Effect of compassion rounds on nurses' professional quality of life on a COVID‐19 unit

Abstract

Introduction

Evidence suggests that support groups enhance nurses' professional quality of life (QOL), and positive professional QOL is associated with better patient and nurse outcomes. This study examined the effect of a unit‐level support group on the professional QOL of nurses working on a progressive care unit‐turned‐dedicated‐COVID‐19 unit.

Aim

We hypothesized that a professionally facilitated Compassion Rounds (CR) support group would improve compassion satisfaction (CS) and reduce compassion fatigue (CF) among COVID‐19 unit nurses.

Methods

For this pre/post, within‐group trial we recruited an inclusive, convenience sample of 84 nurses on a COVID‐19 unit within a 377‐bed, Magnet®‐designated hospital. The 10‐week, CR consisted of biweekly meetings, and the ProQOL version 5 measured pre/post CS and CF.

Results

Paired t‐testing showed that CS scores fell after CR (n = 10; p = .005), while scores rose for CF burnout (p = .05) and secondary traumatic stress (p = .008). Results were similar for unpaired analysis (N = 38; p < .05).

Implications/Conclusion

Although pandemic‐related challenges likely overwhelmed CR's potential to improve professional QOL, CR may have prevented worse deterioration of work‐life quality. CR may also create clinically meaningful improvements for groups or individual nurses, and thus enhance nurse and patient outcomes.

1. INTRODUCTION

Nurses' professional quality of life (QOL) undergirds sound practice. Decades of prepandemic evidence suggests that a more fulfilling work life is associated with better nurse and patient outcomes, including nurse retention, safer practice, and positive patient experiences. ¹ , ² Additionally when work life is more emotionally exhausting than rewarding, negative outcomes like nurse disengagement, medication errors, or adverse events may increase. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸

These relationships between work stress, professional QOL, and outcomes are recognized by both international and US organizations, often with a focus on burnout. In 2019 the World Health Organization (WHO) categorized burnout as an occupational phenomenon. According to WHO's International Classification of Diseases (ICD‐11), burnout is an outcome of poorly managed, “chronic workplace stress…characterized by…: feelings of energy depletion or exhaustion; increased mental distance from one's job, or feelings of negativism or cynicism related to work; and reduced professional efficacy.” ⁹

Additionally, the American Association of Critical Care Nurses (AACN) issued a “Call for Action” to address burnout syndrome (BOS) among critical care health professionals, ⁴ who include nurses working in progressive care units (PCUs). ¹⁰ AACN described BOS as resulting from the difference between professionals' expectations of a work environment and the sometimes disappointing reality of overwhelming workplace demands—a discrepancy that threatens quality care, patient satisfaction, and professional retention and mental health. The Association proposed a number of strategies to reduce BOS, including debriefing, support groups, team building, and consultation with ethics and palliative care services. ⁴

More recently, the National Academies of Sciences, Engineering, and Medicine (NASEM) published a consensus report on promoting clinicians' professional well‐being and preventing BO. ¹¹ NASEM identified that when job demands outweigh positive job resources, such as social support, compatible “values and expectations,… [and] work‐life integration,” ^{11 ,p.84} the results may be clinician burnout and associated negative nurse and patient consequences. ¹¹

These WHO, ⁹ AACN, ⁴ and NASEM ¹¹ conclusions are supported by hundreds of publications that document the importance of professional QOL among the helping professions, often using Stamm's Compassion satisfaction‐Compassion fatigue (CS‐CF) framework. ² , ¹² , ¹³ For example, CF was greater among oncology nurses who set aside their own needs in order to care for patients, while higher secondary traumatic stress scores were associated with personal characteristics of age less than 40, headaches, depression, posttraumatic stress, and difficult finances (N = 549). ¹⁴ Those researchers further found that a collegial environment lowered the risk of CF and seemed to safeguard CS against the negative effects of high work hours and numerous patient deaths. ¹⁴

In addition Nahrgang et al. ⁵ used a job demands and job resources model of workplace safety to document the negative outcomes of poor professional QOL. Their meta‐analysis of 179 studies with 203 independent employee samples, including nurses, suggested that job demands such as risks, hazards, and complexity related positively to BO, and BO in turn was associated with more accidents, injuries, and adverse events. Nonetheless, a work environment that sustained knowledgeable and autonomous workers was associated with increased worker engagement. ⁵

Beyond these findings, other prepandemic authors documented an array of positive outcomes in facilities that actively supported professional QOL by tangibly rewarding professionals' caring behaviors, ¹ establishing a “culture of compassion” from senior leaders downward, ¹⁵ or conducting supportive group interventions as described in a recent systematic review of 43 studies. ¹⁶ Thematic analysis within that systematic review revealed that safe space interventions as diverse as group rounds, critical incident stress debriefing, and peer‐supported storytelling were associated with improved provider, patient, collegial, and organizational outcomes. Errors decreased, while teamwork, empathy, communications, patient‐centered care, resilience, retention, and more improved. ¹⁶

In summary, prepandemic literature documents that personal and work environment characteristics, including PCU work, are associated with the risk of negative professional QOL; caregiver support groups may promote professional QOL; and professional QOL is directly related to nurse and patient outcomes. Yet, when COVID‐19 hospitalizations surged during winter 2020–2021, no research was available about how to sustain nurses' professional QOL during the extreme workplace demands precipitated by the pandemic. The aim of this study was to address that knowledge gap after our facility's PCU was abruptly designated a COVID‐19 unit. The specific hypothesis was that investigator‐designed, unit‐level, professionally‐facilitated Compassion Rounds (CR) would improve CS and reduce CF among these PCU‐turned‐COVID‐19 nurses.

1.1. Conceptual framework

Stamm's CS‐CF model ¹² guided this study. Stamm defined CS as the positive feelings derived from being able to do a good job at work, including positive collegiality and a sense of contributing not only to the work setting but also to the larger world. ¹² In contrast work‐related CF consists of two components: burnout (BO) and secondary traumatic stress (STS). BO may emerge slowly as a sense of disengagement from work, being less than one's ideal self, and hopelessness about being able to do a good job or make a difference; BO may be aggravated by lack of “sustaining beliefs,” ^{12 ,p.21} an unsupportive environment, high workload, and a stressful context. Unlike BO, STS may emerge rapidly when a professional helper fixates on and is devastated by the traumatic experiences of those helped. The result of STS is a complex of sleep disruption, fear, “intrusive images,” ^{12 ,p.13} forgetfulness, avoiding reminders of the trauma, and difficulty setting boundaries between work and personal life. When BO and STS outweigh CS, the result is low professional QOL.

2. METHODS

2.1. Design, setting, and sample

The hypothesis was tested using a prospective, pre/post, within group design. After the protocol was approved by the facility's institutional review committee as exempt (IRB ID: STUDY2020000138), the principal investigator (PI) invited participation from an inclusive, convenience sample of 84 registered nurses (RNs) working on the PCU‐turned‐COVID unit. The unit was located within a 377‐bed, Magnet®‐designated, non‐profit hospital on the US west coast, and the sample included all full‐time, per diem, and traveler nurses providing direct care on day and night 12‐hour shifts. The nurse manager, assistant nurse manager, and PI were excluded, and the study was conducted during the height of the winter 2020–2021 COVID‐19 surge.

2.2. Instruments

Two paper‐and‐pencil instruments measured outcomes: an investigator‐designed demographic questionnaire and the well‐established Professional Quality Of Life scale, version 5 (ProQOL5). ¹² First, the demographic tool measured sample characteristics and self‐reports of informants' participation in CR. Second, the 30‐item ProQOL5 measured nurses' self‐reported CS and CF on three, 10‐item subscales: CS, BO, and STS. The ProQOL5 asked respondents to rate how often they experienced each item on a 1 to 5 scale as never, rarely, sometimes, often, or very often; each subscale score ranged from 10 to 50. Examples of items include “I am happy that I chose to do this study” (CS), “I feel connected to others” (BO), and “Because of my [helping], I have felt ‘on edge’ about various things” (STS). ¹²

The ProQOL5 is a screening tool rather than a diagnostic one, and cut scores identify professionals at risk for low, moderate, or high levels of CS, BO, or STS. A low score is 22 or less, a moderate score is 23–41, and a high score is 42 or more. ¹² Because the ProQOL5 is a screening tool, cut scores are more likely to identify false positives, and in research this tends toward creating type 1 statistical errors. Based on 1289 cases from multiple studies, no statistical differences are identified across potential ProQOL5 covariates of gender, White vs non‐White populations, age, years with employer, or years in present field. ¹² The ProQOL5 is validated for use with helping professionals, including nurses. Construct validity is well‐established, and internal subscale reliability is strong (CS α = .88; BO α = .75; STS α = .81). ¹²

2.3. Intervention

The CR intervention was a 10‐week series of daytime, biweekly, 30–45 min RN meetings facilitated by either the Director of Palliative Care or a Spiritual Care team member, who were already known to the nurses and were neither nurses nor members of the research team. CR structure and processes were adapted to our setting from effective group strategies used by others ⁴ , ¹⁶ For example, during small group CR, the skilled facilitators used open‐ended communication to acknowledge employee experiences and to create a safe, confidential, nonjudgmental venue for nurses' emotional disclosure, debriefing, and self‐reflection about work‐related rewards and stressors. CR was also designed to be less resource‐intensive and more focused than the well‐established Schwartz Rounds (SR) that are facility‐wide, interprofessional, and structured. ¹⁶ Attendance was not recorded, the PI attended, and she provided tea and snacks during CR.

Each combined live and online CR meeting was held in the quiet unit conference room s between 1400 and 1500 when unit activity usually slowed. Because facilitators were unavailable for nighttime meetings, nurses who worked nights were invited to attend CR at 1400 either live or simultaneously using online Microsoft® Teams software. Night shift RNs already regularly used Microsoft® Teams to attend daytime staff meetings and so were familiar with the technology. Nurses working days thus participated during paid hours, while nurses working night hours attended on their own time. Attendance was voluntary.

2.4. Procedures

Procedures included four phases: (1) 3 weeks of informing staff of the upcoming study and coding questionnaires, (2) 3 weeks of baseline data collection, (3) 10 weeks conducting CR, and (4) 3 weeks of postintervention data collection. During the first phase, the PI explained study details to and answered questions of unit nurses during daytime and nighttime unit huddles; huddles were brief meetings before the start of each shift that facilitated care consistency among staff. She also provided study information on a unit poster. All nurses were invited to participate in CR whether or not they voluntarily completed questionnaires.

Also during this first phase, a co‐PI randomly coded each staff member's questionnaires to facilitate paired pre/post analysis. She also prepared envelopes for questionnaire distribution marked on the outside with only the staff member name. Inside each envelope were IRB‐approved consent information, instructions, an unmarked envelope for questionnaire return, and coded demographic sheet and ProQOL5. The PI was blinded to code assignment, and the co‐PI did not know PCU/COVID‐19 unit staff.

The PI then collected baseline data, implemented 10 weeks of CR, and gathered postintervention data. Pre‐ and postdata collection procedures were the same. First, the PI distributed the prepared envelopes either directly to each staff member or to individual mailboxes. Participants' return of coded questionnaires in unmarked, sealed envelopes to a designated unit container documented their voluntary consent, and the PI gathered returned envelopes twice weekly from the container. Additionally, the PI sent weekly email reminders to staff of their opportunity to participate, and the assistant nurse manager reminded staff daily during huddles.

2.5. Statistical analysis

After postintervention data collection, the PI and co‐PI entered the data into the secure, facility‐hosted Research Electronic Data Capture® (REDCap) database system for management and analysis. Descriptive analysis of demographic data was completed within REDCap, and ProQOL5 data were exported to Microsoft Excel for descriptive and inferential analysis. Established subscale cut scores were used to determine whether the sample reported a low, moderate, or high risk for CS, BO, and STS, ¹² while two‐tailed paired and unpaired t‐tests were used to detect significant pre‐CR to post‐CR changes.

3. RESULTS

Overall response rate was 45% (N = 38). Thirty nurses completed the pretest, but only 10 also completed the posttest. Eight more completed the posttest only. Almost half of respondents (n = 18; 46%) reported attending CR 1–10 times (M = 3; SD = 1.4). Most participants held a baccalaureate degree or higher (87%) with less than a quarter holding specialty certification (see Table 1 for demographics). Years of total nursing experience varied widely (M = 9.3; SD = 7; Median = 10), as did years in PCU nursing (M = 5.6; SD = 5.2; Median = 5). Almost all respondents were women, and three quarters self‐identified as Asian or White. Respondents' ages ranged from 21 to 55 years (M = 35; SD = 24; Median = 32; Mo = 29), and when divided into generational groups, all but two were either “GenX‐ers” or Millennials. No demographics were available for unit nonrespondents.

Table 1.

Sample self‐reported demographics

	N	%a
Age in yearsb and generation (n = 34)
21 (Generation Z)	1	2
23–36.5 years (Millenials)	23	68
41–53 (Generation X)	9	27
55 (Baby boomers)	1	2
Ethnicity (n = 36)
White	18	50
Asian	9	25
Hispanic/Latinx	4	11
Mixed ethnicity	2	6
Black	1	3
Hawaiian/Pacific Islander	1	3
Native American	1	3
Gender (n = 37)
Female	33	89
Male	3	8
Prefer not to state	1	3
Highest Nursing Degree (n = 36)
Associate	5	14
BSN	28	78
Master's	2	6
Doctorate	1	3
Specialty certification (n = 37)
CCRN, PCCN, or other	9	24
None	28	76
Years in Nursing (n = 35)b
<2	2	6
2–4	6	17
5–10	16	46
11–20	7	20
25–30	3	9
Years in PCU (n = 37)b
<2	6	16
2–4	16	43
5–10	12	32
15–27	3	8

Abbreviations: BSN, Bachelor of Science in Nursing; CCRN, critical care registered nurse; PCCN, progressive care certified nurse; PSU, progressive care unit.

^a Percentages are rounded so some variables may not equal 100%.

^b Year ranges are not continuous because they are based on respondents' actual reported years.

The hypothesis of whether CR would improve these nurses' CS and reduce their CF was then examined. Ten nurses submitted paired pre and post ProQOL5 questionnaires, and analysis revealed that after CR, their CS scores fell from moderate to low levels t(8) = 3.7, p = .005 (see Table 2). Moreover, both BO scores t(8) = −2.3, p = .05 and STS scores t(8) = −3.5, p = .008 rose from low to moderate levels postintervention.

Table 2.

ProQOL mean scores^a and risk level (low, moderate, high) based on cut scores

	Paired sample (n = 10)		All respondents (N = 38)
	Baseline Mean (CI)	After CR Mean (CI)	Baseline Mean (CI)	After CR Mean (CI)
Compassion satisfaction subscaleb	28 95% (21, 36)	18 95% (15, 22)	24 95% (22, 28)	17 95% (14, 22)
Risk level	Moderate	Low	Moderate	Low
Secondary traumatic stress subscaleb	21 95% (13, 31)	33 95% (26, 40)	28 95% (24, 33)	34 95% (30, 39)
Risk level	Low	Moderate	Moderate	Moderate
Burnout subscalec	22 95% (19, 26)	27 95% (24, 31)	25 95% (23, 27)	28 95% (26, 30)
Risk level	Low	Moderate	Moderate	Moderate

Abbreviations: BO, burnout; CR, Compassion Rounds; STS, secondary traumatic stress.

^a All mean subscale scores were rounded down for reporting in this table.

^b CS and STS changes significant with two‐tailed t‐test at p ≤ .05 for paired and unpaired data.

^c Paired BO change significant with two‐tailed t‐test at p ≤ .05; unpaired BO significant only with one‐tailed t‐test at p < .05.

When data from the entire sample (N = 38) were analyzed, results showed again that CS scores fell after CR from moderate to low levels t(39) = 2.8, p = .008, while STS scores rose significantly but remained in the moderate range t(41) = −1.99, p = .05. In contrast, moderate BO scores remained stable t(39) = −1.8, p = .08, and rose only when a less rigorous one‐tailed standard was applied (p = .04).

4. DISCUSSION

To determine how representative this study's respondents were of the larger RN population, respondents were compared to the surrounding state's employed RNs and nationally to PCU nurses. State RNs were much like respondents: mostly women (87%) under 50 years of age (63%) and roughly three quarters White (42%) and Asian/Pacific Islander (32%). ¹⁷ Educationally the state population differed; 78% of this study's sample held a BSN compared to only 48% of practicing nurses in the state, ¹⁷ perhaps because of the facility's hiring related to Magnet® recognition.

Respondents were less similar to PCU nurses nationally. ¹⁸ Participants' mean age was 9 years younger than the national US PCU nurse population with 9% more women, 3% more Latinxs, 17% more Asians, 30% fewer Whites, and 9% fewer Blacks. No summary demographics were available for US nurses working on COVID‐19 units.

The hypothesis that CR would improve CS and reduce CF was rejected. Nonetheless, this project contributes new knowledge. First, researchers found only three other intervention studies examining support for COVID‐19 nurses during the 2020–2021 pandemic. Additionally, even though findings from the present study are not generalizable given the small sample, they may be at least transferable.

Second, readers should consider that CR may have yielded benefits to professional QOL that are impossible to verify. Specifically, CR may have prevented an even worse deterioration of these nurses' work‐life quality into what Stamm ¹² identified as a worst case scenario: low CS with high BO and high STS. No counterfactual exists.

Third, according to Polit and Beck, readers must consider potential clinical significance of findings as well as their statistical significance. ¹⁹ CR may have produced “genuine, palpable effects on the daily lives” of some individuals. ^{19 ,p.449} For example, one nurse's CS score rose post‐CR, while two respondents' BO scores fell; and when considering the entire sample BO remained stable. If indeed even a few participants' professional QOL improved, then they may have delivered better care and collegial support, given the findings of prior studies. ³ , ⁴ , ⁵ , ⁶ , ⁸ , ¹¹ , ¹⁴ Such individual changes hold the potential to produce a positive ripple effect across the unit. ¹⁴

Nonetheless, several possibilities exist for why CR proved ineffective among these nurses as a whole. First may have been the design of CR. No evidence existed on how to tailor content, process, or length of a nursing support group for maximum benefit during a global pandemic. Thus in a time of high demands on facility resources, researchers planned 10 weeks of focused, unit‐level CR that avoided resource‐intensive strategies of facility‐wide, interprofessional SR. Perhaps 10 weeks allowed only enough time for staff to surface important issues, but not enough time to resolve them. Too, virus‐related hospitalizations and deaths continued to surge during the 3 weeks of post‐CR data collection, thus creating a threat to internal validity of reversing any unconsolidated improvements after CR stopped.

CR content and process might have also been inadequate to address participants' need to express their distress. Some respondents exhibited unhappiness in their volunteered, handwritten comments on questionnaires, such as “I feel I do more harm [than] ‘helping’… 80% of my time” or “I usually think I'm NOT helping but forced to ‘keep someone alive’ or turn them into a vegetable. I feel futile.” Also, a few individuals reported post‐CR demographics that were noticeably different from their pre‐CR demographics, thus suggesting their concern about being identified with ProQOL5 responses that might reflect negatively on them. At the same time, such social desirability bias likely had minimal effect on scores because responses to some positively worded ProQOL5 items (e.g., “I feel happy”) were reversed for scoring. ¹²

Additionally, the pandemic radically altered these nurses' work and personal environments in ways that may have overwhelmed the positive effect of CR. At work the unit abruptly switched from the status quo of PCU care to novel COVID‐19 care, while the larger hospital organization reallocated resources and reassigned staff. Thus, study participants confronted unexpected, unchosen practice changes, including a new patient population, evolving and uncertain treatment protocols, and increased frequency of deaths among isolated patients for whom they became almost sole support.

Moreover, the pandemic's impact on these respondents' hospital system and unit may have aggravated this group's existing risks for lower professional QOL. Sacco et al.'s ²⁰ prepandemic study documented the negative impact on critical care nurses' professional QOL of major system and practice changes, working in mixed‐acuity PCU practice (as compared to single acuity ICU practice), and BSN education (N = 221). And, Kelly et al. found that being in the millennial generation and years of experience predicted worse BO. ²¹ Notable then is that all respondents worked in PCU, 68% were “millennials,” 78% held a BSN, and almost a third (29%) had over 10 years of practice (Table 1).

Also, in their personal lives, these participants like other healthcare providers faced pandemic‐related social distancing, lockdowns, isolation, uncertainty about the future, PPE shortages, stigmatization, high census, and their own rising fears of both personal infection or death and bringing the virus home to family. ²² , ²³ Researchers have since documented that for many providers this context led to burnout, psychological distress, sleep disorders, suicidal ideation, depression, posttraumatic stress, reduced self‐efficacy, anxiety, and depression. ²³ , ²⁴ One sample of nurses, who expressed altruistic sentiments during the pandemic, also reported being four times more likely than other professionals to consider resigning. ²² Such was the environment in which this study began and ended.

Limiting CR scope only to nurses on a single unit may also have contributed to worse than hypothesized outcomes. SR achieved success using an ongoing, organization‐wide, and interprofessional group intervention. ¹⁶ Nonetheless, those SR successes occurred during nonpandemic times and contradictory evidence exists; Dawson et al. ²⁵ found that SR did not improve compassion, engagement, or reflection although SR reduced psychological distress among UK nurses (N = 274). Reports of SR modifications and outcomes within the COVID‐19 context are limited. ²⁶ , ²⁷

Like SR, CR might have improved professional QOL under nonpandemic circumstances. At baseline these respondents reported less BO than a national prepandemic sample of physicians and nurses ¹¹ ; nonetheless their physical and psychological exhaustion grew during the 2020–2021 COVID‐19 surge despite CR. Their falling CS and rising BO and STS suggested an increasing sense of ineffectiveness and being trapped, overwhelmed, or even frightened by work. ¹² This despite the ongoing infrastructure of nurse recognition and engagement that the facility maintained as part of their Magnet® status.

Finally, ProQOL5 is a screening tool that may over‐estimate CS, BO, or STS. Despite that, the ProQOL5 has demonstrated stability across time, so that changes in scores reflect an actual change in persons, not spurious findings related to the questionnaire itself. ¹² Thus, because this study examined trends over time within a single group, readers can have additional confidence in study findings.

5. IMPLICATIONS FOR PRACTICE

This study addressed how to support nurses' professional QOL during a time of extreme workplace stress and resultant nurse distress. Implications are related to the support of nurses in other substantial crisis situations, such as war, ongoing mass casualty events, or future pandemics. In such situations, CR may prove effective if continued for the duration of the crisis rather than limiting it to 10 weeks, and ProQOL5 measurements may provide more insight if taken before CR ends so that any temporary gains are captured in data, and leaders can take action to consolidate them.

Additionally, the effectiveness of CR in supporting professional QOL during “normal” times and non‐COVID‐19 units is unknown, and so evaluating CR in new settings, samples, and times is reasonable. Deets ²⁸ argued that even one study provides enough information to apply an intervention elsewhere as long as patient safety is not jeopardized and outcomes are evaluated. In this instance, the well‐established, no‐cost ProQOL5 is readily available to measure gains. Moreover, CR is low risk, easily adapted to other settings, uses available in‐house resources during work hours, has lower cost in training and personnel time than SR, creates connections between nurses and palliative care/ethics experts that may reduce BO, ⁴ and holds potential for meaningful recognition of staff that is associated with positive nurse and patient outcomes. ²¹ Too, CR may enhance reflection, learning, collegiality, autonomy, and job control—important resources in dealing with job demands. ¹¹ , ¹⁴ CR limitations are its restriction to one discipline (unlike interprofessional SR), the cost of facilitator time, limited availability to nurses working nights, and yet‐to‐be documented positive outcomes.

Second, this project adds to research evidence that completing the ProQOL5 may be a useful first step toward improving work‐life quality. For future projects, the ProQOL5 and information on scoring and interpretation are available online at no cost. ¹² At an individual level, nurses can complete the online ProQOL5 for use in career planning, and at a group level nurse leaders can use ProQOL5 with staff as a basis for assisting them to identify strategies that support CS. This is true whether professional QOL is good or poor. For example, in an ideal unit scenario of high CS and low BO and STS, the manager might focus on meaningful recognition for staff efficacy, safety practices, collegiality, and so on as a path toward positive outcomes. Stamm ¹² suggests that such providers may also “benefit from engagement, opportunities for continuing education, and other opportunities to grow in their position."^p.22 In contrast, low CS and high BO or STS scores, suggest the need to attend to individual or group risks for CF by leadership compassion and engaging staff in assessing issues and developing solutions. ⁴

Finally, from a research perspective, further CR studies are warranted. Replication under nonpandemic conditions would yield perspective on CR value, and qualitative interviews of both CR participants and facilitators would provide a more in‐depth understanding of CR's strengths and weaknesses in timing, process, and content. Like the study conducted by Kelly et al. ²¹ on meaningful recognition, a larger project on CR might use regression analysis to identify the relative contributions of CR and of personal and work factors to nurses' professional QOL. Also, replication in the present study's unit after nurses return to care for mixed acuity PCU patients would contribute to a better understanding of CR and PCU nurses.

6. LIMITATIONS

The project had several limitations. First, its small, convenience sample from a single setting may not be representative of either unit nurses or the larger, US PCU or COVID‐19 nurse populations. Second, fewer than half of potential respondents completed any questionnaire, and third, CR may have produced temporary improvements in CS and CF during the 10‐weeks of CR that were not captured by post‐CR measurement. Finally CR attendance was limited and not uniform among respondents.

7. CONCLUSION

Working in PCU already put these respondents at risk for lower professional QOL, and their risk was further raised by the pandemic's abrupt impact on resources, care expectations, and perhaps even the nurses' sense of personal safety. Thus, when PCU converted to a COVID‐19 unit, the pandemic's corrosive effects on these nurses' work and personal lives likely overwhelmed CR's potential to improve professional QOL. Nonetheless, CR may have prevented even worse deterioration of work‐life quality and created clinically meaningful improvements for some individual nurses. No counterfactual exists, and further evaluation of CR is warranted as a supportive strategy for nurses in varied settings.

CONFLICTS OF INTEREST

During the study Ms. Bhatnagar was employed fulltime as a staff RN on the PCU at PHCMC. Dr. Highfield is employed parttime as the Advisor for Nursing Research Fellows at PHCMC.

Bhatnagar D, Highfield MEF. Effect of compassion rounds on nurses' professional quality of life on a COVID‐19 unit. Nurs Forum. 2022;57:1365‐1372. 10.1111/nuf.12821

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author.