Health Care Worker Burnout and Perceived Capacity to Address Social Needs

Abstract

Health care organizations are increasingly incorporating social care programs into medical care delivery models. Recent studies have identified burnout as a potential unintended consequence of this expansion. Successful implementation of these programs requires investment in the health care team, although understanding the impact of this expansion on nonphysician team members remains limited. Utilizing a theory-informed model for organizational behavior change, the authors aim to characterize the perceived elements of capacity to address patients' social needs within a heterogenous group of health care workers, and to examine the association of these behavioral conditions with burnout. A cross-sectional analysis was conducted of a survey of ∼1900 health care staff from 46 organizations in a large delivery system. Exploratory factor analysis identified factors contributing to the “Perceived Capacity to Address Social Needs” domain; Motivation, Organizational Reinforcement, and 3 task-specific capacities (Identification of social needs, Providing care for patients with social needs, and Linkage to social needs resources). Logistic regression found both a lower sense of motivation (OR 0.71, 95% CI 0.59–0.86), and organizational reinforcement (OR 0.51, 95% CI 0.42–0.62) associated with a higher rate of burnout. These associations with burnout differed by organizational role, suggesting role-specific relationships between these behavioral conditions. As health care has evolved into team-based interventions, staff across the care spectrum are now tasked with addressing larger social issues that affect their patients. A systems approach, aligning organizational priorities and staff motivations, in addition to task-specific skill sets is likely necessary to prevent burnout in this setting.

Introduction

A growing body of literature has documented the impact of unmet social needs (eg, unemployment, housing quality and availability, food insecurity) on health outcomes of patients.^1,2 In accordance with many professional associations, payers, and governmental bodies,^3–7 health care organizations have begun to address these risks through a variety of interventions that integrate social care into medical care delivery models. Evidence from these programs addressing patients' social needs has shown that light-touch, low-cost “screen and refer” models have not demonstrated improvements in downstream health outcomes.⁸ In contrast, team-based integrated health systems interventions, while significantly more resource intensive, have led to improved health outcomes (as cited in Butler et al⁹). In this context of expanding the already thinly stretched roles of health care workers, studies have identified physician burnout as a potential unintended consequence of this expansion, with lower perceived clinic capacity to address social needs associated with higher rates of burnout, independent of individual skills or attitudes around social needs.^10–14 Successful implementation of these programs relies on investment in the health care team as a whole,¹⁵ although studies examining associations with burnout among nonphysician team members are lacking. From the inception of the Institute for Healthcare Improvement Triple Aim, there have been calls for an expansion to a Quadruple Aim, improving the work life of health professionals, as an equally important arm in improving health care.¹⁶ More recently, the World Health Organization has designated 2021 as the Year of the Health and Care worker, acknowledging the ongoing burnout crisis in the health care workforce.¹⁷ In efforts to integrate social care into medical care delivery, an awareness of the impact this may have on the work life of all health care team members involved is critical to achieving the Quadruple Aim.

Safety net health care settings serve vulnerable communities with high rates of unmet social needs, and often are under-resourced,¹⁸ with the demands placed on staff often exceeding the available resources. This tension likely contributes to stress and high rates of burnout through “professional dissonance,” the conflict between professional values and the requirements of the workplace.^19,20 The current burnout crisis in health care has had profound effects on individual health and job satisfaction,^21,22 resulted in poorer quality of care,^23–25 and had significant financial costs associated with turnover and lost revenue from decreased productivity.²⁶ To address this crisis, new programs in health care settings must continue to develop implementation strategies to mitigate burnout, and be aware of this as an often unintended consequence of doing more.

Mitigating these unintended consequences requires using evidence-based behavior change models throughout program implementation.^27,28 In the Capability, Opportunity, and Motivation Behavioral system (COM-B) model of behavior change, an understanding of both the internal (psychological and physical), and external (resources and support) environments impacting behavior is necessary for successful behavior change. In this model, Capability, Opportunity, and Motivation are the 3 conditions necessary for any volitional behavior to occur.²⁹ In health care settings, the tension inherent in balancing these internal and external environments is often between professional and system values, known as professional dissonance.³⁰ The psychological discomfort that occurs from this conflict may lead to symptoms of stress, burnout, or “moral injury.”^31,32 The capacity of health care workers to address patients' social needs can be understood within this behavior change framework, including the ways different behavioral conditions relate to the professional dissonance often seen in health care settings.

This study aims to characterize the perceived elements of capacity to address patients' social needs within a heterogenous group of health care workers, and to examine the association of these behavioral dimensions with burnout. The hypothesis is that different behavioral conditions will relate to burnout differently for different team members. While previous studies have focused mostly on burnout among physicians alone, the present study population sample provides perspectives from health care workers in a diversity of roles and organizations on their expanding role in addressing unmet social needs. Understanding health worker perspectives will inform developing effective implementation strategies that dampen the impact of these expanded roles on burnout.

Methods

Study sample

From February through April of 2018, staff from 46 diverse stakeholder provider organizations within one of the 25 Performing Provider Systems in the New York State Delivery System Reform Incentive Payment (DSRIP) program completed a voluntary and anonymous self-administered survey as part of a DSRIP funding requirement. Participants had the option to enter a raffle to win $50 gift cards upon survey completion. Each individual organization identified a convenience sample of at least 10% of their staff email list for survey distribution. Individuals with complete data on primary study variables were included in the analysis (N = 1022). The survey was approved by the Montefiore Medical Center/Albert Einstein College of Medicine Institutional Review Board.

Measures

Demographic characteristics of survey respondents were collected, including age, race/ethnicity and sex. Race/ethnicity was categorized as non-Hispanic White, non-Hispanic Black, Hispanic, Asian, other, or multiracial/ethnic. Respondents' place of employment was asked from a list of organizations within the Montefiore DSRIP Performing Provider System, and then categorized according to type of health care setting and services provided. Survey items addressing organizational role, burnout, and perceived capacity to address social needs, as will be outlined in the following sections, were included in the analysis (Supplementary Data S1).

Organizational role

Respondents were asked to identify their role in their organization from a list of 20 potential roles. These responses were then recategorized, through an iterative process with the research team, into 7 options based on nature and extent of patient contact. These included the following roles: case manager/social worker; nurse (registered nurse, licensed practical nurse, nurse practitioner) and physician assistant; physician; community health worker (CHW), peer navigator, health educator; clerical staff, food services, environmental services; program manager; and senior leadership (Table 1).

Table 1.

Respondent Characteristics

Characteristic	N (%)
Age (yr), median [IQR]*	46 [35–56]
Race/ethnicity †
Non-Hispanic Black	167 (14.0)
Non-Hispanic White	703 (58.9)
Hispanic	190 (15.9)
Asian	50 (4.2)
Other	28 (2.4)
Multiracial	55 (4.6)
Sex ‡
Male	246 (20.8)
Female	939 (79.2)
Organizational role §
Case manager/Social Worker	359 (30.8)
RN/LPN/PA/NP	170 (14.6)
Physician	71 (6.1)
CHW/peer navigator/educator	153 (13.1)
Clerical/food/environmental	136 (11.7)
Program managers	178 (15.3)
Senior Leadership	98 (8.4)
Organizational Type ¶
Mental Health	352 (29.3)
Substance Use	100 (8.3)
Skilled Nursing Facility	52 (4.3)
Primary Care Practice	126 (10.5)
Care Management	110 (9.1)
Federally Qualified Health Center	13 (1.1)
Hospital	394 (32.8)
Other Community Based Organization	56 (4.7)

^* N = 1150.

^†N = 1193.

^‡N = 1185.

^§N = 1165.

^¶N = 1203.

CHW, community health worker; IQR, interquartile range; LPN, licensed practical nurse; NP, nurse practitioner; PA, physician assistant; RN, registered nurse; yr, years.

Burnout

This study uses the single-item burnout question, adapted from the Physician Worklife Study,³³ as the burnout outcome variable. This item asks respondents to classify their level of burnout according to their own definition of burnout. The 5 response categories included: (1) I enjoy my work. I have no symptoms of burnout; (2) I am under stress, and don't always have as much energy as I did, but I don't feel burned out; (3) I am definitely burning out and have one or more symptoms of burnout (eg, emotional exhaustion); (4) The symptoms of burnout that I'm experiencing won't go away. I think about work frustrations a lot; and (5) I feel completely burned out. I am at the point where I may need to seek help. This survey item has been previously validated against the 22-item standard Maslach burnout inventory,³⁴ correlating closest with the emotional exhaustion domain of the larger measurement tool.^35–37 For the present analysis, this question was dichotomized into no burnout (1 and 2) vs burnout (3, 4, and 5) consistent with previous studies using this burnout measure.^10,38

Perceived capacity to address social needs

Capacity to address patients' social needs was measured using a novel 15-item survey based on the Prosci ADKAR model for organizational change management,³⁹ as well as the COM-B Behavior Change Model.²⁹ Items were grouped into 3 subdomains, including (1) Identification of social needs; (2) Providing health care to patients with unmet social needs; and (3) Linking patients to available resources to address their identified social needs. Within each of these subdomains were 5 questions addressing (1) Awareness, (2) Desire/Motivation, (3) Knowledge, (4) Ability/Preparedness, and (5) Reinforcement and alignment with organizational priorities, all on a 5-point Likert scale from Not at all (1) to Fully (5).

Statistical analysis

Of the 1947 individuals who began the survey, 1576 completed responses beyond the initial demographic questions. A complete Perceived Capacity to Address Social Needs domain was defined by at least 4 (of 5) questions answered in at least 1 (of 3) subdomains. Individuals with complete data in the Perceived Capacity to Address Social Needs domain, the burnout item, and the organizational role item were included in the analysis (N = 1022; 52% of those who began the survey). A further analysis of partially missing data is available in Supplementary Table 1, available with the article online. Descriptive statistics and a correlation matrix were calculated for all 15 survey items in the Perceived Capacity to Address Social Needs domain. The internal consistency of this domain was assessed with the Cronbach α coefficient and the Keyser-Myer Olkin measure of sample adequacy. Given that the survey was designed to capture several dimensions, an exploratory factor analysis also was performed to assess the interrelationships among items in this domain. Factors were extracted using an iterated principal factors method, and then rotated orthogonally with the Varimax method to achieve a simple structure. A least squares regression approach was used to estimate factor scores.⁴⁰

Bivariate associations among variables were assessed. For continuous variables (age, factor scores), assumptions of normality and equal variance were tested, and t tests (or nonparametric Mann-Whitney U tests) were used to assess the degree of association with burnout outcome. For dichotomous and categorical variables (race/ethnicity, sex, organizational role), chi-square tests (or Fisher exact tests) were used to assess association with burnout. Analysis of variance with post hoc Bonferroni correction was used to assess association between factor score and organizational roles.

A hierarchical modeling approach was used for the logistic regression, considering respondents' age, sex, race/ethnicity, and organizational role as covariates in each model first, followed by the 5 perceived capacity factor scores together. To test the hypothesis that behavioral conditions do not act in isolation, first-order interaction effects were tested between task-specific capacities (Factors 3, 4, and 5) and both motivation (Factor 1) and organizational reinforcement (Factor 2). When significant interaction effects were discovered, the means of the post-estimation y-hats for each model were plotted for each combination of factor score tertiles. Tertile factor score variables were used here for ease of interpretation of interaction effects. Based on the a priori hypothesis that there would be role-specific differences between perceived capacity to address social needs and burnout, the model was stratified by organizational role. The assumption of linearity in the logit for continuous factor scores was assessed with the Lowess (locally weighted scatterplot smoothing) Smoother. No multicollinearity of perceived capacity factor scores was found within each role-specific subgroup. All models were significant at P < 0.05, unless otherwise noted. Statistical analysis was done using Stata Statistical Software: Release 16.0 (StataCorp LLC, College Station, TX).

Results

Survey responses were collected from a heterogenous group of health care employees. Median age was 46 years, with the majority identifying as non-Hispanic white (58.9%), followed by Hispanic. Slightly more than 79% of respondents identified as female (Table 1). Overall, ∼31% of respondents identified their role within their organization as case managers or social workers, with program managers as the next largest group represented. Respondents from hospital systems completed the largest portion of the data, followed by mental health facilities, and then primary care practices (Table 1).

Exploratory factor analysis

For the 15-item Perceived Capacity to Address Social Needs domain, Cronbach α coefficient and the Keyser-Myer-Olkin measure of sample adequacy were 0.96 and 0.93, respectively. Exploratory factor analysis of this domain yielded 5 factors, with the first factor accounting for 77.5% of the shared variance prior to rotation. A Varimax orthogonal rotation yielded a simple structure, with the factor loadings for each factor displayed in Table 2.

Table 2.

Exploratory Factor Analysis Loadings

Variable	Factor 1	Factor 2	Factor 3	Factor 4	Factor 5
Identification of Social Needs
Awareness	0.25	0.22	0.67	0.29	0.29
Motivation	0.67	0.27	* 0.43	0.13	0.22
Knowledge	0.28	0.20	0.66	0.32	0.32
Ability	0.23	0.29	0.57	0.38	0.35
Organizational Reinforcement	0.20	0.80	0.33	0.14	0.26
Providing Care for Patients with Social Needs
Awareness	0.32	0.24	0.40	0.59	0.31
Motivation	0.75	0.27	0.17	* 0.43	0.19
Knowledge	0.26	0.25	0.31	0.68	0.35
Ability	0.24	0.31	0.26	0.71	0.36
Organizational Reinforcement	0.25	0.77	0.09	* 0.41	0.26
Linkage to Resources to Address Social Needs
Awareness	0.23	0.22	0.29	0.26	0.75
Motivation	0.63	0.24	0.17	0.15	* 0.53
Knowledge	0.16	0.25	0.22	0.25	0.83
Ability	0.20	0.27	0.20	0.26	0.81
Organizational Reinforcement	0.23	0.65	0.10	0.12	* 0.60
Proportion of Variance	0.17	0.20	0.17	0.18	0.28

^* Split loadings.

Bolded numbers represent factor loadings >0.5.

Five factors were identified that correlated with the COM-B model of behavior change.²⁹ Factors 3, 4 and 5 relate to Capability, and can be explained as task-specific awareness, knowledge, and ability, with specific tasks defined as Identification of social needs (Factor 3), Providing care to patients with social needs (Factor 4), and Linkage to social needs resources (Factor 5). Factor 1 aligns with Motivation, defined by desire across tasks. Factor 2 aligns closely with Opportunity, defined by Organizational Reinforcement (feeling supported and in alignment with organizational priorities) across tasks. Split loadings (as seen in Factors 3, 4, and 5) suggest domain-specific motivation and reinforcement, in addition to the aforementioned 3 main components. In light of the split loadings, orthogonal factors were retained to limit multicollinearity and allow for operationalizing these factors in the multivariable regression. The factor scores from the aforementioned 5 factors are used as individual variables of the Perceived Capacity to Address Social Needs domain.

Bivariate results

Bivariate associations were assessed between organizational role and the factor scores of perceived capacities to address social needs (Table 3). Physicians and nurses felt least confident in their capacity to link patients to available resources to address social needs, in comparison to the higher scores seen among case managers/social workers and senior leadership. Perceived sense of organizational reinforcement to address unmet social needs of patients/clients was highest among senior leadership, CHWs/peer navigators/health educators, and program managers, while it was lowest among nurses and physicians. Confidence in ability to provide care for patients/clients with unmet social needs was highest among physicians, nurses, case managers/social workers, and CHWs/peer navigators/health educators, and lowest among clerical/food/environmental services and senior leadership. Motivation to address unmet social needs was lowest among physicians and clerical/food/environmental services, and highest among senior leadership. Capacity to identify unmet social needs was highest among case managers/social workers and senior leadership, and lowest among clerical/food/environmental services and physicians.

Table 3.

Association Between Perceived Capacity to Address Social Needs and Organizational Role

	Case manager, social worker	Nurse (RN, LPN, NP), PA	Physician	CHW, peer navigator, educator	Clerical, food, environmental	Program managers	Senior leadership
1. Motivation	.08 (.83)	.06 (.92)	-.39 (.93)	.01 (.85)	-.35 (1.08)	.05 (.92)	.20 (.88)
2. Reinforcement	-.05 (.97)	-.20 (.94)	-.17 (1.00)	.15 (.81)	-.02 (.99)	.14 (.92)	.22 (.90)
3. Identification of Social Needs	.19 (.78)	-.15 (.78)	-.28 (.83)	.03 (.85)	-.41 (1.11)	.03 (.82)	.17 (.98)
4. Providing care	.04 (.87)	.03 (.80)	.08 (.78)	.04 (.91)	-.24 (1.00)	.06 (.86)	-.12 (1.05)
5. Linkage to resources	.23 (.83)	-.32 (.90)	-.41 (.92)	-.16 (1.17)	-.17 (1.02)	.07 (.87)	.16 (.91)

^* All values mean (standard deviation), with P values for all row analysis of variance <0.001.

CHW, community health worker; LPN, licensed practical nurse; NP, nurse practitioner; PA, physician assistant; RN, registered nurse.

Also assessed were bivariate associations of burnout with demographic characteristics and perceived capacity factor scores (Table 4). Respondent age differed significantly by level of burnout, with higher burnout reported among slightly younger respondents (median of 43 vs 47 years). Level of burnout did not differ significantly by race/ethnicity, sex, or organizational role in the full data set. Lower perceived organizational reinforcement (P = 0.07), as well as lower motivation to address patients' social needs (P = 0.05) also were associated with higher burnout.

Table 4.

Bivariate and Multivariate Relationships of Perceived Capacity Variables, Covariates, and Burnout

Characteristic	Bivariate relationships, N (%)			Full logistic regression model*		Logistic regression model including interaction terms*
	High burnout	Low burnout	P value	OR [95% CI]	P value	OR [CI]	P value
Age (yr), median [IQR]	43 [32–53]	47 [36–57]	0.0001	0.98 [0.96, 0.99]	<0.001	0.98 [0.97, 0.99]	0.001
Race/ethnicity Non-Hispanic White Non-Hispanic Black Hispanic Asian Other Multiracial	177 (25.3) 45 (27.3) 48 (25.5) 10 (20) 4 (14.3) 15 (27.3)	522 (74.7) 120 (72.7) 140 (74.5) 40 (80) 24 (85.7) 40 (72.7)	0.72	[REF] 0.96 [0.62, 1.51] 0.96 [0.63, 1.46] 0.73 [0.33, 1.63] 0.45 [0.14, 1.47] 0.81 [0.40, 1.63]	0.87 0.85 0.44 0.19 0.56	[REF] 0.94 [0.60, 1.50] 0.98 [0.64, 1.50] 0.67 [0.30, 1.53] 0.50 [0.15, 1.65] 0.80 [0.39, 1.65]	[REF] 0.81 0.94 0.35 0.26 0.55
Sex Male Female	57 (23.2) 235 (25.3)	189 (76.8) 695 (74.7)	0.50	1.03 [0.70, 1.52]	0.86	1.08 [0.73, 1.60]	0.70
Organizational role Case manager, Social Worker Nurse (RN, LPN, NP), PA Physician CHW, peer navigator, educator Clerical, food, environmental Program managers Senior Leadership	101 (28.3) 48 (28.6) 11 (15.5) 37 (24.3) 30 (22.2) 45 (25.4) 17 (17.7)	256 (71.7) 120 (71.4) 60 (84.5) 115 (75.7) 105 (77.8) 132 (74.6) 79 (82.3)	0.13	1.31 [0.69, 2.49] 1.44 [0.72, 2.91] 0.66 [0.26, 1.70] 1.18 [0.57, 2.41] 0.91 [0.42, 1.97] 1.30 [0.65, 2.61] [REF]	0.42 0.31 0.39 0.66 0.82 0.46 [REF]	1.17 [0.61, 2.24] 1.32 [0.65, 2.67] 0.57 [0.21, 1.50] 1.08 [0.52, 2.23] 0.91 [0.42, 1.97] 1.16 [0.57, 2.33] [REF]	0.64 0.45 0.25 0.84 0.81 0.69 [REF]
Perceived Capacity Variables, mean (SD) 1. Motivation 2. Reinforcement 3. Identification of Social Needs 4. Providing Care 5. Linkage to Resources	-.14 (.97) -.35 (.98) .06 (.84) .04 (.87) -.03 (1.0)	.05 (.88) .12 (.89) -.02 (.89) -.01 (.90) .01 (.93)	0.05 0.07 0.25 0.51 0.17	0.78 [0.66, 0.92] 0.55 [0.47, 0.65] 1.13 [0.95, 1.35] 1.18 [0.99, 1.40] 0.96 [0.81, 1.12]	0.004 <0.001 0.17 0.06 0.58	0.71 [0.59, 0.86] 0.51 [0.42, 0.62] 1.17 [0.95, 1.44] 1.20 [0.98, 1.48] 1.00 [0.84, 1.20]	<0.001 <0.001 0.14 0.08 0.97
Interaction Effects of Perceived Capacity Variables Motivation x Identification of Social Needs Motivation x Providing Care Motivation x Linkage to Resources Reinforcement x Identification of Social Needs Reinforcement x Providing Care Reinforcement x Linkage to Resources Motivation x Reinforcement						1.03 [0.84, 1.20] 0.72 [0.60, 0.87] 0.93 [0.78, 1.11] 0.78 [0.63, 0.97] 1.19 [0.97, 1.47] 1.25 [1.02, 1.51] 1.07 [0.88, 1.29]	0.77 0.001 0.43 0.03 0.10 0.03 0.49

N = 1022.

^* Model P value <0.001.

Bolded numbers represent statistically significant results (at P <0.05).

CHW, community health worker; CI, confidence interval; IQR, interquartile range; LPN, licensed practical nurse; NP, nurse practitioner; OR, odds ratio; PA, physician assistant; RN, registered nurse; SD, standard deviation; yr, years.

Regression models

Overall model

In the overall sample, demographics accounted for 2% of the variance (pseudo R²) of burnout, primarily related to age. After controlling for demographics, perceived capacity to address social needs variables (factor scores) accounted for an increase of 8% of the variance. Perceived lack of Motivation (odds ratio [OR] 0.78, 95% confidence interval [CI] 0.66 - 0.92) and lack of organizational Reinforcement (OR 0.55, 95% CI 0.47 - 0.65) were both associated with higher burnout. Additionally, there was a trend toward significance of lower Perceived capacity to provide care for patients with social needs associated with higher levels of burnout (Table 4).

Interaction effects among perceived capacity behavioral conditions

A significant interaction effect was noted between Motivation and Perceived capacity to provide care for patients with social needs (Table 4). Post-hoc testing revealed that for those with the lowest level of motivation, higher perceived capacity to provide care for patients with social needs was associated with higher burnout (Supplementary Fig 1a). On the other hand, for those with higher levels of motivation, levels of burnout remained similar across levels of perceived capacity to provide care. A significant interaction effect also was found between Reinforcement and Perceived capacity to Identify social needs of patients in the overall regression model (Table 4). For respondents with the lowest level of organizational reinforcement, higher perceived capacity to identify social needs was associated with higher burnout, similar to the aforementioned interaction (Supplementary Fig 1b). This association of greater burnout with greater perceived capacity to identify social needs held true across levels of organizational reinforcement, although the highest rates of burnout remained for those with the lowest level of organizational reinforcement. Finally, an interaction effect also was found between Reinforcement and Perceived capacity to link patients to social needs resources (Table 4). Unlike the aforementioned interactions, for respondents with the lowest levels of organizational reinforcement, higher perceived capacity to link patients to social needs resources was associated with lower burnout. Respondents with higher levels of organizational reinforcement, however, had relatively similar levels of burnout, regardless of perceived capacity to link patients to social needs resources (Supplementary Fig 1c).

Stratified model

Stratified analyses by organizational role revealed that younger age was associated with higher burnout only among CHWs/peer navigators/health educators, adjusting for all specified covariates (OR 0.95, 95% CI 0.91–0.99) (Table 5). Female respondents were also 2.74 times as likely to report higher levels of burnout than males (95% CI 1.20-.27) only among case managers/social workers. Lower Reinforcement was significantly associated with higher burnout for almost all roles, and lower Motivation was associated with higher burnout among only case managers/social workers and senior leadership (Table 5).

Table 5.

Multivariate Regressions of Perceived Capacity Variables Predicting Burnout, Stratified by Organizational Role (OR, 95% CI)

	Case manager, social worker	Nurse (RN, LPN, NP), PA	Physician	CHW, peer navigator, educator	Clerical, food, environmental	Program managers	Senior leadership
Stratified Models N Pseudo R2 P value	326 0.14 <0.001	144 0.10 0.51	56 0.59 0.01	134 0.27 0.003	109 0.11 0.72	153 0.23 0.004	82 0.35 0.05
Age	0.99 [0.97, 1.01]	0.99 [0.95, 1.02]	1.22 [0.91, 1.63]	*0.95 [0.91, 0.99]	0.97 [0.92, 1.02]	0.99 [0.95, 1.03]	0.92 [0.83, 1.01]
Race Non-Hispanic White Non-Hispanic Black Hispanic Asian Other Multiracial	[REF] 1.06 [0.50, 2.25] 1.56 [0.79, 3.07] 0.91 [0.14, 6.13] 1.01 [0.16, 6.46] 0.24 [0.04, 1.35]	[REF] 1.06 [0.34, 3.29] 0.39 [0.09, 1.68] 0.46 [0.12, 1.84] —- 1.49 [0.30, 7.35]	[REF] —- —- 15.23 [0.01, 70841] —- 334.59 [0.01, 2.03]	[REF] 1.45 [0.38, 5.58] 1.97 [0.52, 7.47] 1.79 [0.05, 58.90] —- 14.93 [0.97, 229.87]	[REF] 0.60 [0.11, 3.30] 0.45 [0.13, 1.56] —- —- 0.58 [0.05, 6.42]	[REF] 0.14 [0.02, 1.02] 1.21 [0.30, 4.92] 1.60 [0.23, 11.26] 0.59 [0.04, 8.55] 0.04 [0.01, 1.26]	[REF] 2.26 [0.14, 36.42] 2.32 [0.15, 36.20] —- —- 0.07 [0.01, 92.78]
Sex	*2.74 [1.20, 6.27]	0.55 [0.16, 1.82]	0.46 [0.01, 25.09]	0.32 [0.10, 1.04]	0.72 [0.17, 3.02]	1.15 [0.38, 2.87]	4.96 [0.37, 65.62]
Perceived Capacity Variables 1. Motivation 2. Reinforcement 3. Identification of Social Needs 4. Providing Care 5. Linkage to Resources	**0.57 [0.39, 0.82] ***0.44 [0.29, 0.66] 1.19 [0.79, 1.80] 1.09 [0.75, 1.59] 1.30 [0.89, 1.88]	1.25 [0.63, 2.48] **0.50 [0.30, 0.83] 0.92 [0.53, 1.60] 1.10 [0.63, 1.94] 0.83 [0.51, 1.34]	3.98 [0.01, 4603] *0.04 [0.01, 0.72] 0.96 [0.01, 151.8] 0.24 [0.01, 92.76] 0.58 [0.02, 20.24]	0.51 [0.25, 1.02] 0.57 [0.25, 1.30] 1.12 [0.53, 2.36] 2.04 [0.96, 4.34] 0.95 [0.58, 1.60]	1.43 [0.74, 3.61] 0.47 [0.21, 1.33] 1.87 [0.75, 3.50] 1.00 [0.39, 1.95] 1.19 [0.58, 2.20]	0.67 [0.41, 1.08] ***0.25 [0.12, 0.54] 1.42 [0.72, 2.80] 1.51 [0.69, 3.12] 1.45 [0.70, 2.98]	*0.18 [0.04, 0.84] *0.27 [0.08, 0.89] 0.59 [0.08, 4.17] 7.97 [0.43, 147.59] 0.68 [0.16, 2.93]
Interaction Effects Motivation x Identification of Social Needs Motivation x Providing Care Motivation x Linkage to Resources Reinforcement x Identification of Social needs Reinforcement x Providing Care Reinforcement x Linkage to Resources Motivation x Reinforcement	0.93 [0.60, 1.44] **0.56 [0.39, 0.82] 0.74 [0.49, 1.10] 0.87 [0.60, 1.25] 1.08 [0.76, 1.56] 1.24 [0.86, 1.81] 1.22 [0.85, 1.76]	1.22 [0.60, 2.49] 0.78 [0.44, 1.40] 1.03 [0.55, 1.96] 0.71 [0.42, 1.19] 1.21 [0.74, 1.96] 0.98 [0.60, 1.61] 1.32 [0.77, 2.28]	0.09 [0.01, 5.65] 1.00 [0.10, 10.34] 3.9 [0.10, 129.71] 2.04 [0.04, 100.45] 0.99 [0.01, 97.98] 4.20 [0.03, 540.25] 0.55 [0.01, 294.68]	1.02 [0.40, 2.57] 0.90 [0.42, 1.93] 0.73 [0.38, 1.41] **0.20 [0.06, 0.66] 2.89 [0.83, 10.04] 1.22 [0.57, 2.62] 0.74 [0.30, 1.85]	1.62 [0.96, 2.72] 0.73 [0.43, 1.23] 1.39 [0.79, 2.44] 1.07 [0.36, 3.17] 0.80 [0.29, 2.20] 0.77 [0.40, 1.50] 1.14 [0.42, 3.05]	0.93 [0.52, 1.67] 0.58 [0.31, 1.09] 1.00 [0.44, 2.24] *0.35 [0.14, 0.86] 1.95 [0.83, 4.59] *2.48 [1.01, 6.09] 1.55 [0.72, 3.34]	9.99 [0.24, 417.80] 0.06 [0.01, 2.65] 1.61 [0.18, 13.96] 2.36 [0.13, 44.06] 0.07 [0.01, 3.45] 2.86 [0.29, 28.70] 1.24 [0.32, 4.86]

^* P < 0.05; ^**P < 0.01; ^***P < 0.001.

—- Group dropped because of <5 responses.

Bolded numbers represent statistically significant results (at P <0.05).

CI, confidence interval; CHW, community health worker; LPN, licensed practical nurse; NP, nurse practitioner; OR, odds ratio; PA, physician assistant; RN, registered nurse.

When first-order interaction effects were examined in the stratified models, there was a significant interaction among case managers and social workers between Motivation and Perceived capacity to provide care for patients with social needs (Table 5). For those with lower motivation, higher perceived capacity to provide care for patients/clients with social needs was associated with higher burnout. An opposite trend was observed for those with higher motivation, with lower perceived capacity to provide care for patients/clients with social needs associated with higher burnout (Supplementary Fig 1d). A significant interaction effect was also noted between Reinforcement and Perceived capacity to Identify social needs of patients among CHWs and peer navigators, as well as program managers (Table 5). A post-estimation analysis of these groups showed that for those with a lower sense of organizational reinforcement, higher perceived capacity to identify social needs was associated with higher burnout. However, for those with a higher sense of organizational reinforcement, similar degrees of burnout were seen regardless of one's perceived capacity to identify social needs. Looked at differently, if respondents did not feel confident in their capacity to identify social needs, higher reinforcement was associated with higher burnout (Supplementary Fig 1e). Finally, there was a significant interaction between Reinforcement and Perceived capacity to Link patients to social needs resources among program managers (Table 5). Similar to this interaction in the full model, for those with the lowest level of organizational reinforcement, higher capacity to link to social needs resources was associated with lower burnout. Similar levels of burnout were seen for those with higher levels of organizational reinforcement, regardless of capacity to link to social needs resources (Supplementary Fig 1f).

Discussion

This cross-sectional study of health care provider survey responses found that a lower sense of organizational reinforcement, as well as motivation for addressing patients' social needs, were both associated with a higher rate of burnout. In the models stratified by organizational role, this inverse relationship between reinforcement and burnout was found among most health care worker roles (including physicians), while motivation was only inversely associated with burnout among case managers/social workers and senior leadership. These findings are consistent with prior studies among physicians, which highlight the relationship between organizational reinforcement (through clinic resources) and burnout.^10–12 This study also adds the important dimension of motivation among other health care workers, supporting the COM-B model of behavior change. Although this study cannot speak to causality or directionality in these relationships, in theory a connection is plausible. Addressing unmet social needs may uncover traumatic experiences for patients, with ensuing conversations that are often difficult and uncomfortable, confronting the realities of living with poverty, racism, and intimate partner violence. Secondary trauma (to health care providers) is a known consequence of providing care for individuals with a history of trauma, and may be the underlying mechanism by which perceived capacity to address unmet social needs is associated with burnout.

The factor analysis also clarifies how motivation, organizational reinforcement, and 3 task-specific capabilities (identifying, providing care, and linking to resources) provide a structure that can define a Perceived Capacity to Address Social Needs domain (Table 2), and fit well with previously defined models of behavior change,²⁹ as well as assessments of capacity to address social needs.¹² The instrument used by Olayiwola et al, adapted from an alternative behavioral framework, includes both internal factors (knowledge, attitudes, confidence and skills), and external factors (organizational capacity and resources) as contributory to one's perceived capacity to address patients' social needs. The present analysis and framework add to this literature, providing further evidence for behavior change models to shape workforce development initiatives in this field, across multiple roles and a variety of organizational types.

Not surprisingly, this study found significant differences in perceived capacity domains among respondents in different organizational roles (Table 3). Other studies have found that physicians report lower confidence in addressing their patients' social needs than other health care team members.¹⁴ Present study results support these findings, with physicians in this study reporting some of the lowest motivation, sense of organizational reinforcement, and capabilities to identify social needs or to link to available resources. They did, however, report a greater capacity to provide care to patients with social needs, as one would expect in comparison to respondents from other roles. Case managers and social workers reported the highest capacities to identify social needs, and to link to available resources, although reported lower sense of organizational reinforcement. Interestingly, CHWs, peer navigators, and health educators reported a high sense of organizational reinforcement, but had lower capacity to link patients to available resources. Taken together, these findings suggest that designing implementation strategies and workforce trainings that acknowledge these differences in role may be important, and better than a one-size-fits-all approach to implementation of social needs screening programs.

In the present study data set, although ∼25% of the sample reported symptoms of burnout, only ∼17% of physicians reported burnout, which is lower than other studies have reported with similar metrics (27%),¹⁰ as well as lower than other surveys of physicians within the authors' own institution (21%; personal communication with Damara Gutnick, MD; 9/2/21). However, there were no significant differences in burnout by organizational role, suggesting that all members of the health care team in this study were subject to burnout at similar rates. This provides further validation that studies examining health care system capacities to tackle new challenges should not only look at unintended consequences such a burnout, but also should include team members other than physicians. Although no differences were noted in burnout by sex, the stratified model shows that for case managers and social workers alone, women were 2.74 times as likely to experience symptoms of burnout than men.

As shown by Schickedanz et al, accounting for time, resources, and professional scope is critical in designing implementation strategies for social needs screening programs.¹⁴ The present study adds to this literature as well, highlighting not only these tangible resources that are needed for skill building, but even more fundamentally, the motivation and organizational reinforcement that are necessary to succeed in building these programs. The implications of these data extend beyond the literature, providing evidence for the inclusion of strategies to build organizational reinforcement into the implementation of these programs. Additionally, these data suggest that understanding where different team members are in relation to their motivation to do this work is also critical to avoid unintended burnout. As momentum for addressing the work life of health professionals continues to build, this study provides empiric evidence of the need to incorporate behavior change principles, addressing skills as well as motivations and organizational reinforcement, into health system strategies to achieve the Quadruple Aim.

This study has limitations. First, as this survey was done as part of a larger implementation planning process through the New York State DSRIP program, sampling of respondents was determined by individual organizations, and so the data are from a convenience sample. Similarly, there was a high rate of respondents beginning the survey but not completing it (Supplementary Table 1). The supplemental data show that while there was variability in which groups of respondents completed the survey, the highest rates of completion were from those in roles traditionally more likely to screen patients for unmet social needs. This selection bias and item nonresponse bias both limit the generalizability of these results. Also, a single-item burnout indicator was used as the outcome in this study. This has been shown to be a close approximation of the emotional exhaustion dimension of the full Maslach burnout Inventory⁴¹; however, the authors acknowledge that important components of burnout are missing in this analysis (namely, personal accomplishment and depersonalization). Finally, this study is cross-sectional in nature, and so the directionality or causality of the associations could not be interpreted.

Despite these limitations, this study highlights multiple important analytic approaches to consider in future studies. First, the overall analysis masked many of the within-role differences that were seen in the stratified analysis. Although role was not significantly associated with level of burnout in the bivariate analysis, this study stratified by role because of the a priori hypothesis, and was able to detect multiple noteworthy differences in associations. Second, in studying multiple dimensions of capacity in a work environment, it is known that these variables often act in concert with one another, and not in isolation. Given this multidimensional approach, examining interaction effects can be an important way to understand these nonlinear, higher order associations, and was useful here in providing explanatory models for these relationships. Additionally, data from this unique, heterogenous sample of health care workers provide an opportunity to use real-world data from a learning health system to expand the current literature and inform implementation efforts.

Conclusions

These analyses provide further evidence strengthening the association between burnout and perceived capacities to address unmet social needs in clinical contexts. The role-specific findings from this study also may have broad implications for addressing training and workflow in different settings. As early as the mid-19^th century, Rudolf Virchow, the founder of the Social Medicine movement, described physicians as the natural attorneys of the poor, well-positioned to solve their social problems.⁴² As health care has evolved into team-based interventions, staff across the care spectrum are now tasked with addressing larger social issues impacting their patients. Present study findings expand prior studies beyond physicians and primary care settings alone to reflect the reality of how these interventions are being implemented across health care networks. A systems approach, aligning organizational priorities and staff motivations, in addition to task-specific skill sets is likely necessary to prevent exacerbation of burnout in this setting, and can serve as a model for transformation of a clinical practice environment.

Authors' Contributions

Conception or design of the work: Drs. Telzak, Chambers, Gutnick, Flattau, McAuliff, and Rapkin, and Ms. Chaya. Data collection: Drs. Gutnick, McAuliff, and Rapkin, and Ms. Chaya. Data analysis and interpretation: Drs. Telzak, Chambers, Gutnick, Flattau, McAuliff, and Rapkin, and Ms. Chaya. Drafting the article: Dr. Telzak. Critical review and revision of the article: all authors. Final approval of the version to be published: all authors.

Author Disclosure Statement

The authors declare that there are no conflicts of interest.

Funding Information

The research described was supported by National Institutes of Health/National Center for Advancing Translational Science Einstein-Montefiore CTSA Grant Number UL1TR002556, as well as by Empire Clinical Research Investigator Program at the Albert Einstein College of Medicine and the NY State Delivery System Reform Incentive Payment Program.

Supplementary Material

Supplementary Data S1

Supplementary Figure S1

Supplementary Table S1