Effectiveness of psychological interventions for compassion fatigue: a systematic review and meta-analysis

Abstract

Objectives: Constant exposure to trauma and death adversely affects the physical, emotional, and mental stability of the helping professionals, leading to compassion fatigue (CF). Although research has explored the effect of various psychological interventions (PIs) on CF, no studies have quantitatively synthesized their effectiveness. Thus, the current systematic review and meta-analysis address this gap by examining the efficacy of PIs in reducing CF among helping professionals.

Methods: Databases such as Scopus, Web of Science, PubMed, JSTOR, ScienceDirect, and Google Scholar were searched to identify studies published between 2004 and 2023. Quantitative studies published in English that used any PI to reduce CF of helping professionals were selected for analysis. The risk of bias in studies was evaluated using Joanna Briggs Institute (JBI) checklists and the National Heart, Lung, and Blood Institute quality assessment tool.

Results: Of the 1995 records identified from databases, 82 intervention studies were included in the systematic review, and 11 randomized controlled trials were chosen for meta-analysis. Post-intervention means and SDs of the intervention group and the control group were used to run the meta-analysis. Random effects meta-analysis results revealed that PIs are highly effective in reducing CF (standardized mean difference (SMD) = −0.95; 95% CI, −1.63 to −0.27; P = .006).

Conclusion: Evidence suggests that PIs, especially online-delivered PIs, could reduce the CF of helping professionals. Health policymakers, concerned authorities, and intervention designers should focus on reducing the CF of helping professionals, as they need to work with vulnerable populations efficiently.

Key points

• What is already known on this topic: Fifteen percent of working adults experience mental illness, with helping professionals at higher risk. Compassion fatigue (CF) is prevalent among helping professionals due to the prolonged exposure to secondary trauma from their patients or clients. The extensive literature on CF investigates the various risk factors and the interventions to reduce CF among professionals. However, there is a lack of literature on the quantitative synthesis of the effectiveness of psychological interventions (PIs) on CF among helping professionals.

• What this study adds: Findings of the meta-analysis suggest that PIs are effective in reducing CF of helping professionals. Subgroup meta-analysis on the mode of interventions found that online-delivered PIs showed a higher effect size in reducing CF when compared with offline-delivered PIs. The systematic review revealed that mindfulness-based interventions, psychoeducational interventions, and resiliency interventions were commonly used PIs.

• How this study might affect research, practice, or policy: Results highlight that integration of online-delivered PIs into the training of helping professionals will reduce their CF and lead to better productivity. Therefore, the government and other health care authorities should consider incorporating these technology-based PIs into the interventions of helping professionals, as reducing CF is indispensable for their mental well-being and workplace productivity.

1. Introduction

Globally, 15% of working adults experience mental illness in a year, and 12 billion working days are lost every year due to various mental health conditions such as anxiety and depression.¹ In particular, the same organization reports that helping professionals are at a higher risk of mental illness. Professionally, people who engage in helping practices include nurses, psychologists, psychotherapists, social workers, cancer care providers, physicians, home nurses, emergency care providers, and police officers.² Helping professionals are indispensable to patients’ or clients’ well-being as they offer quality care, empathy, compassion, and emotional energy. On the other hand, previous studies reported that helping professionals experience high stress levels in the workplace,³ and continual exposure to trauma and death can affect their physical, emotional, and mental stability.⁴ Notably, such difficulties might lead to the development of a condition called compassion fatigue (CF),⁵ similar to post-traumatic stress disorder, with symptoms such as depression, low energy, increased irritability, anxiety, and disruption in sleep patterns.⁶

CF is “a state of exhaustion and dysfunction biologically, psychologically, and socially as a result of prolonged exposure to compassion stress and all it invokes.”⁷ Stamm and colleagues⁸ revealed that CF comprises burnout and secondary traumatic stress. Further, it is characterized by irritability and anger, decreased ability to feel empathy and sympathy, reduced sense of job satisfaction, increased absenteeism, unhealthy coping mechanisms such as substance abuse, and impaired decision-making regarding patients or clients.⁷ Besides, CF affects 7 spheres of an individual, including behavioral, emotional, interpersonal, spiritual, occupational, cognitive, and physiological aspects.² Literature shows that the prevalence of CF ranges from 7.3% to 40% among emergency care workers,² from 21.6% to 44.8% among nurses,⁹ and is reportedly 75.96% among oncology health care professionals,¹⁰ highlighting the importance of addressing CF among helping professionals.

Many studies have focused on identifying the causal factors of CF among helping professionals. Dasan and colleagues¹¹ found that the professional’s past traumatic life events and past instances of poor mental health contribute to the development of CF. In their study, Beaumont and colleagues¹² reported that a lack of compassion toward self could be a major causal factor for CF. Additionally, various organizational factors such as lack of social support from colleagues and increased exposure to secondary trauma from the clients,¹³ physically and emotionally taxing tasks, and more workdays¹⁴ act as risk factors for CF among helping professionals.

Reducing CF among helping professionals has been a major interest of researchers over the years, with most psychological interventions (PIs) targeted at the risk factors of CF. For instance, the literature showed the effectiveness of several mindfulness interventions,^15‑17 psychoeducational interventions,^18‑20 resiliency interventions,^21,22 and self-compassion training programs²³ to reduce CF among helping professionals. Although a few systematic reviews are available of CF among various helping professionals,^7,24,25 to the best of the authors’ knowledge, no studies aimed to quantitatively synthesize the effectiveness of PIs for their CF. Therefore, through a systematic review and meta-analysis, the current study examined the effectiveness of PIs in reducing CF among helping professionals.

2. Methods

This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)²⁶ and registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the registration ID: CRD42022373763.

2.1. Eligibility criteria

The study inclusion criteria adhered to the Population, Intervention, Comparator, and Outcomes (PICO) guidelines. Population: studies with helping professionals as samples were included. A helping profession refers to “professional interaction between a helping expert and a client, initiated to nurture the growth of, or address the problems of a person's physical, psychological, intellectual, or emotional constitution including medicine, nursing, psychotherapy, psychological counseling, social work, education, or coaching.”²⁷ No limits were placed on the race and sex of helping professionals. Intervention: the review targeted studies that employed PIs for helping professionals in reducing their CF. Both in-person (exercise led by a therapist) and online therapies (using downloadable audio tapes) were included. Comparator: studies comparing the effect of PIs on reducing CF to a control condition were chosen for meta-analysis. Outcome: studies were eligible for meta-analysis if they contained a validated, continuous clinical measure of CF and were a randomized controlled trial (RCT) providing a post-intervention mean and SD score of the experimental and control groups. RCTs, quasi-experimental studies with control groups, and pre-post-design studies were included for systematic review. Other study designs, such as qualitative studies, secondary analysis, observational studies, animal trials, research programs, conference papers, books, and ongoing trials, were excluded.

2.2. Information sources and search strategy

Two reviewers (J.M.L. and A.G.) in December 2023 independently searched electronic databases such as PubMed, Web of Science, Scopus, JSTOR, ScienceDirect, and Google Scholar for studies published from 2004 to 2023. Studies of CF have grown since 2004. Three groups of search terms were used. Group 1 contained CF-related words, such as secondary traumatic stress or vicarious traumatization. Group 2 key terms included study design, for which the study employed words for identifying trials and interventions (eg, training or intervention or efficacy or program or randomized control group), and Group 3 included employment types who are at risk of frequent exposure to CF risk factors, such as emergency, frontline health, and community health workers. Keywords were combined in each database using Boolean operators (AND/OR). Search strategies used in each database are provided in Table S1.

2.3. Selection process

The results identified from each database were exported to Zotero reference management software²⁸ to deduplicate records. Two authors (J.M.L. and A.G.) screened the titles and abstracts of the remaining records according to the study eligibility criteria. Then, studies that did not meet inclusion criteria (such as theoretical articles, case studies, and qualitative studies) were eliminated. The same authors then independently evaluated these eligible studies with full-text accessibility. Discussion with a third author (E.R.) assisted in resolving disagreements during the selection process.

2.4. Data item and data collection process

The present study considered CF as its data item. CF is “a state of exhaustion and dysfunction biologically, psychologically, and socially as a result of prolonged exposure to compassion stress and all it invokes.”⁷ Following the full-text analysis of eligible reports, significant information from the finalized studies of CF was recorded into a data extraction sheet developed by the first author (J.M.L.). Two reviewers (J.M.L. and A.G.) independently extracted the following data: (1) author name and publication year; (2) country; (3) study design; (4) sample characteristics (sample and sample size); (5) intervention descriptors (type of PI being used, treatment modality, duration of intervention); (6) outcome measures; (7) follow-up, and (8) conclusions.

2.5. Study risk-of-bias assessment

The quality assessment of included studies was independently done by 2 reviewers (J.M.L. and A.G.) using the Joanna Briggs Institute (JBI) checklist for RCT and quasi-experimental studies and the National Heart, Lung, and Blood Institute’s quality assessment tool for pre-post studies with no control group. For each fulfilled criterion, a point was given. The total score range based on the number of “Yes” assessments is between 0 and 13 for RCTs and between 0 and 9 for quasi-experimental studies. For studies using RCTs, a score of 1-4 indicated "low quality,” a score of 5-8 indicated " medium quality,” and a score of 9-13 indicated "high quality."²⁹ For the quasi-experimental studies, a quality score of 6 and above was considered for inclusion in the current review.³⁰ For studies using pre-post design, a quality score of 9-12 was considered high quality, 7-8 moderate quality, and a score of ≤6 low quality.³¹

2.6. Effect measures and statistical analyses

The outcome measure’s mean and SD scores of the intervention and control conditions during the post-intervention phase were used to perform meta-analyses by calculating standardized mean difference (SMD). Version 5.3 of Review Manager (RevMan)³² was used to perform the appropriate statistical analyses. Forest plots were constructed to pool data from studies with comparable interventions and results. For each pooled estimate, the I² value was used to evaluate between-study heterogeneity. The study used a random-effects model due to the presence of heterogeneity. A subgroup analysis was conducted based on the mode of administration of the intervention.

2.7. Publication bias

To measure publication bias, funnel plots were used. The Egger test was performed to test small study effects in the finalized reports, with P < .05.

3. Results

3.1. Study selection

Of the 1995 results identified from the database search, 1483 remained after eliminating 512 duplicates (Figure 1). Among these 1483 studies, 1339 failed to fulfill the requirements for inclusion when titles and abstracts were screened, thus leaving 144 articles for full-text analysis. Of the 144 articles sought for full-text retrieval, 7 articles were not accessible. During the full-text screening of 137 articles, 55 failed to fulfill the requirements for inclusion due to various reasons such as not assessing CF (n = 19), non-intervention studies of CF (n = 16), not psychological interventions (n = 4), not helping professionals (n = 6), no standardized CF outcome measure (n = 3), qualitative analysis (n = 2), duplicate (n = 1) and protocol documents for intervention studies (n = 4). Thus, the final data extraction process comprised 82 reports for systematic review. However, for the meta-analysis, RCTs providing post-intervention mean and SD scores of the experimental and control groups were included. RCTs adhere to randomization of participants and control the inequalities between the groups, making it advisable to ignore the baseline scores for meta-analysis. However, of 13 RCTs in the finalized reports, 3 did not report the required values,^33‑35 leading to 10 studies in the meta-analysis. Additionally, the present study could extract 2 PIs for CF from the same study,³⁶ contributing 11 data in total for the meta-analysis. Although the present review intended to examine the interventions’ effectiveness using change mean scores and SDs from baseline to post-intervention phase, only limited studies provided these parameters. Despite the reviewers contacting the corresponding authors to obtain the required scores, only 5 responses were received, of which 1 provided the relevant data.

Figure 1.

PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram.

3.2. Study characteristics

The final sample for systematic review included 82 intervention studies published during 2004-2023 (Table 1). The helping professionals included nurses (n = 39),^{18‑23,33,34,36–66} health care professionals (n = 23),^16,67–88 social workers (n = 8),^{15,17,89–94} mental health professionals (n = 4),^35,95–97 doctors (n = 3),^98‑100 special educators (n = 1),¹⁰¹ hospice workers (n = 1),¹⁰² residential caregivers (n = 1),¹⁰³ air medical crew members (n = 1),¹⁰⁴ and child welfare workforce (n = 1).¹⁰⁵ The total review sample size was 4859, in which the study by Orsi-Hunt and colleagues¹⁰⁴ contained the largest sample (n = 375), of the child welfare workforce, whereas Best and colleagues⁴¹ contained the smallest sample (n = 4), of military nurses.

Table 1.

Summary of study characteristics (n = 82).

Author(s) and year	Country	Study design	Sample	Sample size	Type of intervention	Mode	Duration	Measures of CF	Follow-up	Conclusions
Potter et al (2013a)22	United States	Pre-post	Oncology staff nurses	13	CF resiliency program	Offline	5 wk	ProQOL-4	3 mo, 6 mo	Reduction in CF
Gentry et al (2004)95	Canada	Pre-post	Mental health professionals	83	Certified CF specialist training (CCFST)	Offline	17 h and 20 h	Compassion Satisfaction/Fatigue Self-Test (CSFST)	Nil	Reduction in CF
Qualls et al (2022)37	United States	Pre-post	Oncology nurses	12	Mindfulness-based stress reduction (MBSR)	Online	8 meditations	ProQOL-V	Nil	Reduction in CF
Han & Kim (2021)89	South Korea	Quasi	Social workers	105	Simulation-based empathy enhancement program	Offline	2 wk	ProQOL-V	Nil	Reduction in CF
Van Kirk (2021)96	United States	Pre-post	Mental health professionals	56	Employee wellness program	Offline	9 wk	ProQOL-V	Nil	No significant reduction
Pandya (2021)90	South Asian countries	RCT	Geriatric social workers	144	Meditation program	Online	3 mo	ProQOL-V	Nil	Reduction in CF
Phillips et al (2020)38	United States	Quasi	Oncology nurses	43	Storytelling through music (STM)	Offline	—	ProQOL-V	1 mo	Reduction in CF
Weinlander et al (2020)67	Not mentioned	Quasi	Health care workers	281	Mind–body skills professional training programs	Offline	5 d	ProQOL	3 mo, 12 mo	Reduction in CF
Kinman et al (2020)17	United Kingdom	Pre-post	Social workers	18	Mindfulness training course	Offline	8 wk	ProQOL	Nil	Reduction in CF
Blackburn et al (2020)21	United States	Quasi	Oncology nurses	164	THRIVE program	Offline	6 wk	Compassion Fatigue Short Scale (CFSS)	Nil	Reduction in CF
Yilmaz et al (2018)66	Turkey	Pre-post	Oncology nurses	43	Nurse-led intervention	Offline	5 wk	ProQOL-IV	2 wk	Reduction in CF
Klein et al (2018)79	Not mentioned	Pre-post	Health care professionals	15	Resiliency program	Offline	3 sessions	ProQOL-V	6 mo	No significant reduction
Kiley et al (2018)35	Not mentioned	RCT	Mental health professionals	69	Prerecorded guided imagery (GI)	Online	4 wk	ProQOL	Nil	No significant reduction
Kinman and Grant (2016)92	United Kingdom	Quasi	Early-career social workers	56	Multi-modal intervention	Offline	2 mo	ProQOL	Nil	No significant reduction
Duarte and Pinto-Gouveia (2017)47	Portugal	Quasi	Oncology nurses	48	Mindfulness-based intervention	Offline	6 wk	ProQOL-V	Nil	Reduction in CF
Crowder and Sears (2017)15	Canada	Quasi	Social workers	14	Mindfulness-based intervention	Offline	8 wk	ProQOL-V	1 wk, 13 wk, 26 wk	No significant reduction
Jakel et al (2016)54	United States	Quasi	Oncology nurses	25	Provider resilience mobile application	Online	6 wk	ProQOL-V	Nil	Reduction in CF
Gregory (2015)91	Not mentioned	Quasi	Social workers	11	Yoga and mindfulness	Offline	3 wk	ProQOL-V	Nil	No significant reduction
James et al (2014)97	Haiti	Quasi	Mental health professionals	8	Psychoeducational intervention	Offline	1 wk	ProQOL-V	Nil	No significant reduction
Hilliard (2006)76	Not mentioned	Pre-post	Professional hospice workers	17	Music therapy	Offline	6 wk	Compassion Satisfaction/Fatigue Self-Test for Helpers (CFS)	Nil	No significant reduction

(Continued)

Table 1.

Continued

Author(s) and year	Country	Study design	Sample	Sample size	Type of intervention	Mode	Duration	Measures of CF	Follow-up	Conclusions
Best et al (2020)41	United States	Pre-post	Military nurse	4	Mindfulness	Online	8 wk	ProQOL	Nil	Reduction in CF
Sharp Donahoo et al (2018)101	United States	Pre-post	Staff of special education institute	27	Educational intervention	Offline	3 h	ProQOL	4 wk	No significant reduction
Abernathy and Martin (2019)39	United States	Pre-post	Pediatric ICU nurses	19	Mindfulness and self-care	Online	30 d	ProQOL-V	Nil	Reduction in CF
Dreher et al (2019)19	United States	Pre-post	Nursing assistants	45	Educational intervention	Offline	90 min	ProQOL-V	1 mo, 3 mo	Reduction in CF
Pfaff et al (2017)81	Canada	Pre-post	Interprofessional staff at a regional cancer center	12	CF resiliency program	Offline	6 wk	ProQOL-V	Nil	No significant reduction
Tucker et al (2017)85	Canada	Pre-post	Medical trainees	109	Self-reflection exercise	Offline	—	ProQOL	Beginning, middle, and end of academic year	No significant reduction
Flarity et al (2013)20	United States	Pre-post	Emergency nurses	59	Educational intervention	Offline	—	ProQOL-V	Nil	Reduction in CF
Flarity et al (2016)50	United States	Pre-post	Graduate nurse residents	94	CF resiliency program	Offline	4 h	ProQOL-V	2 mo	Reduction in CF
Allen et al (2017)98	Australia	Pre-post	Gynecology doctors	25	Work-focused discussion groups	Offline	6 mo	ProQOL-V	3 mo, 6 mo	Reduction in CF
Owens et al (2020)56	United States	Pre-post	Nurses	32	Mindfulness	Online	4 wk	ProQOL-V	Nil	Reduction in CF
Copeland (2021)43	Not mentioned	Quasi	Nurses	23	Wellness program	Offline	6 wk	ProQOL-V	Nil	Reduction in CF
Delaney (2018)45	Not mentioned	Quasi	Nurses	13	Mindful self-compassion (MSC) training	Offline	8 wk	ProQOL-V	Nil	Reduction in CF
Rajeswari et al (2020)60	India	RCT	Nurses	120	Accelerated recovery program	Offline	5 wk	ProQOL-V	3 mo, 6 mo, 9 mo, 12 mo	Reduction in CF
Pérez et al (2022)58	Spain	Quasi	Geriatric nurses	74	Mindfulness-based therapies	Online	6 wk	ProQOL-IV	3 mo, 6 mo	Reduction in CF
Heeter et al (2017)75	Not mentioned	Quasi	Hospice and palliative professionals	36	Yoga therapy meditation program	Online	6 wk	ProQOL-V	Nil	Reduction in CF
Wahl et al (2018)64	United States	Pre-post	Nurses	20	Peer support network (PSN)	Offline	1 d	ProQOL-V	Nil	No significant reduction
Beres et al (2022)68	Not mentioned	Pre-post	Emergency and intensive care health care professionals	25	Structured debriefing	Online	12 wk	ProQOL-V	Nil	No significant reduction
Maegli (2014)93	Not mentioned	Pre-post	Social service workers	7	Appreciative inquiry (AI) intervention	Offline	6 h	ProQOL	Nil	No significant reduction
Potter et al (2013b)59	Not mentioned	Pre-post	Oncology staff	85	Resiliency program	Offline	8 h	ProQOL-IV	Nil	Reduction in CF
Shaker et al (2022)83	Iran	Quasi	Family caregivers	58	Online support group	Online	5 wk	ProQOL	Nil	No significant reduction
Goktas (2022)51	Turkey	RCT	Emergency nurses	60	Daily motivational messages	Online	21 d	Compassion Fatigue Scale	Nil	Reduction in CF
Trowbridge et al (2017)94	United States	Pre-post	Pediatric medical social workers	21	Compressed mindfulness-based stress reduction (cMBSR) course	Offline	6 wk	ProQOL	Nil	No significant reduction

(Continued)

Table 1.

Continued

Author(s) and year	Country	Study design	Sample	Sample size	Type of intervention	Mode	Duration	Measures of CF	Follow-up	Conclusions
Chia-Yun et al (2021)18	Taiwan	Quasi	Nurses	124	Educational intervention	Offline	4 wk	ProQOL-V	4 wk, 12 wk	Reduction in CF
Slatyer et al (2018)62	Australia	Quasi	Nurses	91	Brief mindful self-care and resiliency intervention	Offline	8 wk	ProQOL-V	6 mo	Reduction in CF
Craigie et al (2016)44	Australia	Pre-post	Nurses	20	Mindful self-care and resiliency (MSCR) intervention	Offline	4 wk	ProQOL-V	1 mo	No significant reduction
Vermeesch et al (2022)34	United States	RCT	Undergraduate nursing students	51	Nature-based intervention (NBI)	Online	4 wk	ProQOL-V	Between 1 wk, 2 wk, 3 wk, 4 wk	No significant reduction
O’Riordan et al (2020)100	Ireland	Pre-post	Doctors and midwives	5	Short work-focused discussion group	Offline	—	ProQOL	Nil	Reduction in CF
Wylde et al (2017)65	United States	Quasi	Novice pediatric nurses	95	Mindfulness	Both	4 wk	Compassion Fatigue Self Test	Nil	No significant reduction
Kaur et al (2021)16	Not mentioned	Pre-post	Palliative cancer care professionals	25	Mindfulness cognitive behavioral interventions (MICBI)	Offline	6 wk	ProQOL-V	3 mo	Reduction in CF
Moreno-Jiménez et al (2020)99	Mexico	Quasi	Physicians of intensive care units	8	Psychoeducational intervention	Offline	5 wk	Secondary Traumatic Stress Scale	Nil	Reduction in CF
Powell and Yuma-Guerrero (2016)82	United States	Pre-post	Community health workers	45	Resilience and coping for the health care community (RCHC) intervention	Offline	1 d	ProQOL-V	3 wk	No significant reduction
Berger and Gelkopf (2011)40	Gaza	Quasi	Baby clinic nurses	80	Psychoeducational intervention	Offline	12 wk	ProQOL	Nil	Reduction in CF
Franco and Christie (2021)23	United States	Quasi	Pediatric nurses	48	Self-compassion training	Offline	1 d	ProQOL	3 mo	No significant reduction
Kang et al (2021)78	Not mentioned	Quasi	Hospice palliative care	13	Meaning-centered, spiritual care training program (McSCTP)	Offline	4 wk	—	4 wk	Reduction in CF
Singh et al (2020)84	Not mentioned	Pre-post	Caregivers of intellectually disabled and autistic individuals	123	Mindfulness-based positive behavior support (MBPBS) and positive behavior support (PBS)	Offline	10 wk	ProQOL	Nil	Reduction in CF
Sullivan et al (2019)63	United States	Quasi	Pediatric oncology nurses	59	Unit-based self-care program intervention	Offline	—	ProQOL-V	2 mo, 4 mo, 6 mo	Reduction in CF
Hevezi (2016)53	United States	Pre-post	Nurses	15	Structured meditations	Online	4 wk	ProQOL-V	Nil	Reduction in CF
Bhardwaj et al (2023)69	India	RCT	Health care providers	98	Mobile health intervention	Online	12 wk	ProQOL	Nil	Reduction in CF
Pehlivan and Güner (2020)36	Turkey	RCT	Oncology-hematology nurses	91	Compassion fatigue resiliency program	Offline	5 wk	ProQOL-IV	3 mo, 6 mo, 1 y	Reduction in CF
Orsi-Hunt et al (2023)105	United States	RCT	Child welfare workforce	375	Resilience program	Offline	6 mo	Secondary Traumatic Stress Scale	Nil	No reduction in CF

(Continued)

Table 1.

Continued

Author(s) and year	Country	Study design	Sample	Sample size	Type of intervention	Mode	Duration	Measures of CF	Follow-up	Conclusions
Wayment et al (2019)87	United States	Pre-post	Health care professionals	37	Educational workshop	Offline	2 wk	ProQOL	Nil	Reduction in CF
Johansson et al (2022)77	Not mentioned	RCT	Health care professionals	21	Internet-based compassion course	Online	5 wk	ProQOL-V	10 wk, 15 wk, 6 mo	Reduction in CF
Grabbe et al (2020)52	Not mentioned	Pre-post	Nurses	32	Resiliency intervention	Offline	3 h	Secondary Traumatic Stress Scale (STSS)	1 wk, 3 mo, 1 y	Reduction in CF
Fiore (2021)102	United States	Pre-post	Hospice workers	13	Mindfulness-based intervention	Online	20 min	ProQOL	Nil	No reduction in CF
Halamová et al (2022)73	Not mentioned	RCT	Health care professionals	253	Emotion-focused training	Online	2 wk	Secondary Traumatic Stress Scale (STSS)	2 mo	Reduction in CF
Santos et al (2023)103	Portugal	RCT	Residential caregivers	108	Compassionate mind training	Offline	12 wk	ProQOL-V	3 mo, 6 mo	No reduction in CF
Asadollah et al (2023)33	Iran	RCT	Nurses	66	Loving-kindness meditation	Offline	6 wk	Nursing Compassion Fatigue Inventory (NCFI)	Nil	Reduction in CF
Watts et al (2021)86	Australia	Pre-post	Health care professionals	31	Mindfulness-based compassion training intervention	Offline	6 wk	ProQOL-V	8 wk	Reduction in CF
Partlak Günüşen et al (2022)57	Turkey	RCT	Nurses	48	Cognitive behavioral approach	Offline	4 wk	ProQOL-IV	6 mo	No reduction in CF
Zajac et al (2017)88	United States	Pre-post	Health care professionals	91	Debriefs	Offline	3 mo	ProQOL-V	Nil	Reduction in CF
Hayakawa et al (2022)74	United States	Pre-post	Health care professionals	71	Classical musical virtual reality	Offline	5 wk	ProQOL	Nil	Reduction in CF
Kestler et al (2020)55	Not mentioned	Pre-post	Nurses	25	Educational intervention	Offline	3 wk	Secondary Traumatic Stress Scale (STSS)	Nil	Reduction in CF
Charlescraft et al (2010)42	United States	Pre-post	Pediatric intensive care unit nurses	6	Self-care intervention	Offline	5 wk	Compassion Satisfaction and Fatigue Test	Nil	Reduction in CF
Flanders et al (2020)49	United States	Pre-post	Pediatric intensive care unit nurses	75	Resilience program	Offline	Not mentioned	ProQOL	Nil	Reduction in CF
McCall (2023)104	United States	Quasi	Air medical crew members	36	Peer support program	Offline	Not mentioned	ProQOL	Nil	No reduction in CF
Burnett et al (2023)70	Georgia	Pre-post	Emergency department health care workers	14	Community resilience model	Offline	1 h	ProQOL	2 wk, 12 wk	Reduction in CF
Drury et al (2022)46	United States	Pre-post	Oncology nurses	10	Structured mentorship program	Offline	12 mo	ProQOL	Nil	Reduction in CF
Cain and Gautreaux (2022)71	United States	Pre-post	Health care providers	84	Educational intervention	Offline	1 h	ProQOL	6 mo	Reduction in CF
Saribudak et al (2020)61	Turkey	Pre-post	Nurse managers	16	Compassion fatigue resiliency program	Online	2 d	ProQOL-IV	Nil	Reduction in CF
Egami et al (2023)48	United States	Pre-post	Nurses	9	Mindfulness phone application	Online	3 wk	ProQOL-V	Nil	Reduction in CF
Dobrina et al (2023)72	Italy	Pre-post	Health care professionals	48	Narrative medicine intervention	Offline	5 wk	ProQOL-V	Nil	Reduction in CF
McCool et al (2022)80	United States	Pre-post	Health care professionals	12	Peer support (buddy system)	Online	6 mo	ProQOL-V	Nil	No reduction in CF

Abbreviations: CF, compassion fatigue; pre-post, pre-post design; ProQOL, Professional Quality of Life; quasi, quasi-experimental design; RCT, randomized controlled trial.

The PIs included mindfulness-based interventions (n = 14),^{15,16,37,41,47,48,56,58,64,84,86,92,94,102} in which various studies adopted various principles of mindfulness. For instance, the study by Trowbridge and colleagues⁹⁴ adopted the principles of a mindfulness-based stress reduction program by Santorelli¹⁰⁶ and included exercises like meditation practice time, didactic periods, body scan, mindful movement, and individual sharing time. Other PIs were resiliency intervention (n = 14),^{21,22,36,49,50,52,54,59,61,70,79,81,82,105} psychoeducational interventions (n = 11),^{18–20,40,55,66,71,87,97,99,101} multicomponent interventions (n = 10),^{17,43,44,60,62,69,77,95,96,103} self-care enhancement interventions (n = 5),^{39,42,63,67,85} self-compassion training programs (n = 2),^24,45 yoga (n = 2),^75,91 meditation (n = 3),^33,53,90 music interventions (n = 3),^38,74,76 guided imagery (n = 1),³⁵ spiritual interventions (n = 1),⁷⁸ nature-based interventions (n = 1),³⁴ debriefing sessions (n = 2),^68,88 sending motivational messages (n = 1),⁵¹ empathy enhancement program (n = 1),⁷⁸ emotion-focused training (n = 1),⁷³ cognitive-based approach (n = 1),⁵⁷ narrative medicine intervention (n = 1),⁷² and appreciative inquiry (n = 1).⁹³ Seven studies focused on group work and included peer support group interventions.^{46,64,80,83,98,100,104}

The included studies were conducted in the United States (n = 33),^{19–23,34,37–39,41,42,46,48–50,53,54,56,63–65,71,74,80,82,87,88,94,96,102,104,105} Turkey (n = 5),^{36,51,57,61,66} Canada (n = 4),^15,81,85,95 Australia (n = 4),^44,62,86,98 India (n = 2),^60,69 United Kingdom (n = 2),^17,92 Portugal (n = 2),^47,103 Iran (n = 2),^33,83 South Korea (n = 1),⁸⁹ South Asian countries (n = 1),⁹⁰ Haiti (n = 1),⁹⁷ Spain (n = 1),⁵⁸ Taiwan (n = 1),¹⁹ Ireland (n = 1),¹⁰⁰ Mexico (n = 1),⁹⁹ Gaza (n = 1),⁴⁰ Italy (n = 1),⁷² and Georgia (n = 1)⁷⁰; some studies did not mention the country details (n = 18).^{16,35,43,45,53,55,59,67,68,73,75–79,84,91,93}

The study designs of the finalized studies were RCTs (n = 13),^{33–36,51,57,60,69,73,77,90,103,105} pre-post design (n = 45), and quasi-experimental design (n = 24). The mode of administering the intervention was offline (n = 60), online (n = 21),^{34,35,37,39,41,48,51,53,54,56,58,61,68,69,73,75,77,80,83,90,102} or both (n = 1).⁶⁴ Among the 82 studies, 69 used a version of the Professional Quality of Life Scale (ProQOL) to measure CF. Other measures used were the Compassion Fatigue Short Scale (n = 2),^21,51 Compassion Satisfaction and Fatigue Test (n = 3),^37,42,74 Compassion Fatigue Self-Test (n = 3),^64,75,94 and Secondary Traumatic Stress Scale (n = 4)^52,55,73,99; 1 study did not mention the outcome measure.⁷⁸ For instance, Partlak Günüşen and colleagues⁵⁷ assessed the effect of an offline-administered nurse-led educational intervention in reducing the CF of nurses. The study used an RCT design and ProQOL was used to measure the effectiveness of intervention on levels of CF among the nurses.

3.3. Risk of bias in studies

The quality assessment of RCTs found that 12 studies were of high quality and 1 study was of moderate quality. The methodological quality was limited for a few studies as they did not report clear information regarding the concealed method of allocation of intervention groups, whether participants, those delivering the intervention, and outcome assessors were blinded to treatment assignment groups, and participants were not analyzed within the allocated groups. The quality assessment of quasi-experimental studies found that 22 studies were of high quality. The quality assessment of pre-post experimental studies found that 20 studies were of high quality and 27 were of moderate quality. The included studies were limited in quality due to unclear information about whether the outcome assessors were blinded to participants’ interventions and the usage of individual level data to determine the effects at the group level. Tables S2, S3, and S4 provide detailed information about study quality.

Upon the qualitative synthesis of the included studies, it was found that CF was mainly studied in the nursing population, and mindfulness-based interventions were mostly given to the helping professionals, followed by resiliency and psychoeducational interventions. It was also found that many PI studies on CF were conducted in the United States. A majority of the included studies used a quasi-experimental design, and the mode of delivery of the intervention was primarily offline. It was also found that various versions of ProQOL were widely used in CF research.

3.4. Effect of PIs on CF

Of the 82 studies, 11 were suitable for meta-analysis as the reported outcomes of removed studies did not meet the study eligibility criteria.

Overall meta-analysis (see Figure 2) of a random effect model revealed a significant pooled effect size of the various PIs with a large effect size (SMD = −0.95; 95% CI, −1.63 to −0.27; P = .006), suggesting that PIs significantly reduce CF among the helping professionals. The study results showed significant heterogeneity (I² = 96%; P < .00001), which was to be expected given the variety of support intervention strategies.

Figure 2.

Effect of psychological interventions for compassion fatigue.

3.4.1. Subgroup analysis on the mode of intervention administration

The subgroup analysis was done based on the mode of intervention, that is, online or offline (see Figure 3). Subgroup meta-analysis found a significant large effect size for the online-administered interventions (SMD = −1.22; 95% CI, −2.02 to −0.42; P = .003) and a nonsignificant medium effect size for the offline-administered interventions (SMD = −0.73; 95% CI, −1.65 to 0.18; P = .12). Both online-administered (I² = 91%) and offline-administered interventions (I² = 97%) had a significant high heterogeneity.

Figure 3.

Comparison of mode of intervention—online and offline—on compassion fatigue.

3.5. Publication bias

The funnel plot of the effectiveness of the PIs (see Figure 4) did not appear to be totally symmetrical, suggesting that there might be a possible publication bias.¹⁰⁷ However, results from the Egger's test (see Figure S1) revealed that there is no small study effect (P = .1726). The asymmetry might be due to the high between-study heterogeneity (I² = 96%). The asymmetry might also be due to the language bias as only literature in English was considered.

Figure 4.

Funnel plot to estimate publication bias.

4. Discussion

The present review identified 82 PI studies between 2004 and 2023 with diverse study designs (eg, randomized group comparisons, quasi-experimental design, and pre-post study design) for reducing CF among helping professionals. Most studies concluded that PIs could reduce CF, with the majority including mindfulness-based, psychoeducational, and resiliency interventions. However, some reports concluded that increased rigor was needed to evaluate appreciative inquiry, nature-based interventions, debriefing, compressed mindfulness, and guided imagery interventions, as they did not significantly reduce CF. The use of a quasi-experimental design in many of the studies limits their ability to conclude a causal association between PI and CF. When individuals are not randomly assigned to intervention versus control groups, fundamental biases may affect group participation, which makes quasi-experimental studies problematic.¹⁰⁸ Besides, findings highlight a disproportionate representation of nursing professionals in the review sample, limiting the generalizability of results to other helping professionals who are similarly at risk of CF, such as emergency medical care professionals, police officers, mental health professionals, fire and safety employees, and other individuals who work in the health and community services. Also, included studies used various versions of the ProQOL to assess the CF of various helping professionals.

Following a systematic review, the current study performed a meta-analysis to quantitatively determine the effectiveness of PIs in reducing CF. As per the authors’ knowledge, this is the first meta-analysis to show the effectiveness of PIs for CF among helping professionals. Of 82 studies that met the selection criteria, 11 RCT studies comparing PIs’ effect on reducing CF to a control condition were deemed eligible for meta-analysis. A significant large pooled effect size for PIs in reducing CF favored the experimental group compared with the control group because the included PIs focused on developing self-compassion, resiliency, emotion regulatory training, and teaching various relaxation methods. These techniques are found to be essential to improving mental health and professional quality of life and in lowering CF among helping professionals.¹⁰⁹ Further, such methods support professionals in reducing negative self-critical judgment, cultivating a positive self-compassionate attitude, and enhancing their general well-being.¹¹⁰ Furthermore, PIs improve overall professional performance, client assessment of the therapeutic relationship, and professionals' empathy in addition to preventing burnout in helping professionals.¹¹⁰

Subgroup analyses were done to determine whether the CF outcome of PIs varied based on the mode of intervention administration. Interventions delivered online were found to have a greater SMD than interventions with offline administration. Relatedly, a previous systematic review revealed that online-delivered interventions are most effective for dealing with job-related stressors like burnout.¹¹¹ This is because helping professionals can obtain information and assistance at their own pace and convenience using online interventions, which can be especially helpful for those with busy schedules.¹¹² Online platforms also provide anonymity, which helps to foster a feeling of privacy and lessen the stigma attached to asking for assistance.¹¹³ These factors may motivate people to interact with the intervention more fruitfully. Online interventions can reach more health care practitioners due to their flexibility, which increases the accessibility of assistance.¹¹⁴ Additionally, the interactive features of online interventions can boost participation and offer customized feedback, allowing the intervention to be tailored to each participant's requirements.¹¹⁵ The increased heterogeneity found in the meta-analysis results is alarming and could be attributed to factors like clinical factors and statistical factors.¹¹⁶ Major sources of clinical heterogeneity are the differences in the content of the interventions included in each study, and the diverse employment backgrounds of participants, who included nurses, doctors, and mental health professionals. Further, differences in the mode of delivery of intervention, the duration of administration of intervention, and mechanisms to handle losses to follow-up would be possible sources of statistical heterogeneity.

Although the study offers insight into the effectiveness of PIs for CF, certain limitations must be mentioned. Firstly, the study design of many of the included studies was pre-post without a control group, and quasi-experimental, making them ineligible for meta-analysis. Secondly, since the meta-analysis was conducted based on the post mean and SD scores of the intervention and the control groups, it could not take into consideration the baseline scores of the included studies. Additionally, the use of diverse CF scales limited the comparability of studies although most administered ProQOL, which is a reliable measure and has been validated in earlier studies. Lastly, only articles published in English were included, which could lead to selection bias.

Several recommendations can be made for future research. First, although PI studies have grown from 2004 to 2023, the efficacy of less popular PIs, such as resiliency interventions, self-compassion training programs, yoga, music interventions, guided imagery, spiritual interventions, nature-based interventions, and sending motivational messages in reducing CF in helping professionals needs to be tested in future studies. The present meta-analysis did not allow a subgroup analysis based on intervention types, given the small number of RCTs on mindfulness interventions, psychoeducational interventions, resiliency interventions, self-compassion training interventions, yoga, music interventions, guided imagery, and spiritual interventions. Therefore, future research with RCTs is required to get accurate pre- and post-intervention ratings for the control and intervention groups. The preponderance of studies involving the nurse population highlights the need to extend CF intervention research to other emergency professions, including police, firefighters, and child protection and disability service workers. Finally, findings revealed that PIs delivered online tend to impact CF significantly. Therefore, the current study advises that researchers, managers, and employers could consider this when designing CF interventions in the future.

5. Conclusion

To the best of the authors’ knowledge, the current study is the first to provide a quantitative synthesis of evidence on PIs’ effect on CF in helping professionals. The findings highlight the potential of various PIs in reducing CF. Subgroup analyses based on the mode of administration showed higher SMDs in the online mode of delivery. The current study suggests that governments and other health care authorities should include PIs in health care programs for helping professionals, as reducing CF is indispensable for their mental well-being and workplace productivity.

Acknowledgments

The protocol for the systematic review and meta-analysis has been registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the registration ID: CRD42022373763.

Author contributions

J.M.L. and E.R. conceived the ideas; J.M.L. and A.G. collected the data; J.M.L., E.R., and A.G. analyzed the data; J.M.L. and A.G. led the writing; and E.R. and J.R. the supervision.

Supplementary material

Supplementary material is available at Journal of Occupational Health online.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Conflicts of interest

The authors report that there are no competing interests to declare.

Data availability

The data that supports the findings of this study are included in the article and are available in the supplementary material of this article.