Abstract
Objective
The Fear-Avoidance Model (FAM) of chronic pain posits that pain catastrophizing and fear-avoidance beliefs are prognostic for disability and chronicity. In acute low-back pain, early physical therapy (PT) is effective in reducing disability in some patients. How early PT impacts short- and long-term changes in disability for patients with acute pain is unknown. Based on the FAM, we hypothesized that early reductions in pain catastrophizing and fear-avoidance beliefs would mediate early PT’s effect on changes in disability (primary outcome) and pain intensity (secondary outcome) over 3 months and 1 year.
Subjects
Participants were 204 patients with low-back pain of <16 days duration, who enrolled in a clinical trial (NCT01726803) comparing early PT sessions or usual care provided over 4 weeks.
Methods
Patients completed the Pain Catastrophizing Scale (PCS), Fear-Avoidance Beliefs Questionnaire (FABQ work and physical activity scales), and outcomes (Oswestry Disability Index and Numeric Pain Rating Scale) at baseline, 4 weeks, 3 months, and 1 year. We applied longitudinal mediation analysis with single and multiple mediators.
Results
Early PT led to improvements in disability and pain over 3 months but not 1 year. In the single mediator model, 4-week reductions in pain catastrophizing mediated early PT’s effects on 3-month disability and pain intensity improvements, explaining 16% and 22% of the association, respectively, but the effects were small. Pain catastrophizing and fear-avoidance beliefs did not jointly mediate these associations.
Conclusions
In acute low-back pain, early PT may improve disability and pain outcomes at least partly through reducing patients’ catastrophizing.
Keywords: Acute Pain, Physical Therapy, Disability, Back Pain, Mediation Analysis
Introduction
With growth in disability caused by low back pain (LBP) in the United States and worldwide [1, 2], effectively managing LBP in the acute stage is a pressing population health concern. Physical therapy (PT) is one strategy that may mitigate the transition to chronicity and promote more rapid recovery in some patients [3, 4]. Limited evidence is available, however, on the mechanisms by which PT leads to functional improvements in patients with acute LBP.
Psychosocial risk factors are more prognostic for pain-related outcomes than biomedical factors in LBP [5]. The most common model developed to explain how psychosocial factors influence pain-related outcomes is the Fear-Avoidance Model (FAM) [6], which posits that pain catastrophizing and fear-avoidance beliefs are causally linked to disability. Research has generally upheld their key roles in persistent disability [7].
Pain catastrophizing, or exaggerated negative thoughts and feelings about a current or anticipated pain experience [8, 9] is prognostic for chronicity [5, 10]. It also mediates outcomes from treatments for chronic LBP, including PT [11, 12]. Fear-avoidance beliefs, or beliefs that activity will exacerbate pain or cause reinjury, predict LBP persistence and chronicity [13, 14]. There has been robust work on risk stratification in acute LBP, particularly work related to the STarT Back screening tool in primary care [15, 16], that measure these factors early on to identify patients at greater risk for disability for additional treatment. Pain catastrophizing and fear-avoidance beliefs are also modifiable treatment targets, however [11, 17]. Studies based on the FAM have found that reduced fear-avoidance beliefs mediate treatment effects on clinical outcomes including return to work [14], and that pain catastrophizing may play a central role in the early, acute stage of LBP [18]. To the extent that early changes in these factors impact patients’ outcomes, they may therefore be used to inform treatment monitoring. There is a dearth of work, however, on the timing of treatment-related changes in these factors and their impacts on treatment efficacy. Furthermore, most research examining these factors as prognostic for pain-related outcomes has focused on chronic LBP. Their role as mediators of treatment outcomes in acute LBP has not been adequately evaluated [19], particularly for treatments like early PT, which can improve some patients’ outcomes [20, 21] while reducing costs, health utilization, and the likelihood of future invasive treatments [22, 23]. Specifically, the extent to which early changes in these factors within the first month of consultation for acute LBP explain how early PT works is unknown.
A randomized controlled trial found that early PT reduced patients’ disability after 3 months and improved quality-of-life 1 year later relative to education alone [21]. Although its effects were small, similar to other small-to-moderate effects reported for PT interventions for LBP [24], this early PT protocol was also cost-effective over 1 year [25]. Another recent trial among patients with sciatica showed clinical benefits of early PT over 6 months [20]. Additionally, patients in the parent trial who were retrospectively stratified as high risk at baseline based on risk factors for poor outcome, including catastrophizing and fear-avoidance, benefited more from early PT at 3 months [26]. These findings are consistent with work showing that physical treatments that do not specifically target psychosocial factors can nevertheless impact these factors [11], and further exploration of a possible mediating role is warranted. Gaining understanding of mediating processes is critical for informing modifications that can lead to larger treatment effects.
We hypothesized that reductions in pain catastrophizing and fear-avoidance beliefs over 4 weeks would mediate the association between early PT (compared to usual care) and subsequent reductions in disability and pain intensity at 3-month and 1-year follow-up assessments.
Methods
Participants
The present study was a secondary analysis of a randomized controlled trial (ClinicalTrials.gov number: NCT01726803) evaluating the efficacy and cost-effectiveness of an early PT intervention (delivered in the first four weeks of primary care consultation) relative to usual care in nonspecific acute LBP. The trial’s design, methods, and main results have been described elsewhere [21]. Trial participants included 220 patients aged 18–60 years who consulted primary care in Salt Lake City, Utah between 2011 and 2014 for an acute LBP episode and whose clinical presentations made them ideal candidates for the PT protocol based on a published decision rule [27]. Specifically, these patients sought care within 16 days of symptom onset; presented with moderate disability (Oswestry disability scores ≥ 20); had not experienced pain distal to the knee in the 72 hours prior; and had not received any treatment for LBP in the past 6 months. This secondary analysis of these data was approved by the local University Institutional Review Board. All procedures performed in this study were in accordance with the ethical standards of the University Institutional Review Board and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
The present study’s full mediation analysis included 204 (early PT group: N = 104; usual care group: N = 1001) participants who had complete data on the baseline covariates, proposed mediators, and outcomes. Participants’ mean age was 37.36 years; (SD = 10.42), 52% were female, 80% were non-Hispanic white, and 67% had previously sought treatment for an LBP episode. There were no substantial differences on baseline measures of patient demographics, putative mediators, or outcomes between the treatment groups. Females were marginally more likely to have been randomly assigned to early PT and gender was kept as a covariate. Of 204 participants in the 3-month full mediation analysis sample, 97% (N = 197; early PT: N = 102; usual care: N = 95) of patients also provided outcome data at the 1-year follow-up and were included in the 1-year full mediation analysis sample. The demographics of the full 1-year analysis sample did not significantly differ from the 3-month analysis sample. Comparisons between the characteristics of participants in the final 3-month analysis sample versus those who were excluded list-wise from the analysis suggested that participants who were excluded due to missing outcome data reported lower educational attainment at baseline (P = .037) and had less disability improvement over 3 months (P = .017).
Procedures
All participants received an educational session with a physical therapist and a copy of The Back Book [28] and were subsequently randomized to early PT or usual care. Early PT consisted of standardized protocols of spinal manipulation and exercise during 4 sessions over 4 weeks. The usual care group received the educational session only and were told to consult their primary care provider as needed during the 4-week treatment period. Patients completed online surveys at four time points: baseline, or the initial evaluation and educational session (i.e., prior to treatment randomization), at 4 weeks (immediately following the treatment period), and 3-month and 1-year follow-ups.
Measures
Covariates
In a trial in which treatment is randomized, the validity of mediation analyses depends primarily on the assumption that common causes of the mediators and outcome can be blocked by covariate adjustment. Accordingly, we identified participant demographics and pain-relevant factors from the baseline assessment as potential confounders. Covariates were selected a priori, based on theory [6] and predictive associations from prior empirical work [29]. Factors that differed between the early PT and usual care groups at baseline (i.e., gender) were also controlled for in the analyses. Covariates included participants’ age in years and body mass index (BMI); gender (0= male; 1= female); education level (1= less than high school; 2 = graduated from high school/GED; 3 = some college; 4= graduated from college; 5 = some post-graduate course work; 6 = completed post-graduate degree); marital status (0= single/divorced/widowed; 1= married or cohabitating); prior treatment-seeking for an LBP episode (0= no; 1= yes); and lastly, baseline scores of disability (Oswestry), pain intensity, and the proposed mediators.
Pain Catastrophizing (Proposed Mediator)
The Pain Catastrophizing Scale (PCS) [30] assessed the extent to which participants experience magnification, rumination, and helplessness in response to their pain episode. The PCS has 13 items, each scored on a 5-point Likert scale from 0 to 4 (0= not at all; 4= all of the time). The PCS is reported as a total composite score (range 0–52), with higher scores indicating greater catastrophic thoughts about pain. The internal consistency of the PCS is high in adult outpatient samples (Cronbach’s α =0.92) [8]. Pain catastrophizing was measured at each assessment. We focused on change between the baseline and 4-week measures (Δ 4-week) in mediation analyses.
Fear-Avoidance Beliefs (Proposed Mediator)
The Fear Avoidance Beliefs Questionnaire [31] includes two subscales: a 7-item work subscale (FABQW) and 4-item physical activity subscale (FABQPA) which assess individuals’ beliefs about how work and physical activity, respectively, may affect their pain, and about their perceived risk for worsening symptoms or reinjury. Items are scored 0–6, with higher scores indicating greater fearful beliefs about symptom exacerbation or re-injury and greater avoidance of activity. The FABQW composite score ranges from 0 to 42 and the FABQPA from 0 to 24. The internal consistency of these measures are high (Cronbach’s α =0.82 and 0.70, respectively) [32]. Fear-avoidance beliefs was measured at each assessment. We focused on change between the baseline and 4-week measures (Δ 4-week) in mediation analyses.
Disability (Primary Outcome)
The Oswestry Disability Index (ODI) [33] is a pain-specific disability measure that includes 10 items on the extent to which LBP interferes with participants’ ability to perform typical daily activities (e.g., walking, sitting, lifting objects, etc.). Total scores range from 0 to 100, with higher numbers indicating greater disability. This scale has high internal consistency (Cronbach’s α =0.86) [34]. Mediation analyses included ODI changes from baseline to 3 months (Δ 3-month) and from baseline to 1 year (Δ 1-year) as outcomes.
Pain Intensity (Secondary Outcome)
The Numeric Pain Rating Scale (NPRS) [35] asks participants to rate their pain intensity on a scale from 1 to 10, with higher scores reflecting greater pain intensity. The responsiveness of this scale is high [36]. Mediation analyses included NPRS changes from baseline to 3 months (Δ 3-month) and from baseline to 1 year (Δ 1-year) as outcomes.
Statistical Analyses
Bivariate relationships among the predictors, mediators, and outcomes were first examined. Univariate analysis of covariance (ANCOVA) were applied to compare mean levels of the three proposed mediators (i.e., PCS, FABQW and FABQPA) at the 4-week time point between the randomized treatment groups (early PT vs usual care), adjusted for baseline scores [37]. The statistical significance level for all tests was set at an alpha probability value of .05.
Mediation Analysis
Mediation analyses within an ordinary least squares (OLS) regression framework [38] were applied to examine whether early changes in pain catastrophizing and fear-avoidance beliefs during the treatment period (from baseline to 4 weeks) mediated the association between early PT (relative to usual care) and outcomes of changes in disability and pain intensity. We calculated early changes in the three putative mediators by taking the raw difference of participants’ scores between baseline assessment and 4-week follow-up (Δ 4-week= 4-week score—baseline score). Negative scores reflected reductions (improvements) in these factors. In all models, a complete case analysis was conducted, such that only participants with complete data on the mediators and outcomes were included in the analysis. The outcomes were relatively short-term changes from baseline to 3-month follow-up (Δ 3-month= 3-month score—baseline score) in disability (primary outcome) and pain intensity (secondary outcome); again, negative scores reflected reductions in disability and pain. Analyses were repeated with longer-term outcomes of changes from baseline to 1-year follow-up (Δ 1-year= 1-year score—baseline score) in disability and pain intensity.
We first evaluated the three proposed mediators individually. The validity of these analyses requires the assumption that the mediator being analyzed is not causally affected by either of the other two mediators. Because we evaluate each mediator separately, these analyses ignore the impact of other potential mediators on the mediatory pathway modeled. Although existing theory does not definitively stipulate causal relationships among the mediators, such dependence is hard to rule out, so these analyses are interpreted cautiously. In part to address uncertainty in the causal independence assumption, we also evaluated the three mediators jointly in analyses which evaluate the overall indirect effect jointly mediated by all three mediators.
Longitudinal mediation analyses were conducted using the Hayes PROCESS macro (model 4, version 3.5) [38] in IBM SPSS for Windows, version 26.0 (IBM Corporation, Armonk, NY, USA). We estimated the direct, indirect, and total effects of early PT on changes in disability and pain intensity, with 5,000 bootstrap resamples used to calculate 95% confidence intervals around these parameter estimates. In this approach, the total effect in each model represents the unstandardized regression coefficient or path (denoted c) from early PT to the outcome after adjusting for covariates. The total effect is the sum of the direct and indirect effects. The direct effect is the unstandardized path (denoted c’) from early PT to the outcome once the putative mediator has been added to the model. The indirect effect (denoted ab) represents the product of the unstandardized regression paths: 1) from early PT to the mediator (a path); and 2) from the mediator to the outcome (b path). The path diagrams for the full mediation models are shown in the Figure 1. All parameter estimates were reported as unstandardized regression coefficients (b). Indirect effects were identified as statistically significant if their 95% confidence intervals did not include zero.
Figure 1.
Path Diagram of the Full (Multiple Mediator) Models Illustrating the Direct Effects and Causal Paths Linking Early PT to Short-Term Outcomes of: (A) Change in Disability from Baseline to 3 Months; and (B) Change in Pain Intensity from Baseline to 3 Months. These two models were also fitted for long-term outcomes of change in disability and pain intensity from 3 months to 1 year. a1, a2, a3= the unstandardized path coefficients of each proposed mediator regressed on early PT (X1, X2, X3), adjusted for covariates; b1, b2, b3= the unstandardized path coefficients of the outcome regressed onto each proposed mediator (M1, M2, M3), adjusted for covariates; c= total effect (unstandardized path coefficient) of early PT on the outcome, adjusted for covariates but without controlling for the mediators. c’= total direct effect (unstandardized path coefficient) of early PT on outcomes, after controlling for covariates and the mediators. The total indirect effect= a1*b1+a2*b2+a3*b3.
Results
Participants
Participants’ baseline characteristics and mediator and outcome scores across assessments by treatment group are presented in Table 1. There were no significant between-group differences in baseline demographics, mediators, or outcomes. Baseline PCS and FABQ-W scores reflected mild levels of catastrophizing thoughts and work-related fear-avoidance across both groups, whereas baseline FABQ-PA scores reflected relatively high levels of physical activity-related fear-avoidance. Participants’ average ODI and NPRS scores at baseline reflected moderate disability and pain intensity across both groups. Early (Δ 4-week), short-term (Δ 3-month) and long-term (Δ 1-year) change scores in pain catastrophizing, fear-avoidance beliefs, disability, and pain intensity are summarized in Table 2.
Table 1.
Summary of patients’ baseline characteristics, mediators and outcomes across assessments by treatment group
| Variable | Early PT Group |
Usual Care Group |
||||
|---|---|---|---|---|---|---|
| (N = 104) |
(N = 100) |
|||||
| M (SD) or n (%) | Min | Max | M (SD) or n (%) | Min | Max | |
| Baseline covariates | ||||||
| Age (years) | 38.10 (10.35) | 19.00 | 60.00 | 36.59 (10.50) | 18.00 | 60.00 |
| Body mass index | 28.66 (7.33) | 17.69 | 52.49 | 29.12 (8.54) | 17.80 | 71.82 |
| Gender | – | – | – | – | ||
| Female | 61 (58.65%) | 45 (45.00%) | ||||
| Male | 43 (41.35%) | 55 (55.00%) | ||||
| Education | ||||||
| ≤ Some college | 46 (44.23%) | – | – | 55 (55.00%) | – | – |
| ≥ College degree | 58 (55.77%) | 45 (45.00%) | ||||
| Marital status | – | – | – | – | ||
| Married/cohabitatingSingle/divorced/widowed | 66 (63.46%) | 60 (60.00%) | ||||
| 38 (36.54%) | 40 (40.00%) | |||||
| Prior history of low-back pain treatment | 70 (67.31%) | – | – | 66 (66.00%) | – | – |
| Mediators (across assessments) | ||||||
| PCS (baseline) | 13.61 (10.74) | 0.00 | 46.00 | 13.51 (9.96) | 0.00 | 40.00 |
| PCS (4 weeks) | 4.80 (7.25) | 0.00 | 42.00 | 6.96 (8.21) | 0.00 | 40.00 |
| PCS (3 months) | 2.89 (5.77) | 0.00 | 39.00 | 4.20 (6.21) | 0.00 | 29.00 |
| PCS (1 year) | 2.95 (5.43) | 0.00 | 36.00 | 3.45 (6.82) | 0.00 | 34.00 |
| FAB-W (baseline) | 11.00 (8.75) | 0.00 | 37.00 | 12.02 (9.78) | 0.00 | 42.00 |
| FAB-W (4 weeks) | 7.52 (8.64) | 0.00 | 42.00 | 9.11 (9.07) | 0.00 | 42.00 |
| FAB-W (3 months) | 4.78 (7.06) | 0.00 | 29.00 | 7.15 (8.23) | 0.00 | 42.00 |
| FAB-W (1 year) | 4.55 (6.73) | 0.00 | 29.00 | 5.48 (9.04) | 0.00 | 42.00 |
| FAB-PA (baseline) | 14.86 (4.90) | 0.00 | 24.00 | 15.32 (4.82) | 0.00 | 24.00 |
| FAB-PA (4 weeks) | 7.12 (5.70) | 0.00 | 24.00 | 8.09 (5.52) | 0.00 | 23.00 |
| FAB-PA (3 months) | 5.04 (5.68) | 0.00 | 20.00 | 5.54 (4.74) | 0.00 | 19.00 |
| FAB-PA (1 year) | 5.32 (5.98) | 0.00 | 23.00 | 5.21 (6.16) | 0.00 | 24.00 |
| Outcomes (across assessments) | ||||||
| ODI (baseline)* | 41.38 (14.11) | 20.00 | 74.00 | 40.43 (12.52) | 20.00 | 72.00 |
| ODI (4 weeks) | 10.86 (11.97) | 0.00 | 70.00 | 13.70 (12.69) | 0.00 | 54.00 |
| ODI (3 months) | 6.29 (8.25) | 0.00 | 36.00 | 8.74 (10.66) | 0.00 | 44.00 |
| ODI (1 year) | 6.47 (10.37) | 0.00 | 54.00 | 7.34 (9.45) | 0.00 | 38.00 |
| NPRS (baseline)* | 5.26 (1.80) | 1.33 | 9.33 | 4.95 (1.88) | 0.33 | 9.67 |
| NPRS (4 weeks) | 1.71 (1.65) | 0.00 | 7.00 | 2.02 (2.00) | 0.00 | 8.00 |
| NPRS (3 months) | 1.34 (1.60) | 0.00 | 7.67 | 1.68 (1.79) | 0.00 | 7.00 |
| NPRS (1 year) | 1.18 (1.59) | 0.00 | 8.33 | 1.20 (1.57) | 0.00 | 7.33 |
SD = Standard deviation. ODI = Oswestry Disability Index (possible scale range= 1–100). NPRS = Numeric Pain Rating Scale (full scale range = 1–10). PCS = Pain Catastrophizing Scale (possible scale range = 0–52). FAB-W = Fear-Avoidance Beliefs Questionnaire (work subscale; possible score range = 0–42). FAB-PA = Fear-Avoidance Beliefs Questionnaire (physical activity subscale; possible score range = 0–24).
Variable was also included as a covariate in regression models.
Table 2.
Summary of raw change scores on the mediators and outcomes across 4 weeks, 3 months, and 1 year for the full 3-month sample (N = 204)
| Variable | Time Period | N | M (SD) | Min | Max |
|---|---|---|---|---|---|
| PCS | Δ 4-week | 204 | −7.70 (9.18) | −33.00 | 19.00 |
| Δ 3-month | 203 | −9.93 (10.15) | −46.00 | 18.00 | |
| Δ 1-year | 196 | −10.02 (10.64) | −46.00 | 27.00 | |
| FAB-W | Δ 4-week | 204 | −3.20 (7.58) | −25.00 | 23.00 |
| Δ 3-month | 203 | −5.52 (8.63) | −30.00 | 22.00 | |
| Δ 1-year | 196 | −6.26 (8.10) | −28.00 | 14.00 | |
| FAB-PA | Δ 4-week | 204 | −7.49 (6.57) | −22.00 | 13.00 |
| Δ 3-month | 204 | −9.80 (6.67) | −24.00 | 10.00 | |
| Δ 1-year | 197 | −9.81 (7.02) | −24.00 | 10.00 | |
| ODI | Δ 4-week | 202 | −28.82 (16.89) | −72.00 | 14.00 |
| Δ 3-month | 204 | −33.42 (16.60) | −74.00 | 6.00 | |
| Δ 1-year | 195 | −34.10 (16.48) | −72.00 | 16.00 | |
| NPRS | Δ 4-week | 204 | −3.25 (2.16) | −7.67 | 7.00 |
| Δ 3-month | 204 | −3.61 (2.19) | −8.00 | 3.00 | |
| Δ 1-year | 197 | −3.89 (2.10) | −8.67 | 1.83 |
SD = standard deviation. PCS = Pain Catastrophizing Scale. FAB-W = Fear-Avoidance Beliefs Questionnaire (work subscale). FAB-PA= Fear-Avoidance Beliefs Questionnaire (physical activity subscale). ODI = Oswestry Disability Index. NPRS = Numeric Pain Rating Scale. Δ 4-week = 4-week change from baseline; Δ 3-month = 3-month change from baseline; Δ 1-year= 1-year change from baseline.
Associations Among Predictors, Mediators, and Outcomes
Changes from baseline to 4 weeks (Δ 4-week) in pain catastrophizing and fear-avoidance beliefs were positively correlated with changes from baseline to 3 months (Δ 3-month) in disability and pain intensity (Table 3). Early changes in PCS and FABQPA (Δ 4-week) were also positively correlated with changes from baseline to 1 year (Δ 1-year) in disability and pain intensity; whereas changes in FABQW (Δ 4-week) were positively correlated with 1-year changes in disability (r (195) = 0.20, P = .005) but were not significantly associated with changes in pain intensity (r (197) = 0.06, P = .438). The three mediators were modestly, positively correlated with one another (rs ranged from 0.13 to 0.30; Ps from 0.074 to <0.001). Results from the univariate ANCOVAs showed that early PT led to significant mean reductions in PCS scores at the 4-week assessment (Early PT group: M difference = −8.78, 95% CI [12]; Usual Care group: M difference = −6.58, 95% CI [−7.87, −5.28]; F(1,201)=5.59; P = .018), after adjusting for baseline PCS scores. Early PT did not significantly reduce FABQW or FABQPA at the 4-week assessment (Ps > .05).
Table 3.
Zero-order correlations among the mediators and outcomes
| 3-Month ODI Change | 1-Year ODI Change | 3-Month NPRS Change | 1-Year NPRS Change | |
|---|---|---|---|---|
| 4-week PCS change | .18* | .24*** | .17* | .22** |
| 4-week FAB-W change | .28*** | .20** | .16* | .06 |
| 4-week FAB-PA change | .30*** | .25*** | .27*** | .16* |
PCS = Pain Catastrophizing Scale. FAB-W = Fear-Avoidance Beliefs Questionnaire (work subscale). FAB-PA = Fear-Avoidance Beliefs Questionnaire (physical activity subscale). ODI = Oswestry Disability Index. NPRS = Numeric Pain Rating Scale.
P < .05.
P <.01.
P <.001.
Indirect Effects of 4-Week Reductions in Pain Catastrophizing and Fear-Avoidance Beliefs on Outcomes at 3 Months and 1 Year
Disability as the Outcome
In the 3-month disability full mediation model, with the three potential mediators tested jointly, the total, indirect and direct effects were, respectively: −3.06 (95% CI: −5.57, −0.55), P = .017; −0.72 (95% CI: −1.88, 0.37); and −2.34 (95% CI: −4.71, 0.04), P = 0.054 (Table 4). Early PT’s treatment effect (a 3-point difference on the ODI) was small and did not reach the minimum clinically significant difference, which was defined as a 6-point difference on the ODI [21]. The magnitude of the point estimate of the indirect effect is 23.5% as large as the total effect, suggesting that a relatively small portion of the total effect of early PT treatment on disability was mediated by the three potential mediators. In the full mediation model, early PT led to significant 4-week reductions in PCS (i.e., significant a1 path in Figure 1, panel A: a1= −2.33 (95% CI: −4.23, −0.44, P = 0.016), and changes in FABQW and FABQPA were prognostic for improved disability over 3 months (i.e., significant b2 and b3 paths in Figure 1, panel A; b2= 0.29 (95% CI: 0.08, 0.50), P = .003; and b3= 0.42 (95% CI: 0.16, 0.70), P = .002). There were significant relationships between the 3-month change in disability and the three individual mediators when each mediator was considered separately (Table 4). These relationships translated into similar, but relatively small point estimates for the indirect effects, which ranged from −0.46 to −0.56 ODI points across the three mediators. Only the estimated indirect effect through the 4-week change in PCS reached significance (indirect effect: b= −0.56 (95% CI: −1.48, −0.04). In this model, early PT led to a greater reduction in disability from baseline to 3 months than usual care, and 4-week reductions in PCS mediated 16% of the treatment effect, supporting mediation. In the two models evaluating fear-avoidance beliefs individually as mediators, early PT did not lead to significant reductions in either factor, but 4-week changes in FABQW and FABQPA remained significant predictors of disability improvement over 3 months. In the 1-year disability change models (as shown in Electronic Supplementary Material 1), early PT was not significantly associated with reductions in disability from baseline to 1 year, and results did not support mediation by 4-week changes in any of the psychosocial variables.
Table 4.
Longitudinal mediation analyses of early PT’s effect on 3-month changes in disability and pain intensity by 4-week changes in pain catastrophizing and fear-avoidance beliefs: unstandardized path regression coefficients
| Model | N | Total Effect | Indirect Effect | Direct Effect | Path a | Path b |
|---|---|---|---|---|---|---|
| c [95% CI] | a*b [95% CI] | c’ [95% CI] | a1–a3 [95% CI] | b1–b3 [95% CI] | ||
| 3-month disability (ODI) change | ||||||
| Δ 4-week PCS | 206 | −3.46** | −0.56 | −2.90* | −2.46* | 0.23* |
| [−6.05, −0.87] | [−1.48, −0.04]+ | [−5.50, −0.29] | [−4.26, −0.63] | [0.04, 0.47] | ||
| Δ 4-week FAB-W | 206 | −3.27* | −0.55 | −2.71* | −1.25 | 0.44*** |
| [−5.78, −0.75] | [−.148, 0.34] | [−5.11, −0.32] | [−3.08, 0.73] | [0.25, 0.65] | ||
| Δ 4-week FAB-PA | 208 | −3.87** | −0.46 | −3.41** | −0.88 | 0.53*** |
| [−6.47, −1.28] | [−1.36, 0.34] | [−5.89, −0.93] | [−2.39, 0.63] | [0.30, 0.75] | ||
| Full model | 204 | −3.06* | −0.72 | −2.34 | – | – |
| [−5.57, −0.55] | [−1.88, 0.37] | [−4.71, 0.04] | ||||
| 3-month pain intensity (NPRS) change | ||||||
| Δ 4-week PCS | 206 | −0.52* | −0.11 | −0.41 | −2.48* | 0.05** |
| [−0.96, −0.08] | [−0.29, −0.01]+ | [−0.85, 0.03] | [−4.30, −0.67] | [0.01, 0.09] | ||
| Δ 4-week FAB-W | 206 | −0.48* | −0.09 | −0.40 | −1.20 | 0.07 *** |
| [−0.92, −0.04] | [−0.24, 0.04] | [−0.82, 0.03] | [−3.08, 0.62] | [0.04, 0.10] | ||
| Δ 4-week FAB-PA | 208 | −0.56* | −0.09 | −0.47* | −0.88 | 0.10*** |
| [−1.00, −0.12] | [−0.25, 0.06] | [−0.89, −0.06] | [−2.38, 0.63] | [0.06, 0.14] | ||
| Full model | 204 | −0.49* | −0.14 | −0.35 | – | – |
| [−0.93, −0.05] | [−0.36, 0.04] | [−0.77, 0.07] | ||||
CI = Confidence interval. Models include covariates of age, body mass index, gender, education, marital status, and prior LBP treatment.
b = unstandardized beta regression coefficient. 95% CI = bootstrap estimates of the upper and lower limits of the 95% confidence interval around each estimate. Full model = Multiple mediator model with 4-week changes in pain catastrophizing and work- and physical activity-related fear-avoidance beliefs. Δ 4-week PCS = Single mediator model with 4-week changes in pain catastrophizing. Δ 4-week FAB-W = Single mediator model with 4-week changes in work-related fear-avoidance beliefs. Δ 4-week FAB-PA = Single mediator model with 4-week changes in physical activity-related fear-avoidance beliefs.
P < .05.
P < .01.
P <.001. + Bootstrap estimate of 95% confidence interval does not include zero.
Pain Intensity as the Outcome
In the 3-month pain intensity full mediation model with the three mediators tested jointly, the total, indirect, and direct effects were: −0.49 (95% CI: −0.93, −0.05), P = .028; −0.14 (95% CI: −0.36, 0.04); and −0.35 (95% CI: −0.77, 0.07), P = .098, respectively (Table 4). Again, the total effect of early PT relative to usual care was small and not clinically impactful. The magnitude of the indirect effect suggests that the three potential mediators explained 28.6% of the total effect of early PT on 3-month changes in pain intensity, reflecting a small effect. In the full mediation model, early PT led to significant 4-week reductions in PCS (i.e., significant a1 path in Figure 1, panel B: a1= −2.32, 95% CI: −4.17, −0.39, P = .016), and 4-week change in FABQW and FABQPA predicted 3-month improvements in pain intensity (i.e., significant b2 and b3 paths in Figure 1, panel B; b2= 0.04 (95% CI: 0.0004, 0.08), P = .022; and b3= 0.07 (95% CI: 0.02, 0.12), P = .003; respectively). When each mediator was evaluated separately, there were significant relationships between 3-month changes in pain intensity and each mediator (Table 4). The indirect effect estimates across the three mediators ranged from −0.09 to −0.11 NPRS point differences, suggesting that the three factors had similar, but small effects. The only indirect effect point estimate to reach significance in these single mediation models was through the 4-week change in PCS (b= −0.11, 95% CI: −0.29, −0.01). In this model, the indirect effect of 4-week reductions in PCS explained 22% of the total effect of early PT on 3-month changes in pain intensity. In the 1-year pain intensity change models (as shown in Electronic Supplementary Material 1), early PT was not associated with reductions in pain from baseline to 1 year, and results did not support mediation by 4-week changes in any of the psychosocial variables.
Discussion
We found small and statistically significant total effects of early PT on 3-month changes in disability and pain intensity, with the early PT group showing greater reductions in ODI and NPRS scores than the usual care group. These results are consistent with previously reported results from intention-to-treat analyses [21]. The multiple mediator model in this study suggested that the joint indirect effects of pain catastrophizing and fear-avoidance beliefs about work and physical activity on the 3-month outcomes were relatively small (−0.72 for change in disability/ODI scores, and −0.14 for change in pain intensity/NPRS ratings). In the single mediator analyses, these effects again translated into small indirect effects, which were similar across the three mediators. Four-week reductions in pain catastrophizing mediated the association between early PT and 3-month improvements in disability and pain intensity in the single mediator models. Although the upper bounds of the bootstrap-estimated confidence intervals around the indirect effects were close to including zero (−0.04 and −0.01, respectively), these reductions explained 16% and 22% of early PT’s association with 3-month improvements in disability and pain intensity, respectively. We also demonstrated that early PT reduced patients’ pain catastrophizing in the first four weeks after initial consultation with a primary care provider in both models. Conversely, early PT did not reduce patients’ fear-avoidance beliefs about work or physical activity. Independent of early PT intervention, however, reductions in fear-avoidance beliefs remained prognostic for outcome improvements.
The proposed mediation models were guided by the Fear-Avoidance Model of chronic pain (FAM) and prior empirical work supporting the mediating role of these factors on patient-centered outcomes [10–12, 14]. The present study advances the field’s understanding of how early PT affects patients’ pain-related outcomes in acute LBP, through: 1) consideration of multiple, psychosocial mediating pathways jointly and independently; and 2) an assessment of the relations among changes across the initial treatment period and short- and longer-term follow-ups in the mediators and outcomes.
The statistically significant total effect of early PT on both the primary and secondary outcomes at 3 months but not at 1 year, supports trial results [20, 21] and other observational work reporting on the clinical benefits of early PT in acute LBP [27, 39]. Although early PT’s treatment effects were small, this study’s point estimates were similar to others reported for PT for acute LBP [24]. Further, although the traditional causal steps approach to mediation from Baron and Kenny posits that mediation analysis should not proceed in the absence of a significant total or direct effect, recent frameworks suggest that significance is not a necessary condition for such tests, and mediational processes can still be identified [40]. Many factors impact significance (e.g., sample size, a stronger treatment effect on a mediator as compared to the outcome, the presence of other, uncontrolled mediators). We therefore built upon the extant work by demonstrating that the indirect effect of early PT on 3-month changes in disability and pain via the joint mediating paths of 4-week reductions in pain catastrophizing and fear-avoidance beliefs about work and physical activity was small, if present, and not maintained over 1 year. Although the indirect effect of early PT on both outcomes mediated by the path of 4-week change in pain catastrophizing was nominally significant when considered individually, this effect was similarly small.
The proportion of early PT’s effect on the outcomes explained by changes in pain catastrophizing is consistent with other studies in chronic LBP treatment. For example, studies of multi-disciplinary treatments that target patients’ cognitions (e.g., combined PT and cognitive-behavioral therapy) have also supported mediation [11, 41, 42], with pain catastrophizing reductions explaining 34% of the treatment’s association with pain [41] in one study, and 37% of the treatment’s association with disability in another [11]. Still, it is possible that other factors play more of a mechanistic role in explaining how early PT improves patients’ functioning. Recent work has highlighted the potential role of modifiable resilience factors (e.g., positive affect, self-regulation, self-efficacy, and coping behaviors) in explaining variability in patients’ pain and disability [43, 44]. In one study, changes in mood resulting from an affective intervention for experimentally-induced acute pain reduced participants’ pain catastrophizing, which in turn reduced pain intensity [45]. Although some resilience factors have been found to impact psychological risks within the FAM, resilience and risk processes can also be distinct. Individuals may experience high or low levels of both risk and resilience factors concurrently; high pain catastrophizing and fear-avoidance levels do not imply low resilience [43]. Although a study among primary care patients with chronic LBP did not find resilience and coping resources to be prognostic for patients’ disability over 1 year [46], how these factors operate in the context of early intervention for acute LBP is unknown.
Our finding that early PT led to reductions in pain catastrophizing over 4 weeks and that these reductions mediated early PT’s link with 3-month outcomes in acute LBP aligns with empirical findings on catastrophizing’s mediating role in chronic pain treatments on patients’ functioning [11], even for physical interventions that do not explicitly target cognitions. This finding also suggests that educating and reassuring patients with acute LBP without physical intervention may not be sufficient to address pain catastrophizing early on, and rather, incorporating hands-on physical interventions may reduce patients’ catastrophizing in the first month of an acute LBP episode. An observational study of a pain education intervention for chronic LBP found that although increases in patients’ knowledge reduced pain catastrophizing over 6 months, these reductions did not mediate the effect of knowledge on functioning and pain from 6 months to 1 year [47]. Our finding that early PT’s indirect effect on 3-month outcomes through reductions in pain catastrophizing further reduced in significance in the full model may suggest that changes in fear-avoidance beliefs confounded the effect of early PT on pain catastrophizing.
Our finding that early PT did not impact reductions in fear-avoidance beliefs about physical activity is surprising, given prior work showing reductions in fear-avoidance beliefs about physical activity among patients receiving PT for LBP [48, 49]. Previous studies have included different PT protocols and patient presentations (i.e., acute, subacute, chronic, and LBP with sciatica). The present study focused on patients receiving PT for an acute episode without sciatica. Other work has found that changes in fear-avoidance and other indicators of distress mediate the effect of psychologically informed PT on disability but only among high-risk patients [16]. Thus, early PT’s effect on disability via the mediating pathway of 4-week reductions in fear-avoidance beliefs about physical activity may be contingent upon patients’ initial risk status.
Strengths of this study include its experimental, longitudinal design with multiple time points of mediator and outcome data. Two post-treatment follow-up assessments permitted the evaluation of the proposed temporal associations. The measurement of the analyses’ key constructs at baseline occurred in close proximity in time to symptom onset, which limited bias in patients’ reports and improved our ability to reliably capture their pain experiences. Also, the response rate for the trial’s 1-year assessment was high (90%), with very little attrition. Furthermore, mediation models were adjusted for patients’ baseline scores on mediators and outcomes, residualizing the change scores to adjust for random error. Finally, our a priori analysis plan with theoretically and empirically derived proposed mediation models were strengths.
Limitations of this study are that it included a select cohort of patients, most of whom showed improved functioning over time and relatively low levels of pain catastrophizing. Consequently, the variability in our outcomes of interest may be restricted in this highly acute sample. Another limitation is the potential for violation of mediation analysis assumptions, namely, that there is no unmeasured confounding. This, however, is an untestable assumption and a common concern of almost all mediation analyses. Because the covariates in our analyses were selected carefully and a priori, based on prior literature, and the mediators were also assessed relatively early in follow-up, violating this assumption may have been less likely. It is still possible, however, that we did not adjust for all common causes of the mediators and outcomes. If a confounder was omitted, the most likely direction of bias would have been an underestimation of the direct effect and overestimation of the indirect effect; thus, this study’s conclusion of a relatively small indirect effect is likely robust to this issue of unmeasured confounding. The single mediator analyses not only share the assumption of no uncontrolled confounding, but they also require a more stringent assumption that the mediator being tested is not affected by the other mediators. Therefore, these analyses should be interpreted with caution. Furthermore, the presence of measurement error in the mediators may have biased results in the opposite direction, toward underestimation of the indirect effect [50]. Finally, this was a complete case analysis, which relies on the stringent assumption data are missing completely at random; however, given the study’s low attrition rate, high completion rate of mediator and outcome measures, and careful selection and inclusion of potential confounders, biases due to missing data may be less likely.
Conclusion
Our findings show that a small proportion of early PT’s effects on short-term disability and pain intensity was mediated by early reductions in pain catastrophizing and fear-avoidance beliefs within the first month of initial primary care consultation for nonspecific acute LBP. Results also suggested that physical intervention, beyond education and reassurance, may impact patients’ catastrophic thoughts in the acute stage of LBP.
Supplementary Material
Acknowledgments
This research was supported through the Agency for Healthcare Research and Quality (AHRQ) (R18HS018672; PI: J.M. Fritz).
Footnotes
One patient in the trial was randomly assigned to the early PT group but received the usual care protocol instead of the intervention. This participant was included and analyzed with the usual care group.
Contributor Information
Brittany L Sisco-Taylor, Department of Population Health Sciences, University of Utah, Salt Lake City, Utah, USA.
John S Magel, Department of Physical Therapy and Athletic Training, University of Utah, Salt Lake City, Utah, USA; Department of Physical Therapy, Intermountain Healthcare, Salt Lake City, Utah, USA.
Molly McFadden, Department of Physical Therapy and Athletic Training, University of Utah, Salt Lake City, Utah, USA.
Tom Greene, Department of Population Health Sciences, University of Utah, Salt Lake City, Utah, USA; Department of Internal Medicine, University of Utah, Salt Lake City, Utah, USA.
Jincheng Shen, Department of Population Health Sciences, University of Utah, Salt Lake City, Utah, USA; Department of Internal Medicine, University of Utah, Salt Lake City, Utah, USA.
Julie M Fritz, Department of Physical Therapy and Athletic Training, University of Utah, Salt Lake City, Utah, USA.
Funding source: This research was supported through the Agency for Healthcare Research and Quality (AHRQ) (R18HS018672; PI: J.M. Fritz).
Conflict of interest: Dr. Greene reports grants from National Institutes of Health (NIH) during the conduct of the study; personal fees from Janssen Pharmaceuticals, DURECT Corporation, Pfizer Inc. and Astrazeneca, outside the submitted work; Dr. Fritz reports grants from Agency for Healthcare Research & Quality (R18HS018672; PI: J.M. Fritz), during the conduct of the study. All other authors have no conflicts of interest to report.
Supplementary Data
Supplementary data are available at Pain Medicine online.
References
- 1. Hartvigsen J, Hancock MJ, Kongsted A, et al. ; Group LLBPSW. What low back pain is and why we need to pay attention. Lancet 2018;391(10137):2356–67. [DOI] [PubMed] [Google Scholar]
- 2.Collaborators GDaIIaP. Global, regional, and national incidence, prevalence, and years lived with disability for 310 diseases and injuries, 1990-2015: A systematic analysis for the Global Burden of Disease Study 2015. Lancet 2016;388(10053):1545–602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Fritz JM, Brennan GP, Hunter SJ, Magel JS. Initial management decisions after a new consultation for low back pain: Implications of the usage of physical therapy for subsequent health care costs and utilization. Arch Phys Med Rehabil 2013;94(5):808–16. [DOI] [PubMed] [Google Scholar]
- 4. Ojha HA, Wyrsta NJ, Davenport TE, Egan WE, Gellhorn AC. Timing of physical therapy initiation for nonsurgical management of musculoskeletal disorders and effects on patient outcomes: A systematic review. J Orthop Sports Phys Ther 2016;46(2):56–70. [DOI] [PubMed] [Google Scholar]
- 5. Grotle M, Foster NE, Dunn KM, et al. Are prognostic indicators for poor outcome different for acute and chronic low back pain consulters in primary care? Pain 2010;151(3):790–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Vlaeyen JWS, Kole-Snijders AMJ, Boeren RGB, van Eek H. Fear of movement/(re)injury in chronic low back pain and its relation to behavioral performance. Pain 1995;62(3):363–72. [DOI] [PubMed] [Google Scholar]
- 7. Edwards RR, Dworkin RH, Sullivan MD, Turk DC, Wasan AD. The role of psychosocial processes in the development and maintenance of chronic pain. J Pain 2016;17(9, Suppl):T70–T92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Osman A, Barrios FX, Gutierrez PM, Kopper BA, Merrifield T, Grittmann L. The pain catastrophizing scale: further psychometric evaluation with adult samples. J Behav Med 2000;23(4):351–65. [DOI] [PubMed] [Google Scholar]
- 9. Sullivan MJ, Thorn B, Haythornthwaite JA, et al. Theoretical perspectives on the relation between catastrophizing and pain. Clin J Pain 2001;17(1):52–64. [DOI] [PubMed] [Google Scholar]
- 10. Wertli MM, Eugster R, Held U, Steurer J, Kofmehl R, Weiser S. Catastrophizing-a prognostic factor for outcome in patients with low back pain: A systematic review. Spine J 2014;14(11):2639–57. [DOI] [PubMed] [Google Scholar]
- 11. Smeets RJEM, Vlaeyen JWS, Kester ADM, Knottnerus JA. Reduction of pain catastrophizing mediates the outcome of both physical and cognitive-behavioral treatment in chronic low back pain. J Pain 2006;7(4):261–71. [DOI] [PubMed] [Google Scholar]
- 12. Mansell G, Storheim K, Løchting I, Werner EL, Grotle M. Identification of indirect effects in a Cognitive Patient Education (COPE) intervention for low back pain. Phys Ther 2017;97(12):1138–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Fritz JM, George SZ, Delitto A. The role of fear-avoidance beliefs in acute low back pain: Relationships with current and future disability and work status. Pain 2001;94(1):7–15. [DOI] [PubMed] [Google Scholar]
- 14. Wertli MM, Rasmussen-Barr E, Weiser S, Bachmann LM, Brunner F. The role of fear avoidance beliefs as a prognostic factor for outcome in patients with nonspecific low back pain: A systematic review. Spine J 2014;14(5):816–36.e4. [DOI] [PubMed] [Google Scholar]
- 15. Hill JC, Whitehurst DGT, Lewis M, et al. Comparison of stratified primary care management for low back pain with current best practice (STarT Back): A randomised controlled trial. lancet 2011;378(9802):1560–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Mansell G, Hill JC, Main C, Vowles KE, van der Windt D. Exploring what factors mediate treatment effect: Example of the STarT back study high-risk intervention. J Pain 2016;17(11):1237–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Baez S, Hoch MC, Hoch JM. Evaluation of cognitive behavioral interventions and psychoeducation implemented by rehabilitation specialists to treat fear-avoidance beliefs in patients with low back pain: A systematic review. Arch Phys Med Rehabil 2018;99(11):2287–98. [DOI] [PubMed] [Google Scholar]
- 18. Ramírez-Maestre C, Esteve R, Ruiz-Párraga G, Gómez-Pérez L, López-Martínez A. The key role of pain catastrophizing in the disability of patients with acute back pain. Int J Behav Med 2017;24(2):239–48. [DOI] [PubMed] [Google Scholar]
- 19. Mansell G, Kamper SJ, Kent P. Why and how back pain interventions work: What can we do to find out? Best Pract Res Clin Rheumatol 2013;27(5):685–97. [DOI] [PubMed] [Google Scholar]
- 20. Fritz JM, Lane E, McFadden M, et al. Physical therapy referral from primary care for acute back pain with sciatica: A randomized controlled trial. Ann Intern Med 2021;174(1):8–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Fritz JM, Magel JS, McFadden M, et al. Early physical therapy vs usual care in patients with recent-onset low back pain: A randomized clinical trial. JAMA 2015;314(14):1459–67. [DOI] [PubMed] [Google Scholar]
- 22. Arnold E, La Barrie J, DaSilva L, Patti M, Goode A, Clewley D. The effect of timing of physical therapy for acute low back pain on health services utilization: A systematic review. Arch Phys Med Rehabil 2019;100(7):1324–38. [DOI] [PubMed] [Google Scholar]
- 23. Fritz JM, Childs JD, Wainner RS, Flynn TW. Primary care referral of patients with low back pain to physical therapy: Impact on future health care utilization and costs. Spine 2012;37(25):2114–21. [DOI] [PubMed] [Google Scholar]
- 24. Kuczynski JJ, Schwieterman B, Columber K, Knupp D, Shaub L, Cook CE. Effectiveness of physical therapist administered spinal manipulation for the treatment of low back pain: A systematic review of the literature. Int J Sports Phys Ther 2012;7(6):647–62. [PMC free article] [PubMed] [Google Scholar]
- 25. Fritz JM, Kim M, Magel JS, Asche CV. Cost-effectiveness of primary care management with or without early physical therapy for acute low back pain: Economic evaluation of a randomized clinical trial. Spine 2017;42(5):285–90. [DOI] [PubMed] [Google Scholar]
- 26. Magel J, Fritz JM, Greene T, Kjaer P, Marcus RL, Brennan GP. Outcomes of patients with acute low back pain stratified by the STarT Back screening tool: Secondary analysis of a randomized trial. Phys Ther 2017;97(3):330–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Childs JD, Fritz JM, Flynn TW, et al. A clinical prediction rule to identify patients with low back pain most likely to benefit from spinal manipulation: A validation study. Ann Intern Med 2004;141(12):920–8. [DOI] [PubMed] [Google Scholar]
- 28. Martin R, Waddell G, Klaber Moffett J, Burton K, Main C. The Back Book. London, UK: The Stationary Office; 1996. [Google Scholar]
- 29. Traeger AC, Henschke N, Hübscher M, et al. Estimating the risk of chronic pain: Development and validation of a prognostic model (PICKUP) for patients with acute low back pain. PLoS Med 2016;13(5):e1002019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Sullivan MJL, Bishop SR, Pivik J. The pain catastrophizing scale: Development and validation. Psychol Assess 1995;7(4):524–32. [Google Scholar]
- 31. Waddell G, Newton M, Henderson I, Somerville D, Main CJ. A Fear-Avoidance Beliefs Questionnaire (FABQ) and the role of fear-avoidance beliefs in chronic low back pain and disability. Pain 1993;52(2):157–68. [DOI] [PubMed] [Google Scholar]
- 32. Swinkels-Meewisse EJCM, Swinkels RAHM, Verbeek ALM, Vlaeyen JWS, Oostendorp RAB. Psychometric properties of the Tampa Scale for kinesiophobia and the fear-avoidance beliefs questionnaire in acute low back pain. Man Ther 2003;8(1):29–36. [DOI] [PubMed] [Google Scholar]
- 33. Fairbank JC, Pynsent PB. The Oswestry Disability Index. Spine 2000;25(22):2940–52. [DOI] [PubMed] [Google Scholar]
- 34. Wittink H, Turk DC, Carr DB, Sukiennik A, Rogers W. Comparison of the redundancy, reliability, and responsiveness to change among SF-36, Oswestry disability index, and multidimensional pain inventory. Clin J Pain 2004;20(3):133–42. [DOI] [PubMed] [Google Scholar]
- 35. Childs JD, Piva SR, Fritz JM, Childs JD, Piva SR, Fritz JM. Responsiveness of the numeric pain rating scale in patients with low back pain. Spine 2005;30(11):1331–4. [DOI] [PubMed] [Google Scholar]
- 36. Herr KA, Spratt K, Mobily PR, Richardson G. Pain intensity assessment in older adults: Use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults. Clin J Pain 2004;20(4):207–19. [DOI] [PubMed] [Google Scholar]
- 37. Vickers AJ, Altman DG. Statistics notes: Analysing controlled trials with baseline and follow up measurements. BMJ 2001;323(7321):1123–4. Article.) [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Hayes AF. Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach. New York, NY: Guilford Press; 2013. [Google Scholar]
- 39. Cleland JA, Fritz JM, Kulig K, et al. Comparison of the effectiveness of three manual physical therapy techniques in a subgroup of patients with low back pain who satisfy a clinical prediction rule: A randomized clinical trial. Spine 2009;34(25):2720–9. [DOI] [PubMed] [Google Scholar]
- 40. Rucker DD, Preacher KJ, Tormala ZL, Petty RE. Mediation analysis in social psychology: Current practices and new recommendations. Soc Pers Psychol Compass 2011;5(6):359–71. [Google Scholar]
- 41. Craner JR, Sperry JA, Evans MM. The relationship between pain catastrophizing and outcomes of a 3-week comprehensive pain rehabilitation program. Pain Med 2016;17(11):2026–35. [DOI] [PubMed] [Google Scholar]
- 42. Spinhoven P, ter Kuile M, Kole-Snijders AMJ, Mansfeld MH, den Ouden D-J, Vlaeyen JWS. Catastrophizing and internal pain control as mediators of outcome in the multidisciplinary treatment of chronic low back pain. Eur J Pain 2004;8(3):211–9. [DOI] [PubMed] [Google Scholar]
- 43. Wideman TH, Asmundson GGJ, Smeets RJEM, et al. Rethinking the fear avoidance model: Toward a multidimensional framework of pain-related disability. Pain 2013;154(11):2262–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Slepian PM, Ankawi B, France CR. Longitudinal analysis supports a fear-avoidance model that incorporates pain resilience alongside pain catastrophizing. Ann Behav Med 2020;54(5):335–45. [DOI] [PubMed] [Google Scholar]
- 45. Lefebvre JC, Jensen MP. The relationships between worry, happiness and pain catastrophizing in the experience of acute pain. Eur J Pain 2019;23(7):1358–67. [DOI] [PubMed] [Google Scholar]
- 46. Jegan NRA, Brugger M, Viniol A, et al. Psychological risk and protective factors for disability in chronic low back pain - a longitudinal analysis in primary care. BMC Musculoskelet Disord 2017;18(1):114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Lee H, McAuley JH, Hübscher M, Kamper SJ, Traeger AC, Moseley GL. Does changing pain-related knowledge reduce pain and improve function through changes in catastrophizing? Pain 2016;157(4):922–30. [DOI] [PubMed] [Google Scholar]
- 48. Beneciuk JM, Bishop MD, Fritz JM, et al. The STarT back screening tool and individual psychological measures: Evaluation of prognostic capabilities for low back pain clinical outcomes in outpatient physical therapy settings. Phys Ther 2013;93(3):321–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. George SZ, Fritz JM, Bialosky JE, Donald DA. The effect of a fear-avoidance-based physical therapy intervention for patients with acute low back pain: Results of a randomized clinical trial. Spine 2003;28(23):2551–60. [DOI] [PubMed] [Google Scholar]
- 50. Fritz MS, Kenny DA, MacKinnon DP. The combined effects of measurement error and omitting confounders in the single-mediator model. Multivariate Behav Res 2016;51(5):681–97. Article.) [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.