Introduction
Cancer is a common disorder worldwide, and there were an estimated 19.3 million new cancer cases in 2020 globally (Sung et al. Reference Sung, Ferlay and Siegel2021). A review study indicated that stress from living with cancer has negative impacts on patients’ emotional status: 4–49% of cancer patients suffer from depression and 3.4–43.0% of cancer patients suffer from anxiety (Brandenbarg et al. Reference Brandenbarg, Maass and Geerse2019; Niedzwiedz et al. Reference Niedzwiedz, Knifton and Robb2019). Approximately 20–30% of cancer patients’ psychological distress remains for a long time after the initial diagnosis (Campos et al. Reference Campos, Besser and Ferreira2012). Psychological distress influences cancer patients’ quality of life, self-concept, and emotional well-being; it is associated with poor disease progression, cancer recurrence, and lower cancer survival rates (Niedzwiedz et al. Reference Niedzwiedz, Knifton and Robb2019; Walker et al. Reference Walker, Magill and Mulick2020; Wang et al. Reference Wang, Wang and Zhong2020).
The protective factor of self-compassion is associated with a decrease in psychopathological symptoms, and it can also improve the quality of life of cancer patients (Pinto-Gouveia et al. Reference Pinto-Gouveia, Duarte and Matos2014). Furthermore, a growing number of studies show that compassion-based intervention leads to positive outcomes for cancer patients, such as reducing depression and anxiety (Brooker et al. Reference Brooker, Julian and Millar2020; Campo et al. Reference Campo, Bluth and Santacroce2017; Dodds et al. Reference Dodds, Pace and Bell2015; Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018; Trindade et al. Reference Trindade, Ferreira and Pinto-Gouveia2020) and increasing body image satisfaction, self-compassion, mindfulness (Brooker et al. Reference Brooker, Julian and Millar2020; Campo et al. Reference Campo, Bluth and Santacroce2017), physical health, and the quality of social relationships (Trindade et al. Reference Trindade, Ferreira and Pinto-Gouveia2020). However, there were no significant improvements in anxiety, depression, fear of recurrence, or psychological distress in other studies (Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). As the results were inconsistent, the effectiveness of compassion-based intervention on cancer patients remains unclear. Thus, a comprehensive systematic review and meta-analysis examining the effectiveness of compassion-based intervention in cancer patients are necessary. The aim of this study was to provide an overview of compassion-based intervention studies and to synthesize their effectiveness in cancer patients.
Background
The concept of compassion has been rooted in religious, spiritual, or philosophical traditions for thousands of years (Gilbert et al. Reference Gilbert, Catarino and Duarte2017). Compassion is defined as a sensitivity to suffering, with a commitment to try to alleviate and prevent it (Gilbert Reference Gilbert2014). Self-compassion is defined as individuals focusing on their inner self while in a stress event, and it consists of 3 main components: self-kindness, a sense of common humanity, and mindfulness (Neff and Germer Reference Neff and Germer2013). According to the 3-circle model of emotion, there are at least 3 types of emotion regulation systems, namely threat and protection systems, drive resource seeking and excitement systems, and soothing and safeness systems (Depue and Morrone-Strupinsky Reference Depue and Morrone-Strupinsky2005; Gilbert Reference Gilbert2014). Compassion is linked with attachment theory and plays a key role in soothing and safeness systems, helping individuals to maintain emotional balance even when facing stressful events (Gilbert Reference Gilbert2009, Reference Gilbert2014).
Compassion-based intervention is designed to cultivate cognitive, emotional, or motivational compassionate habits by using specific techniques and procedures. It is a combination of developmental, social, neuroscience, and Buddhist psychologies (Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018; Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018). According to Kirby (Reference Kirby2017), at least 6 types of compassion-based interventions have been developed: compassion-focused therapy (CFT; Gilbert Reference Gilbert2014), mindful self-compassion (MSC; Neff and Germer Reference Neff and Germer2013), compassion cultivation training (Jazaieri et al. Reference Jazaieri, Jinpa and McGonigal2013), cultivating emotional balance (Kemeny et al. Reference Kemeny, Foltz and Cavanagh2012), loving-kindness meditation (Hofmann et al. Reference Hofmann, Grossman and Hinton2011), and cognitively based compassion training (CBCT; Pace et al. Reference Pace, Negi and Adame2009). The previous studies on compassion-based intervention focused on nonclinical adult populations (Kirby et al. Reference Kirby, Tellegen and Steindl2017) and patients with psychological disorders (Leaviss and Uttley Reference Leaviss and Uttley2015; Shonin et al. Reference Shonin, Van Gordon and Compare2014). There is a review study on patients with long-term physical conditions, which includes those with both cancer and persistent pain (Austin et al. Reference Austin, Drossaert and Schroevers2020). To explore the effectiveness of compassion-based intervention focusing solely on cancer patients, we conducted a systematic review and meta-analysis to examine the effectiveness of depression, anxiety, self-compassion, and other outcomes relating to mental health.
Objectives
The objectives of this study were (a) to provide a comprehensive systematic review of compassion-based interventions for cancer patients from randomized controlled trials (RCTs) and (b) to examine the effectiveness of compassion-based interventions among cancer patients
Methods
Design
A systematic review and meta-analysis were performed in this study, and they complied with the recommendations for the latest Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Page et al. Reference Page, McKenzie and Bossuyt2021).
Search methods
Eleven bibliographic databases were searched: PubMed, CINAHL, MEDLINE, PsycINFO, WOS, Cochrane, Embase, Scopus, ProQuest Dissertations, Airiti Library, and the National Digital Library of Theses and Dissertations in Taiwan. We employed the following search strategies: compassion * [Title/Abstract] OR Compassion-Focused Therapy [Title/Abstract] OR Mindful Self-Compassion [Title/Abstract] OR Cultivating Compassion Training [Title/Abstract] OR Cognitively-Based Compassion Training [Title/Abstract] OR loving kindness meditation [Title/Abstract] combination with “cancer” [Title/Abstract] OR “neoplasms” [Title/Abstract] OR “neoplasms” [MeSH Terms]. The search had no year range limit, and English and Chinese were the selected languages. The last search took place on March 1, 2022. Databases were searched by title and abstract by the first author, and the results of the search were exported to EndNote X9 for further duplication management. After the removal of duplicated studies, the abstract and title were screened by 2 authors to select the studies based on eligibility criteria, and full-text reviews were also conducted by the 3 authors to confirm the inclusion.
Inclusion criteria
Studies had to meet the following inclusion criteria: (a) studies had RCT designs or were RCT pilot studies for compassion-based intervention in cancer patients, (b) studies were published in a peer-review journal or in a dissertation, (c) interventions included compassion-based activities or practices for cancer patients, and (d) outcomes included at least one psychological measurement with quantitative methodology. Exclusion criteria were (a) outcome measurements without quantitative methodologies and (b) documents published in conference papers or protocols. Participants included those with any type or stage of cancer, and there were no limits for age or geographical location. The types of interventions mainly focused on cultivating cognitive, emotional, or motivational compassionate habits, such as CFT, MSC, compassion cultivation training, CBCT, or any intervention including the following components: (a) an open awareness of suffering and (b) an intention or motivation to relieve the suffering with emotional empathy or a noncritical approach. There was no limit to the duration, frequency, or delivery method. Types of comparisons were not limited to any type of control group, including a treatment as usual group or any kind of active treatment control group. Outcomes included at least one self-report measure relating to mental health (depression and anxiety), self-compassion, mindfulness, or well-being.
Quality appraisal
Risk of bias was assessed with version 2 of the Cochrane risk-of-bias tool for randomized trials (Higgins et al. Reference Higgins, Sterne and Savovic2016; Sterne et al. Reference Sterne, Savović and Page2019). The bias domains included randomization process, deviations from the intended intervention, missing outcome data, measurement of outcome, selection of the reported results, and overall bias. First, 2 authors categorized the risk of bias into low risk, high risk, or some concerns for each domain independently. Afterwards, these authors discussed any inconsistencies in their judgments by repeated reviews of the studies and comprehensive discussion until the judgments of risk of bias from authors achieved consistency.
Data abstraction
The information on the study characteristics included author, year, country, participants’ characteristics, study method, contents, homework, instructor, comparison type, and outcome measured as suggested by the Cochrane Collaboration (Higgins et al. Reference Higgins, Thomas and Chandler2019). The means, standard deviations, and sample sizes were extracted for meta-analysis. For studies without available data, authors were contacted for additional results via email. Finally, studies were excluded if the authors were unable to provide data (e.g., mean and standard deviations of outcome variables) or did not respond to our emails.
Data analyses
Comprehensive Meta-Analysis version 3 was used to analyze the effects of compassion-based intervention. Primary outcome variables included depression, anxiety, and self-compassion, which were analyzed by Hedges’s g and 95% confidence intervals (CIs) due to small sample sizes (Hedges and Olkin Reference Hedges and Olkin1985). The interpretations of effect sizes were small (0.2), moderate (0.5), and large (0.8) (Cohen Reference Cohen1988). Due to large variations in follow-up length among compassion-based interventions, we could only pool data immediately after the intervention.
Heterogeneity testing among the studies was assessed by Q value and I 2 statistics. When the Q value was statistically significant, p < 0.10 was interpreted as heterogeneity, while I 2 demonstrated low (25%), moderate (50%), or high (75%) degrees of heterogeneity (Higgins et al. Reference Higgins, Thompson and Deeks2003). The subgroup moderation analysis was conducted with sufficient available data to explore the root of heterogeneity (Card Reference Card2012). The random-effect model was used in this study, as some of the studies’ outcome measurements were not identical. Publication bias was examined by a funnel plot. An asymmetrical funnel indicates potential publication bias.
Sensitivity analysis (Bown and Sutton Reference Bown and Sutton2010) was conducted to examine the robustness of the results by using leave-one-out test. A pooled effect size was estimated by removing studies judged as at high risk of bias or studies with small sample sizes (<50) to evaluate the influence of such studies.
The Hartung–Knapp–Sidik–Jonkman (HKSJ) adjustment was applied due to the small number of studies in the meta-analysis. The HKSJ method, based on t distributions, provides a more robust estimation of the CIs. Some have suggested applying HKSJ method for random-effects meta-analyses when 5 or fewer trails are included (Bender et al. Reference Bender, Friede and Koch2018; Friede et al. Reference Friede, Röver and Wandel2017; Knapp and Hartung Reference Knapp and Hartung2003; Saueressig et al. Reference Saueressig, Pedder and Bowe2021). The HKSJ pooled effects were calculated by Excel conversion from DerSimonian and Laird method for random effects (IntHout et al. Reference IntHout, Ioannidis and Borm2014).
Results
This study was guided by the PRISMA 2020 flow diagram in Figure 1. The search yielded a total of 945 studies, and 10 studies went into the systematic review. After excluding one study without available data (Milbury et al. Reference Milbury, Weathers and Durrani2020), 9 studies were included in the meta-analysis.
Fig. 1. PRISMA flow diagram.
Study characteristics
Table 1 presents the characteristics of the included studies. Ten studies were published between 2015 and 2021. Eight of them were published in the last 5 years. Seven studies were designed as 2-armed, and 3 studies (Cheung et al. Reference Cheung, Cohn and Dunn2017; Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Wren et al. Reference Wren, Shelby and Soo2019) had 3-armed designs.
Table 1. Characteristics of included studies
Notes: BDI-13, Beck Depression Inventory; BAI, Beck Anxiety Inventory; BAS, Body Appreciation Scale; BIS, body image scale; BP, blood pressure; HR, heart rate; BSI-18, Brief Symptom Inventory; BPI, Brief Pain Inventory; CAMS-R, Cognitive and Affective Mindfulness Scale – Revised; CBCT, cognitively based compassion training; CFT, compassion-focused therapy; CES-D, Center for Epidemiologic Studies – Depression questionnaire; COMP, Compassion Scale; DASS, Depression, Anxiety and Stress Scale; FW, follow-up; FFMQ, Five Facets of Mindfulness Questionnaire – Short Form; FACIT, Functional Assessment of Cancer Therapy; FCRI, Fear of Cancer Recurrence Inventory; FACT, Functional Assessment of Cancer Therapy; GQ-6, Gratitude Questionnaire-6; IES-R, Impact of Events Scale – Revised; LKM, loving-kindness Meditation; MDASI-BT, MD Anderson Symptom Inventory-Brain Tumor; MAAS, Mindful Attention Awareness Scale; MyCB, My Changed Body: self-compassion-focused writing activity; MyCB+M, My Changed Body plus meditation; PSS-4, Perceived Stress Scale; PARI, Personal Assessment of Intimacy in Relationships Inventory; PANAS, Positive and Negative Affect Schedule; R-UCLA, Revised Loneliness Scale Version 3; SF-12, Medical Outcomes Study Short Form 12-Item Health Survey; SCS, Self-Compassion Scale Short Form; STAI, State-Trait Anxiety Inventory; SCA, self-compassionate attitude; TAU, treatment as usual control group; QoL, quality of life.
Population characteristics
A total of 771 participants were involved across the included studies. The mean age was 55.51 years (range 38–60). Most of the interventions targeted women with breast cancer, while 2 studies targeted people with metastatic brain tumors (Milbury et al. Reference Milbury, Weathers and Durrani2020) and skin cancer (Latifi et al. Reference Latifi, Soltani and Mousavi2020). One study selected mainly anxiety and depression participants according to the Beck Depression and Anxiety Inventory (2 standard deviations) (Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018). Approximately 98% of intervention participants were female.
Intervention characteristics
The interventions fall into 2 types: constructive compassion-based interventions with 4–12 weeks multiple sessions (n = 6) and brief compassion-based interventions with a single session of approximately 30 minutes (n = 4). The constructive compassion-based interventions were theoretical: 2 CBCT (Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018), 1 CFT (Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018), 1 mindful compassion meditation (Milbury et al. Reference Milbury, Weathers and Durrani2020), 1 lessons in linking affect and coping intervention (Cheung et al. Reference Cheung, Cohn and Dunn2017), and 1 self-healing training intervention (Latifi et al. Reference Latifi, Soltani and Mousavi2020). Four out of 6 constructive interventions were conducted for patients who were undergoing treatment (Cheung et al. Reference Cheung, Cohn and Dunn2017; Latifi et al. Reference Latifi, Soltani and Mousavi2020; Milbury et al. Reference Milbury, Weathers and Durrani2020; Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018). Depression was the most measured outcome: 5 out of 6 constructive interventions measured depression (Cheung et al. Reference Cheung, Cohn and Dunn2017; Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018; Milbury et al. Reference Milbury, Weathers and Durrani2020; Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018). Four out of 6 constructive interventions were delivered in a face-to-face group format with 8–16 sessions (90–120 minutes per session) (Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018; Latifi et al. Reference Latifi, Soltani and Mousavi2020; Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018). One study (Cheung et al. Reference Cheung, Cohn and Dunn2017) used both an online and a face-to-face delivery format to examine the comparison benefit. The interventions were led by experienced, qualified, and well-trained psychologists or social workers.
More than half the brief compassion-based interventions were developed based on Neff’s concept of self-compassion, which includes self-kindness, common humanity, and mindfulness (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Neff and Germer Reference Neff and Germer2013; Przezdziecki and Sherman Reference Przezdziecki and Sherman2016; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). The brief interventions included 1 single compassionate writing session and 1 audio-based loving-kindness meditation. Three out of 4 brief interventions used writing guided by self-compassionate prompts: participants described distressing events they experienced after breast cancer treatment (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Przezdziecki and Sherman Reference Przezdziecki and Sherman2016; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). One study provided an audio-based loving-kindness meditation intervention during which participants listened to an MP3 during a biopsy procedure and they were encouraged to continue this as a daily practice afterward (Wren et al. Reference Wren, Shelby and Soo2019). Most of the brief interventions were conducted at the posttreatment cancer survivorship stage (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Przezdziecki and Sherman Reference Przezdziecki and Sherman2016; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). Anxiety was the most measured outcome: 3 out of 4 brief interventions measured anxiety (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018; Wren et al. Reference Wren, Shelby and Soo2019). Three out of 4 brief interventions were delivered in a non-face-to-face format, which included web-based or paper-based compassion writing prompts (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Przezdziecki and Sherman Reference Przezdziecki and Sherman2016; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). The interventions were implemented by playing an MP3 or providing self-guided writing prompts.
Comparison group
In constructive compassion-based interventions, 4 of the 6 studies (Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018; Latifi et al. Reference Latifi, Soltani and Mousavi2020; Milbury et al. Reference Milbury, Weathers and Durrani2020) used waitlist-controlled groups or treatment as usual groups as comparison groups. The typical procedure included pharmacological treatment, outpatient care, and/or psychological counseling. Two studies used active control groups: Sadeghi et al. (Reference Sadeghi, Yazdi-Ravandi and Pirnia2018) provided motivational enhancement therapy for participants in the control group; Cheung et al. (Reference Cheung, Cohn and Dunn2017) provided a one-on-one in-person attention-matched intervention for participants in the control group, which encouraged them to express their life histories, use of complementary and alternative medicine, diet and exercise, social networks, and meaning and spirituality.
In brief compassion-based interventions, all the control groups were active control group, including those writing without compassion prompts (Przezdziecki and Sherman Reference Przezdziecki and Sherman2016), using expressive writing (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018), or listening to music and developing a supportive dialogue as a comparison group (Wren et al. Reference Wren, Shelby and Soo2019).
Treatment outcomes
Treatment outcomes can be categorized into psychological (e.g., depression and anxiety), process-related (e.g., mindfulness and self-compassion), and cancer-related measures (e.g., quality of life, symptom distress, fear of cancer recurrence, body image distress, and body appreciation). Overall, the primary and the most frequent outcome measures in the included studies were depression (n = 5), anxiety (n = 4), and self-compassion (n = 7). There was heterogeneity in the outcome measurement scales. Depression was measured by the Center for Epidemiologic Studies Depression Scale (Cheung et al. Reference Cheung, Cohn and Dunn2017; Dodds et al. Reference Dodds, Pace and Bell2015), Brief Symptom Inventory (Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018), Beck’s Depression Inventory (Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018), and Depression Anxiety Stress Scales (Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). Anxiety was measured by the Brief Symptom Inventory (Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018), Beck Anxiety Inventory (Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018), Depression Anxiety Stress Scales (Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018), and State-Trait Anxiety Inventory (Wren et al. Reference Wren, Shelby and Soo2019). Self-compassion was measured by the Self-Compassion Scale (Latifi et al. Reference Latifi, Soltani and Mousavi2020), Self-Compassion Scale – Short Form (Cheung et al. Reference Cheung, Cohn and Dunn2017; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018; Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018; Wren et al. Reference Wren, Shelby and Soo2019), and a self-designed Self-Compassion Scale (Przezdziecki and Sherman Reference Przezdziecki and Sherman2016).
Quality assessment
Figure 2 presents the summary of risk of bias for each study. Among the 10 studies, 2 studies (20%) were judged as high risk of bias, 6 studies (60%) were judged as some concerns, and 2 studies (20%) were judged as low risk of bias.
Fig. 2. The summary of risk of bias for each study.
Post-treatment effects of compassion-based interventions on depression, anxiety, and self-compassion
Figure 3 shows the results of depression, anxiety, and self-compassion for cancer patients by forest plots. Small to medium effect sizes were achieved for reducing depression (Hedges’s g = –0.497, 95% CI = –0.864 to –0.130, p = 0.008) with moderate heterogeneity (Q = 8.739; p = 0.068; I 2 = 54.229%), and large effect sizes were achieved for increasing self-compassion (Hedges’s g = –0.869, 95% CI = –1.325 to –0.414, p < 0.001) with high heterogeneity (Q = 36.271, p = 0.000, I 2 = 83.458%). Conversely, the results showed no significant reduction in anxiety (Hedges’s g = –0.375, 95% CI = –0.814 to –0.064, p = 0.094) with moderate heterogeneity (Q = 8.919; p = 0.030; I 2 = 66.365%). A leave-one-out analysis showed that Sadeghi et al. (Reference Sadeghi, Yazdi-Ravandi and Pirnia2018) was an influential outlier. This study used prescreening to select participants suffering from anxiety and depressive symptoms (2 standard deviations on the Beck Depression and Anxiety Inventory). After excluding this study, the effects on reducing anxiety became significant (Hedges’s g = –0.211, 95% CI = –0.414 to –0.008, p = 0.041) with no heterogeneity (Q = 0.691, p = 0.708, I 2 = 0%).
Fig. 3. Post-treatment effects of compassion-based interventions on depression, anxiety, and self-compassion.
Due to the small number of studies in our meta-analysis, we applied HKSJ adjustment for robust CI and to reduce the likelihood of a Type I error. The result from the HKSJ method does not violate the result from the DerSimonian and Laird method. Medium effect sizes were achieved for reducing depression from HKSJ adjustment (SMD = –0.5088, 95% CI = –1.00 to –0.018, t = –2.897, p = 0.045), and medium to large effect sizes were achieved for increasing self-compassion from HKSJ adjustment (SMD = –0.884, 95% CI = –1.65 to –0.12, t = –2.83, p = 0.030). The HKSJ adjustment method did not reduce anxiety significantly (SMD = –0.381, 95% CI = –1.29 to 0.53, t = –1.34, p = 0.27).
The moderation effects of compassion-based intervention on self-compassion outcomes were further examined by using 2 subgroups analysis: intervention type and intervention delivery format. In intervention type, constructive compassion-based intervention (Hedges’s g = –1.669, 95% CI = –2.800 to –0.539, p = 0.004) showed a significantly higher effect size for self-compassion than the brief compassion-based intervention (Hedges’s g = –0.371, 95% CI = –0.549 to –0.194, p = 0.000) with a Q value of 4.941 (p = 0.026). The heterogeneity test showed no heterogeneity in the brief intervention (Q = 0.861, p = 0.835, I 2 = 0%) but high heterogeneity in constructive intervention (Q = 14.315, p = 0.001, I 2 = 86.028%). In the intervention delivery format, there was no significant difference (Q = 2.170, p = 0.141) between face-to-face delivery format (Hedges’s g = –1.011, 95% CI = –1.853 to –0.169, p = 0.019) and non-face-to-face delivery format, which indicated web or paper-based intervention (Hedges’s g = –0.363, 95% CI = –0.548 to –0.179, p = 0.00). The heterogeneity test showed no heterogeneity in the non-face-to-face format (Q = 0.773, p = 0.679, I 2 = 0%) but high heterogeneity in the face-to-face format (Q = 9.744, p = 0.008, I 2 = 79.475%).
Publication bias was examined by funnel plots, which were asymmetric in all outcomes (Appendix A). Sensitivity analysis was estimated by removing studies judged as at high risk of bias or studies with small sample sizes. The results did not change in depression and self-compassion outcomes. In anxiety outcome, after removing one study judged as at high risk of bias (Sadeghi et al. Reference Sadeghi, Yazdi-Ravandi and Pirnia2018), the effect of reducing anxiety changed from nonsignificant to significant.
Follow-up effectiveness of compassion-based interventions on depression, anxiety, and self-compassion
Due to the limited numbers of studies and the wide range of follow-up periods, we were unable to conduct a meta-analysis of follow-up effectiveness. We analyzed the follow-up effectiveness through systematic review. Two studies with constructive intervention reported improvements in depression in the follow-up period. Cheung et al. (Reference Cheung, Cohn and Dunn2017) reported that depression was below the clinical threshold by the 1-month follow-up, while Milbury et al. (Reference Milbury, Weathers and Durrani2020) found an improvement in depressive symptoms from 6 to 12 weeks after completing the intervention with a marginally significant effect (p = 0.06). Two studies with brief interventions reported no significant changes after intervention or at the 1-month and 3-month follow-up periods (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). Only Wren et al. (Reference Wren, Shelby and Soo2019) showed that anxiety significantly reduced after 2 weeks of brief compassion-based intervention.
For self-compassion, the effectiveness lasted longer in constructive intervention than in brief intervention. Three out of 4 constructive interventions showed significantly increased self-compassion after 2–6 months of intervention (Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018; Latifi et al. Reference Latifi, Soltani and Mousavi2020; Milbury et al. Reference Milbury, Weathers and Durrani2020), while 1 study showed no significant change at the 1-month follow-up (Cheung et al. Reference Cheung, Cohn and Dunn2017). All the brief interventions showed significantly increased self-compassion after 2 weeks to 1 month, but it did not last for 3 months (Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018; Wren et al. Reference Wren, Shelby and Soo2019). Sherman et al. (Reference Sherman, Przezdziecki and Alcorso2018) further examined the mediator role of self-compassion and found that self-compassion mediated the effects of body image distress and body appreciation.
The effectiveness of mindfulness and cancer-related outcomes
No meta-analysis was performed for the intervention effects on mindfulness and cancer-related outcomes due to the limited numbers of studies and the insufficient data. Mindfulness was assessed in 4 studies featuring constructive interventions immediately after intervention, 2 of which found improvements (Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018), while 2 did not (Cheung et al. Reference Cheung, Cohn and Dunn2017; Milbury et al. Reference Milbury, Weathers and Durrani2020). In follow-up results, mindfulness observation was maintained for 1 month but did not achieve significant changes in the 6th month (Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018).
Quality of life was assessed in 2 studies featuring constructive interventions, neither of which found significant changes (Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018). Moreover, Milbury et al. (Reference Milbury, Weathers and Durrani2020) found that the 4-week mindful compassion meditation intervention yielded a significant improvement in cancer-specific symptom distress. Wren et al. (Reference Wren, Shelby and Soo2019) used an audio-based loving-kindness meditation intervention in their study, and they found a significant improvement in pain and heart rate when women with breast cancer received biopsy procedures.
The Fear of Cancer Recurrence Inventory was used in 2 studies featuring constructive interventions, both of which found a significant improvement immediately after intervention in some symptoms, such as psychological distress (Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018) and functioning impairments (Dodds et al. Reference Dodds, Pace and Bell2015). Only one showed significant reductions in fear of cancer recurrence within the 6-month follow-up period (Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018). Moreover, 2 studies found that body image distress reduced immediately after constructive intervention (Latifi et al. Reference Latifi, Soltani and Mousavi2020) and brief compassionate sessions (Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). In follow-up results, Mifsud et al. (Reference Mifsud, Pehlivan and Fam2021) found that body image distress reduced significantly at 1 month after brief compassionate intervention. Body appreciation was assessed in 2 brief interventions. One found enhanced body appreciation immediately after intervention, and the appreciation could be maintained for 3 months. (Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018), while the other showed nonsignificant changes at 1 month after intervention (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021).
Discussion
This systemic review and meta-analysis has examined the effectiveness of compassion-based interventions from 10 RCTs among cancer patients with a total of 771 participants. The meta-analysis showed positive effects at the end of compassion-based interventions, in terms of decreasing depression and increasing self-compassion.
Moderate effects were found in reduced depression after compassion-based interventions. This finding was similar to one systematic review, which indicated that compassion-based interventions reduced depression and anxiety among patients with cancer or persistent pain (Austin et al. Reference Austin, Drossaert and Schroevers2020). Cancer patients commonly suffer from depression during conventional chemotherapy (Pitman et al. Reference Pitman, Suleman and Hyde2018). In our study, we found that depression was the most measured outcome when patients underwent treatment. Most of the interventions were theoretically based, and they demonstrated the mechanisms of decreased depression. Gilbert’s CFT aims to enhance patients’ emotional regulation system by having them practice compassion exercises. These exercises can increase the affiliative and soothing-oxytocin/endorphin system, which could help cancer patients to tolerate distress during the treatment period and cultivate sympathy and empathy toward themselves and others with a non-judgmental attitude, effectively reducing depression (Gilbert Reference Gilbert2010, Reference Gilbert2014). Dr. Lobsang Tenzin Negi developed CBCT, which consists of intrapersonal and interpersonal domains. CBCT uses cognitive and analytic approaches to help patients to explore personal insights by practicing self-reflection on their own life experience, resulting in diminishing stress reactions and depression and enhancing compassion (Dodds et al. Reference Dodds, Pace and Bell2015; Gonzalez-Hernandez et al. Reference Gonzalez-Hernandez, Romero and Campos2018).
The results from the meta-analysis on reducing anxiety were not significant, which was inconsistent with a previous meta-analysis of compassion-based interventions in nonclinical populations (Kirby et al. Reference Kirby, Tellegen and Steindl2017). However, after excluding one potential outlier, small to moderate effects were found in reduced anxiety after compassion-based intervention; therefore, more studies are necessary to examine the effects on anxiety. In our systematic review, we found that anxiety was the most measured outcome in brief compassion during the posttreatment cancer survivorship stage. A systematic review showed that anxiety, rather than depression, is most likely to be a problem in long-term cancer survivors compared with healthy controls (Mitchell et al. Reference Mitchell, Ferguson and Gill2013).
Moderate to large effects were found in increased self-compassion after compassion-based intervention. This result is similar to a meta-analysis of 27 RCTs, which showed that compassion-based intervention produced a moderately significant improvement in self-compassion among both nonclinical people and patients with mental health symptoms (Ferrari et al. Reference Ferrari, Hunt and Harrysunker2019). In our subgroup analysis, although both constructive and brief intervention could increase self-compassion compared to the control condition, constructive intervention showed more benefits via increased self-compassion than brief intervention. A possible reason might be because constructive interventions last longer and have more comprehensive compassion practices to help patients to cultivate their self-compassion abilities. Cultivating self-compassion needs a deep awareness of self and others. By bringing mindfulness into patients’ daily lives, patients start to accept their physical or mental suffering from cancer without avoidance; furthermore, common humanity could help cancer patients not to feel so lonely or isolated from others (Neff Reference Neff2015; Neff and Germer Reference Neff and Germer2013). The self-compassion motivation and actions become firmer and more stable after several weeks of constructive compassion-based intervention. Cancer patients learn to treat themselves with kindness and compassion, which could facilitate emotional regulation and protect them from psychopathological symptoms (Pinto-Gouveia et al. Reference Pinto-Gouveia, Duarte and Matos2014). We found that most of the studies used self-compassion scales as a measurement of compassion; previous studies have reported that there are 3 orientation flows of compassion: self-compassion, compassion from others, and compassion for others, which are moderately correlated with one another (Gilbert et al. Reference Gilbert, Catarino and Duarte2017; Neff and Germer Reference Neff and Germer2013). More compassion orientation flows could be explored in the future.
Both non-face-to-face and face-to-face delivered formats had benefits for increased self-compassion over the control condition. There was no difference between these 2 groups in the subgroup analysis. Online-delivered formats such as meditation interventions via FaceTime and online compassion-based writing are acceptable due to their low cost and the minimal user time (Mifsud et al. Reference Mifsud, Pehlivan and Fam2021; Milbury et al. Reference Milbury, Weathers and Durrani2020; Sherman et al. Reference Sherman, Przezdziecki and Alcorso2018). Sotirova et al.’s (Reference Sotirova, McCaughan and Ramsey2021) systematic review also indicated that internet-based interventions could increase acceptability and satisfaction, and they are cost-effective. With a face-to-face format intervention, it is easier to build a trusting relationship between therapists and patients (Ash et al. Reference Ash, Harrison and Pinto2021). Patients might be able to concentrate better on the present moment because both security and nurturing are foundational components of compassion-based intervention. Integrating an online approach with a face-to-face format might be a future challenge.
Implications and recommendations
The implication of this study is that it is important to develop constructive compassion-based or brief interventions for cancer patients. Since cancer is a chronic disease, patients have to learn how to live with it from the active cancer treatment stage to the posttreatment cancer survivorship stage. Most brief compassion-based interventions were conducted during the posttreatment cancer survivorship stage, and anxiety was the most measured outcome. Most constructive compassion-based interventions were conducted when patients were undergoing treatment, and depression was the most measured outcome. Recommendations for future researchers include measuring biological outcomes, home practice times, and long-term follow-up effects and examining the different compassion orientations. In our review, we found that self-compassion was the most measured concept in compassion-based interventions; therefore, more compassion orientations such as compassion for others or compassion from others could be explored. Recommendations for health-care providers include developing personalized interventions for different cancer treatment stages or different delivery formats (online, face-to-face, or a combination of both) for cancer patients.
Limitations
First, this review was limited by the small number of studies and a small number of studies; only one outcome had sufficient data to conduct subgroup analysis. Second, all the studies were based on Western cancer patients and approximately 98% of participants were female; therefore, the results of the meta-analysis may not direct apply to non-Western or to male cancer patients. Third, most of the included studies were on female breast cancer patients. The effect of compassion-based interventions on other types of cancer or different gender still needs further research. Fourth, the heterogeneity of some moderator effects remained high. Finally, the follow-up effects of compassion-based intervention were not established in this meta-analysis due to the wide range of follow-up time in the included papers. More studies are necessary to identify the long-term effectiveness of compassion-based interventions.
Conclusion
This is the first systematic review and meta-analysis of compassion-based intervention studies with RCT designs focused on cancer patients. Most compassion programs were developed and examined their effects in female breast cancer patients. The systematic review has identified the constructive compassion-based and brief interventions in online or face-to-face formats. The meta-analysis was based on small sample sizes and a small number of studies, and it suggests that compassion-based interventions might provide an acceptable and effective strategy for improving self-compassion and depression among female patients with breast cancer.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1478951522001316.
Conflicts of interest
None.
Open access
