Process-Outcome Associations in an Interdisciplinary Treatment for Chronic Pain and Comorbid Mental Disorders Based on Acceptance and Commitment Therapy

Abstract

Objective

Numerous studies support the effectiveness of acceptance and commitment therapy (ACT) for chronic pain, yet little research has been conducted about its underlying mechanisms of change, especially regarding patients with comorbid mental disorders. The present investigation addressed this issue by examining associations of processes targeted by ACT (pain acceptance, mindfulness, psychological flexibility) and clinical outcomes (pain intensity, somatic symptoms, physical health, mental health, depression, general anxiety).

Subjects

Participants were 109 patients who attended an ACT-based interdisciplinary treatment program for chronic pain and comorbid mental disorders in a routine care psychiatric day hospital.

Methods

Pre- to posttreatment differences in processes and outcomes were examined with Wilcoxon signed-rank tests and effect size r. Associations between changes in processes and changes in outcomes were analyzed with correlation and multiple regression analyses.

Results

Pre- to posttreatment effect sizes were mostly moderate to large (r between 0.21 and 0.62). Associations between changes in processes and changes in outcomes were moderate to large for both, bivariate correlations (r between 0.30 and 0.54) and shared variances accounting for all three processes combined (R² between 0.21 and 0.29).

Conclusion

The present investigation suggests that changes in pain acceptance, mindfulness, and psychological flexibility are meaningfully associated with changes in clinical outcomes. It provides evidence on particular process-outcome associations that had not been investigated in this way before. The focus on comorbid mental disorders informs clinicians about a population of chronic pain patients who often have a severe course of illness and have seldom been studied.

Introduction

Chronic pain is a complex and highly distressing condition [1]. It is widely disseminated throughout the population [2–4] and associated with extensive societal costs [4–6]. According to the Global Burden of Disease Study, it is the most important contributor to morbidity and disability worldwide [7]. Chronic pain is often intertwined with severe mental health conditions, such as addiction [8], affective disorders [9], psychological trauma [10], and personality psychopathology [11]. Mental disorders increase the risk of developing chronic pain and vice versa, while their comorbidity usually leads to mutual negative effects on the courses of illness [12]. Consequently, chronic pain treatments are preferably interdisciplinary and typically include psychological interventions [13–15]. Given the overlap of chronic pain and mental dysfunction, calls for deeper integration of psychiatry and pain management procedures are increasing [16].

Acceptance and commitment therapy (ACT) is a promising psychological treatment with increasing research support for chronic pain [17] and a broad range of mental disorders [18]. ACT is a contextual, transdiagnostic, process-oriented, third-wave cognitive behavior therapy approach [19]. Its objective is to increase psychological flexibility via six core processes: acceptance, contact with the present moment, cognitive defusion, self-as-context, values, and committed action [19]. The first four processes are also referred to as mindfulness [20]. ACT for chronic pain does not rely primarily on the goal of pain reduction or control but on pain acceptance and values-based actions that contribute to a more meaningful life despite experiencing pain [21]. Meta-analytic evidence supports its effectiveness for chronic pain as single therapy [22] and as part of interdisciplinary programs [23]. However, its mechanisms of change are still not well understood [17], especially for chronic pain patients with comorbid mental disorders.

Addressing mechanisms of change, a series of studies on ACT-based interdisciplinary treatments for chronic pain generated important findings on process-outcome associations [24–31], determining small to large associations between changes in outcomes (pain intensity, depression, pain-related anxiety, physical health, mental health) and changes in processes. In these studies, the most investigated processes were pain acceptance [24–31] and values-based or committed action [24–30]. Whereas associations between changes in these processes and changes in outcomes are well studied, evidence for other vital processes is lacking. For example, mindfulness is considered crucial for alleviating pain-related suffering [32], yet it was examined in only one process-outcome study [27]. A similar argument applies to psychological flexibility [33], which has been investigated in only two studies of this kind, including the aforementioned study on mindfulness [27] and another study with patients aged 65 years and older [29]. It is further noteworthy that the study on mindfulness and psychological flexibility [27] examined associations of changes that referred to different periods: pre- to posttreatment changes in processes were correlated with pretreatment to follow-up changes in outcomes. While this methodological approach allows predictions at follow-up, mechanisms that occur within treatment are not explicitly covered. Finally, previous studies [24–29, 31] also analyzed associations of combined changes in several processes and changes in outcomes via multiple regression analyses, yielding moderate to large shared variances. However, the particular combination of pain acceptance, mindfulness, and psychological flexibility has not yet been investigated in this context.

Furthermore, it is noteworthy that previous process-outcome studies [24–31] included a broad range of chronic pain patients, while an important subpopulation, those with comorbid mental disorders, has not been explicitly examined in terms of process-outcome associations so far. As this subpopulation’s pain suffering is particularly constituted by psychological factors, mechanisms of change in treating these individuals might be different compared with chronic pain patients who do not suffer from severe mental health conditions. Therefore, evidence in this regard seems warranted and may inform clinicians about a group of difficult-to-treat patients with an often severe course of illness.

The present study addressed the aforementioned issues by examining associations between pre- to posttreatment changes in processes (pain acceptance, mindfulness, psychological flexibility) and pre- to posttreatment changes in outcomes (pain intensity, somatic symptoms, physical health, mental health, depression, general anxiety) in patients with chronic pain and comorbid mental disorders. In accordance with a recent meta-analysis [23], it was assumed that pre- to posttreatment differences in processes and outcomes had at least moderate effect sizes. Based on existing evidence [24–31], it was hypothesized that changes in processes had small to large associations with changes in outcomes. It was further hypothesized that the combined changes in the three investigated processes shared moderate to large variances with changes in outcomes. The present investigation is the first ACT study on process-outcome associations to focus on patients with chronic pain and comorbid mental disorders. Furthermore, it is the first ACT study on chronic pain to associate pre- to posttreatment changes in outcomes with pre- to posttreatment changes in mindfulness and psychological flexibility (for patients younger than 65 years), as well as the first to examine changes in the particular combination of pain acceptance, mindfulness, and psychological flexibility with changes in outcomes.

Methods

Participants

Participants were 109 patients of a routine care psychiatric day hospital with a specialized program for patients with chronic pain and comorbid mental disorders, in Berlin, Germany, admitted between January 2018 and September 2020. Inclusion criteria were 1) pain for more than 6 months, 2) at least one comorbid mental disorder, 3) sufficient German language skills, and 4) at least 18 years of age. Exclusion criteria were a) acute substance abuse, 2) acute psychosis, 3) severe organic brain diseases, 4) intellectual disability, 5) acute suicidality, and 6) less than 4 weeks of treatment. The presence of a mental disorder, the absence of acute substance abuse, the absence of acute psychosis and the absence of acute suicidality, as well as the minimum age of 18 years, were requirements for admission in the psychiatric day clinic and assessed in an admission interview by a physician or psychologist (and confirmed by a senior psychiatrist). Abstinence from drug or alcohol abuse was monitored by urine or blood tests in patients with a history of addiction. The inclusion criterion of having pain for more than 6 months was determined at admission with the German Pain Questionnaire [34]. Intellectual disability was identified by clinical judgment, patient history, review of patient records, and neuropsychological testing if needed. German language skills were evaluated by clinical judgment. A minimum of 4 weeks of treatment duration was chosen as an exclusion criterion because it seemed a reasonable time frame to expect measurable treatment effects, based on clinical experience. The study was approved by the ethics committee of Charité – Universitätsmedizin Berlin.

Between January 2018 and September 2020, a total of 223 patients began treatment. Based on a relatively high number of first-generation Vietnamese patients [35], 44 (19.7%) were excluded for poor German language skills. A further 20 (9.0%) patients were excluded for not attending treatment for at least 4 weeks, 16 (7.2%) for not suffering from pain for more than 6 months, and 1 (0.4%) for intellectual disability. Data for 33 (14.8%) patients were lost due to the pitfalls of everyday practice. Thus, 109 (48.5%) patients were included.

Sociodemographic characteristics of the sample are displayed in Table 1. Mean age was 48.9 (11.4) years. The majority were female (66.1%), had a higher secondary school degree (56.7%), were employed (53.3%), and lived with a partner (53.3%). Clinical characteristics are provided in Table 2. Treatment lasted on average 6.4 (1.3) weeks. Most patients (89.7%) had received some kind of pain treatment before admission. The majority suffered from pain for more than 5 years (63.0%), experienced pain at two (38.1%) or more (38.1%) sites, and had a medium (41.2%) to high (27.8%) degree of pain chronicity, according to Mainz Pain Staging System (MPSS) [36]. Most patients took psychopharmaceutic or analgesic medication regularly at admission (66.1%) or discharge (84.4%). Diagnoses were coded by the attending physicians and psychologists according to the 10th edition of the International Classification of Diseases (ICD-10) [37] (see Supplementary Data). The mean number of diagnosed mental disorders was 2.7 (1.0). The most common (86.2%) single condition was chronic pain disorder with somatic and psychological factors (F45.41) [38]. Other prevalent mental disorders were depressive (90.8%), anxiety (24.8%), and personality (16.5%) disorders. Most patients (89.0%) received a chronic pain diagnosis that is not considered as a mental disorder according to ICD-10, including conditions related to the back (65.1%), head (45.0%), joints (37.6%), nervous system (5.5%), and other pain-related physical conditions (18.4%). The mean number of physical conditions related to chronic pain was 2.4 (1.4).

Procedure

At admission to the day hospital, patients gave informed written consent for the scientific use of their data. Most sociodemographic and clinical characteristics were recorded as part of routine evaluation, using the German Pain Questionnaire [34], MPSS [36], and a standard questionnaire from the clinic. Information on diagnoses, medication, and treatment duration was retrieved from the patients’ medical discharge letters. The process and outcome variables were assessed during the first and last weeks of treatment.

Measures

Mainz Pain Staging System

Pain chronicity and number of pain sites at admission were captured with the MPSS [36]. The MPSS includes 10 items on four axes (temporal aspects of pain, spatial aspects of pain, addictive behavior, utilization of the health care system), which are rated by a clinician and can be aggregated to a global level (low, medium, high) of pain chronicity. Among patients with different pain syndromes, higher MPSS stages were associated with higher rates of psychosocial distress and depression [36].

Chronic Pain Acceptance Questionnaire

Pain acceptance was determined with the Chronic Pain Acceptance Questionnaire (CPAQ) [39]. It contains 20 statements (e.g., “My life is going well, even though I have chronic pain”) with a seven-point Likert scale from 0 (“never true”) to 6 (“always true”). A systematic review confirmed its reliability, internal consistency, and construct validity [40].

Acceptance and Action Questionnaire II

Psychological flexibility was assessed with the Acceptance and Action Questionnaire II (AAQ-II) [41]. Participants are asked to rate seven statements on a scale from 1 (“never true”) to 7 (“always true”). The items are phrased as negative sentences (e.g., “My painful memories prevent me from having a fulfilling life”) and reflect psychological inflexibility. For the present analyses, items were reversed; therefore, higher scores indicate higher psychological flexibility. The validity of AAQ-II has been confirmed in patients with chronic pain [42] and general psychopathology [43].

Freiburg Mindfulness Inventory

The 14-item version of the Freiburg Mindfulness Inventory (FMI) [44] was administered to capture mindfulness. The frequency of mindfulness experiences (e.g., “I am open to the experience of the present moment”) is rated on a scale from 1 (“rarely”) to 4 (“almost always”). The FMI has demonstrated good psychometric properties and ability to distinguish individuals with and without psychopathology [45].

Pain Intensity

Pain intensity (PI) was determined with two items from the German Pain Questionnaire [34], covering PI at the current moment and PI over the last 4 weeks, on a numerical rating scale from 0 (“no pain”) to 10 (“extreme pain”). Both items were averaged to obtain a stable yet treatment-responsive score.

Patient Health Questionnaire

The Patient Health Questionnaire (PHQ) [46] was used to assess symptoms of somatization (PHQ-15), depression (PHQ-9), and general anxiety (GAD-7). The PHQ-15 assesses somatic symptoms over the last 4 weeks, including pain (e.g., stomach pain) and physical illness (e.g., nausea, gas, or indigestion). It consists of 13 items with a three-point scale, ranging from 0 (“not bothered at all”) to 2 (“bothered a lot”) and two more items retrieved from the PHQ-9. It has demonstrated good psychometric properties as a severity measure [47]. In the current study, values for the fourth item, asking about “menstrual cramps or other problems with your periods”, were systematically missing for men and women aged 60 years and older. As this item did not apply to them for biological reasons, it was set to 0 (“not bothered at all”) for these subgroups.

The PHQ-9 measures the frequency of nine depressive symptoms over the last 2 weeks (e.g., “little interest or pleasure in doing things”) with response categories of 0 (“not at all”) to 3 (“nearly every day”). It is sensitive to change in treatment and shows good psychometric properties [48]. The PHQ-9 has also been proven to be a valid instrument for patients with chronic spinal pain [49] and general psychosomatic pathology [50].

The GAD-7 is a screening instrument for generalized anxiety disorder, also sensitive to impairments in related disorders, including panic, social anxiety, and posttraumatic stress disorder [51]. Participants are asked to assess the appearance of anxiety symptoms (e.g., feeling nervous, anxious or on edge) over the last 2 weeks. The most widely used version of the GAD-7 [52] comprises a four-point scale, ranging from 0 (“not at all”) to 3 (“nearly every day”). In the current study, an earlier version of the GAD-7 [46] with a slightly different set of items was used, with a three-point scale, ranging from 0 (“not at all”) to 2 (“more than half the days”). Psychometric validity for measuring symptom severity has been confirmed for anxiety [52] and general disability [53].

Short Form 12

The Short Form 12 [54] measures physical and mental health with its subscales: the Physical Component Summary (PCS-12) and the Mental Component Summary (MCS-12). Both scales comprise 12 items covering different health facets (e.g., “During the past 4 weeks, how much did pain interfere with your normal work?” with five response options from “not at all” to “extremely”). Evidence supports the validity of the Short Form 12 [55], including its responsiveness to change in pain intensity [56].

Treatment

The treatment was interdisciplinary and based on ACT. It focused on treating chronic pain and comorbid mental disorders in groups of 10 to 12 patients in a psychiatric day hospital. The standard treatment length was 6 weeks for admitted patients, with the possibility of an extension to 8 weeks if indicated. Interdisciplinary treatment interventions were delivered by a psychiatrist/psychotherapist, a pain specialist/neurologist, a general practitioner, two psychologists, two nurses, an occupational therapist, a physiotherapist, a dance therapist, and a social worker. The treatment consisted of body-oriented therapies (physical conditioning, movement therapy, Nordic Walking, physical exercises), experience-oriented therapies (dance therapy, occupational therapy, creative therapy), and ACT-oriented therapies (ACT-based group psychotherapy, mindfulness training, ACT-Matrix group, ACT-based occupational therapy). Additionally, every patient received 20 to 50 minutes of individual psychotherapy per week to deepen the ACT processes. According to the existing guidelines and based on clinical judgment, medication was prescribed by a senior pain specialist and a senior psychiatrist, both with more than 10 years of professional experience.

The principles of ACT did pervade not only the ACT-oriented psychological therapies but also the entire treatment. The whole team was trained and regularly supervised in ACT by the head of the department (RB), who is an ACT trainer recognized by the Association of Contextual Behavioral Sciences. For a better understanding of the current treatment program and its implementation in the clinic, see Burian et al. [57] (note that the current treatment program is slightly modified and tailored to patients with chronic pain and comorbid mental disorders).

The ACT-oriented therapies were based on an adaptation of the manual Life with Chronic Pain: An Acceptance-Based Approach [58]. Every week, the treatment program was orientated toward one of five ACT processes (acceptance, values, committed action, cognitive defusion, self-as-context). Mindfulness was integrated as a weekly 50-minute training. One week contained three ACT-based group-psychotherapy sessions and one ACT-based occupational therapy session, each lasting 50–75 minutes. The ACT-Matrix group was based on the ACT-Matrix [59], a tool to enhance psychological flexibility and valued living, and was delivered in two 50-minute sessions per week. A notable feature of the treatment program was that some of the ACT-oriented therapies, namely the mindfulness training and ACT-Matrix group, were not conducted by mental health professionals but rather by nurses and social workers trained in ACT.

Statistical Analyses

Small proportions of data were missing for sociodemographic (3.3%) and clinical characteristics (3.7%), as well as for processes and outcomes at pre- (2.2%) and posttreatment (2.2%). Missing values for processes and outcomes were imputed with a nonparametric random forest algorithm, which computes missing data in an iterative process that is based on regression trees [60]. This method uses and combines information from categorical variables (e.g., sociodemographic and clinical characteristics) and continuous variables (e.g., the processes and outcomes) and accounts for complex and nonlinear interactions between them. The current analyses were conducted with 10,000 regression trees in each forest. The normalized root mean square error was 0.220, indicating an imputation error of 22.0%. This magnitude of imputation error appeared adequate, as other psychotherapy studies that used this imputation method reported normalized root mean square errors of 0.09 [61], 0.28 [62], or 0.50 [63].

Pre- to posttreatment differences in processes and outcomes were analyzed with one-sided Wilcoxon signed-rank tests and effect size r (0.10 = small, 0.30 = moderate, 0.50 = large) [64]. This nonparametric approach was chosen because some of the variables were not normally distributed, and it was deemed more informative to report uniformly interpretable effect sizes for all variables. Associations between changes in processes and changes in outcomes were examined with residualized change scores [65], which have the beneficial feature that correlations among them are corrected for variable properties that are not related to change. Bivariate associations between change scores were examined with Pearson or Spearman (zero-order) correlation coefficients, depending on normality. Multiple associations of the combined processes (pain acceptance, mindfulness, psychological flexibility) and single outcomes were analyzed with multiple regression analyses, with the change scores of the processes as independent variables and the change scores of the outcomes as dependent variables. The main coefficient of interest was R² (0.02 = small, 0.13 = moderate, 0.26 = large [66]). Multiple regression model assumptions were checked and satisfied, according to current standards of diagnostics [67]. As the analyses contained 51 significance tests, multiple comparison problems were addressed by adjusting the threshold of the P value from 0.05 to 0.001 with Bonferroni correction (0.05/51 = 0.00098).

All statistical analyses were performed with R version 4.0.2 [68]. The complete syntax and detailed results, along with Supplementary Data, are provided on the Open Science Framework (https://osf.io/sc7uk/).

Results

Pre- to posttreatment differences in processes and outcomes, analyzed with one-sided Wilcoxon signed-rank tests and effect size r, are displayed in Table 3. The difference in PCS-12 was not statistically significant and small (Z = 2.21, P = 0.014, r = .21). All other differences were statistically significant (P < 0.001), with either moderate effect sizes, including CPAQ (Z = 4.01, r = 0.39), AAQ-II (Z = 3.86, r = 0.37), PI (Z = -4.21, r = -0.40), PHQ-15 (Z = -3.98, r = -0.38), and GAD-7 (Z = -4.94, r = -0.47), or large effect sizes, including FMI (Z = 5.46, r = 0.53), MCS-12 (Z = 5.52, r = 0.53), and PHQ-9 (Z = -6.43, r = -0.62).

Table 4 displays the Pearson and Spearman correlation coefficients for associations between change scores of processes and change scores of outcomes. Almost all correlations were moderate, apart from two large correlations of ΔAAQ-II, with ΔPHQ-9 (r = -0.54) and ΔGAD-7 (r = -0.50). Most correlations (ranging from 0.33 to 0.54) were statistically significant (P < 0.001), yet two correlations, between ΔFMI and ΔPHQ-15 and between ΔFMI and ΔGAD-7, failed to reach significance after Bonferroni correction (both r = -0.30, P = 0.002). The whole correlation matrix (covering all possible correlations between the change scores) is available in Supplementary Data.

Table 5 summarizes the results of multiple regression analyses, determining multiple associations of change scores for the combined processes and change scores for the outcomes. All shared variances were statistically significant (P < 0.001). One shared variance was large, targeting ΔPHQ-9 [R² = 0.29, F(3, 105) = 15.61]. The remaining were moderate, including ΔPI [R² = 0.22, F(3, 105) = 11.08], ΔPHQ-15 [R² = 0.23, F(3, 105) = 11.53], ΔPCS-12 [R² = 0.21, F(3, 105) = 10.62], ΔMCS-12 [R² = 0.23, F(3, 105) = 11.74], and ΔGAD-7 [R² = 0.24, F(3, 105) = 12.35].

Notably, bivariate correlations of ΔPHQ-9 and ΔGAD-7 were clearly higher for ΔAAQ-II (ranging from -0.54 to -0.50) than for ΔCPAQ (ranging from -0.39 to -0.35) and ΔFMI (ranging from -0.34 to -0.30). In addition, of all multiple regression coefficients, only two were statistically significant (P < 0.001), which corresponded to the bivariate correlations and reflected associations of ΔAAQ-II with ΔPHQ-9 [β = -0.45, t(105) = -4.63] and ΔGAD-7 [β = -0.43, t(105) = -4.24].

Discussion

The present study examined process-outcome associations in an ACT-based interdisciplinary treatment for chronic pain and comorbid mental disorders in a routine care psychiatric day hospital. It was assumed that pre- to posttreatment differences had at least moderate effect sizes. This was mostly confirmed, apart from a small effect for physical health, contrasting meta-analytic evidence, which suggests a moderate effect for physical disability [23]. Results on the other outcomes were in line with existing evidence [23]. The pre- to posttreatment effect sizes of the processes, however, differed from previous findings. Changes in pain acceptance were typically large in previous studies [23] but moderate in the present. Conversely, change in mindfulness was large here, yet negligible [69] to small [27] elsewhere. Additionally, change in psychological flexibility was moderate in the present study, which was only in line with one previous study [27], whereas others reported small to nonsignificant effects [29, 69, 70]. In the case of mindfulness, the inconsistent findings might be explained by its measurement with the FMI, in contrast to the usage of the Mindfulness Attention Awareness Scale [71] in the other studies [27, 69]. Given that findings on pain acceptance and psychological flexibility were likewise deviating, an alternative explanation could be differences in treatment and patient characteristics. Previous studies [24–31, 69] examined patients who attended pain rehabilitation or management in London or southwest England for 3 to 4 weeks. In contrast, the present study examined chronic pain patients with comorbid mental disorders for on average 6.4 weeks in a psychiatric day hospital in Berlin, Germany. Moreover, the treatment program in the present study included more experience-oriented therapies (dance therapy, creative therapy) and was characterized by a diversity of ACT-based treatment modalities (ACT-based group psychotherapy, ACT-based individual psychotherapy, ACT-Matrix group, mindfulness training, ACT-based occupational therapy). Considering these various differences between the current study and previous studies, some variation in results is not surprising.

It was hypothesized that associations between changes in processes and changes in outcomes were small to large. This was fully confirmed and even surpassed, as all correlations exceeded r = 0.30, and were thereby moderate to large. Regarding changes in pain acceptance, its moderate associations with changes in outcomes were in line with previous evidence [24, 26–28, 30]. Associations between changes in mindfulness and changes in outcomes were likewise moderate, which was consistent with the only other ACT study for chronic pain that examined mindfulness in this context [27]. In that study, however, changes in outcomes referred to a follow-up period, which reduces comparability. Changes in psychological flexibility had moderate to large associations with changes in outcomes, which also was in line with a study previously examining follow-up changes in outcomes [27], while another study of patients aged 65 years and older yielded only small associations with changes in outcomes [29]. Overall, the present study confirms the already established process-outcome associations regarding pain acceptance and extends evidence on mindfulness and psychological flexibility.

It was hypothesized that multiple regression analyses yielded moderate to large shared variances of combined changes in processes and changes in outcomes. This hypothesis was confirmed and is mainly in line with the literature [24–29, 31]. The particular combination of pain acceptance, mindfulness, and psychological flexibility generated shared variances that were similar to those of other combinations of processes examined in previous studies [24–29, 31]. As ACT-specific processes are hypothesized to be highly interrelated [19], it seems plausible that different combinations of processes do not differ regarding their total shared variances with outcomes. A more detailed analysis of different combinations, however, may uncover differential qualities of single processes to be associated with outcomes. In the present study, psychological flexibility shared variance uniquely with depression and general anxiety. Correspondingly, bivariate associations of depression and general anxiety were higher with psychological flexibility than with the other processes. It is subject to debate, however, whether the construct of psychological flexibility is indeed more closely associated with general psychopathology or whether the measure with which it was assessed, the AAQ-II, lacks discriminant validity [72]. In fact, factor analytic studies suggest that the AAQ-II captures depression and anxiety [73], neuroticism/negative affect [74], or psychological distress [75] rather than psychological inflexibility. The field is therefore currently shifting from general measures, such as the AAQ-II, to context-specific measures of psychological flexibility, as they appear to show more discriminant validity [72, 76]. Against this background, the results of the current study should be interpreted with caution.

While results on process-outcome associations were generally in accordance with previous studies, one inconsistency between current and prior findings is worth noting. Associations of changes in pain intensity with changes in processes were moderate in the present study but small to negligible in previous studies [24, 26–29]. This contrast might be explained by differing operationalizations of pain intensity. In the present study, it was determined by averaging the severity of pain at the current moment and the last 4 weeks, whereas in other studies, it was referred to as pain in “the past week” [26–29] or assessed as “usual pain intensity” [24]. However, as mentioned before, the inconsistent findings may also be explained by systematic differences in treatment and patient characteristics. For example, the subjects of previous studies [24, 26–29] attended pain rehabilitation or management, while the current study included patients of a psychiatric clinic who suffered from comorbid mental disorders. Thus, it might be possible that the pain experience of the patients with comorbid mental disorders was more shaped by psychopathological factors (e.g., severity of depression and anxiety) and thereby easier to access by ACT-specific processes, resulting in higher process-outcome associations for changes in pain intensity.

The present study has a number of limitations. First, the study was conducted under conditions of routine care, which implies various shortcomings in terms of scientific quality and potential for bias when compared with psychological basic research. For example, the therapeutic staff changed over time and the prescription of individual treatment components, namely physical pain treatment and medication, was based on clinical judgment. Certain confounder variables could, therefore, not be controlled. However, the current study thus reflected the reality of clinical practice, which supports its external validity.

Second, the choice of the examined processes was not only based on theoretical considerations but also driven by clinical pragmatism. A conceptual imprecision of the current study is that the three examined processes are treated as equivalents, while they actually occur in different levels of the ACT model: acceptance is one of the six core processes, while mindfulness is a more comprehensive construct and psychological flexibility constitutes the overarching concept. While this commingling could be criticized conceptually, it seems justifiable from a clinical point of view: assessing pain acceptance accounts for the specific pathology of the sample, measuring mindfulness reflects the presence of the mindfulness training in the treatment program, and applying a short instrument for a preferably comprehensive construct (psychological flexibility) minimizes the study-related workload for the severely ill patients. In this light, the current study represents an effort to reconcile the demands of science and clinical practice. Beyond that, it has to be kept in mind that the ACT processes are by definition not strictly delimitable entities but rather clinically pragmatic constructs, which interact and overlap with each other [20].

Third, the proportion of data loss is notable, as 51.5% of the admitted patients were not included in the analyses. While this limitation is diminished by the fact that 36.7% of the admitted patients were excluded for not meeting inclusion criteria (or meeting exclusion criteria), it has still to be acknowledged that data of 14.8% eligible patients were lost due to the pitfalls of everyday practice. This restricts the generalizability of the results.

Fourth, most of the diagnoses were not captured by structured interviews but by unstructured clinical assessment of the attending physicians and psychologists, as is usual in clinical practice.

Fifth, treatment fidelity was not determined. This would have been especially advantageous for those ACT-oriented therapies (mindfulness training, ACT-Matrix group) that were led by nurses and social workers.

Sixth, the study design does not allow any causal conclusions. The associations of processes and outcomes were correlational, that is, based on the current study, it cannot be concluded that changes in processes caused changes in outcomes. Consequently, it cannot be ruled out that changes in outcomes caused changes in processes. In this regard, it is worth noting that some studies performed mediation analyses and determined pain acceptance and psychological flexibility as mediators of change for various outcomes [77–81]. However, evidence on mediation is sparse and far from conclusive. Future studies should therefore focus on directed interrelations between processes and outcomes by implementing more measurement timepoints and conducting analyses that have the potential to reveal causal mechanisms of change. Randomized controlled trials would furthermore be helpful to rule out that changes in processes and outcomes are consequences of a natural course. In addition, mixed method studies are needed to gain more insight into the subjective perspective of patients.

Seventh, it is crucial to consider that the examined processes are difficult to grasp empirically, as current self-report measures for pain acceptance, mindfulness, and psychological flexibility are limited. A systematic review on chronic pain acceptance questionnaires concluded that none of the reviewed measures met several important quality criteria, such as criterion validity, agreement, responsiveness, floor/ceiling effects, and interpretability [40]. A review on the assessment of mindfulness states that none of the examined self-report measures is able to cover all facets of mindfulness adequately and that there remains a fundamental issue concerning the construction of items that can be understood semantically across different individuals [82]. A further review suggests that psychological flexibility as a general construct may be too complex to assess and that it may be more appropriate to capture specific manifestations of psychological flexibility instead [72]. In this light, studies of these processes have to be interpreted with caution in general. It remains to the next generation of instruments to assess these processes with greater confidence.

Conclusion

The present study examined mechanisms of change in an ACT-based interdisciplinary treatment for patients with chronic pain and comorbid mental disorders in a routine care psychiatric day hospital. The results suggest that pain acceptance, mindfulness, and psychological flexibility are meaningfully related to symptom change. Well-studied associations of changes in pain acceptance and changes in outcomes were confirmed, and valuable evidence was generated regarding mindfulness and psychological flexibility, which have rarely been studied in this context before. By examining patients with comorbid mental disorders, the present study informs clinicians about an important subpopulation of chronic pain patients who often have a severe course of illness. As some of the ACT-oriented therapies of the interdisciplinary treatment program were conducted by ACT-trained nurses and social workers, the current investigation further adds a novel aspect of treatment delivery to the literature. Given the variety of processes and outcomes and the complexity of their associations, more studies are warranted for a deepened understanding of the mechanisms of change in ACT for chronic pain.

Funding source: This study did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflicts of interest: The authors declare that they have no conflicts of interest.

Acknowledgments

Special thanks go to Lena Fliedner, who played an important role in the establishment phase of the study. Gratitude is owed to Thi Hoan Nguyen, Andrea Flatow, Kerstin Britting, Daniela Koch, Mai Thy Phan-Nguyen, Linn Kriedel, and Sarah Dreßel, who contributed to the recruitment and data collection process. Further thanks go to Annette Burns for her proofreading of this paper. Finally, credit is given to all other colleagues of the staff of Evangelisches Krankenhaus Königin Elisabeth Herzberge, who supported the study under the strain of everyday practice.

Supplementary Data

Supplementary data are available at Pain Medicine online.

References