Mindfulness-Based Cognitive Therapy for Fatigue in Patients with Inflammatory Bowel Disease: Results of a Randomized Controlled Trial

Patients were eligible for inclusion if they had a diagnosis of either Crohn’s disease or ulcerative colitis, as confirmed by a medical specialist. In addition, patients needed to report elevated levels of fatigue, i.e., a score ≥ 27 on the subjective fatigue scale of the Checklist Individual Strength (CIS-20), indicating at least mild fatigue (Vercoulen et al., 1994). Their disease had to be in remission at the start of the intervention, as defined by a score < 4.99 on the mHealth Index Colitis Ulcerosa or < 6.38 on the mHealth Index Crohn’s Disease (Van Deen et al., 2016). Patients also needed to report a need for care concerning their fatigue and indicate motivation to participate in MBCT, which was assessed by open-ended questions during the intake. Additionally, eligible patients were aged between 18 and 75 years old, able to attend the eight weekly group sessions of 2.5 hr, and able to read, write, and speak Dutch and had no expectation of surgery in the upcoming 3 months. Exclusion criteria were severe cognitive, neurological, and psychiatric co-occurring conditions that could interfere with patients’ participation or warranted other treatment, specifically psychotic complaints or diagnosis of schizophrenia, neurological disorders including severe cognitive limitations, substance use disorder, and acute suicidal ideations or behavior. These comorbidities were examined separately via open-ended questions during the intake. Other exclusion criteria were pregnancy, anemia (i.e., Hb < 7.4 for women, Hb < 8.1 for men), change in IBD medication within 1 month before study entry, and receiving psychological treatment for fatigue or psychological/psychiatric problems at the time of recruitment. The latter was implemented in order to minimize possible confounding by other psychological treatments.

This two-arm multicenter randomized controlled trial (RCT) was conducted in the outpatient gastroenterology departments of the University Medical Centre Groningen, Medisch Spectrum Twente, and Isala Clinics Zwolle in the Netherlands, between July 2017 and September 2018. A consecutive sample was recruited via treating physicians in the three hospitals. All IBD patients who were under treatment in one of the three hospitals at the time of recruitment received a short screening questionnaire by mail, including the subscale subjective fatigue of the CIS-20, the mHealth Index, and a question about their current need for care, with an accompanying letter from their physician. Based on this screening, patients with IBD in remission, elevated levels of fatigue, and a possible need for care to manage their fatigue were invited for an intake by telephone. During the intake, patients were informed about the goals and setup of the study and intervention, and assessed for eligibility by a brief standardized interview. Patients who fulfilled our criteria received an information letter and informed consent form, as well as the baseline questionnaire. Those patients that provided written informed consent for participation were included in the study. Assessments in both groups were performed before randomization (i.e., baseline) and directly after the intervention period (i.e., post-measurement, approximately 3 months after baseline).

The baseline questionnaire included questions regarding demographic, socioeconomic, and disease-specific information. All psychological outcomes were assessed at baseline and post-measurement (on average, 3 months after the baseline assessment).

Sample size calculation for the primary research question was performed based on a pilot study (Drent et al., 2016), which assessed pre- and post-measurement fatigue levels in IBD patients participating in a mindfulness program similar to MBCT, specifically Mindfulness-Based Stress Reduction (MBSR) as developed by Kabat-Zinn (1990). The MBCT and MBSR program both involve eight weekly 2.5-hr sessions, focused on exercise and inquiry, one silent session of 3 hr, and homework exercises. Also, the content of the sessions is similar, except for the explicit focus on negative thoughts in MBCT. Given the similarities in the content, structure, and intensity of the two programs, we used this pilot study for the sample size calculation. With a statistical power of 0.80 and an alpha of 0.05, 64 patients were required per group (128 in total) to be able to detect differences with an effect size of at least 0.50.

Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) for Windows (version 26). Descriptive statistics are presented as means and standard deviations, medians and interquartile, or numbers and percentages, as appropriate (Table 1). Independent samples t-tests (or Mann-Whitney U tests in case of non-parametric data) and χ² tests were used to examine group equivalence at baseline for demographic and clinical variables, as well as primary and secondary outcomes.

All inferential analyses were based on the intention-to-treat principle. To account for missing data, multiple imputations were performed with the use of predictive mean matching. Twenty imputed datasets were specified and results were pooled over all datasets. Based on Little’s MCAR test and visual inspection of the data, we considered the data missing at random and multiple imputations were assumed to provide unbiased results. We performed sensitivity analyses in patients who completed both baseline and post-measurement questionnaires to demonstrate the robustness of the intention-to-treat findings (i.e., complete case analysis, see Supplementary Table 1).

Paired samples t-tests were conducted to examine within-group change in primary and secondary outcomes from baseline to post-measurement. Pooled standard deviations were calculated following Cohen’s (1988) rule. Analysis of covariance (ANCOVA) was performed to examine the effect of MBCT on fatigue and secondary outcome variables. Post-measurement levels of the primary and secondary outcomes were used as dependent variables, condition as factor and baseline levels of the outcomes were considered covariates. Since no other possible confounding variables correlated significantly with the outcomes (e.g., changes in medication use and experiencing a flare-up during the study period), only these variables were entered in the model. Per-protocol analyses were performed in patients who completed five or more sessions of MBCT. Assumptions for all analyses were met and the impact of potential outliers was examined using winsorizing. Two-tailed p-values < 0.05 were considered statistically significant.

Effect sizes were calculated using Cohen’s d. Values of 0.20–0.50 indicated small effects, 0.50–0.80 moderate effects, and > 0.80 large effects (Cohen, 1988). For the subscale subjective fatigue, we defined clinically relevant outcomes as having improved and being recovered. Improvement was calculated by the Reliable Change Index (RCI), which refers to the difference between an individuals’ baseline and post-measurement score, divided by the standard error of the difference (Jacobson & Truax, 1991). Each participant was categorized as improved (RCI > 1.96), no change (−1.96 to 1.96), or deteriorated (RCI < −1.96). Recovery was defined as a post-measurement score of subjective fatigue < 27, i.e., a level of fatigue comparable to healthy people (Vercoulen et al., 1994). χ² tests were used to compare the number of clinically relevant improved and recovered patients between the MBCT and waiting-list control conditions.

In total, 3896 patients were approached for screening (Fig. 1). Of the 2376 patients who returned the screening questionnaire, 1440 patients reported an elevated fatigue score (CIS — subjective fatigue ≥ 27). Around one-third of the patients with elevated fatigue could be approached for the intake (n = 453). Based on the intake, 125 patients were not eligible for participating in the study, mostly because they did not feel the need for treatment after all. Half of the eligible patients (n = 215) declined participation because of various reasons.

In the end, 113 patients gave consent and were randomized into either the MBCT condition or the waiting-list control condition (n = 56 and 57, respectively). Although the overall response rate was 3% (i.e., approached vs. randomized), 34% of the eligible patients participated in the study. Patients included in the study did not differ significantly in age, gender, and type of IBD compared to all approached patients (n = 3896). Eighteen patients (32%) in the MBCT condition did not complete the post-measurement questionnaire, in contrast to five patients (9%) in the waiting-list condition (χ²(1, n = 113) = 9.52, p < .01). Within the intervention group, seventeen patients (30%) did not complete MBCT (i.e., participated in less than five sessions). Reasons for dropout were mainly medical or related to scheduling; four patients did not provide any reason. Patients did not report any harm related to the study. Dropout of the intervention was only significantly associated with being a female (χ²(1, n = 56) = 6.10, p = .01) and not with other demographic or clinical variables, nor with fatigue or secondary outcomes.

Table 1 shows an overview of the baseline characteristics of our sample. Patients in the MBCT and waiting-list condition did not differ significantly on demographic and clinical baseline characteristics. Averages of primary and secondary outcome measures at baseline are presented in Table 2. Descriptive statistics show severe levels of fatigue and reduced QoL, whereas levels for fatigue interference, depressive symptoms, and anxiety indicate little impairments. No significant differences in these outcome measures were found between the two conditions.

Intention-to-treat analyses revealed significant differences in post-measurement subjective fatigue scores between the MBCT and waiting-list control conditions (Table 2). Patients in the intervention group reported significantly lower subjective fatigue scores at post-measurement than patients in the waiting-list control group, when adjusted for baseline values of the outcome (p = 0.03). The effect size was small to moderate (Cohen’s d = 0.46). Although within-group time effects were significant for both conditions, the decline in subjective fatigue was stronger in the intervention group compared to the control group (Cohen’s d = 0.62 and 0.19, respectively). Regarding the other aspects of fatigue, no significant differences were found between the two conditions in the post-measurement scores (neither in the total fatigue score, nor in the other subscales). Per-protocol analyses (i.e., in patients who completed five or more sessions of MBCT) yielded similar results, except for an additional significant effect on total fatigue (p < 0.05), with a small to moderate effect size (Cohen’s d = 0.40) (Table 3).

Intention-to-treat analyses showed no significant differences between the two conditions in the post-measurement scores of fatigue interference, anxiety, depression, and QoL, when controlled for baseline levels of the outcome (Table 2). Effect sizes were small (Cohen’s d < 0.25). Per-protocol analyses showed that patients receiving MBCT reported significantly lower depression at post-measurement than patients in the waiting-list control condition, when adjusted for baseline levels. The effect size was small (p = 0.04; Cohen’s d = 0.29; Table 3).

Clinically relevant improvement of subjective fatigue (i.e., RCI score > 1.96) was found in 36% of the participants after MBCT versus 10% in the waiting-list control condition (χ²(1, n = 86) = 9.37, p < 0.01) (Table 4). After the intervention period, 21% of the patients in the MBCT condition were considered recovered (i.e., CIS — subjective fatigue < 27), compared to 9% in the control condition (χ²(1, n = 90) = 4.54, p = 0.03). Sixteen percent of patients in the MBCT condition both improved and recovered, whereas no patients in the waiting-list control group fulfilled this criterion (χ²(1, n = 90) = 10.39, p < 0.01). Rates of improvement and recovery in patients completing five or more MBCT sessions were similar (see Supplementary Table 2).

Although our study was carefully designed, the findings should be considered in light of several limitations. First, as the current sample size allowed us to reveal medium to large effects, we might have overlooked small effects. Second, consistent with previous MBCT studies in IBD patients, only 70% of the patients completed the intervention (Schoultz et al., 2015). Most patients dropped out because of scheduling issues with the group sessions. Offering individual MBCT could solve this issue (Schroevers & Fleer, 2019). Previous studies have also suggested that treatment adherence could be increased by offering MBCT online, as this type of treatment would be more accessible and flexible (Bruggeman-Everts et al., 2017; Compen et al., 2018). These studies showed, however, that dropout rates of online MBCT programs were higher compared to face-to-face MBCT. Blending face-to-face with online sessions might be a suitable alternative. Other reasons for dropout might have been the intensity and time-consuming nature of the program or a disease flare-up (Schoultz et al., 2015). As dropout was unrelated to the severity of fatigue and psychological symptoms at baseline, we assume that the results are not affected by attrition bias. Another limitation is that we were limited in examining the long-term effects of MBCT for IBD-related fatigue, since the waiting-list control condition received the intervention after the post-measurement. Long-term effects of MBCT should be assessed to examine stability of effects (Farrell et al., 2020). Moreover, although sleep problems are often associated with fatigue, we did not examine sleep disturbances and we thus do not know whether sleep quality has interfered with the effects of MBCT on fatigue. It can be hypothesized that severe sleep problems might hinder the effect of MBCT on fatigue, but it is also known that mindfulness can positively affect sleep disturbances (Shallcross & Visvanathan, 2016), which could have enhanced the effect of MBCT on fatigue. Future research is needed to examine these associations and effects. A last limitation concerns the self-report measurement of disease activity. Although the mHealth Indexes show reasonable measurement properties for use in clinical trials, they demonstrate limited validity compared to more recent self-report instruments or objective measures of inflammation (de Jong et al., 2018). We recommend the additional examination of fecal calprotectin or endoscopic assessment, or the use of a newly validated and reliable short screening instrument such as the Monitor IBD At Home questionnaire (MIAH; de Jong et al., 2019) for future research.

Together with other national and international IBD professionals and researchers, we emphasize the need and importance of appropriate psychological care for fatigue in IBD patients in remission. Fatigue is a multifaceted concept with a complex and multidimensional etiology. Therefore, the treatment of fatigue may require a multimodal and personalized approach, in which patients are screened for their symptoms, and different interventions can be offered that target not only fatigue, but also psychological functioning, sleep, nutrition, and physical activity, depending on the individual patients’ complaints, needs, and wishes (Borren et al., 2019). Researchers in the Netherlands have developed and implemented the telemedicine tool MyIBDCoach (de Jong et al., 2017), which successfully monitors and screens for IBD symptoms, including fatigue, on a regular basis and from home. A next step is to provide suitable and effective interventions based on the outcomes of these screening tools.

Given the low uptake of care and high dropout rates when psychological interventions are offered to patients in a medical setting, the current literature, including our study, asks for a better understanding of patients’ needs, wishes, and possibilities to participate in interventions when experiencing severe fatigue. When developing new interventions, content, format, and duration could be adjusted accordingly to increase beneficial outcomes, for instance, by offering one-on-one treatment or online sessions as part of the intervention (Bruggeman-Everts et al., 2017). Alternatively, including physical activity components to the treatment of fatigue seems promising (Davis et al., 2020). We also suggest replication studies with larger sample sizes to give more robust insights into the benefits of MBCT for treating fatigue, in comparison to other (psychological) interventions. In addition, the examination of demographic, clinical, and psychological moderators, as well as underlying mechanisms of the effects of MBCT, would allow us to better understand which IBD patients benefit most from MBCT and how MBCT reduces fatigue.