Exploration of Individual and System-Level Well-being Initiatives at an Academic Surgical Residency Program A Mixed-Methods Study

IMPORTANCE Physician well-being is a critical component of sustainable health care. There are few data on the effects of multilevel well-being programs nor a clear understanding of where and how to target resources. OBJECTIVE To inform the design of future well-being interventions by exploring individual and workplace factors associated with surgical trainees’ well-being, differences by gender identity, and end-user perceptions of these initiatives. DESIGN, SETTING, AND PARTICIPANTS This mixed-methods study among surgical trainees within asingleUSacademicsurgicaldepartmentincludedaquestionnaireinJanuary2019(98participants, includinggeneralsurgeryresidentsandclinicalfellows)andafocusgroup(9participants,allclinical residents who recently completed their third postgraduate year [PGY 3]) in July 2019. Participants self-reported gender (man, woman, nonbinary). EXPOSURES Individual and organizational-level initiatives, including mindfulness-based affective regulation training (via Enhanced Stress Resilience Training), advanced scheduling of time off, wellness half-days, and the creation of a resident-driven well-being committee. MAIN OUTCOMES AND MEASURES Well-being was explored using validated measures of psychosocial risk (emotional exhaustion, depersonalization, perceived stress, depressive symptoms, alcohol use, languishing, anxiety, high psychological demand) and resilience (mindfulness, social support, flourishing) factors. End-user perceptions were assessed through open-ended responses and a formal focus group. RESULTS Of 98 participants surveyed, 64 responded (response rate, 65%), of whom 35 (55%) were women. Women vs men trainees were significantly more likely to report high depersonalization (odds ratio [OR], 5.50; 95%


Introduction
Physician well-being is a critical component of sustainable health care. 1 Its absence affects the personal and professional lives 2 of surgeons and trainees, with resultant distress in the form of burnout, 3 depression, 4 alcohol abuse, 4,5 and attrition, which carry a profound economic impact. 6 While risk factors have become clearer, factors that enhance physician well-being remain poorly understood. 3,4,7 Increasing evidence suggests that both individual and organizational-level interventions are necessary 3,8,9 but likely need tailoring to meet the needs of diverse individuals, groups, and settings. [10][11][12] To date, there are few data on multilevel well-being programs among surgical trainees, nor a nuanced understanding of where and how to tailor interventions to optimize the use of limited resources.
To address this gap, we drew on both Broaden-and-Build 13 and Job Demand-Resource 14 theories to conceive of ways in which individual and workplace factors might be associated with surgical trainee well-being. We then surveyed surgical trainees in the University of California San Francisco (UCSF) Department of Surgery (DOS) using published measures of individual and occupational risk and resilience factors. We additionally held a focus group exploring surgical residents' perceptions of existing well-being initiatives at our institution and aspects of their work experience. Our goal was to help to inform the design of future multilevel well-being initiatives by exploring individual and workplace factors associated with well-being, evaluating differences by gender identity, and assessing end-user experience.

Study Design and Population
We conducted a mixed-methods study comprised of 2 components. The first component was a All 83 residents (including those engaged in clinical duties and those conducting research as part of our academic general surgery residency program) and 15 clinical fellows in the UCSF DOS received the online questionnaire. Because of stigma surrounding mental illness in medicine, [15][16][17] respondents were guaranteed anonymity. The American Association for Public Opinion Research (AAPOR) standard definitions were followed for calculating survey response rate.
The focus group involved off-site dinner but no compensation. A semistructured script was used, expanding on questions from the questionnaire with elaboration prompts. By participant request, no recording was made, but extensive field notes were taken. The Standards for Reporting Qualitative Research (SRQR) reporting guideline was followed. The study was approved by UCSF's institutional review board, and informed consent was obtained from participants.

Individual and Organizational Well-being Initiatives
In the UCSF DOS, resident well-being interventions began in response to data reflecting a high prevalence of burnout and distress. In 2016, we designed ESRT, an individually based intervention tailored to surgeons that teaches cognitive skills for affective regulation (ie, self-awareness, perspective taking, and control of emotional reactions to thoughts and events). We tested ESRT in 2 randomized clinical trials with a total of 65 surgery trainees in PGY 1 (21 categorical trainees) between January 2016 and December 2018 and reported feasibility 18 ; improvements to cognitive function, burnout, mindfulness, and physiologic stress 19 ; and activation of neural substrates associated with top-down executive function and interoception (ie, awareness of one's own thoughts, sensations, and emotions). 19,20 In 2018, multiple organizational initiatives were introduced: an administrative chief resident role (allowing for advanced scheduling of resident time off) and representation on education committees (influencing organizational-level changes). Results included biannual wellness half-days, a resident research fund, personal finance seminars, social gatherings, and healthy workplace food.

Well-being Measures
Scales and measures used in our prior work, related to performance and/or distress in surgeons, 4,5,21,22 were used and scored according to published methods. These include the Maslach Burnout Inventory, a validated 2-item screening that assesses emotional exhaustion and depersonalization 23 ; Cohen Perceived Stress Scale, 4 a criterion-standard measure of stress, with normative data for men and women aged 18 to 34 years, used to determine high cutoffs (ie, Questionnaire, a 17-item measure of job strain with subdomains for psychological demand, control, and social support. 35 Only demand and social support (12 items) were assessed in this study. High subdomain cutoffs were defined by convention as scores in the upper tertile of possible scores. 36

End-User Experience
We explored end-user experience through evaluation of responses to 2 open-ended survey questions and thematic analysis of qualitative content from a 3-hour focus group (moderated by C.C.L., P.O., and N.L.A., surgical and medical education faculty). The focus group followed a semistructured script exploring resident perceptions of influences, personal changes, and residency program elements, with 9 volunteer surgical trainees in PGY 3.

Statistical Analysis
Survey response data were converted to numeric scores and used as continuous variables of psychosocial risk and resilience factors. Binary categorical variables were established for select factors (cutoffs discussed previously) to determine the prevalence of high levels of these factors, with odds ratios (ORs) created to assess gender associations. Because distributions of several factors were skewed and discrete, the nonparametric measure, Kendall τ-b, was used to estimate the association among the different risk and resiliency factors, given that this test is not influenced by extreme measures. All tests were 2-tailed, with significance set at P < .05. All computations were performed with SAS version 9.4 (SAS Institute) by a senior biostatistician (K.L.D. ).
The open-ended survey responses were compiled. Notes from the focus group were input into a spreadsheet and analyzed using inductive and deductive thematic analysis techniques 37 following published guidelines and a 6-step approach. 38 Codes, concepts, and themes were iteratively reviewed, refined, discussed, and described in the context of past work 18 and current aims (by C.C.L. and A.L.G.). Qualitative results are reported according to SRQR 39 guidelines.

Respondents and Participants
Among the 98 DOS trainee survey recipients (83 residents and 15 clinical fellows), 64 responded, yielding a survey response rate of 65% (

Differences by Gender Identity
As shown in Table 2, women trainees were significantly more likely to report high depersonalization (OR, 5.50; 95% CI, 1.38-21.85) and less likely to report high mindfulness tendencies (OR, 0.17; 95% CI, 0.05-0.53) compared with men trainees. Gender differences were not significant for other factors.

Associations Between Risk and Resilience Factors
Individual and occupational risk and resilience factors by gender (Figure 1) revealed patterns and trends that suggest both shared and unique associations with well-being. Among women, factors influencing psychological risk were suggested by significant positive correlations between negative affect (such as depressive symptoms, high perceived stress, and anxiety), burnout (emotional exhaustion and depersonalization), and high workplace psychological demand (eg, depressive symptoms and emotional exhaustion, r = 0.34; perceived stress and workplace demand, r = 0.37).    memberships, additional wellness days, affordable day care) and social (eg, team-building and family-inclusive events, more mentoring).

JAMA Network Open | Surgery
Analysis of the focus group data led us to identify 4 emergent themes ( Table 3). Demand was referred to as necessary and motivating in the form of a challenge (quotation 1) but was associated with dread and resentment when it hindered rewarding aspects of work (quotation 2) or was unequally shared by team members (quotation 3). Control, referred to variably as negative or positive, was associated with frustration and anger when it led to inefficiency (quotation 4) but was referred to as positive in the form of thoughtful and/or effective scheduling (quotation 5). Control in the form of choice was associated with greater work satisfaction (eg, residents' choosing to extend hours for educational reasons; quotation 6). Support, both social (made possible by thoughtful and/or effective scheduling; quotation 7) and practical (in the form of well-being resources provided by the residency program; quotation 8), was referred to as essential but often absent or inaccessible.
Affective regulation skills, in the form of reframing, acceptance, or perspective taking, were referred to as necessary and transformative (quotations 9 and 10), with multiple references to overwhelming negative emotions when such skills were absent (quotation 11).

Discussion
The results of this mixed-methods study of well-being among surgical trainees at a single institution support 3 main findings. First, the prevalence of specific risk and resilience factors differs by gender identity; second, there are targetable individual characteristics and workplace elements that are associated with trainee well-being; and third, work quality and context, compared with work quantity, may be more associated with well-being. Our first finding, that the prevalence of risk and resilience factors differed by gender identity, is supported by the significantly increased odds of high depersonalization among women trainees and of high inherent mindfulness among men trainees. Emotional exhaustion and depersonalization are increasingly recognized as representing different processes within burnout. 3 In 1 of the few longitudinal causal studies of burnout in physicians, 40 results suggest that depersonalization may be relatively protective against stress and anxiety, possibly through the mechanism of ego-defense. This corresponds with Maslach's original work, in which the development of what she termed detached concern was interpreted as a protective mechanism, with depersonalization representing a dysfunctional extreme. 41 Later studies have found that a high degree of concern for the patient, coupled with high detachment, can allow for emotional distance, which is protective regarding burnout development. 42 This role for depersonalization is further supported by the Job Demand-Resources model of occupational strain, 43 wherein women are particularly prone to disengagement and/or detachment when workplace stress is high and social support is poor. 36 As such, provision of resources that support healthier coping skills (eg, available and accessible ESRT) and socialization (eg, mentoring, advance scheduling, and organized events) may be particularly important.
Relatedly, increased odds of higher mindfulness among men may be a marker of a local culture in which men feel greater ease. Although the CAMS-R instrument was designed to capture attention, present focus, awareness, and acceptance, 33,44 these attributes are not exclusive to mindfulnessmeditation training and can be found in individuals with resilience and healthy coping skills. 45 Increased odds of high mindfulness tendencies in men may reflect a local culture in which men are more readily able to thrive. This possibility is further supported by the tendency for flourishing to be more prevalent in men (Table 2) and a growing body of work demonstrating greater mistreatment of women in surgical training nationally. 4,46 Our second main finding, that there appear to be targetable elements that are associated with trainee well-being, is supported by the patterns and trends observed in correlations between individual risk and resilience factors and mindfulness, workplace social support, and workplace psychological demand. Overall, findings from both men and women reflect Broaden-and-Build theory, 13 in that high positive affect (higher positive emotions) was correlated with higher mindfulness tendencies, while high negative affect (higher negative emotions) was correlated with higher burnout. These observations echo a strong body of theoretical work and empirical data suggesting reciprocity between positive affect and mindfulness and the ability of these internal states to promote resilience, meaning making, and higher life and work satisfaction. 47 Findings from both genders also reflect Job Demand-Resource theory in that higher workplace psychological demand was correlated with psychological risk factors, and higher workplace social support was correlated with psychological resilience factors. Extensive work in occupational science demonstrates that job strain increases with greater psychological demand 43 and is mitigated by greater social support, particularly among women. 36 Although correlation is not causation, this constellation of findings is important because it allows first-pass construction of a conceptual framework for how well-being may be affected (and potentially targeted) in the setting of surgical residency (Figure 2). Because of the cross-sectional nature of our study, these nascent models are not causal nor definitive, but they do allow us to begin a targeted approach to designing multilevel well-being initiatives. For example, mindfulness is a resilience factor that occurs naturally but can also be trained, as seen in our work with ESRT 19  decreasing administrative burden 51 ) rather than simply quantity (eg, work-hour restriction 52 ). The previously described correlations and potential gender differences are intriguing but should be confirmed in future studies in light of our limited sample size and single-institution setting.
One gender-based finding worth noting is that none of the individual factors or workplace elements evaluated here appeared to have a significant correlation to flourishing in males. This is Therefore, our findings suggest that additional factors should be evaluated in the interest of identifying the most effective well-being targets for men trainees. Increased control is a wellcharacterized way to mitigate job strain, particularly in men, 35 making this a critical metric to evaluate in future work.

JAMA Network Open | Surgery
Our third finding, that work quality may be more strongly associated with resident well-being than work quantity, was supported by the findings of our focus group, which reflected the importance of context (Table 3). For example, results described demanding work as a challenging opportunity when demand is manageable but a source of resentment and distress when demand is inequitably shared or subsumes occupational rewards. Similarly, control in the form of selfdetermination was described as making additional work feel worthwhile, whereas it bred frustration in the context of inefficient hierarchical systems. Finally, our results reflected the potency of both social support and affective regulation skills in their ability to mitigate negative emotional influences on trainee work satisfaction. When social support was facilitated and affective regulation (ie, coping) skills were provided, trainees described the ability to navigate the intense emotional landscape of surgical training. These findings are supported by studies of job strain that show, for example, that workplace demand is not homogeneous and can promote or diminish work engagement when seen as a challenge vs a hindrance, respectively. 53

Limitations
While potentially provocative, our findings should be viewed in the context of several limitations.
Specifically, while response rates and respondent demographic characteristics suggest representation of our population, our sample size is small owing to the fixed sized of our residency