement) for clinically important changes in NPRS and ODI scores provided similar results. A greater percentage of patients in the stratified care group achieved the clinically relevant change threshold for NPRS (47.8%) and ODI (61.2%) scores compared with the standard care group (21.2% and 33.3%, respectively). These differences were not trivial, as patients who received stratified care were approximately 2 times more likely to achieve ≥30% improvement on the pain intensity and disability measures. Furthermore, compared with the standard care group, there was a consistent trend for a higher cumulative percentage of patients in the stratified care group who achieved a wide range of NPRS (Fig. 3A) and ODI (Fig. 3B) thresholds for improvement across the continuum, potentially providing further validation for the outcomes reported in this study. The greatest percentage of positive responders observed in the current study (71.0%) and the STarT Back trial (71.0%) were patients at SBT medium risk who received stratified care. Several factors need to be considered prior to interpreting these findings related to treatment response. For example, treatment response in this study was based on 4-week clinical outcomes; however, 73.0% of the patients were still receiving physical therapy care at that time. Therefore, we cannot speculate on long-term implications.
Strengths and Limitations
One strength of this study is that there was a planned combination of internal (phase 1) and external (phase 2) validity. Clinicians were randomized to different training and education groups during phase 1, and we intentionally incorporated a pragmatic approach to be consistent with typical clinical management for LBP during phase 2. Neither stratified care nor standard care clinicians were required to follow rigid treatment protocols for patients with LBP. Rather, clinicians utilized the knowledge and strategies learned during training sessions (stratified care group) or their normal management strategies (standard care group) for patients with LBP. Another strength of the study is that, compared with previous studies,2 the stratified care education content provided was brief, but similar outcomes were obtained. Previous studies53,54 have indicated that one of the most significant barriers to attending postprofessional education courses is time. Therefore, it was our intent to provide the necessary education for this study in reasonable intervals that also accommodated clinicians' schedules (ie, 8 hours over multiple sessions) in order to improve retention rates and be consistent with our pragmatic approach.
This study also had several limitations that have not already been mentioned. First, most physical therapists participating in this study were either orthopedic residency trained (n=11, 91.7%) or manual therapy fellowship trained (n=8, 66.7%), or both (n=8, 66.7%), which may limit the generalizability of these findings. In addition, only 12 of 63 eligible physical therapists (19%) volunteered to participate in this study; therefore, the influence of volunteer bias should be considered, which also may limit the generalizability of these findings. However, the overall influence of this bias was likely minimized by randomly assigning therapists to training groups (ie, stratified or standard care). Second, patients older than 65 years were not eligible to participate in this study, which may limit the generalizability of our findings, especially considering that 35.8% of the patients were not eligible based on the age criterion. In addition, 2 patients were consulting with a lawyer due to work-related LBP and refused to participate. Future studies are needed to determine if the risk-stratified care approach used in this study is generalizable across all age groups and for individuals receiving work-related disability. Third, in phase 2, patients were not randomized to different treatment groups in this observational study, nor did we assess for clinician selection bias of patients; therefore, our outcome findings should be interpreted with caution. Planning of future larger-scale studies (eg, comparative effectiveness research) may not necessarily require randomization of patients to treatment groups, particularly if the intent is for pragmatic implementation in routine clinical practice.55 Finally, 4-week outcomes provide a short-term perspective on the impact of patient management. Ideally, longer follow-up data would have been available.
Clinical Implications
Overall, our results indicated that implementation of stratified care based on the STarT Back approach and APTA Orthopaedic Section CPGs is feasible in outpatient physical therapy settings. However, estimates of specific treatment effects for the clinical outcomes reported in this article should be interpreted with caution based on this study design. Findings from this preliminary study can be used for planning of future studies that more definitively determine the impact of stratified care on patient outcomes.
Appendix.
Stratified Care Education Content
Footnotes
Both authors provided concept/idea/research design and writing. Dr Beneciuk provided data collection and analysis, project management, and fund procurement. Dr George provided consultation (including review of manuscript before submission). The authors acknowledge the following clinicians from Brooks Rehabilitation in Jacksonville, Florida: Trent Harrison, Renata Salvatori, Monika Beneciuk, Raine Osborne, Brian Hagist, Tasha Mouton, Timothy Shreve, Jenny Hagist, Sara Cristello, Sara Bertrand, Nate Moore, and Steve Pompilio. The authors also acknowledge the following administrative personnel from Brooks Rehabilitation: Douglas Baer, Michael Spigel, Amanda Osborne, Victor Derienzo, Robert Rowe, Holly Morris, Flo Singletary, and Mallorey Smith.
This study was funded by the 2012 Brooks Rehabilitation Collaborative Grant. The manuscript was written while Dr Beneciuk received support from the National Institutes of Health Rehabilitation Research Career Development Program (K12-HD055929).
- Received September 25, 2014.
- Accepted April 2, 2015.
- © 2015 American Physical Therapy Association
References
Issue highlights
