The association between self-efficacy on function and pain outcomes among patients with chronic low back pain managed using the McKenzie approach: a prospective cohort study

ABSTRACT

Introduction

Self-efficacy is a determinant of function and pain outcomes in patients with chronic low back pain receiving physiotherapy. The McKenzie approach is an effective intervention for patients with back pain that may affect self-efficacy. Study aims were to determine if, among patients with back pain being managed by McKenzie-credentialed physiotherapists: intake self-efficacy is correlated with intake function and pain; intake self-efficacy is associated with changes in function and pain during treatment; self-efficacy improves during treatment; and improvements in self-efficacy during treatment are associated with improvements in function and pain at discharge.

Methods

Two-hundred-eighty-two subjects with chronic low back pain seen by McKenzie-credentialed clinicians provided data on self-efficacy, function and pain at intake and discharge.

Results

Self-efficacy was correlated with function and pain at intake; however, intake self-efficacy was not associated with function or pain outcomes. Self-efficacy increased during treatment. This increase was associated with improvements in function and pain at discharge.

Conclusion

While intake self-efficacy was associated with function and pain when initiating physiotherapy, it did not result in improved treatment outcomes. Specific interventions may be necessary to improve self-efficacy. The increases in self-efficacy observed during treatment were associated with improvements in function and pain outcomes at discharge.

Introduction

The American Psychological Association defines self-efficacy as an individual’s confidence in the ability to exert control over one’s own motivation, behavior, and social environment [1]. In relation to patients with chronic pain, including chronic musculoskeletal pain, the definition of self-efficacy is refined to indicate a patient’s overall confidence in the ability to cope and self-manage symptoms, stresses or limitations associated with a painful condition [2,3].

Self-efficacy is hypothesized to affect chronic musculoskeletal disability and pain by influencing how situations such as performing physical activities that are associated with pain are managed in an attempt to mitigate painful experiences [4]. For example, patients with low self-efficacy might avoid activities that are accompanied by pain, whereas those with high self-efficacy might instead independently identify and implement changes in movement patterns that reduce pain with these activities [4].

There is a growing consensus in the literature of a significant inverse association between baseline self-efficacy; and disability [3,5–15] and pain intensity assessed at intake [3,7–10,15]. In fact, we were able to identify only one study in which investigators reported that there were no significant differences between self-efficacy measure and pain intensity at baseline [12]. These findings imply that self-efficacy is associated with a patient’s physical function and pain intensity before a patient with chronic low-back pain begins treatment.

Several studies assessed the relation between intake self-efficacy and changes in disability and/or pain outcomes measured at the end of treatment among subjects with chronic low back pain. In 2 studies, self-efficacy was significantly and inversely associated with disability [14,15] and pain [15] outcomes among patients receiving physical therapy. In another study, performed on patients in primary care, disability outcomes were not affected by intake self-efficacy measures [16]. It is therefore not confirmed that patients’ level of self-efficacy measured at the onset of treatment for chronic low-back pain predicts treatment outcomes.

One study addressed this question of whether self-efficacy affects treatment outcomes by analyzing whether changes in self-efficacy were associated with changes in disability and pain over a 12-month period among subjects with chronic low-back pain receiving treatment from primary care physicians. Greater increases in self-efficacy were associated with greater decreases in disability and pain at 12-months follow-up [8].

This research on self-efficacy supports suggestions by rehabilitation experts that physiotherapists use a biopsychosocial approach to manage their patients with chronic low-back pain [17–21]. Nevertheless, many physiotherapists have traditionally been trained to examine and treat patients using a biomechanical model [20]. The biopsychosocial approach encourages clinicians to balance patients’ biomechanical and psychosocial or cognitive and behavioral issues when prescribing interventions to produce more effective and efficient treatment outcomes [22]. Components of a biopsychosocial approach could potentially include interventions to improve patients’ confidence in their ability to cope with and self-manage symptoms, stresses and/or functional limitations associated with their chronic disability and pain.

One approach to evaluating and treating patients with low back pain is the McKenzie method. Several recent critical reviews support the effectiveness of this mechanical diagnosis and therapy (MDT) strategy when delivered by trained clinicians following an adherent approach to the core principles of MDT [23–25]. Briefly, the McKenzie approach is a comprehensive method used by physical therapists during routine clinical care that categorizes patients with low back pain into biomechanical subgroups [26] while emphasizing self-management of pain in relation to specific biobehavioral spinal movements and posture. [27] Even though treatment is based on categorization into biomechanical subgroups, a 2014 study suggested that McKenzie-trained physical therapists in Japan have a more psychosocial perspective on the management of patients with low back pain than general non-trained MDT physical therapists [28]. This discrepancy could be explained by the component of the McKenzie approach that focuses on changing patients’ pain behaviors and cognitions through patient education [28]. McKenzie-based patient education could empower patients to attain active self-management strategies and problem-solving techniques to self-control symptoms; all components of self-efficacy.

We identified 2 studies that addressed changes in self-efficacy measured after completing treatment using McKenzie methods for patients with low back pain [29,30]. The primary aims of these studies were to compare changes in self-efficacy in patients experiencing low back pain who were being treated with McKenzie-based interventions to 1) McKenzie plus back strengthening and endurance exercises [29]; and 2) McKenzie vs. a brief pain management approach using cognitive behavioral principles with or without an educational booklet [30]. In both studies, subjects’ self-efficacy measures improved at discharge from rehabilitation within all subgroups regardless of intervention based on McKenzie-based classification [29,30]. However in one study, the addition of back endurance exercises with McKenzie interventions had a significantly higher positive effect on self-efficacy compared to McKenzie intervention alone (p = 0.001) [29].

Prior studies have suggested that the McKenzie-based self-management approach may partially explain the positive effects on improving physical function and decreasing psychosocial distress and pain intensity when patients were managed by clinicians with advanced training in McKenzie methods [27,31–33]. In addition, 2 studies suggest that self-efficacy improves among patients with low back pain who are managed using McKenzie methods [29,30], although this improvement in self-efficacy does not appear to occur to a greater extent when compared with other interventions. Prior research has raised concerns about the clinical utility of the McKenzie classification system among therapists that did not complete training at the credentialed or diploma levels [34]. One possible reason that greater improvements in self-efficacy did not consistently occur among patients being managed with McKenzie methods could therefore be the level of therapist training in the McKenzie method: in the earlier study, participating therapists attended a 1-day training in McKenzie methods and had completed some post-graduate training courses in the McKenzie approach [30], whereas in the second study, therapist credentials were not described [29].

There is therefore a need to investigate associations between patients’ self-efficacy and outcomes among patients with chronic low back pain being managed by physical therapists with advanced training in the McKenzie approach. Study aims were therefore to examine whether:

1. Intake self-efficacy is correlated with intake physical function and pain. We hypothesize that self-efficacy will be positively correlated with physical function and negatively correlated with pain.

2. Intake self-efficacy is associated with changes in physical function and pain intensity by the end of treatment. We hypothesize that patients with higher self-efficacy will have greater improvements in function and pain.

3. Self-efficacy improves by the end of treatment. We hypothesize self-efficacy will improve after completing treatment.

4. Changes in self-efficacy by the end of treatment are associated with changes in physical function and pain when measured after completing treatment. We hypothesize that changes in self-efficacy during treatment will be positively associated with improvements in physical function and pain intensity assessed at discharge from rehabilitation.

Methods

We performed a retrospective analysis of a prospective cohort study. Cohort study designs are recommended as a viable alternative to randomized controlled trials to identify clinical relationships in real‐world settings, thereby enhancing external validity [35,36]. The Rutgers University Institutional Review Board approved this study. Since the study did not require changes in clinical practice or data documentation, patient informed consent was not required.

Two-hundred-eighty-two subjects older than 17 years experiencing chronic (pain > 90 days) nonspecific low back pain who had completed an episode of care between 1 January 2013 and 30 June 2019 at one of 4 participating outpatient physiotherapy clinics were included in our study. Practice settings included private practice or hospital‐based outpatient physiotherapy clinics. Broad inclusion criteria were chosen to replicate the diverse patient case‐mix typically seen in outpatient physiotherapy clinics.

We excluded patients whose pain had occurred less than 90 days before beginning treatment since patients with chronic low back pain have been shown to have a different association with self-efficacy than those with less chronic symptoms [37]. Furthermore, patients with chronic low back pain are less likely to recover spontaneously, and therefore improvements in patient outcomes may be more likely attributable to treatment effects. Patients were also excluded due to pregnancy or suspicion of serious spinal pathology.

Four physiotherapists credentialed in McKenzie methods [26,38] collected patient data. Participating physiotherapists routinely collect patient demographics, health characteristics, and outcomes using the Patient Inquiry software developed by Focus on Therapeutic Outcomes (FOTO), a Net Health company that provides outcomes management software solutions for rehabilitation therapists [39]. All patient data were deidentified using the Safe Harbor method [40].

When collecting patient information in the FOTO system, patients enter their data using an iPad or laptop computer. Therapists are therefore not involved in tabulating outcome scores. To decrease the potential for patient selection bias and the biasing of patient responses when completing the outcome surveys, FOTO has implemented quality audits focused on increasing outcome data completion rates and providing therapists with standards for the administration of outcome measures such that surveys are administered in a neutral manner [41].

Subjects completed functional status, pain and self-efficacy questionnaires at intake. These questionnaires were repeated at discharge, which coincided with the final physiotherapy treatment session.

We quantified subjects’ functional status using FOTO’s Lumbar Computer Adaptive Test. This measure has been previously described in detail [42–44]. Functional status scores range from 0 (low) to 100 (high functioning) on a linear metric. Questions in the item bank have demonstrated internal consistency, reliability, validity, sensitivity, and responsiveness (minimal clinically important improvement averaged 5 points) [42].

Maximal pain intensity during the past 24 hours was assessed using an 11‐point Numeric Pain Rating Scale ranging from 0 (no pain) to 10 (worst imaginable pain). This measure has been reported to be reliable and valid in this population [45,46]. The minimal clinically important improvement for patients with chronic low back pain has been shown to be 2 points [45].

The Chronic Pain Self-Efficacy Scale was used to quantify self-efficacy. This scale is a 22-item questionnaire [2]. Each question asks the patient, ‘How certain are you that you can … ’. Certainty is rated on a 10-point Likert scale anchored by a rating of 10 indicating ‘very uncertain’ and a rating of 100 indicating ‘very certain’ with higher scores representing higher self-efficacy. The Chronic Pain Self-Efficacy Scale is comprised of 3 subscales: pain management, coping with symptoms and physical function. We collected data on the pain management (5 questions) and coping with symptoms (8 questions) subscale exclusively, since physical function self-efficacy using the Chronic Pain Self-Efficacy Scale and perceived physical ability assessed using FOTO’s Lumbar Computer Adaptive Test provides similar information. The reliability of the 2 subscales has been reported to be 0.88 for self-efficacy for pain management, and 0.90 coping with symptoms [2].

Overall rehabilitation interventions were guided by symptom responses from repeated end-range movement tests and/or positioning techniques observed during the physical examination. For instance, patients whose symptoms centralized were prescribed specific exercises based on directional preference [47,48]. For patients whose symptoms neither centralized nor demonstrated a directional preference, an individualized active rehabilitation plan was developed by the treating physical therapist. All patients received the same educational approach following self-management principles. The overall treatment plan focused on return to physical function, empowering the patient to become actively involved in his or her recovery and to enhance their self-efficacy beliefs. There was no attempt to standardize care beyond these guidelines.

Analyses were performed using SAS v.9.4. We first examined for potential loss‐to‐follow‐up bias by comparing patient characteristics for subjects with complete functional status, pain and self-efficacy data at intake with those with any of these measures missing at discharge. To address our first study aim, we calculated partial correlation coefficients. All other study aims were evaluated using multiple linear regression analyses. We calculated beta coefficients and 95% confidence intervals/p-values for all associations. A p-value of 0.05 or less was considered statistically significant.

In all models, the following 9 constructs were evaluated for potential confounding, since each has been shown to influence functional status and/or pain outcomes: age (18 to 45 years, vs. >45 years to <65 years, vs. 65 to < 75 years, vs. >/ = 75 years), gender (male vs. female), third-party payer (Medicaid vs. litigation, no fault, auto insurance or indemnity vs. Medicare vs. workers’ compensation vs. HMO or PPO vs. self-pay, other or no charge), medication use for back pain at intake (no vs. yes), exercise history (seldom or never vs. 1–2 times a week vs. at least 3 times a week), lumbar surgical history (no vs. yes), prior treatment episode (no vs. yes), medical comorbid conditions (none vs. 1 or 2 vs. 3 vs. 4 or more) and directional preference (absent vs. present). Since all covariates were categorized, parametric assumptions of linear regression analyses were met. Confounding effects were identified using the criteria that a 10% change in the effect measure when adding the potential confounder(s) to the baseline model represents meaningful confounding [49]. Variables include in the models were checked for collinearity by evaluating the condition index. If the condition index was higher than 30, variables were eliminated based on the variable in the model with the highest variance inflation factor.

Results

Our sample consisted of 398 patients [73% female; mean age 54.91 (95% CI 47.39,62.44)]. Of those, 282 patients (71%) had complete intake and discharge outcome data. Subjects that did not contribute outcome data at discharge were more likely to be younger, be male, have lower coping skills self-efficacy scores, be covered by Medicaid or be classified as ‘self-pay’ or ‘no charge’, exercise more and/or have a directional preference. (data available upon request) Of the 282 subjects retained for analyses, 26.95% were male and the mean age was 61.49 (95% CL 59.91,63.08)

Correlation coefficients for the explained variability between intake self-efficacy and intake function and pain are presented in Table 1 and range from r² = 0.005 to 0.09. While statistically significant, explained variability was small. Table 2 describes the association between intake self-efficacy; and change in function and pain measures. None of the associations between intake self-efficacy and outcomes were significant (p-values between 0.24 and 0.95). There was evidence of significant improvement in self-efficacy scores at rehabilitation discharge (p-values between <0.0001 and 0.003). (Table 3) Furthermore, these improvements in self-efficacy were associated with significant improvements in both physical function and pain intensity outcomes assessed at discharge from rehabilitation (p-values between <0.0001 and 0.003). (Table 4)

Table 1.

Correlation between intake self-efficacy, and intake function and pain measures (n = 282).

	Intake pain management Self-efficacy	Intake coping with symptoms Self-efficacy
Intake function unadjusted analyses	r = 0.20 95% CI of r = (0.08, 0.34) r2 = 0.04 p-value = 0.0009	r = 0.27 95% CI of r = (0.16, 0.37) r2 = 0.07 p-value < 0.0001
Intake function Adjusted analyses	r = 0.12 95% CI of r = (0.00, 0.23) r2 = 0.01 p-value = 0.05 variables retained in the model: age, gender, third-party payer, exercise history condition index = 21.79	r = 0.21 95% CI of r = (0.10, 0.32) r2 = 0.04 p-value = 0.0004 variable retained in the model: exercise history condition index = 10.86
Intake pain Unadjusted analyses	r = −0.16 95% CI of r = (−0.28, −0.05) r2 = 0.03 p-value = 0.005	r = −0.21 95% CI of r = (−0.32, −0.09) r2 = 0.04 p-value = 0.004
Intake pain Adjusted analyses	r = −0.10 95% CI of r = (−0.22, 0.02) r2= 0.01 p-value = 0.09 variables retained in the model: age, gender, third-party payer, exercise history condition index = 18.44	r = −0.17 95% CI of r = (−0.28, −0.05) r2 = 0.03 p-value = 0.005 variables retained in the model: exercise history condition index = 5.23

Table 2.

Association between intake self-efficacy, and change in function and pain measures (n = 282).

	Intake pain management Self-efficacy	Intake coping with symptoms Self-efficacy
Change in function unadjusted analyses	beta = 0.01 95% CI of beta = (−0.05, 0.07) p-value = 0.82	beta = −0.03 95% CI of beta = (−0.09, 0.03) p-value = 0.28
Change in function Adjusted analyses	beta = 0.002 95% CI of beta = (−0.05, 0.06) p-value = 0.95 variables retained in the model: age, third-party payer, prior treatment episode condition index = 19.39	beta = −0.01 95% CI of beta = (−0.07, 0.47) p-value = 0.75 variables retained in the model: age, gender, third-party payer, medication use at intake, exercise history, prior treatment episode condition index = 29.70
Change in pain Unadjusted analyses	beta = 0.0003 95% CI of beta = (−0.01, 0.01) p-value = 0.96	beta = −0.002 95% CI of beta = (−0.01, 0.01) p-value = 0.69
Change in pain Adjusted analyses	beta = −0.002 95% CI of beta = (−0.01, 0.009) p-value = 0.78 variables retained in the model: age, third-party payer, exercise history, lumbar surgical history, prior treatment episode, directional preference condition index = 25.02	beta = −0.01 95% CI of beta = (−0.02, 0.004) p-value = 0.24 variables retained in the model: age, third-party payer, exercise history, directional preference condition index = 17.72

Table 3.

Change in self-efficacy during treatment (n = 282).

	Intake pain management Self-efficacy	Intake coping with symptoms Self-efficacy
Discharge self-efficacy Unadjusted analyses	beta = 0.20 95% CI of beta = (0.10, 0.30) p-value = 0.0002	beta = 0.39 95% CI of beta = (0.31, 0.48) p-value < 0.0001
Discharge self-efficacy Adjusted analyses	beta = 0.15 95% CI of beta = (0.05, 0.25) p-value = 0.003 variables retained in the model: third-party payer, exercise history condition index = 11.7	beta = same as unadjusted 95% CI of beta = same as unadjusted p-value = same as unadjusted variables retained in the model: none, there was no meaningful confounding

Table 4.

Association between changes in self-efficacy during treatment and changes in physical function and pain levels during treatment (n = 282).

	Change in pain management Self-efficacy	Change in coping with symptoms self-efficacy
Change in function unadjusted analyses	beta = 0.12 95% CI of beta = (0.07, 0.16) p-value < 0.0001	beta = 0.16 95% CI of beta = (0.11, 0.22) p-value < 0.0001
Change in function Adjusted analyses	beta = same as unadjusted 95% CI of beta = same as unadjusted p-value = same as unadjusted variables retained in the model: none, there was no meaningful confounding	beta = same as unadjusted 95% CI of beta = same as unadjusted p-value = same as unadjusted variables retained in the model: none, there was no meaningful confounding
Change in pain Unadjusted analyses	beta = −0.03 95% CI of beta = (−0.04, −0.02) p-value < 0.001	beta = −0.02 95% CI of beta = (−0.03, −0.01) p-value < 0.001
Change in pain Adjusted analyses	beta = same as unadjusted 95% CI of beta = same as unadjusted p-value = same as unadjusted variables retained in the model: none, there was no meaningful confounding	beta = −0.02 95% CI of beta = (−0.03, −0.01) p-value = 0.003 variables retained in the model: age, third-party payer, medical comorbidities, directional preference condition index = 27.23

Discussion

Our study confirmed findings from previous studies demonstrating that among patients with chronic low back pain, higher self-efficacy at baseline is correlated with lower disability [3,5–15] and pain [3,7–10,15] upon entry into a physiotherapy episode of care. Contrary to our hypothesis, yet consistent with 2 prior studies [14,15], intake self-efficacy measures were not associated with improvements in physical function or pain intensity outcomes assessed at discharge from rehabilitation. These results suggest that, while self-efficacy might have an impact on disability and pain before physiotherapy treatment begins, higher measures of self-efficacy at intake are not necessarily associated with improved treatment outcomes. It appears that specific interventions are necessary to improve self-efficacy. This is consistent with findings addressed in our third hypotheses which demonstrated that self-efficacy improves during treatment using the McKenzie approach. Two prior studies addressing changes in self-efficacy in patients being managed with the McKenzie approach [29,30] reported similar findings. Our fourth hypothesis, addressing whether these improvements in self-efficacy were associated with improvements in both function and pain outcomes during treatment was also confirmed. This finding is consistent with one prior study [8]. Owing to the nature of our research design, we were unable to determine if these changes in self-efficacy that occurred during treatment were a major cause of the function and pain outcomes observed or if the improvements in function and pain produced the increase in self-efficacy that were observed during treatment.

Several studies have addressed the hypothesis that self-efficacy is a mediating variable in the association between treatment, and disability and pain outcomes among patients with chronic low back pain [8,37,50]. Under this hypothesis, the effect of treatment on disability and pain outcomes would be explained primarily by the changes in self-efficacy that occur as a result of treatment. Our study design does not permit a formal analysis of the mediator effects of self-efficacy on disability and pain outcomes. Nevertheless, our analyses partially support this hypothesis since self-efficacy was only associated with improvements in function and pain once the subjects received treatment. In these analyses, changes in self-efficacy explained 26.19% and 28.79% of the variance in function, and 24.33% and 28.70% of the variance in pain for pain management self-efficacy and coping with symptoms self-efficacy respectively (adjusted analyses when applicable). Several studies have suggested that treatment monitoring may enhance clinicians’ ability to adjust management strategies, thereby resulting in better treatment decisions and patient outcomes [51–53]. From a clinical perspective, our study findings support the practice of monitoring for changes in self-efficacy during treatment since patients whose self-efficacy scores are not improving during treatment might benefit from greater emphasis on interventions to improve self-efficacy measures.

We were not able to identify which component(s) of the McKenzie approach produced the changes in self-efficacy observed in our study. In a 2020 systematic review, the authors concluded that exercise and psychological interventions improved self-efficacy in patients with chronic musculoskeletal pain [54]. Several psychological interventions have been shown to improve self-efficacy, including feedback on the patients’ past performances and comparisons of their performance with that of others [55], the experience of seeing a ‘similar other’ successfully perform the target behavior [55], motivational interventions [56] and cognitive behavioral therapy [57].

Our study has several strengths. We used an observational research design which has been recommended to enhance the external validity of results by reflecting actual data documentation and interpretation by physiotherapists working in clinical practice [35,36,58]. We also included data from therapists working in 2 common outpatient physical therapy settings in the USA i.e. private practice and hospital-based clinics. Finally, our sample size was large enough to accommodate the analysis of multiple covariates while also yielding precise effect estimates.

The main limitation of our study is the risk of unmeasured confounders. This is a concern for examining observational data regardless of analytical methods [59]. To lessen the impact of this limitation we controlled for a wide range of 9 constructs and 27 health and patient demographic item-characteristics at baseline.

In addition, we could not verify if data from all eligible subjects were entered into FOTO. Our study therefore examined a convenience sample, potentially limiting generalizability. Also, slightly less than 30% of subjects were lost to follow-up, although differences between subjects with complete vs. incomplete data do not suggest a systemic selection bias. For examples, patients with incomplete outcomes were younger, exercised more and had a directional preference; characteristics found to be associated with higher functional status outcomes. However, patients also had a higher rate of Medicaid payer type and lower levels of coping self-efficacy scores; both characteristics found to be associated with lower functional status scores.

Finally, our study was not designed to determine the effect of self-efficacy on function and pain outcomes across different physiotherapy methods of evaluating and managing patients with low back pain or among therapists using McKenzie methods who are not credentialed. We can therefore only generalize our study results to patients being seen by physiotherapists with advanced training and credentialing in the McKenzie approach.

Conclusion

When measured during the initial physiotherapy visit, higher self-efficacy correlated with higher physical function and lower pain levels. However, intake self-efficacy measures were not associated with treatment outcomes. Our study results suggest that specific interventions such as patient education addressing self-management and return to work was required to improve self-efficacy beyond the levels measured at the onset of physiotherapy. Improvements in self-efficacy during the treatment episode were associated with increases in function and pain outcomes at rehabilitation discharge. Our findings are generalizable only to patients with chronic low back pain being treated by credentialed McKenzie therapists.

Funding Statement

The author(s) reported there is no funding associated with the work featured in this article.

Disclosure statement

Mark Werneke is a consultant for Net Health Systems, Inc., the company that owns the FOTO outcomes system, the database management company that manages the data analyzed in this research project. However, research projects such as this article are a normal part of Mark’s job tasks. No funds were received in support of this work. No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this article.