Acceptance and commitment therapy for patients with chronic pain: A systematic review and meta-analysis on psychological outcomes and quality of life.

Chronic pain is a prevalent and debilitating health issue worldwide. It refers to pain that persists beyond three months and is associated with significant emotional distress or functional impairment, such as anxiety, depression, cognitive impairment, and reduced physical activity [ 1 ]. A national health interview survey in the United States revealed that about 50.2 million adults (20.5%) experienced pain on most or every day [ 2 ]. The prevalence of chronic pain was slightly higher in the European Union, at about 27% [ 3 ]. Chronic pain not only affects a large proportion of the population, but it also imposes many restrictions on the individual [ 2 ]. Chronic pain impairs work performance and increases absenteeism [ 4 – 6 ]. Chronic pain also hinders daily functioning and social participation, as one study found that 40% of individuals with chronic pain had difficulty walking, and 34% had difficulty socializing [ 7 , 8 ]. Moreover, chronic pain is linked to depression, anxiety, and lower quality of life (QOL) [ 9 – 11 ]. In general, chronic pain is one of the most common health challenges globally, with significant direct or indirect adverse consequences for patients, their families, and society [ 5 , 12 , 13 ]. Psychological interventions are essential for chronic pain management [ 11 , 14 ]. Research indicates that attempts to control pain often result in increased pain severity [ 15 , 16 ]. For instance, efforts to manage pain may trigger additional psychological stress, leading to elevated levels of stress hormones, thereby amplifying the perception of pain [ 17 ]. Moreover, excessive focus on pain can lead to an attentional focusing effect, consequently heightening individuals’ sensitivity to painful stimuli [ 18 ]. Conversely, when patients attempt to accept pain, their pain intensity decreases, and their mental health improves, which is one of the aims of Acceptance and Commitment Therapy (ACT) [ 19 , 20 ]. ACT is a contextually focused form of cognitive behavioral psychotherapy that uses mindfulness and behavioral activation to increase patients’ psychological flexibility [ 20 , 21 ]. It helps patients engage in values-based, positive behaviors while experiencing difficult thoughts, emotions, or sensations [ 21 , 22 ]. ACT has six core processes: acceptance, being present, cognitive defusion, self as context, values, and committed action [ 20 ]. Unlike traditional cognitive behavior therapy (CBT), ACT encourages individuals to accept and confront their pain [ 20 ]. Based on this, individuals are motivated to discover their self-worth and pursue positive actions aligned with their values in order to achieve their life goals and values [ 16 , 23 , 24 ]. Therefore, theoretically, ACT can be an effective alternative treatment for patients with chronic pain to enhance their functioning and reduce the impact of chronic pain [ 15 ]. Our literature review shows that multiple systematic reviews have confirmed that clinical personnel using ACT can effectively improve certain indicators of patients with chronic pain, such as anxiety, depression, and pain acceptance [ 23 , 25 , 26 ]. For instance, a systematic review conducted by Hughes and colleagues, encompassing 11 trials, demonstrated that ACT yielded significant moderate to large effects in terms of pain acceptance and psychological flexibility, along with small to moderate effects in functionality, anxiety, and depression, but no significant improvements were observed in QOL [ 25 ]. In another systematic review by Du and associates, which focused on the efficacy of ACT in improving functioning in individuals with chronic pain, 12 randomized controlled trials (RCTs) were included [ 27 ]. This review found that ACT had an immediate positive impact on the functioning of patients with chronic pain, yet the long-term effects remain unclear [ 27 ]. Inês and colleagues focused on technology-based ACT interventions for patients with chronic pain, incorporating 5 RCTs with a total of 746 participants [ 26 ]. They observed that, compared to control groups, technology-based ACT demonstrated moderate effects in pain intervention and pain acceptance, and smaller effects in depression, mindfulness, and psychological flexibility [ 26 ]. However, it has been six years since the systematic review by Hughes and colleagues, which focused on the comprehensive psychological outcomes of chronic pain patients treated with various forms of ACT by clinical practitioners in RCTs [ 25 ]. Since then, numerous new high-quality RCTs on this topic have emerged. These RCTs have made many improvements in research design, refinement of pain types in subjects, and standardization of ACT interventions. For example, Hansen’s 2023 study demonstrated advancements in design, participant engagement, and intervention customization, especially by having endometriosis patient complete homework for 10 weeks, using anonymous IDs to ensure blinding, and employing a 12-week pain diary to improve data quality [ 28 ]. Moreover, by combining Mindfulness-Based Stress Reduction and ACT in the tailored intervention, the research not only enhanced the intervention’s relevance and effectiveness but also offered valuable methodological insights for future ACT research in chronic pain management [ 28 ]. Lina’s 2022 research used an automated web program for individual-level crossover block randomization to ensure fair allocation across study arms [ 29 ]. A blinded third party unconnected to the research conducted the randomization and allocation, while data collectors were kept unaware of participant groups [ 29 ]. Despite the inability to blind participants, the study effectively minimized bias and optimized follow-up through regular online assessments and detailed flowcharts. In general, including these new high-quality studies will also improve the accuracy of our systematic review. In addition, according to Cochrane’s recommendations, if a systematic review on this topic exceeds a certain number of years based on its academic value and practical value, it is also necessary to update it [ 30 , 31 ]. Secondly, many new and effective pain management concepts have emerged in the field of pain management during these six years, such as emphasizing the importance of pain interference indicators [ 32 ]. However, these indicators were missing from Hughes and his colleagues’ systematic review [ 25 ]. Therefore, in this systematic review we have also included pain interference and other indicators as our outcome indicators so that the intervention effects of ACT can match with the latest pain concepts. Finally, we have also conducted a subgroup analysis in this systematic review. The grouping criterion is the time when the outcome indicators were measured after the participants received ACT intervention. We have divided our analysis into two subgroups: post-treatment and three months post-treatment hoping to improve our research quality through this subgroup analysis research design and reduce bias in research results. The primary objective and problem addressed by this study is to conduct an extensive systematic review and meta-analysis to evaluate the efficacy of ACT in patients with chronic pain, particularly in light of recent research advancements and evolving concepts in pain management. This will encompass: Analyzing all new RCTs on ACT interventions in chronic pain, published in English academic journals, especially those within the last six years, and incorporating them into our systematic review. Investigating novel pain intervention indicators, such as pain interference, in the context of ACT interventions in chronic pain patients, to provide evidence-based recommendations for pain management guidelines. Conducting subgroup analyses with a special focus on the effects at different time points post-ACT intervention (immediately after treatment and three months post-treatment). This will aid in understanding the short-term and long-term effects of ACT on chronic pain patients, the sustainability of these effects, and in formulating effective intervention strategies.

A searching of the databases obtained 7260 articles ( Fig 1 ). 2568 duplicates were excluded, 4578 irrelevant articles were excluded after reading the titles and abstracts, and 114 articles were obtained after the initial screening. 93 articles were excluded after reviewing the full text, and 21 articles were finally included [ 28 , 29 , 42 – 60 ]. The selection of the studies is summarized in the PRISMA diagram in Fig 1 . Table 1 presents the characteristics of the selected studies. Of the 21 studies included, 16 were from Europe, 3 from North America, and 2 from Oceania. The overall sample size was 1298, with a mean age ranging from 29.5 to 82.26, and the majority were female. In addition, three studies reported subjects with a specific type of chronic pain, Fibromyalgia [ 49 , 51 , 58 ]. Thirteen studies reported an average chronic pain duration for the sample, ranging from 6.8 to 25.34 years. The forms of ACT intervention included in the study were face-to-face individual intervention, face-to-face group intervention, online intervention, and self-help intervention. Control group interventions were either TAU or WL. Abbreviations: AAQ-II: the German version of the Acceptance and Action Questionnaire-II; AQoL: The Assessment of Quality of Life; BAI: Beck Anxiety Inventory; BDI: Beck Depression Inventory; CES-D: The Centre for Epidemiological Studies Depression Scale; CPAQ: Chronic Pain Acceptance Questionnaire; CMDI: Chicago Multi-scale Depression Inventory; EHP: The Endometriosis Health Profile; EQ-5D: The European Quality of Life; GAD: Generalized Anxiety Disorder Screen; GDS: the Geriatric Depression Scale; HADS: Hospital And Anxiety Depression Scale; HDI: The Henry Ford Hospital Headache Disability Inventory; LSQ: Life Satisfaction Questionnaire; MPI: Multidimensional Pain Inventory; NRS: The Numeric Rating Scale; ODI: Oswestry Disability Index; PASS-20: The Pain Anxiety Symptoms Scale; PDI: Pain Disability Index; PHQ: the Patient Health Questionnaire; PIPS: The Psychological Inflexibility in Pain Scale; PII: the Pain Interference Index; PPIS: the PROMIS Pain Interference short form; QOLI: Quality Of Life Inventory; QIDS: The Quick Inventory of Depressive Symptomatology; RIQ: Romyalgia Impact Questionnaire; RMDQ: the Roland and Morris Disability Questionnaire; SF-36: The Short Form-36 Health Survey; SFMPQ: Short-Form McGill Pain Questionnaire; SLS: Satisfaction with Life Scale; STAI: The Spielberger Trait-State Anxiety Inventory; VAS: Visual Analogue Scales; WSAS: The WSAS-Work and Social Adjustment Scale The RCTs included in this meta-analysis all had high methodological quality ( Fig 2 ). Specifically, all but one of the studies demonstrated random sequence generation [ 48 ]. On allocation concealment, one study was assessed as "high risk of bias" [ 55 ], eight studies were assessed as "unclear risk of bias" [ 42 , 43 , 45 , 46 , 48 , 50 , 53 , 54 ]. For performance bias, all studies were considered "low risk of bias." This is because in ACT intervention studies, it is not possible to blind participants or treatment providers. Moreover, no evidence of serious bias was found. For detection bias, six studies did not explicitly indicate whether they blinded the interviewers and were therefore assessed as "some concerns" [ 47 , 54 , 56 , 57 , 59 , 60 ]. In terms of attrition bias, twenty studies were rated as "low risk of bias" [ 28 , 29 , 46 , 47 , 49 – 51 , 53 , 55 – 57 , 60 ]. In these studies, even if there was a loss of some participants, the rate of loss to follow-up and the reasons were similar and comparable between the experimental group and the control; one study was rated as "high risk of bias" [ 59 ]; the rest were rated as "unclear risk of bias." For reporting bias, the researchers rated all studies as "low risk of bias" because we can retrieve their registration information or protocols or compare the consistency of methods and results in research. At the same time, the systematic review found no clues about selective reporting. For other biases, all studies were considered "unclear risk of bias" because the authors of these studies were usually professionals in the ACT field. Fig 2A and 2B show the results of their measurements. Abbreviations of the author names were used in the summary for 21 RCTs. Table 1 shows the total score of each study in this assessment. All RCTs gave the ACT rationale and an adequate description of their intervention content, intervention environment, and duration of treatment. Only nine studies reported adherence to the manual [ 28 , 29 , 42 , 43 , 49 – 51 , 55 , 60 ]. Only two studies reported ACT training specifically for the trial [ 46 , 56 ]. Thirteen studies documented whether client engagement checks were conducted during the trial [ 28 , 29 , 46 , 47 , 49 , 50 , 52 , 53 , 56 – 60 ]. This meta-analysis aimed to evaluate the clinical efficacy of ACT compared with control conditions (e.g., WL; TAU). The systematic review performed a subgroup analysis on the follow-up data at three months post-treatment, except for QOL. In assessing the impact of ACT post-treatment, our meta-analysis integrated data across eight outcomes, with the results presented in Table 2 , Figs 3 , and 4 . Specifically, we observed the following: Pain Interference: Among 625 participants, there was a significant improvement with an SMD of -0.50 (CI: -0.66 to -0.34, P < 0.001); Functional Impairment: In 638 participants, we noted significant improvements with an SMD of -0.74 (CI: -1.13 to -0.35, P < 0.001); Pain Intensity: Evaluated in 1114 participants, the SMD was -0.37 (CI: -0.57 to -0.17, P < 0.001); Pain Acceptance: Among 1008 participants, the result was positive with an SMD of 0.68 (CI: 0.50 to 0.87, P < 0.001); Psychological Inflexibility: In a group of 548 participants, there was a notable decrease with an SMD of -0.65 (CI: -0.89 to -0.40, P < 0.001); Anxiety: Assessed in 751 participants, the improvement was reflected by an SMD of -0.47 (CI: -0.69 to -0.24, P < 0.001); Depression: Among 983 participants, the analysis showed a significant reduction with an SMD of -0.59 (CI: -0.82 to -0.35, P < 0.001); QOL: In 294 participants, the SMD was 0.43 (CI: 0.12 to 0.74, P < 0.001), favoring ACT. a ACT indicates acceptance and commitment therapy b SMD, standardized mean difference c CI represents Confidence Interval I 2 indicates statistical heterogeneity *indicated significantly favor ACT intervention Across these outcomes, ACT consistently showed favorable results. Notably, there was no significant heterogeneity observed in the outcomes of pain interference, psychological inflexibility, and QOL. The meta-analysis focused on data collected three months post-treatment, with results presented in Table 3 , Figs 3 and 4 . This analysis covered seven specific outcomes, with the following participant numbers and findings: Pain Interference: Analyzed in 251 participants, showing significant improvement with an SMD of -0.50 (CI: -0.75 to -0.26, P < 0.001); Functional Impairment: In 421 participants, a notable improvement was observed with an SMD of -0.85 (CI: -1.32 to -0.39, P < 0.001); Pain Intensity: Evaluated in 365 participants, the SMD was -0.57 (CI: -0.89 to -0.25, P < 0.001); Pain Acceptance: Among 346 participants, the result showed a positive outcome with an SMD of 0.75 (CI: 0.53 to 0.97, P < 0.001); Psychological Inflexibility: With 277 participants, there was a significant reduction with an SMD of -0.51 (CI: -0.75 to -0.26, P < 0.001); Anxiety: Assessed in 397 participants, showing improvement with an SMD of -0.53 (CI: -0.93 to -0.14, P = 0.008); Depression: In 434 participants, the analysis indicated a significant decrease with an SMD of -0.67 (CI: -0.98 to -0.37, P < 0.001). a ACT indicates acceptance and commitment therapy b SMD, standardized mean difference c CI represents Confidence Interval I 2 indicates statistical heterogeneity *indicated significantly favor ACT intervention These results consistently demonstrated the efficacy of ACT across all outcomes. It’s noteworthy that there was no significant heterogeneity observed in the outcomes of pain interference, pain acceptance, and psychological inflexibility. The systematic review plotted funnel plots for the studies that reported pain interference (post-treatment) as an outcome measure to assess publication bias. The results showed that most of the studies were located in the middle and upper part of the funnel plot, and the studies on both sides of the vertical line were roughly symmetrical ( Fig 5 ). Therefore, the systematic review concluded that the risk of publication bias in the current study was low. The systematic review switched between different modes of effect (random effects model and fixed effects model) and found no significant change in the ES. In addition, after each study was excluded in order, the results of the meta-analysis were not significantly changed compared with those before the exclusion, all p < 0.05, which indicates that the results of the meta-analysis were stable.

Chronic pain can profoundly influence various facets of an individual’s life. ACT strives to enable patients to participate in numerous meaningful activities, especially when chronic pain is ineliminable, thereby reducing its adverse effects. Our findings demonstrate that ACT exhibited effects ranging from small to large on all outcome measures among patients with chronic pain, and these effects sustained up to three months post-treatment. However, further evidence is still required to ascertain whether these positive impacts persist across specific pain conditions and more extended periods. We hope that future RCTs will be conducted in more specific chronic pain settings with longer follow-ups to draw more precise and relevant conclusions.

This systematic review was guided by the preferred reporting items for systematic review and meta-analysis protocols (PRISMA) recommendations [ 33 ]. From inception to October 1, 2023, the systematic review systematically searched six databases: the Cochrane Library, Web of Science, PubMed, EMBASE, PsycINFO, and CINAHL. In addition, peer-reviewed manuscripts published in English were included within the systematic review. The search term was a combination of medical subject headings (MeSH) terms along with free terms and was adjusted according to the specific database, mainly including “acceptance and commitment therapy” AND “chronic pain” AND “randomized controlled trial.” S1 Appendix presents the search strategies utilized for each database. Studies were eligible for inclusion if they met the following PICOS elements: (1) P (Population) : Participants were ≥16 years of age and suffered from chronic pain lasting more than three months and beyond [ 34 ]. The type of pain was not limited, but the pain associated with malignancy was excluded [ 35 ]. (2) I (Intervention) : The intervention of qualified studies should be ACT that explicitly includes both acceptance and commitment components [ 25 ]; studies of the intervention focusing on only one of these components would be excluded. Therefore, studies based on mindfulness were not included. (3) C (Control) : The intervention in the control group was treatment as usual (TAU) or waiting list (WL). When the interventions in the control group were other interventions such as relaxation therapy, CBT, or exercise, they were excluded. (4) O (Outcome) : The primary purpose of ACT is to help individuals live more productive and meaningful lives and improve their functional status, not to control symptoms [ 23 ]. Therefore, the systematic review used pain interference and functional impairment as the primary outcomes in this study. Pain intensity, pain acceptance, psychological inflexibility, anxiety, depression, and QOL were used as secondary outcome measures. (5) S (Study Design): RCTs. The researchers used EndNote X9 software to manage the database search results. The first step was to remove duplicate studies using the software. In this study, two independent reviewers (YH and QC) completed the screening, review, and data extraction for studies included in the research. As a preliminary trial, both reviewers independently reviewed 50 papers. During this process, YH included 46 articles, while QC included 45. They reached a consensus on including 45 papers and unanimously excluded 4. This high level of agreement is reflected in a Kappa value of approximately 0.88, indicating a strong consistency in the application of screening criteria between the two assessors. This result confirms the reliability and effectiveness of our screening process. Then, the two independent reviewers (YH and QC) continued to assess the titles, abstracts, and keywords of the remaining studies using predefined inclusion and exclusion criteria, eliminating those that did not meet the standards. Subsequently, these reviewers carefully read the full texts of the studies that passed the initial screening, further excluding any that failed to meet the established criteria. This process resulted in the final selection of studies. Throughout this review process, each reviewer worked independently, and any disagreements were discussed collectively. If a consensus could not be reached, a third reviewer (LY) was consulted to resolve the disputed decisions. The researchers used a prepiloted, standardized form to extract information for each experiment. Information extracted included study characteristics (author, country, year of publication), participant information (age, gender, pain type, pain duration), intervention details (mode of intervention, intervention duration, treatment of the control group), and outcome measures. During this process, if missing data are identified, necessary conversions of the data was conducted or the authors were directly contacted to acquire the data, in order to ensure the quality of the systematic review results. Throughout this stage, the extraction of information was conducted independently by two reviewers (YH and QC). If there was a disagreement in the extracted information, the disagreement was resolved by discussion or a third reviewer (LY). The researchers used the Cochrane risk of bias tool [ 36 ] and part one of the Yates quality rating scale [ 37 ] to assess the methodological quality of the included studies. Two reviewers (YH and QC) independently evaluated the included studies. Discussions or a third reviewer (LY) resolved any discrepancies in the evaluation. The Cochrane Assessment of Bias tool [ 36 ] was used to assess the methodological quality and to capture the risk of bias in RCTs [ 36 ]. Each item was assessed as having low, unclear, or high risk of bias. Items were not summed to obtain a total score, as recommended by the Centre for Reviews and Dissemination CRD [ 38 ]. The Cochrane risk of bias tool [ 36 ] evaluated the included studies in the following seven areas: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), other bias. Part one of the Yates quality rating scale [ 37 ] is a tool used to assess specific sources of bias for psychological RCTs that the Cochrane risk of bias tool [ 36 ] failed to capture. Items were scored (1 or 2 points) according to whether the construct was reported. The tool evaluated the included studies in the following six areas: rationale, treatment duration, treatment manualization, adherence to the manual, therapist training, and client engagement. The researchers used RevMan 5.3 software to conduct a meta-analysis, generate forest plots, and calculate statistics for all data. RevMan is software provided by the Cochrane Collaboration for systematic reviewers. It is mainly used to create and save the protocol or full text of Cochrane systematic reviews, perform meta-analysis on the input data, and display the meta-analysis results in more intuitive formats, such as forest plots, and to update systematic reviews [ 39 ]. The systematic review used chi-square tests to determine if heterogeneity existed between studies. If P > 0.1, and I 2 < 50%, the study was not statistically heterogeneous; if P 50%, the study was statistically heterogeneous. When heterogeneity was significant, sensitivity analysis or subgroup analysis was used to determine the source of heterogeneity. To address the heterogeneity detected, we not only explored its potential sources, such as through subgroup analyses, but also implemented sensitivity analyses to test the robustness of our results. The sensitivity analyses involved switching between random-effects and fixed-effects models to assess the impact of different effect models on the estimation of Effect Sizes (ES). Furthermore, we investigated the stability of the results by sequentially excluding each study, a process that helped us ascertain that no single study exerted undue influence on the overall outcomes. In conducting subgroup analyses, we focused on evaluating the effects of ACT interventions on patients with chronic pain at two critical time points: the end of treatment and three months post-treatment. The selection of these time points was grounded in the emphasis on short-term versus long-term effects noted in prior research, as well as the importance of understanding the persistence of intervention effects. We anticipated that the various outcome measures in patients with chronic pain might exhibit different effects immediately post-treatment and at three months post-treatment. Such variations could reveal insights into the durability and evolving trends of ACT interventions, thereby providing a basis for clinicians to develop intervention strategies. For continuous data, weighted mean differences (WMDs) with 95% confidence interval (CI) were used if each study used the same measurement tool to measure the same outcome indicator. Conversely, standardized mean differences (SMDs) with 95% CI were used. For the purpose of this study, the random effects model was selected within the systematic review to calculate ES [ 40 ]. According to the viewpoint of Cohen, ES < 0.2 is considered a negligible effect, 0.2 ≤ ES < 0.5 is small, 0.5 ≤ ES < 0.8 is medium, and ES ≥ 0.8 is large [ 41 ]. At the same time, the systematic review used RevMan 5.3 to draw a funnel plot to measure its publication bias.

(DOC) (XLSX) (DOCX)