Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Eur J Pain . 2026 Apr 14;30(4):e70275. doi: 10.1002/ejp.70275 Search in PMC Search in PubMed View in NLM Catalog Add to search Does Cognitive Functional Therapy Reduce the Amount of Opioids Dispensed for Chronic Low Back Pain Over 12 Months: A Secondary Analysis From the RESTORE Trial Lena Thiveos Lena Thiveos 1 Department of Health Sciences, Faculty of Medicine, Health and Human Sciences, Maquarie University, Sydney, Australia Find articles by Lena Thiveos 1, ✉ , Peter Kent Peter Kent 1 Department of Health Sciences, Faculty of Medicine, Health and Human Sciences, Maquarie University, Sydney, Australia 2 School of Allied Health, Curtin University, Bentley, Australia Find articles by Peter Kent 1, 2 , Danijela Gnjidic Danijela Gnjidic 3 School of Pharmacy, Sydney University, Sydney, Australia Find articles by Danijela Gnjidic 3 , Natasha C Pocovi Natasha C Pocovi 1 Department of Health Sciences, Faculty of Medicine, Health and Human Sciences, Maquarie University, Sydney, Australia Find articles by Natasha C Pocovi 1 , Anne Smith Anne Smith 2 School of Allied Health, Curtin University, Bentley, Australia Find articles by Anne Smith 2 , Peter O'Sullivan Peter O'Sullivan 2 School of Allied Health, Curtin University, Bentley, Australia Find articles by Peter O'Sullivan 2 , Mark J Hancock Mark J Hancock 1 Department of Health Sciences, Faculty of Medicine, Health and Human Sciences, Maquarie University, Sydney, Australia Find articles by Mark J Hancock 1 Author information Article notes Copyright and License information 1 Department of Health Sciences, Faculty of Medicine, Health and Human Sciences, Maquarie University, Sydney, Australia 2 School of Allied Health, Curtin University, Bentley, Australia 3 School of Pharmacy, Sydney University, Sydney, Australia * Correspondence: Lena Thiveos ( [email protected] ) ✉ Corresponding author. Revised 2026 Feb 18; Received 2025 Aug 18; Accepted 2026 Apr 3; Issue date 2026 Apr. © 2026 The Author(s). European Journal of Pain published by John Wiley & Sons Ltd on behalf of European Pain Federation ‐ EFIC ®. This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc-nd/4.0/ License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. PMC Copyright notice PMCID: PMC13078251 PMID: 41979226 ABSTRACT Background Given the potential harms associated with opioids, interventions that can safely and effectively reduce the persistent use of opioids are important. Biopsychosocial interventions including Cognitive Functional Therapy (CFT) are effective for reducing chronic low back pain (LBP) and associated disability; however, it is unclear if they are effective at reducing or preventing new opioid use. The aim of this study was to investigate whether CFT with or without biofeedback changed the amount of opioid dispensed over 12‐months compared to usual care in people with chronic, non‐specific LBP. Methods This study is a secondary analysis of the RESTORE trial ( n = 492) which randomly allocated participants with chronic LBP to one of three groups (CFT only, CFT‐plus‐biofeedback; usual care). This analysis included 301 participants who consented to their Pharmaceutical Benefits Scheme data being used (CFT only n = 107; CFT‐plus‐biofeedback n = 106; usual care n = 88). The primary outcome was average morphine equivalent dose per day dispensed during each quarter in the first 12 months of the trial. The secondary outcome was the proportion of participants that had any opioids dispensed during each quarter. Results There was only a small percentage of participants taking opioids and we did not find any statistically or clinically significant differences between the two CFT groups and the usual care group ( p ≥ 0.140) for either outcome. Clinical improvements in pain and disability following a biopsychosocial intervention do not necessarily lead to a reduction in opioids dispensed. Interventions targeting opioid reduction may be required to reduce the use of opioids in this population. Significance Statement There is growing evidence that Cognitive Functional Therapy (CFT) is effective for reducing chronic low back pain and associated disability; however, no previous trials have investigated if CFT is also effective at reducing opioid use. This randomised controlled trial found CFT was not effective in reducing opioid use in people with chronic low back pain, despite it having large effects on pain and activity limitation. 1. Introduction Analgesic medications are widely used in the management of low back pain (LBP) (Schreijenberg et al. 2019 ). While some analgesics including non‐steroidal anti‐inflammatory drugs are recommended for people with chronic back pain (Machado et al. 2021 ; Peck et al. 2021 ), there has been long‐standing uncertainty about the value of opioids, which are not currently recommended for people with chronic LBP (WHO 2023 ). Opioids show some short‐term, clinical benefit in reducing pain and disability in people with chronic LBP. Farrar et al. ( 2022 ) found that in a population with chronic non‐cancer pain, clinically important reductions in pain and improvements in function were sustained over a 12‐month titration period (Farrar et al. 2022 ). Despite this, available evidence suggests limited short‐term treatment effects on average and potential harmful effects particularly with higher doses (e.g., dependency, overdose) (Abdel Shaheed et al. 2016 ; Nuckols et al. 2014 ). A recent trial found no significant difference between placebo and opioids for the management of acute low back and neck pain, prompting a recommendation to reduce the prescription of these medicines (Jones et al. 2023 ). Given the potential harms associated with opioids, effective interventions that can reduce the need for opioids, or the persistent, long‐term use of opioids, are important. While the effectiveness of interventions for LBP is typically assessed using the outcomes of pain, disability and measures of patient distress, (Turk and Dworkin 2004 ) an important secondary outcome for trials of people with chronic LBP is analgesic use and particularly opioid use. Zgierska et al. ( 2025 ) found improvements in functional outcomes and reductions in opioid use at 6 and 12 months in people with opioid‐treated chronic LBP who used either mindfulness‐based therapies or cognitive behavioural therapy. Cognitive Functional Therapy (CFT) is a physiotherapist‐led person‐centred biopsychosocial approach for the management of non‐specific LBP, which aims to coach people to self‐manage their condition (O'Sullivan et al. 2018 ). In the 2023 RESTORE trial, (Kent et al. 2023 ) CFT was found to produce large improvements in activity limitation and pain for people with chronic disabling LBP at 12 months, when compared to usual care, which were largely sustained at 36 months, (Hancock et al. 2025 ). Considering these sustained benefits to pain and activity limitation, it seems plausible that the intervention may also be effective in reducing opioid use. A propensity score matched case–control study suggests that CFT provides a similar reduction in opioid use to multi‐disciplinary care, in people with disabling chronic LBP presenting to a secondary pain centre (Vaegter et al. 2020 ). To our knowledge, no randomised controlled trial (RCT) has investigated whether CFT can reduce opioid use when compared to usual care in people with chronic LBP in primary care. Therefore, the aim of this secondary analysis of the RESTORE trial was to investigate whether CFT with or without biofeedback changed the amount of opioid dispensed across a 12‐month intervention period, compared to usual care in a population with chronic, non‐specific LBP. 2. Methods 2.1. Study Design and Setting This study is a secondary analysis of data collected in a three‐arm randomised controlled trial, RESTORE, which aimed to compare the effectiveness and cost‐effectiveness of CFT and CFT with biofeedback to usual care in a population with chronic non‐specific LBP. This secondary analysis was planned after the original trial. A protocol for the current study was published prior to initiating any analyses ( https://osf.io/dn3wx/ ). This secondary analysis used data from those participants ( n = 301) who gave consent for their Pharmaceutical Benefits Scheme data to be used. The full inclusion criteria and methodology of the RESTORE study have been previously reported (Kent et al. 2023 , 2019 ) and are summarised below. The RESTORE study was approved by Curtin University Human Research Ethics Committee (HRE2018‐0062). 2.2. Participants In the original clinical trial, eligible participants were aged ≥ 18 years with chronic LBP lasting more than 3 months, had sought care from a primary care clinician for their LBP for at least 6 weeks previously, had an average pain intensity of 4 or more on a 0–10 numerical pain rating scale, and had at least moderate pain‐related interference with normal work or daily activities measured by item 8 of the 36‐item Short Form Health Survey (Ware Jr 2000 ). Participants were excluded if they had serious spinal pathology (e.g., fracture, infection, or cancer), any medical condition that prevented them being physically active, were pregnant or had given birth within the previous 3 months, inadequate English literacy for the study's questionnaires and instructions, a skin allergy to hypoallergenic tape adhesives, any surgery scheduled within 3 months, or were unwilling to travel to trial sites. The participant flow through the study is displayed in Figure 1 . FIGURE 1. Open in a new tab Study flow chart. N.B. Low dose = up to 10 mg/day morphine dose equivalent, High dose = 10.1 to 123 mg/day. 2.3. Interventions In the original clinical trial, participants meeting the eligibility criteria were randomly allocated to one of three groups (CFT only; CFT‐plus‐movement‐sensor‐feedback; usual care) using a 1:1:1 ratio. The usual care group did not receive any CFT treatment, but rather followed their existing care pathway. Participants in the usual care group were paid a token reimbursement for their time after completion of questionnaires at 3 and 12 months. For both CFT groups, participants received up to 7 treatment sessions over 12 weeks, plus a ‘booster’ session at 26 weeks in physiotherapy clinics in Sydney and Perth, Australia. The initial consultation time was 60 min and follow‐up appointment length ranged from 30 to 40 min. During these sessions, clinicians used a structured approach to address the three key components of the CFT intervention: making sense of pain, exposure with control and lifestyle change. A detailed description of the CFT intervention has been previously reported (O'Sullivan et al. 2018 ). In both intervention groups, ViMove2 movement sensors were worn; however, in the CFT‐only group the sensors were a placebo. In the CFT‐plus‐movement‐sensor‐feedback group, the ViMove2 movement sensors provided live training and assessment to inform clinicians and participants about their patterns, and the device was worn throughout the day to reinforce the functional movement program taught during the treatment session. 2.4. Outcome Measures All outcomes are based on dispensed patient medication prescriptions recorded in the PBS database. While this is a surrogate measure for opioid use, some dispensed opioids may not have been consumed. The PBS data collection contains information on prescription medicines that qualify for a financial subsidy from the Federal government and for which a claim has been processed (AIHW 2024 ). The PBS covers approximately 90% of medicines prescribed in Australia (McManus et al. 1999 ). In this dataset, opioid use was defined as exposure to any medications classified as ‘opioid’ in the Anatomical Therapeutic Chemical (ATC) Classification and PBS schedule, pharmacological level, ‘N02A – Opioids’ which could include Buprenorphine, Codeine, Fentanyl, Oxycodone, Tapentadol and Tramadol (Care 2024 ; WHO 1996 ). We modelled these data as four quarterly time windows (0 to 3 months, > 3 to 6 months, > 6 to 9 months and > 9 to 12 months), with the > 9 to 12 month window being the primary outcome time‐point. We did not have access to baseline Pharmaceutical Benefits Scheme data. The primary outcome was average dispensed morphine equivalent dose per day during each of the quarters (outcome 1). For each participant, the cumulative dispensed morphine equivalent/day dose was summed over each quarter and divided by 90 days to estimate the average dispensed morphine equivalent/day dose. The pre‐planned approach was for this outcome to be modelled as a continuous variable. However, the skewed distribution and excess zero values in the data (with approximately 80% of the cohort being prescribed no opioids in any given quarter) and some extreme outliers with high doses created some instability estimating variance when building the longitudinal analysis models. To manage this, we categorised this variable into three clinically conventional categories: no opioids (0.0 mg morphine equivalents/day), lower dose opioids (0.1 to 10 mg morphine equivalents/day) and higher dose opioids (10.1 to 123 mg/day). More detail of the distribution of morphine equivalent dose is provided in Table S1 . The secondary outcome was the proportion of participants dispensed any opioid (a dichotomous yes/no outcome) during each quarter (outcome 2). A person classified as ‘yes’ could be dispensed one or more types of opioid in a quarter. In the published protocol for this paper, we had planned a third outcome, investigating the proportion of participants who ceased their opioid use in Q4 compared to Q1. Due to the small number of participants dispensed opioids in Q1 ( n = 63) (only these participants had the possibility of ceasing by Q4), and our observation that for many participants their dispensed opioids fluctuated across all four quarters, this outcome was not included as it did not provide a helpful assessment of the effect of the intervention on opioid use. 2.4.1. Sample Size As this was a secondary analysis, we did not conduct a formal sample size calculation. For both outcomes, there were quarterly measures of the sample ( n = 301), creating a dataset of 1204 observations. 2.5. Analysis Analysis followed a modified intention‐to‐treat principle, where all participants who gave permission for their Pharmaceutical Benefits Scheme data to be used were analysed in their randomised group but people who did not give permission were excluded. We compared the baseline characteristic of those who did and did not give permission to access their Pharmaceutical Benefits Scheme data. To estimate between‐treatment group differences in morphine equivalent dose dispensed (outcome 1) across the four quarters, we used an ordinal logistic regression model, using the robust variance estimate that adjusts for within‐cluster correlation of repeated time observations per participant. We included treatment group, time (as categorical variable), and group by time. The main model was adjusted for the baseline use (yes/no) of any pain medication (self‐reported questionnaire) because we observed that this variable was unbalanced across the treatment groups and was also associated with this outcome. We did not stipulate one time window as the primary time point as we were interested in between group differences across the four quarters. Two sensitivity analyses were planned. A pre‐planned sensitivity analysis additionally adjusted for pre‐specified baseline measures (symptom duration, pain intensity, participant sex, self‐reported opioid use (yes/no)) and pain catastrophizing. A second pre‐specified sensitivity analysis adjusting for up to five baseline characteristics that differed between RESTORE participants who provided Pharmaceutical Benefits Scheme data and those that didn't was not performed, as pain catastrophizing was the only additional characteristic that differed, so this variable was simply added to the first sensitivity analysis. The assumption for ordinal logistic regression that the odds ratio equally applies to all levels of category comparisons was confirmed using the Brant Test. To estimate between‐treatment group differences in the proportion of participants being dispensed opioids (outcome 2) across the four quarters, we used a logistic regression model utilising generalised estimating equations to account for within‐person repeated measures, using the same independent variables as were used in the main model for outcome 1. Similarly, the same two sensitivity analyses were conducted as used for outcome 1. All statistical analysis was performed in STATA 18.0 (StataCorp LLC, College Station, Texus USA) and statistical models were bootstrapped resampled 500 times. As widely recommended, all analyses focussed on reporting the size of the effect and its uncertainty (including describing confidence intervals and p ‐values) rather than making judgements only based on an arbitrary p ‐value threshold (Herbert 2019 ; Wasserstein and Lazar 2016 ; Wasserstein et al. 2019 ). 3. Results 3.1. Participant Flow/Characteristics The flow chart of participants included in each of the outcomes is displayed in Figure 1 . Of the 492 participants in RESTORE trial, 301 gave permission for their Pharmaceutical Benefits Scheme data to be accessed and were included in the current study. Their mean age was 49.1 (SD 14.8), and 58.1% were female. Their mean baseline disability (Roland‐Morris Disability Questionnaire, RMDQ) (Baker 2014 ) was 13.4 (SD 5.1), and 65.3% of participants reported taking some form of medication for LBP at study baseline; one‐fifth (21.1%) of participants reported taking opioids at baseline. A comparison of the baseline characteristic of those that did give Pharmaceutical Benefits Scheme permission and those that declined permission is shown in Table 1 . Participants who declined permission were approximately 10 years younger, had a median 2‐year shorter pain episode length and their pain catastrophizing score was 2 points more (on a 0–52‐point Pain Catasrophizing Scale) (Darnall et al. 2017 ). The baseline characteristics of the treatment groups in the current study are shown in Table 2 . TABLE 1. Baseline characteristics of those giving permission for their PBS data to be used and those that declined permission. Gave PBS permission ( n = 301) Declined PBS permission ( n = 191) p Age (years) 49.1 (SD 14.8) 39.2 (SD 72.3) 0.022 Sex (female) 58.1% ( n = 111) 60.1% ( n = 181) 0.657 University education (yes) 49.0% ( n = 146) 51.0% ( n = 97) 0.658 Pain episode length (years) 5.0 (IQR 2.0–14) 3.0 (IQR 1.0–10.0) 0.004 Body Mass Index (kg/m 2 ) 28.8 (SD 7.1) 28.7 (SD 6.8) 0.888 Disability (RMDQ, 0–24 scale) 13.5 (5.1) 13.7 (SD 5.3) 0.682 Pain intensity (1‐item NRS, 0–10 scale) 5.7 (SD 1.7) 5.9 (SD 1.6) 0.150 Pain intensity (3‐item NRS, 0–10 scale) 6.1 (SD 1.6) 6.3 (1.5) 0.406 Pain self‐efficacy (PSEQ, 0–60 scale) 35.5 (SD 13.3) 33.9 (SD 14.6) 0.244 Fear of movement (FABQ, 0–24 scale) 14.6 (SD 5.5) 15.2 (SD 5.9) 0.254 Pain catastrophizing (PCS, 0–52 scale) 21.2 (SD 11.6) 23.8 (SD 11.4) 0.017 Cognitive flexibility (CFS, 12–72 scale) 59.2 (SD 7.1) 58.3 (SD 8.2) 0.187 Opioid use (yes) 21.1% ( n = 62) 15.4% ( n = 29) 0.122 Treatment confidence post‐randomisation (0–10 scale) 2.9 (SD 1.2) 2.9 (SD 1.3) 0.800 Open in a new tab Note: RMDQ, pain intensity 1‐item (average pain intensity over the last 2 weeks); NRS, pain intensity 3‐item (mean score of: average pain intensity over the last 2 weeks, worst pain intensity over the last 2 weeks, pain now); FABQ, physical activity sub‐scale. Abbreviations: CFS = Cognitive Flexibility Scale, FABQ = Fear Avoidance Beliefs Questionnaire (physical activity sub‐scale), NRS = Numeric Rating Scale, PCS = Pain Catastrophizing Scale, PSEQ = Pain Self Efficacy Questionnaire, RMDQ = Roland Morris Disability Questionnaire. TABLE 2. Description of participants' baseline characteristics ( n = 301, those giving permission for PBS access). Usual‐care ( n = 88) CFT‐only ( n = 107) CFT‐biofeedback ( n = 106) Age, years 50.9 (SD 15.9) 48.5 (SD 14.2) 48.4 (SD 14.5) Female 63.6% ( n = 56) 60.8% ( n = 65) 56.6% ( n = 60) BMI, kg/m 2 28.1 (SD 6.6) 29.4 (SD 8.2) 28.9 (SD 6.2) University education (yes) 55.2% ( n = 48) 47.2% ( n = 50) 45.7% ( n = 48) Duration of care‐seeking, years 5.0 (IQR 21.4–11.0) 4.3 (IQR 1.0–13.8) 5.0 (IQR 2.0–11.5) Length of current episode, years 5.0 (IQR 2.0–13.0) 5.0 (IQR 1.1–15.0) 5.0 (IQR 2.0–12.0) Disability (RMDQ) 12.3 (SD 5.4) 13.5 (SD 4.9) 14.4 (SD 4.8) Pain intensity: single‐item, average past 14 days, NRS 5.6 (SD 1.6) 5.9 (SD 1.7) 5.6 (SD 1.8) Pain catastrophizing 19.5 (SD 11.1) 20.8 (SD 11.9) 23.0 (SD 11.5) Treatment confidence post‐randomisation 2.1 (SD 1.3) 3.3 (SD 1.1) 3.1 (SD 1.1) Taking any pain medication for back pain 52.9% ( n = 46) 70.5% ( n = 74) 70.6% ( n = 72) LBP medications taken Opioids 21.6% ( n = 19) 22.4% ( n = 24) 17.9% ( n = 19) Analgesics 29.6% ( n = 26) 33.6% ( n = 36) 29.3% ( n = 31) Anti‐inflammatories 27.3% ( n = 24) 36.5% ( n = 39) 34.9% ( n = 37) Anti‐neuropathic analgesics 10.2% ( n = 9) 5.6% ( n = 6) 11.3% ( n = 12) Open in a new tab Abbreviations: BMI = Body Mass Index, NRS = Numeric Rating Scale, RMDQ = Roland Morris Disability Questionnaire. There was no statistical evidence for between‐group differences in morphine equivalent dose. Compared to usual care, the CFT‐only group's odds ratio for a higher category of dispensed morphine equivalent/day ranged from a low of 0.81 (95% CI 0.37 to a high of 1.74) p = 0.59 to 1.17 (95% CI 0.53 to 2.56) p = 0.24 across the four 3‐month periods. Similarly, for the CFT‐biofeedback group compared to usual care the odds ratios varied from 0.57 (95% CI 0.27 to 1.22) p = 0.15 to 0.94 (95% CI 0.45 to 1.95) p = 0.74. For CFT‐biofeedback group compared to CFT only group the odds ratios ranged from 0.71 (95% CI 0.36 to 1.41) p = 0.34 to 1.44 (95% CI 0.73 to 2.82) p = 0.29. At a whole cohort level, the proportion of all participants that had opioids dispensed was 21% ( n = 63) in Q1, 22% ( n = 66) in Q2, 20% ( n = 60) in Q3 and 20% ( n = 59) in Q4 (Figure S1 ). However, this was associated with a range of different individual opioid use trajectories over time (Table S2 ). Of the 63 participants who were dispensed opioids in Q1, 30% ( n = 19) were not dispensed opioids in Q2 or Q3 and remained ceased, 13% ( n = 8) were not dispensed opioids in one quarter but later were dispensed opioids again, and 56% ( n = 35) had opioids dispensed in all quarters. Of the 238 participants who were not dispensed opioids in Q1, 82% ( n = 196) never had opioids dispensed, 9% ( n = 21) started in a subsequent quarter but had stopped by Q4, and 9% ( n = 22) started in a subsequent quarter and continued with the use of dispensed opioids until Q4. Individuals who were dispensed opioids also moved between low and high doses in an array of trajectories over the four quarters. The results for the main analysis of between‐group differences in the average dispensed morphine equivalent/day dose (outcome 1) are shown in Figure S1 and Table 3 . The estimated odds of a treatment group having a higher category of dispensed morphine equivalent/day dose than the comparison group did vary a little between the main analysis and the sensitivity analysis (Table 3 ), but this did not change the overall interpretation. The unadjusted percentages in morphine equivalent category for each treatment group are shown in Table 4 . TABLE 3. Odds ratios of a treatment group having a higher category of morphine equivalent dose than the comparison group in each quarter. N = 301 CFT‐only versus usual‐care CFT‐biofeedback versus usual‐care CFT‐biofeedback versus CFT‐only Main analysis (adjusted for baseline pain medication use yes/no) Q1 0.80 (95% CI 0.39–1.61) p = 0.529 0.56 (95% CI 0.27–1.18) p = 0.128 0.70 (95% CI 0.35–1.42) p = 0.326 Q2 1.15 (95% CI 0.56–2.34) p = 0.708 0.93 (95% CI 0.45–1.94) p = 0.847 0.81 (95% CI 0.42–1.56) p = 0.532 Q3 0.78 (95% CI 0.36–1.67) p = 0.521 0.64 (95% CI 0.30–1.38) p = 0.258 0.86 (95% CI 0.40–1.68) p = 0.596 Q4 0.62 (95% CI 0.28–1.38) p = 0.242 0.85 (95% CI 0.40–1.80) p = 0.675 1.37 (95% CI 0.67–2.81) p = 0.384 Sensitivity analysis 1 (adjusted for baseline pain medication use (yes/no), LBP episode length (years), biological sex, baseline opioid use (yes/no), treatment confidence post‐randomisation, baseline catastrophizing), age Q1 0.92 (95% CI 0.41–2.09) p = 0.843 0.77 (95% CI 0.32–1.84) p = 0.560 0.84 (95% CI 0.39–1.82) p = 0.658 Q2 1.26 (95% CI 0.49–3.25) p = 0.634 1.27 (95% CI 0.51–3.17) p = 0.608 1.01 (95% CI 0.48–2.14) p = 0.981 Q3 0.73 (95% CI 0.27–1.92) p = 0.519 0.67 (95% CI 0.25–1.75) p = 0.411 0.92 (95% CI 0.38–2.23) p = 0.849 Q4 0.62 (95% CI 0.21–1.82) p = 0.380 1.15 (95% CI 0.41–3.18) p = 0.791 1.86 (95% CI 0.81–4.26) p = 0.141 Open in a new tab Note: Odds < 1 indicate lower odds of a higher dose of opioids for first listed group. Number of participants in each treatment group was usual care 88, CFT only 107, CFT‐biofeedback 106. TABLE 4. Unadjusted percentage of participants whose opioid use was categorised into the three morphine equivalent dose categories based on opioids dispensed, at the treatment group level. Morphine equivalent dose Usual care n = 88 CFT‐only n = 107 CFT‐biofeedback n = 106 Q1 No opioids 77% (68) 77% (82) 83% (88) 0.1–10 mg/day 14% (12) 13% (14) 10% (11) 10.1–123 mg/day 9% (8) 10% (11) 7% (7) Median (IQR) 0 mg/day (0–0) 0 mg/day (0–0) 0 mg/day (0–0) Full range 0–32.2 mg/day 0–123.3 mg/day 0–48.1 mg/day Q2 No opioids 81% (71) 75% (80) 79% (84) 0.1–10 mg/day 13% (11) 15% (16) 13% (14) 10.1–123 mg/day 7% (6) 10% (11) 8% (8) Median (IQR) 0 mg/day (0–0) 0 mg/day (0–0.89) 0 mg/day (0–0) Full range 0–33.3 mg/day 0–66.6 mg/day 0–72.0 mg/day Q3 No opioids 78% (69) 79% (85) 82% (87) 0.1–10 mg/day 14% (12) 10% (11) 10% (11) 10.1–123 mg/day 8% (7) 10% (11) 8% (8) Median (IQR) 0 mg/day (0–0) 0 mg/day (0–0) 0 mg/day (0–0) Full range 0–26.2 mg/day 0–65.3 mg/day 0–61.5 mg/day Q4 No opioids 80% (70) 83% (89) 78% (83) 0.1–10 mg/day 11% (10) 9% (10) 14% (15) 10.1–123 mg/day 9% (8) 7% (8) 8% (8) Median (IQR) 0 mg/day (0–0) 0 mg/day (0–0) 0 mg/day (0–0) Full range 0–28.0 mg/day 0–49.0 mg/day 0–74.7 mg/day Open in a new tab Abbreviation: IQR = interquartile range. The results for the main analysis of between‐group differences in the proportion of participants dispensed any opioid (outcome 2) are shown in Figure 2 and Table 5 . Similar to the results for outcome 1, there was no statistical evidence for differences in proportion across treatment groups at any quarter. The sensitivity analysis also reflected that pattern. The unadjusted percentages in each treatment group are both shown in Figure S1 and Table S3 . FIGURE 2. Open in a new tab Predicted proportions of participants dispensed any opioid (yes/no) by treatment group. TABLE 5. Proportion of participants dispensed any opioid (yes/no) per quarter, n = 301. Quarter CFT‐only versus usual care CFT‐biofeedback versus usual care CFT‐biofeedback versus CFT‐only Main analysis (odds ratios) adjusted for baseline pain medication use (yes/no) 1 0.78 (95% CI 0.38–1.61) p = 0.495 0.55 (95% CI 0.26–1.18) p = 0.126 0.71 (95% CI 0.35–1.44) p = 0.345 2 1.11 (95% CI 0.53–2.32) p = 0.782 0.92 (95% CI 0.43–1.96) p = 0.830 0.83 (95% CI 0.42–1.63) p = 0.586 3 0.75 (95% CI 0.36–1.56) p = 0.437 0.64 (95% CI 0.30–1.37) p = 0.252 0.86 (95% CI 0.42–1.75) p = 0.678 4 0.62 (95% CI 0.29–1.34) p = 0.221 0.90 (95% CI 0.43–1.90) p = 0.782 1.46 (95% CI 0.71–2.97) p = 0.300 Sensitivity analysis 1 (adjusted for baseline pain medication use (yes/no), LBP episode length (years), biological sex, baseline opioid use (yes/no), treatment confidence post‐randomisation, baseline catastrophizing), age 1 0.99 (95% CI 0.40–2.45) p = 0.983 0.77 (95% CI 0.30–1.94) p = 0.578 0.78 (95% CI 0.33–1.83) p = 0.564 2 1.38 (95% CI 0.44–4.35) p = 0.577 1.40 (95% CI 0.46–4.22) p = 0.550 1.01 (95% CI 0.41–2.49) p = 0.981 3 0.75 (95% CI 0.24–2.31) p = 0.616 0.70 (95% CI 0.22–2.25) p = 0.547 0.93 (95% CI 0.34–2.57) p = 0.890 4 0.73 (95% CI 0.21–2.54) p = 0.619 1.37 (95% CI 0.48–4.86) p = 0.628 1.88 (95% CI 0.73–4.82) p = 0.191 Open in a new tab 4. Discussion 4.1. Key Findings We found that CFT only or CFT with biofeedback was not effective in reducing the dispensing of opioids, compared to usual care. This finding contrasts with the large, sustained between‐group reductions in pain and disability when CFT was compared to usual care in the RESTORE trial (Kent et al. 2023 ). Our result of no effect was consistent across both opioid outcomes, all time points and the sensitivity analyses. To our knowledge, this is the first analysis of RCT data to examine whether a CFT intervention reduced the dispensing of opioids and, by extension, the use of opioids in a chronic, musculoskeletal pain population. While CFT is not explicitly designed to reduce opioid use and it was not an explicit target in the RESTORE trial, our hypothesis had been that, with such large clinical effects, positive effects on opioid use may have also been observed. Our findings align with widely reported observations that reductions in opioid use in chronic pain populations are slow and difficult to achieve (Langford, Schneider, et al. 2023 ). Barriers to deprescribing opioids have been described, including fear of withdrawal and worsening or uncontrolled pain (Langford, Schneider, et al. 2023 ). However, in the present study, a reduction in pain was not associated with reduced opioid use, and fear of opioid withdrawal was not assessed. Our finding that reductions in pain and disability did not lead to a reduction in the dispensing of opioids is important and consistent with some previous literature. A recent study aimed at deprescribing opioids found that pain and disability levels were not associated with reduced opioid use (Nielssen et al. 2019 ). Similarly a pilot RCT examining the effect of a meditation‐CBT intervention on pain in participants with opioid‐treated chronic LBP found sustained reductions in pain intensity at 8 and 26 weeks (8‐weeks, mean (95% CI) = 0.9 (0.01 to1.7); 26‐weeks 1.03 (0.2 to 1.9)), but no change in opioid use measured in morphine equivalent dose mg/day (mean (95% CI) 8 weeks = 5.7 (−34.3 to 45.7); 26 weeks = 9.9 (−30.3 to 50.1)) (Zgierska et al. 2016 ). While no previous RCTs have explored the effect of CFT on opioid use, there has been a growing interest in the effectiveness of mindfulness and cognitive behavioural therapies in this space. A recent study compared the effectiveness of mindfulness based therapy and cognitive behavioural therapy for people with opioid‐treated, chronic LBP, and found improvements in opioid use at 6 and 12 months in both groups (Zgierska et al. 2025 ). A previous case–control pilot study by Vaegter et al. ( 2020 ) compared CFT to a multidisciplinary pain intervention. This study found no significant between‐group difference in opioid use (SMD = −0.16 (−0.60 to 0.22)) (Vaegter et al. 2020 ). An important element of that study was that opioid reduction was an explicit target of the multidisciplinary pain management program control group but not the CFT group. Also, while Vaegter et al. ( 2020 ) study did not find a between‐group difference they did report that a proportion of participants in both groups reduced opioid use by the end of the study (absolute pre/post treatment reduction CFT group = 18.2%, multidisciplinary group 27.8%), whereas in our trial there was minimal change in opioid use in either group. This may have been due to the difference in population between both studies, for instance the higher rate of opioid use in the Vaegter study (64% compared to 19% in the present study). There were also distinct differences in the reporting and collection of opioid use data. In the present study, we did not have baseline data on opioid use (i.e., whether opioids were dispensed) from the Pharmaceutical Benefits Scheme database, but rather self‐reported mediation use. Vaegter et al. ( 2020 ) used self‐reported measures for medication consumption outcomes at baseline and follow up time points. 4.2. Implications of the Results In the present study a biopsychosocial intervention (CFT) targeting a range of factors believed to be driving participants' chronic LBP improved their pain and function but not reduce whether opioids were dispensed, suggesting the need for an additional focus on opioid reduction if this is an important outcome. A recent RCT that explicitly targeted opioid reduction found significant reductions in reported opioid use following an online, CBT‐based ‘Pain Course’, compared to a wait list control (Nielssen et al. 2019 ). It is not certain that the positive effects were due to the targeting of opioid reduction but this study provides some support for this approach. Another reason our biopsychosocial intervention may not have reduced opioid use is that it was delivered by physiotherapists alone and did not involve other health professionals. It is outside the scope of practice for physiotherapists in Australia to advise, prescribe, or administer analgesic medications. Recent clinical practice guidelines for the safe deprescribing of opioids provide 11 recommendations, including the gradual tapering of opioids and when available, referral to multidisciplinary care teams and care provision within a biopsychosocial framework (Langford, Lin, et al. 2023 ). This suggests the need for interdisciplinary care, including physician‐guided medication tapering in conjunction with individualised biopsychosocial clinical interventions such as CFT, to most effectively improve outcomes in people with chronic LBP and sustained opioid use. 4.3. Limitations This study has some limitations associated with the data being from secondary analysis of a trial where opioid use was not a planned outcome. We did not have consent for access to Pharmaceutical Benefits Scheme data for the period prior to the study, so we do not have the data on opioid use prior to study commencement. In a future study, it would be ideal if pre‐enrolment Pharmaceutical Benefits Scheme data, for example of the 0 to 6 months prior, were available to establish a true baseline pre‐intervention. Our study also relied on Pharmaceutical Benefits Scheme data that only detailed the dispensing of opioids, and it is not possible to know if or when these opioids were consumed. We also cannot be sure of whether participants were dispensed opioids only for their chronic LBP, or for other co‐morbid pain. Future studies could overcome some of these limitations by collecting self‐reported opioid use specific to LBP during the trial period, in parallel with the use of Pharmaceutical Benefits Scheme data. Many of our study participants did not take any opioids over the period of the study, and future studies where opioid use is a primary outcome should target populations with a higher prevalence of opioid use. The data used in this trial was limited to the participants who consented to the use of their Pharmaceutical Benefits Scheme data. While 301 participants (61%) from the RESTORE trial did provide this consent, as they differed on age, pain episode length, and pain catastrophising, we cannot be certain that the results in this subpopulation generalise to those who did not. There was a 10‐year mean age difference in those who granted Pharmaceutical Benefits Scheme permission (mean age 49.1, SD 14.8) and those who denied Pharmaceutical Benefits Scheme permission (mean age 39.2, SD 72.3), which potentially could have influenced the results. However, the sensitivity analysis included adjustment for age and, as the estimates from the sensitivity analysis are similar to those from the main analysis, they do not change the conclusions. 5. Conclusion Despite large clinical improvements in pain and disability in both CFT groups compared to usual care, there were no statistically significant reductions in opioid dispensed in the CFT groups compared to usual care. Future trials investigating opioid reduction in participants with chronic LBP should consider an interdisciplinary approach, where effective interventions like CFT are provided in conjunction with opioid tapering from general practitioners or pain physicians. Author Contributions This study was designed by P.K., D.G., M.J.H., N.C.P., P.O. and L.T. Data were analysed by P.K. and A.S. The results were critically examined by all authors. L.T. had a primary role in preparing the manuscript, which was edited by P.K., M.H., D.G., P.O., N.C.P. and A.S. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. Supporting information Figure S1: Average adjusted probability of a person in each treatment group being in a category of morphine equivalent dose. Figure S2: Proportion dispensed any opioid at a whole cohort level ( n = 301). Table S1: Distribution of dispensed morphine equivalent/day dose across quarters. Table S2: Description of individual trajectories of opioid use. Table S3: Unadjusted percentage dispensed any opioid at the treatment group level. EJP-30-0-s001.docx (85.2KB, docx) Acknowledgements Open access publishing facilitated by Macquarie University, as part of the Wiley ‐ Macquarie University agreement via the Council of Australasian University Librarians. References Abdel Shaheed, C. , Maher C. G., Williams K. A., Day R., and McLachlan A. J.. 2016. “Efficacy, Tolerability, and Dose‐Dependent Effects of Opioid Analgesics for Low Back Pain: A Systematic Review and Meta‐Analysis.” JAMA Internal Medicine 176, no. 7: 958–968. 10.1001/jamainternmed.2016.1251. [ DOI ] [ PubMed ] [ Google Scholar ] AIHW . 2024. “Pharmaceutical Benefits Scheme data collection.” https://www.aihw.gov.au/about‐our‐data/our‐data‐collections/pharmaceutical‐benefits‐scheme . Baker, A.
2014. “A Study of the Natural History of Back Pain: Part I: Development of a Reliable and Sensitive Measure of Disability in Low‐Back Pain.” In In, 241–244. Springer. 10.1007/978-1-4471-5451-8_59. [ DOI ] [ PubMed ] [ Google Scholar ] Care, D.
2024. “The Pharmaceutical Benefits Scheme: Browse by Body System.” https://www.pbs.gov.au/browse/body‐system?depth=3&codes=n02a#n02a . Darnall, B. D. , Sturgeon J. A., Cook K. F., et al. 2017. “Development and Validation of a Daily Pain Catastrophizing Scale.” Journal of Pain 18, no. 9: 1139–1149. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Farrar, J. T. , Bilker W. B., Cochetti P. T., et al. 2022. “Evaluating the Stability of Opioid Efficacy Over 12 Months in Patients With Chronic Noncancer Pain Who Initially Demonstrate Benefit From Extended Release Oxycodone or Hydrocodone: Harmonization of Food and Drug Administration Patient‐Level Drug Safety Study Data.” Pain 163, no. 1: 47–57. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Hancock, M. , Smith A., O'Sullivan P., et al. 2025. “Cognitive Functional Therapy With or Without Movement Sensor Biofeedback Versus Usual Care for Chronic, Disabling Low Back Pain (RESTORE): 3‐Year Follow‐Up of a Randomised, Controlled Trial.” Lancet. Rheumatology 7, no. 11: e789. 10.1016/S2665-9913(25)00135-3. [ DOI ] [ PubMed ] [ Google Scholar ] Herbert, R.
2019. “Research Note: Significance Testing and Hypothesis Testing: Meaningless, Misleading and Mostly Unnecessary.” Journal of Physiotherapy 65, no. 3: 178–181. 10.1016/j.jphys.2019.05.001. [ DOI ] [ PubMed ] [ Google Scholar ] Jones, C. M. P. , Day R. O., Koes B. W., et al. 2023. “Opioid Analgesia for Acute Low Back Pain and Neck Pain (The OPAL Trial): A Randomised Placebo‐Controlled Trial.” Lancet (British Edition) 402, no. 10398: 304–312. 10.1016/S0140-6736(23)00404-X. [ DOI ] [ PubMed ] [ Google Scholar ] Kent, P. , Haines T., O'Sullivan P., et al. 2023. “Cognitive Functional Therapy With or Without Movement Sensor Biofeedback Versus Usual Care for Chronic, Disabling Low Back Pain (RESTORE): A Randomised, Controlled, Three‐Arm, Parallel Group, Phase 3, Clinical Trial.” Lancet 401, no. 10391: 1866–1877. [ DOI ] [ PubMed ] [ Google Scholar ] Kent, P. , O'Sullivan P., Smith A., et al. 2019. “RESTORE—Cognitive Functional Therapy With or Without Movement Sensor Biofeedback Versus Usual Care for Chronic, Disabling Low Back Pain: Study Protocol for a Randomised Controlled Trial.” BMJ Open 9, no. 8: e031133. 10.1136/bmjopen-2019-031133. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Langford, A. V. , Lin C. C. W., Bero L., et al. 2023. “Clinical Practice Guideline for Deprescribing Opioid Analgesics: Summary of Recommendations.” Medical Journal of Australia 219, no. 2: 80–89. 10.5694/mja2.52002. [ DOI ] [ PubMed ] [ Google Scholar ] Langford, A. V. , Schneider C. R., Lin C.‐W. C., et al. 2023. “Patient‐Targeted Interventions for Opioid Deprescribing: An Overview of Systematic Reviews.” Basic & Clinical Pharmacology & Toxicology 133, no. 6: 623–639. 10.1111/bcpt.13844. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Machado, G. C. , Abdel‐Shaheed C., Underwood M., and Day R. O.. 2021. “Non‐Steroidal Anti‐Inflammatory Drugs (NSAIDs) for Musculoskeletal Pain.” BMJ (Online) 372: n104. 10.1136/bmj.n104. [ DOI ] [ PubMed ] [ Google Scholar ] McManus, P. , Mant A., Birkett D., Hemming M., and Lindner J.. 1999. “Examining the Adequacy of Quantities Available for Subsidized Antidepressant Prescriptions in Australia.” Pharmacoepidemiology and Drug Safety 8, no. 3: 191–195. 10.1002/(SICI)1099-1557(199905/06)8:3<191::AID-PDS417>3.0.CO;2-X. [ DOI ] [ PubMed ] [ Google Scholar ] Nielssen, O. , Karin E., Staples L., et al. 2019. “Opioid Use Before and After Completion of an Online Pain Management Program.” Journal of Consulting and Clinical Psychology 87, no. 10: 904–917. 10.1037/ccp0000407. [ DOI ] [ PubMed ] [ Google Scholar ] Nuckols, T. K. , Anderson L., Popescu I., et al. 2014. “Opioid Prescribing: A Systematic Review and Critical Appraisal of Guidelines for Chronic Pain.” Annals of Internal Medicine 160, no. 1: 38–47. 10.7326/0003-4819-160-1-201401070-00732. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] O'Sullivan, P. B. , Caneiro J. P., and O'Keeffe M.. 2018. “Cognitive Functional Therapy: An Integrated Behavioral Approach for the Targeted Management of Disabling Low Back Pain (Vol 98, Pg 408, 2018).” Physical Therapy 98, no. 10: 903. 10.1093/ptj/pzy087. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Peck, J. , Urits I., Peoples S., et al. 2021. “A Comprehensive Review of Over the Counter Treatment for Chronic Low Back Pain.” Pain and therapy 10, no. 1: 69–80. 10.1007/s40122-020-00209-w. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Schreijenberg, M. , Koes B. W., and Lin C.‐W. C.. 2019. “Guideline Recommendations on the Pharmacological Management of Non‐Specific Low Back Pain in Primary Care ‐ Is There a Need to Change?” Expert Review of Clinical Pharmacology 12, no. 2: 145–157. 10.1080/17512433.2019.1565992. [ DOI ] [ PubMed ] [ Google Scholar ] Turk, D. C. , and Dworkin R. H.. 2004. “What Should Be the Core Outcomes in Chronic Pain Clinical Trials?” Arthritis Research & Therapy 6, no. 4: 151–154. 10.1186/ar1196. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Vaegter, H. B. , Ussing K., Johansen J. V., et al. 2020. “Improvements in Clinical Pain and Experimental Pain Sensitivity After Cognitive Functional Therapy in Patients With Severe Persistent Low Back Pain.” Pain Reports 5, no. 1: E802. 10.1097/PR9.0000000000000802. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Ware, J. E., Jr.
2000. “SF‐36 Health Survey Update.” Spine 25, no. 24: 3130–3139. [ DOI ] [ PubMed ] [ Google Scholar ] Wasserstein, R. L. , and Lazar N. A.. 2016. “The ASA Statement on p‐Values: Context, Process, and Purpose.” American Statistician 70, no. 2: 129–133. 10.1080/00031305.2016.1154108. [ DOI ] [ Google Scholar ] Wasserstein, R. L. , Schirm A. L., and Lazar N. A.. 2019. “Moving to a World Beyond ‘ p < 0.05’.” American Statistician 73, no. 1: 1–19. 10.1080/00031305.2019.1583913. [ DOI ] [ Google Scholar ] WHO . 1996. “Guidelines for ATC Classification and DDD Assignment.” WHO . 2023. WHO Guideline for Non‐Surgical Management of Chronic Primary Low Back Pain in Adults in Primary and Community Care Settings. World Health Organization. [ PubMed ] [ Google Scholar ] Zgierska, A. E. , Burzinski C. A., Cox J., et al. 2016. “Mindfulness Meditation and Cognitive Behavioral Therapy Intervention Reduces Pain Severity and Sensitivity in Opioid‐Treated Chronic Low Back Pain: Pilot Findings From a Randomized Controlled Trial.” Pain Medicine (Malden, Mass.) 17, no. 10: 1865–1881. 10.1093/pm/pnw006. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Zgierska, A. E. , Edwards R. R., Barrett B., et al. 2025. “Mindfulness vs Cognitive Behavioral Therapy for Chronic Low Back Pain Treated With Opioids: A Randomized Clinical Trial.” JAMA Network Open 8, no. 4: e253204. [ 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. Supplementary Materials Figure S1: Average adjusted probability of a person in each treatment group being in a category of morphine equivalent dose. Figure S2: Proportion dispensed any opioid at a whole cohort level ( n = 301). Table S1: Distribution of dispensed morphine equivalent/day dose across quarters. Table S2: Description of individual trajectories of opioid use. Table S3: Unadjusted percentage dispensed any opioid at the treatment group level. EJP-30-0-s001.docx (85.2KB, docx) Articles from European Journal of Pain (London, England) are provided here courtesy of Wiley ACTIONS View on publisher site PDF (861.3 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top