Abstract
Background The STarT Back Screening Tool is a validated multidimensional screening measure and risk stratification tool for people with low back pain.
Objective The study objective was to compare relationships between a modified STarT Back Screening Tool (mSBT) and clinical and psychological measures in people with low back, neck, shoulder, and knee pain. The hypothesis was that the relationships between mSBT scores and clinical and psychological measure scores would be similar across the included musculoskeletal pain conditions.
Design A cross-sectional, secondary analysis was done in this study.
Methods Participants with low back (n=118), neck (n=92), shoulder (n=106), or knee (n=111) pain were recruited, and an mSBT was developed for use across the pain conditions. Separate hierarchical linear regression models were developed, with clinical (health status, pain intensity, and disability) and psychological (kinesiophobia, catastrophizing, fear avoidance, anxiety, depressive symptoms, and self-efficacy) measures as dependent variables. Demographic and pain region variables were entered in the first step, mSBT scores were entered in the second step, and pain region × mSBT interactions were entered in the last step.
Results In the final models, no interactions were identified, suggesting that dependent measure scores did not differ by pain region. The strongest contributor for all dependent variables was mSBT scores (β=|0.32|–|0.68|); higher mSBT scores were associated with poorer health status and self-efficacy and with higher levels of pain intensity, disability, kinesiophobia, catastrophizing, fear avoidance, anxiety, and depressive symptoms.
Limitations Generalizability was restricted to physical therapy outpatients with the included pain conditions. The mSBT used in this study is not ready for clinical implementation.
Conclusions The results of this study support the feasibility of using a single measure for concise risk assessment across different musculoskeletal pain conditions. Further longitudinal studies are needed to better direct the clinical use of an mSBT in people with low back, neck, shoulder, and knee pain.
Screening for factors related to persistent disability and other poor clinical outcomes (eg, elevated pain intensity) provides clinicians with insight into barriers to recovery. Identification of these factors can assist in treatment plan development and support decisions to refer patients to other health care professionals (eg, mental health providers). The STarT Back Screening Tool (SBT) is a validated multidimensional screening measure intended for use as a risk stratification tool for persistent disability in people with low back pain.1,2 The SBT captures perception of physical and psychological functions and, in doing so, provides prognostic information for poor clinical outcomes in people with low back pain.1,2 Specifically, elevated SBT scores predict higher disability and pain intensity ratings in people with low back pain.3–5 The SBT has been successfully used to stratify people with low back pain into categories of low, medium, or high risk for persistent disability1,2; targeted treatment strategies corresponding to each risk category also have been developed.6,7 Recent studies have indicated that this risk stratification strategy is more clinically effective and less costly than usual care for managing low back pain in primary care settings.8,9
Additional studies might provide an initial preliminary evaluation of the clinical utility of the SBT across other commonly occurring musculoskeletal pain conditions, such as neck, shoulder, or knee pain. Testing of the SBT for risk stratification capabilities across other musculoskeletal pain conditions in UK primary care settings is in progress in studies by researchers at Keele University, Keele, United Kingdom. A proof-of-principle study in an outpatient physical therapy setting would provide a preliminary evaluation of a modified SBT (mSBT) in a different practice setting. Modification of the SBT involves changing the wording of the measure to allow for use across other musculoskeletal pain regions (eg, neck, shoulder, and knee).
The purpose of this study was to compare relationships between mSBT scores and clinical and psychological measure scores at intake in people seeking outpatient physical therapy for low back, neck, shoulder, or knee pain. An mSBT was developed to allow for general use across conditions.
Our primary aim was to examine relationships between mSBT scores and clinical measure scores for health status, pain intensity, and disability in people with low back, neck, shoulder, or knee pain. Earlier studies showed a predictive association between pain-related psychological distress and poor clinical outcomes (eg, disability, longer pain duration, and higher pain intensity ratings) across multiple anatomical regions of musculoskeletal pain—low back, neck, lower extremity, and upper extremity.10–12 Specifically, elevated levels of depressive symptoms10,12 and fear-avoidance beliefs11 were found to be positively related to greater disability in these anatomical regions. Therefore, our primary hypothesis was that the relationship between mSBT scores and clinical measure scores for people with low back pain would be similar to that for people with neck, shoulder, or knee pain. Specifically, we expected higher scores on the mSBT to be associated with poorer health status, higher levels of pain intensity, and higher levels of disability.
Our secondary aim was to examine relationships between mSBT scores and established unidimensional, pain-associated psychological measure scores. Our secondary hypothesis was that the relationship between mSBT scores and psychological measure scores would be similar across all included musculoskeletal conditions. Consistent with our primary hypothesis, we expected higher scores on the mSBT to be associated with higher levels of kinesiophobia, pain catastrophizing, fear-avoidance beliefs, anxiety, and depressive symptoms and reduced self-efficacy.
Method
Overview
This article reports a secondary analysis of data from the Optimal Screening for Prediction of Referral and Outcome (OSPRO) cross-sectional development cohort (recruitment period: March 2013 to May 2014). The OSPRO cohort studies consist of specific projects focused on enhancing clinical assessment by physical therapists. The University of Florida Gainesville Health Science Center Institutional Review Board provided approval of the informed consent document used for the OSPRO development cohort, and written informed consent was obtained from each participant.
Clinical Sites
Participants were recruited from Orthopaedic Physical Therapy Investigative Network (OPT-IN) clinical sites. The OPT-IN was formed for the purpose of performing multicenter clinical projects aimed at examining diagnosis/classification, prognosis, or patient-centered treatment outcomes in people with musculoskeletal conditions commonly managed by orthopedic physical therapists. All OPT-IN clinical sites participating in the cross-sectional development phase were located in Florida and included 3 outpatient clinics in the University of Florida Health System (Gainesville, Florida) and 8 in the Brooks Health System (Jacksonville, Florida). Sites within these health systems were selected on the basis of different socioeconomic strata and representations of urban and rural communities.
Participants
Participants were recruited from participating OPT-IN clinical sites while seeking outpatient physical therapy treatment; participating physical therapists at all clinical sites used the same eligibility criteria, and participants were assessed within 1 week of their initial evaluations by the physical therapists. A total of 431 participants recruited for the primary OSPRO study were included in this secondary analysis. In the primary OSPRO study, eligibility criteria for people with low back, neck, shoulder, or knee pain were intentionally broad to enhance the generalizability of the findings across the included pain conditions. Narrower eligibility criteria would have excluded a significant number of people commonly seen by orthopedic physical therapists for these conditions.
Inclusion criteria.
People who were 18 to 75 years of age were eligible to participate in this study if they were seeking outpatient physical therapy treatment for musculoskeletal pain; had primary musculoskeletal pain–related complaints involving the low back, neck, shoulder, or knee region; and were able to read and comprehend the English language.
Exclusion criteria.
People were excluded from study participation for any diagnosis indicative of widespread chronic pain syndrome (eg, fibromyalgia, irritable bowel syndrome), neuropathic pain syndrome (eg, complex regional pain syndrome, diabetic neuropathy), psychiatric history (currently in the care of a mental health care provider or taking multiple psychiatric medications), cancer (currently receiving treatment for active cancer), or a neurological disorder (eg, stroke, spinal cord injury, traumatic brain injury).
Measures
Demographic and historical information.
Data on age, sex, and pain duration (days of current episode) were collected from each participant through self-report. Historical data collected from each participant included the primary anatomical region of pain (low back, neck, shoulder, or knee), current pain being a work-related injury (yes or no), and having had surgery for the current pain episode (yes or no); these data also were obtained through self-report. The participant's anatomical region of pain was verified by a treating clinician.
Clinical measures.
The Medical Outcomes Study 8-item Short-Form Health Survey (SF-8) was used as a general health measure, and the SF-8 Physical Component Score (SF-8 PCS) and SF-8 Mental Component Score (SF-8 MCS) were reported separately for this analysis.13 Pain intensity was assessed with a numerical pain rating scale ranging from 0 (“no pain”) to 10 (“worst pain imaginable”).14–16 Participants rated their current pain intensity and their best and worst pain intensity over the preceding 24 hours. These 3 pain intensity ratings were then averaged and reported as “average pain intensity” for this analysis.
Disability was assessed with region-specific self-report measures. The Oswestry Disability Questionnaire (ODI),17,18 Neck Disability Index (NDI),19,20 reduced-item version of the Disabilities of the Arm, Shoulder and Hand Questionnaire (QuickDASH),21 and International Knee Documentation Committee Subjective Knee Form (IKDC)22 were used for low back, neck, shoulder, and knee pain, respectively, as the primary complaints in this study.
Psychological measures.
The full-length psychological measures included in this study were selected because of their clinical relevance and because they comprise key constructs relevant to the SBT. These measures included the Tampa Scale for Kinesiophobia (TSK-11)23 to assess kinesiophobia, the Pain Catastrophizing Scale24,25 to measure pain catastrophizing, the Modified Fear-Avoidance Beliefs Questionnaire Physical Activity subscale (mFABQ-PA)11 to assess fear-avoidance beliefs, the trait portion of the State-Trait Anxiety Inventory (STAI-T)26,27 to assess anxiety, and the Patient Health Questionnaire (PHQ-9)28,29 to measure depressive symptoms. The original FABQ-PA30 items were revised to make them appropriate across anatomical regions; such modifications were successful in previous studies with the FABQ and its subscales.11,31 Pain self-efficacy was assessed with the Pain Self-Efficacy Questionnaire (PSEQ).32,33
mSBT.
The original SBT was developed and validated for people with low back pain.1 An mSBT was created from items taken directly from the original derivation of the SBT for low back pain1 and items from validated psychological questionnaires; it closely resembled the original SBT but was applicable across different anatomical regions. Table 1 provides more detail on the original SBT items and matched mSBT items. As with the original SBT, our mSBT had an overall score ranging from 0 to 9 and a psychosocial subscore ranging from 0 to 5; it was scored in the same manner as the original SBT, with higher scores indicating a higher risk for disability and the presence of higher levels of psychological distress.
Description of Original STarT Back Screening Tool (SBT) Items and Matched Items on the Developed Modified STarT Back Screening Tool (mSBT)a
During the planning stages of the OSPRO study protocol, we planned a priori for the use of a modified version of the SBT across different anatomical regions. This was done because the development of concise measures was a primary aim of the OSPRO study. As part of this planning, the construction of the psychosocial portion of the mSBT would be based on other psychological measure items to reduce the burden on participants during data collection. Modifying existing questionnaires without thorough psychometric testing was not an ideal approach but was justified given the general nature of the constructs assessed and the fact that, in our experiences, individual item scores often correlate quite strongly with overall questionnaire scores for previously validated measures. Indeed, similar modifications for use in other anatomical regions have been successfully reported for the FABQ and its subscales in the cervical spine,34 knee,31,35 and shoulder.11 Furthermore, we included the low back pain group in the analysis as a reference to determine whether the performance of the mSBT across the included anatomical pain regions was statistically similar.
Sample Size Estimate
There are no definitive parameters for sample size estimates in psychometric studies, but a common recommendation is a minimum of 100 participants, with larger sample sizes being acknowledged as better.36–38 Therefore, the goal for the OSPRO development cohort was to recruit a minimum of 400 total participants or approximately 100 participants per anatomical region (low back, neck, shoulder, and knee).
Data Analysis
Data management.
All analyses were conducted with IBM SPSS Statistics for Windows, version 22.0 (IBM Corp, Armonk, New York). Data from 4 participants were not included in the analyses because of inconsistent reports of primary pain location; therefore, the total study sample included 427 participants. An alpha level of less than .01 was used for all analyses because of the number of comparisons and the sample size of the cohort. Region-specific measures—ODI, NDI, QuickDASH, and IKDC—were used in 2 ways: (1) they were transformed to z scores for use in regression models that also included anatomical regions (subsequently referred to as z-transformed disability scores), and (2) they were kept in original scale form for use in separate regression models for the 4 anatomical regions. Scores from the IKDC were reversed before data analyses to represent a measure of disability that was consistent with other measures (ie, higher scores indicating greater disability).
Regression model assumptions for the primary multivariate model were checked by visual inspection and statistical estimates when appropriate. The sample size was robust with regard to normality concerns. Visual inspection of relationships between each independent variable and dependent measure scores indicated linear relationships for the multivariate models. Variance inflation factor estimates were less than 2 for all independent variables across all models, indicating that multicollinearity was not a concern.
General health, pain, and disability measures.
For descriptive purposes, Pearson correlations with 95% confidence intervals were generated through a bootstrap analysis (5,000 samples) for mSBT and demographics, historical information, general health, average pain intensity, disability, and region-specific disability. To test our primary hypothesis, we designed a series of 4 separate multiple hierarchical regression models to determine the influence of anatomical pain region, mSBT overall score, and their statistical interaction on general health, average pain intensity, and z-transformed disability scores. In the first step, demographic (age and sex) and historical (pain duration, work-related pain, and surgery for pain) variables and anatomical region (low back, neck, shoulder, or knee) were included; in the second step, mSBT overall score (ranging from 0 to 9) was added; and anatomical region × mSBT overall score interaction terms were added in the third step. Demographic and historical variables chosen for these and subsequent multivariate models were based on a theoretical approach; these particular variables are often associated with the selected dependent measures.
For all models, the anatomical region was dummy coded with the low back as the reference category; low back pain was chosen for these and subsequent models because previous studies indicated a positive relationship between SBT overall score and measures of interest (ie, higher SBT scores were associated with greater disability, greater pain intensity, and greater pain-related psychological distress) in people with low back pain.3,5,39 Anatomical region and mSBT overall score variables were centered, and interaction terms for each anatomical region and mSBT overall score were created. The low back anatomical region × mSBT overall score interaction term was used as the reference category.
Region-specific disability.
In a planned further analysis, the influence of mSBT overall score on region-specific disability measures (ODI, NDI, QuickDASH, and IKDC) was examined with separate multiple hierarchical regression models for the 4 anatomical regions. This analysis was performed for each anatomical region and matched region-specific disability measure to provide further descriptive information and to aid in clinical interpretation of the direct influence of mSBT on the nontransformed individual measure scores. A modeling approach similar to that described for previous analyses was used for this analysis, but anatomical region and interaction terms were not included because of the region specificity of the disability measures and the separate analyses by anatomical region.
Psychological measures.
For descriptive purposes, Pearson correlations with 95% confidence intervals were generated through a bootstrap analysis (5,000 samples) for mSBT and psychological measures. As in the analyses for general health and disability measures, a series of 6 multiple hierarchical regression models were designed to determine the influence of anatomical pain region, mSBT psychosocial subscore, and their potential interaction on TSK-11, Pain Catastrophizing Scale, mFABQ-PA, STAI-T, PHQ-9, and PSEQ scores. The mSBT psychosocial subscale score was used for this analysis rather than the mSBT overall score to specifically explore the relationships between the psychosocial components of the mSBT and full measures for different psychological constructs; this method is consistent with those used in previous studies investigating these relationships in people with low back pain.5,40 In the first step, demographic and historical variables and anatomical region were included; in the second step, mSBT psychosocial subscore (ranging from 0 to 5) was added; and interaction terms were added in the third step.
Role of the Funding Source
This article was written while Dr Beneciuk received support from the National Institutes of Health Rehabilitation Research Career Development Program (K12-HD055929). This work also was funded by the Orthopaedic Section of the American Physical Therapy Association and Brooks–PHHP Research Collaboration, University of Florida.
Results
Descriptive Information
The total sample included 427 participants with primary low back (n=118; 27.6%) neck (n=92; 21.5%), shoulder (n=106; 24.8%), and knee (n=111; 26.0%) pain. Table 2 shows descriptive characteristics for the sample as a whole and by anatomical region. Correlations for mSBT scores and general health, pain, z-transformed disability, region-specific disability, and psychological measure scores across the anatomical regions are shown in Table 3. Graphical depictions of selected correlations with mSBT by anatomical region are shown in the Figure.
Descriptive Characteristics of the Sample as a Whole and by Anatomical Regiona
Associations Between Modified STarT Back Screening Tool (mSBT) Scores and General Health, Pain, Disability, and Psychological Measure Scoresa
Comparison, by anatomical region, of unadjusted association between Modified STarT Back Screening Tool (mSBT) overall score and (A) Medical Outcomes Study 8-item Short-Form Health Survey Physical Component Score (SF-8 PCS), (B) Medical Outcomes Study 8-item Short-Form Health Survey Mental Component Score (SF-8 MCS), (C) pain intensity rating, and (D) disability score. The pain intensity rating was the average of the current pain intensity rating, the worst pain intensity rating in the preceding 24 hours, and the best pain intensity rating in the preceding 24 hours (all rated on a numerical scale from 0 to 10). The disability score was the z-transformed disability score generated from region-specific disability measures (low back [Oswestry Disability Index], neck [Neck Disability Index], shoulder [Disabilities of the Arm, Shoulder and Hand Questionnaire], or knee [International Knee Documentation Committee Subjective Knee Form]). Lines represent best-fitting linear regression lines for each anatomical region.
General Health and Pain Measures
The results of the final regression models for general health and pain measures are shown in Table 4. Across all models, no statistical interactions between mSBT overall score and anatomical regions were detected. Demographic and historical variables explained 10.2%, 3.2%, and 7.2% of the variance in SF-8 PCS scores, SF-8 MCS scores, and average pain intensity ratings, respectively. The addition of mSBT overall score resulted in increases in R2 by 30.4% (P<.01), 25.5% (P<.01), and 24.4% (P<.01) for SF-8 PCS, SF-8 MCS, and average pain intensity, respectively. In all final models, the strongest unique contributor was mSBT overall score, with higher mSBT scores indicating lower SF-8 PCS and SF-8 MCS scores and higher average pain intensity ratings. These results were consistent with our primary hypothesis; the lack of interactions indicated similar directions and magnitudes of the associations for all anatomical regions. Additionally, surgery was a unique contributor for lower SF-8 PCS scores.
Contribution of Anatomical Region (AR) and Modified STarT Back Screening Tool (mSBT) Overall Score to General Health, Pain, and Disability Measure Scoresa
Disability Measures
The results of the final regression model for disability measures are shown in Table 4. No statistical interactions between mSBT overall score and anatomical regions were detected. Demographic and historical variables explained 10.1% of the variance in disability. The addition of mSBT overall score resulted in an increase in R2 by 43.1% (P<.01). In the final model, the strongest unique contributor was mSBT overall score, with higher mSBT scores indicating greater disability. These results also were consistent with our primary hypothesis; the lack of interactions indicated similar directions and magnitudes of the associations for all anatomical regions. These findings are shown in the Figure (panel D). Surgery also had a unique relationship with greater disability in the final model.
Region-Specific Disability Measures
The final regression models for nontransformed region-specific disability measures explained 63.7% (P<.01), 69.8% (P<.01), 43.5% (P<.01), and 48.3% (P<.01) of the variance in ODI, NDI, QuickDASH, and IKDC scores, respectively. Demographic and historical variables explained 21.7%, 4.9%, 14.7%, and 10.4% of the variance in ODI, NDI, QuickDASH, and IKDC scores, respectively. The addition of mSBT overall score resulted in increases in R2 by 41.9% (P<.01) for ODI scores, 64.9% (P<.01) for NDI scores, 28.8% (P<.01) for QuickDASH scores, and 37.9% (P<.01) for IKDC scores. In all final models, mSBT overall score was the strongest unique contributor and was positively related to region-specific disability scores. Additionally, surgery was a unique contributor to higher ODI and QuickDASH scores.
Psychological Measures
The results of the final regression models for psychological measures are shown in Table 5. Across all models, no interactions between mSBT psychosocial subscore and anatomical regions were detected. Demographic and historical variables explained 4.8%, 3.5%, 4.7%, 2.7%, 3.2%, and 6.5% of the variance in TSK-11, Pain Catastrophizing Scale, mFABQ-PA, STAI-T, PHQ-9, and PSEQ scores, respectively. The addition of mSBT psychosocial subscore resulted in increases in R2 by 26.0% (P<.01) for TSK-11 scores, 45.0% (P<.01) for Pain Catastrophizing Scale scores, 9.7% (P<.01) for mFABQ-PA scores, 39.4% (P<.01) for STAI-T scores, 36.7% (P<.01) for PHQ-9 scores, and 32.6% (P<.01) for PSEQ scores. In all final models, the strongest unique contributor was mSBT psychosocial subscore, with higher mSBT psychosocial subscores being associated with higher TSK-11, Pain Catastrophizing Scale, mFABQ-PA, STAI-T, and PHQ-9 scores but lower PSEQ scores. These results were consistent with our secondary hypothesis; the lack of interactions indicated similar directions and magnitudes of the associations for all anatomical regions. For TSK-11, the shoulder × mSBT subscore and knee × mSBT subscore interaction terms had P values of less than .05 but greater than .01; although these data did not meet our critical P value set at less than .01, there were potential interactions between these 2 anatomical pain regions and mSBT psychosocial subscore for TSK-11. Additionally, surgery was associated with lower PSEQ scores, the knee anatomical region was associated with higher PSEQ scores, and pain duration was associated with lower mFABQ-PA scores.
Contribution of Anatomical Region (AR) and Modified STarT Back Screening Tool (mSBT) Psychosocial Subscore to Psychological Measure Scoresa
Discussion
The findings of the present study provide a preliminary indication that an mSBT has potential utility as a multidimensional screening measure for people who have neck, shoulder, or knee pain and are at increased risk for poor clinical outcomes. In the multivariate models, elevated mSBT scores showed the strongest unique association with poorer health, higher pain intensity, and greater disability. Elevated mSBT psychosocial subscores showed the strongest unique association with higher negative psychological scores and lower positive psychological scores. Additionally, our findings revealed associations between surgery and poorer physical health, greater disability, and poorer self-efficacy. The associations between surgery and lower clinical measure scores were expected and were similar to previous findings.10,41,42 These strong and consistent findings are supportive of a concise assessment across different anatomical regions, but further testing is needed before an mSBT can be implemented in clinical practice. Before recommendations for clinical practice can be made, longitudinal investigations are needed to fully validate the use of an mSBT in people who have neck, shoulder, or knee pain.
Our primary hypothesis involved interactions, and consideration of the precision of the reported effects is necessary given that analyses for the detection of interactions are often underpowered. The 95% confidence intervals of the nonstandardized regression coefficients suggested that if interactions were assumed, then specific anatomical regions would have corresponded to a 1.59- to 1.69-point difference in SF-8 PCS scores, a 1.98- to 2.11-point difference in SF-8 MCS scores, a 0.44- to 0.46-point difference in average pain ratings, and a difference in the standard deviation of disability scores of 0.17 to 0.18. These small differences are not particularly clinically meaningful for these measures. For example, the range of differences in pain intensity ratings does not exceed commonly reported minimal clinically important differences15,43–45; additionally, differences in health status and disability scores corresponded to a change in the standard deviation of less than 0.25. Although a larger sample may have allowed us to statistically detect interactions, our primary results suggested that the impact of interactions may have had minimal clinical relevance. This trend also was observed for the psychological multivariate models, demonstrating that if interactions were detected in a larger sample, their impact would not have provided clinically meaningful information.
We also investigated the relationship between the mSBT and a psychosocial questionnaire for positive coping: the PSEQ. This examination was important given the recent attention to positive coping in current literature.46–51 Our findings revealed a negative relationship between the mSBT and self-efficacy, with the mSBT contributing to a 4.91- to 6.54-point reduction in PSEQ scores. This moderate effect provided conceptual support for this relationship, suggesting that an mSBT may be used to indirectly screen for positive coping. These results are consistent with those of earlier research identifying self-efficacy as a predictor of greater disability at a 6-month follow-up in people with low back pain.51 Our findings, combined with those of other recent investigations of positive coping,46–51 support the need for further examination of the longitudinal relationship between mSBT scores and other positive factors (such as resilience and optimism). Additionally, the TSK-11 multivariate model suggested that the groups with knee and shoulder pain may have had weaker relationships between the mSBT and kinesiophobia than the group with low back pain. The presence of any interaction between these pain regions and the mSBT, however, would correspond to a TSK-11 score difference of only 2.22 to 2.33 points—again, a difference that likely would not be clinically relevant even if it were statistically detected.
Strengths and Limitations
The inclusion of low back pain as a reference condition in the present study allowed for direct comparisons of the mSBT with neck, shoulder, and knee pain conditions within the same regression models. Analyses were based on a sample comprising approximately 100 people for each anatomical region, a sample size that was adequate for investigating the relationships of interest. The 95% confidence interval findings across the multivariate models were precise enough to not exceed clinically relevant sizes, strengthening our overall conclusion that the mSBT has the potential for application beyond low back pain. Additionally, we used a wide range of measures—many of which are commonly used as primary outcomes in longitudinal studies—to investigate the relationships between the mSBT and anatomical regions. Finally, data were collected in physical therapy settings and could be used as important comparison data in future longitudinal studies in primary care settings, such as those currently in progress at Keele University.
Several limitations should be considered in the interpretation of the findings from this proof-of-principle study. First, our sample was one of convenience—people seeking care in outpatient physical therapy settings in Florida. Additionally, although the inclusion criteria were broad (people with low back, neck, shoulder, or knee pain), our sample did not include people with other common pain conditions (eg, fibromyalgia, neuropathic pain, pain in other anatomical regions) for which the use of an mSBT might be warranted. These factors limit the overall generalizability of our results. Second, the cross-sectional design of the present study would not yield evidence of longitudinal relationships. Therefore, we are unable to provide specific recommendations for the risk assessment capability of an mSBT and cannot recommend the clinical use of an mSBT at this time. Longitudinal studies are needed before such recommendations can be made.
In conclusion, our findings provide preliminary, conceptual support for the potential clinical use of an mSBT as a general multidimensional screening tool for people with low back, neck, shoulder, or knee pain. Our findings revealed similar directions and magnitudes of associations between an mSBT and clinical and psychological measure scores across the included musculoskeletal pain conditions but did not establish whether an mSBT contributes to risk assessment for poor clinical outcomes in these conditions. Indeed, this proof-of-principle study provides conceptual support only for these associations. Before the clinical use of an mSBT for other musculoskeletal pain conditions can be recommended, large-scale longitudinal studies must be conducted to determine the feasibility of risk-stratified care for people with the included musculoskeletal pain conditions.
Footnotes
All authors provided concept/idea/research design and reviewed the manuscript before submission. Dr Butera and Mr Lentz provided data analysis. Dr Beneciuk and Dr George provided consultation for interpreting results. Mr Lentz and Dr Beneciuk provided data collection. Dr George provided project management, fund procurement, and facilities/equipment.
The authors thank the following people for their help with this study: from UF Health—Michael Borut, Christine Martin, Michael Hodges, Penny Goldberg, Derek Miles, Jason Bouwkamp, Jody Davis, Josh Barabas, Brittney Barrie, Christine Eckert, and Ludo DeWolf; and from Brooks Rehabilitation—Raine Osborne, Monika Beneciuk, Renata Salvatori, Brian Hagist, Jenny Hagist, Tasha Mouton, Trent Harrison, Timothy Shreve, Michael Bourassa, Christina Harder, Sara Bertrand, Nate Moore, John Leschitz, and Robert Rowe.
This study was approved by the University of Florida Gainesville Health Science Center Institutional Review Board.
This article was written while Dr Beneciuk received support from the National Institutes of Health Rehabilitation Research Career Development Program (K12-HD055929). This work also was funded by the Orthopaedic Section of the American Physical Therapy Association and Brooks–PHHP Research Collaboration, University of Florida.
- Received June 1, 2015.
- Accepted January 24, 2016.
- © 2016 American Physical Therapy Association
References
Issue highlights
