Healthcare system innovators may seek out physician associates/assistants (PAs) to take on increasing responsibility for leadership tasks in the healthcare workplace.1 PAs may lead in clinical, administrative, academic, or research settings, among others.2 Taking on leadership tasks may require PAs to expand their roles beyond current competence or perceived self-efficacy, triggering occupational stress.3 On the other hand, the responsibility to lead may motivate PAs. This study sought to broaden what is known about the influence of leadership task responsibility on PA professional well-being.

Leadership has been defined and measured in numerous ways in healthcare well-being literature. Tetzlaff and colleagues analyzed leadership effectiveness from the direct report perspective by asking PAs to assess the leadership qualities of their collaborative physicians.4 Ekström and Idvall assessed the influence of leadership training by examining perceived readiness to lead among new nurse team leaders.5 Louwagie and colleagues examined leadership development by exploring the influence of physician support for PA leaders.2 For this study, PA leadership was defined as the sum of all leadership tasks for which PA leaders were responsible. Research has demonstrated that physician and nurse leaders may be assigned leadership tasks before undergoing leadership training and developing leadership competence; the same may hold true for PAs.5-7 By measuring leadership task responsibility, the focus of the study remained on how leadership tasks influenced well-being regardless of leadership readiness or effectiveness.

Outcomes from studies examining how leadership influences the well-being of healthcare leaders suggest a positive relationship.6-11 Nurse managers in a study conducted in Belgium were more likely to be engaged than burned out.9 In a review of European studies on the satisfaction of hospital physicians, Domagała and colleagues found that having a leadership position was associated with higher satisfaction levels.10 In a study of US medical residents, perceived leadership development opportunities negatively correlated with burnout.11

Investigations into how PAs perceive leadership task responsibility are limited. One national study conducted between 2015 and 2016 found that 72.8% of PA respondents felt that they had the authority and competency necessary to lead or influence others.12 A 2002 study by Bell and colleagues of PAs in emergency medicine found a correlation between decreased emotional exhaustion and increased administrative time and concluded that administrative time may serve as a respite from a stressful clinical environment.13 More research is needed to understand how increasing leadership task responsibility may influence PA job attitudes.

In the absence of empirical research, the theory of employee identification put forth by Atewologun and colleagues suggests an interaction between PA career length, leadership responsibility, and professional well-being.14 Atewologun and colleagues found that an employee's identification as a follower at work diminishes over time.14 For PAs, this may manifest as the expectation of an expanded scope of practice as their careers mature. New PAs may initially expect tasks to be delegated to them until they demonstrate the clinical competence to initiate these tasks on their own. Progression beyond here may involve PAs expecting to delegate tasks to others in a leadership role. Consistent with this, Gillette and colleagues theorized that as PA clinical roles expand, PAs may be called on to take more significant levels of responsibility for the delivery of healthcare.1 Accordingly, hypothesis 1 involves the relationship between career length and levels of leadership responsibility.

Hypothesis 1: Increasing PA career length positively predicts levels of leadership task responsibility.

Increasing leadership task opportunity may in turn improve PA well-being. Leadership task responsibility may enhance perceived autonomy, job meaningfulness, and job satisfaction; provide a respite from the rigors of clinical practice; or satisfy career advancement expectations.7,10,13,14 Several aspects of expanded PA leadership responsibility may promote professional fulfillment and attenuate burnout.

Hypothesis 2: Increasing leadership task responsibility positively predicts PA professional fulfillment and negatively predicts PA burnout.

If the relationship between PA experience level and leadership task responsibility is incongruous, however, poor professional well-being may result. Kelly and colleagues demonstrated that less leadership experience predicted higher burnout among nurse leaders.6 Ekström and Idvall found that new nurses in Sweden were stressed by the expectation to lead while simultaneously orienting to their new profession.5 Burnout may develop if PAs are assigned leadership roles before their capacity to lead is fully developed or if they are not offered leadership opportunities as their careers mature.

Hypothesis 3: Increasing levels of leadership task responsibility mediates the relationship between PA career length and professional well-being.

METHODS

The study was approved by the Walden University institutional review board. The study was designed to address gaps in current PA well-being research. Existing studies on PA well-being either primarily used nonvalidated measures or did not consider leadership as a predictor of well-being.15-20 This study examined leadership as an antecedent to well-being and was conducted using a validated outcome measure, making study findings comparable across PA studies and with research from other healthcare disciplines.

Setting, population, and sampling procedures

This quantitative and nonexperimental study used portions of an archival dataset gathered in 2018 as part of a larger survey conducted annually by researchers from the American Academy of Physician Associates (AAPA).21 Permission to use AAPA data was granted through an application process coordinated by the AAPA's research department.

The study population was all actively practicing PAs in the United States. Inclusion criteria were US residence, nonretired status, and valid email address.21 Current members of the AAPA and nonmembers whose email addresses were on file with the AAPA were sent survey links via email.21 Survey links also were posted on social media.

The dataset used in this study was gathered between February 2 and March 2, 2018; items addressed practice conditions in 2017.21 Links to the survey were sent to 78,244 PAs and were posted on social media. An a priori sample size for each path analysis of the linear regression with mediation was calculated using a medium Cohen f2 effect size of 0.15, an alpha error probability of 0.05, and a power of 0.8. The minimum recommended sample size ranged from 55 to 85. Responses varied per item used for this study from 7,150 to 9,906, resulting in a response rate range of 9.14% to 12.66%. Despite the low response rate, the overall survey from which this dataset was extracted had a low margin of error, or +/- 1.05 percentage points at the 95% confidence level.21

Datasets from previous iterations of the AAPA Salary Report have demonstrated good representativeness of the population of certified PAs in the United States.20 To verify the validity of the current study sample, descriptive statistics related to sex, race, and ethnicity were compared with population-level data from the National Commission on Certification of Physician Assistants (Table 1).22 The sample also was validated across specialty categories using NCCPA data (Table 1).23 The representativeness of the sample's sex composition was equivalent to the national population of PAs, with only minor differences in its racial, ethnic, and specialty composition.

TABLE 1. - Descriptive statistics of sample demographic and control variables and comparative population Percentages may not sum to 100 because of rounding. Sample % Sample n NCCPA 2018 % NCCPA 2018 n Sex    Male 30.7 2,721 31.2 40,900    Female 69.3 6,152 68.8 90,239    Total 8,873 131,139 Race    White 85.7 7,624 86.9 94,827    Black 2.5 222 3.6 3,951    Native American/Alaska Native 0.4 35 0.4 411    Asian 4.9 435 5.8 6,295    Native Hawaiian/Other Pacific Islander 0.3 25 0.3 335    Two or more1 2.2 194 --- ---    Other 1.7 149 3 3,321    Prefer not to answer1 2.4 213 --- ---    Total 8,897 109,140 Ethnicity    Hispanic or Latino 5.4 482 6.3 7,078    Not Hispanic or Latino 92.4 8,239 93.7 104,550    Two or more NA NA 0 299    Prefer not to answer 2.2 192 0 4,925    Total 8,913 116,852 Specialty2    Primary care 24 2,161 25.8 25,487    Internal medicine subspecialty 11.8 1,058 10.5 10,463    Pediatric subspecialty 1.1 102 1.1 1,334    Surgical subspecialty 25.4 2,301 23.3 23,015    Emergency medicine 9 813 13 12,860    All other specialties 26.5 2,391 25.9 25,404    No medical specialty3 2.2 201 --- ---    Total 9,027 98,563 Changed in the past year4    Employer (yes) 12.4 984 --- ---    Specialty (yes) 6.4 511 --- ---

1NCCPA does not report PAs identifying as two or more races or those preferring not to answer.

2Data on specialty derived from NCCPA's 2018 Statistical Profile of Certified Physician Assistants by Specialty.

3NCCPA does not report PAs identifying no medical specialty.

4NCCPA does not report PAs identifying no medical specialty or the percentage of PAs who recently changed employer or specialty, so neither of these categories were included.

Instruments and operationalization of variables

The outcome variables were professional fulfillment and burnout. Professional fulfillment develops when a person experiences a job as rewarding, engaging, and meaningful.24 Burnout is characterized by varying degrees of physical and emotional exhaustion and detachment from patients and coworkers.24 Higher levels of burnout among physicians have correlated with reduced productivity, greater intention to leave their positions, higher likelihood of medical errors, and reduced personal physical and psychologic health.25 Additionally, healthcare leaders' level of satisfaction and burnout has implications for how they are perceived as leaders and the well-being of those they lead.4,26

The Professional Fulfillment Index (PFI) was used in this study to measure professional fulfillment and burnout.24 The PFI's 16 items are divided into two subscales, with one measuring professional fulfillment and the other measuring work exhaustion and interpersonal disengagement. The professional fulfillment subscale asks: “How true do you feel the following statements are about you at work during the past 2 weeks?” and a sample item is My work is satisfying to me. The five-point response scale consists of not at all, somewhat true, moderately true, very true, and completely true.

The exhaustion and interpersonal disengagement subscales ask: “To what degree have you experienced the following?” and a sample item is During the past 2 weeks I have felt lacking in enthusiasm at work. Response options were not at all, very little, moderately, a lot, and extremely.

The PFI has been validated against the Maslach Burnout Inventory-Human Services Survey (MBI-HSS), a gold-standard proprietary tool used to measure burnout.24 The internal reliability of the PFI has been established in three previous studies.24,27,28

The predictor variable experience level was used as an indicator of career progression and measured continuously as the number of years that respondents had been practicing as PAs. The validity of experience level as a measure of career progression and a relevant measure of healthcare professional well-being was established in previous studies of PAs, NPs, and physicians.13,29-32

The potential mediator of leadership task responsibility was represented as an increasing number of leadership-related tasks. The 2018 AAPA Salary Report included a list of 13 leadership tasks that respondents were asked to identify as having or not having. Leadership task options on the survey were budgeting, PA credentialing, ensuring accreditation compliance, hiring and firing, managing nonclinical staff, determining workforce requirements, managing PAs, competency assessments, performance assessments, managing other clinical staff, quality improvement activities, committees, and educating others. All leadership tasks selected by each respondent were accumulated into an aggregate score to distinguish increasing levels of leadership task responsibility between respondents. Such aggregate scores represent multiple theoretically relevant dimensions of one construct and enable a simplified method for assessing a complex phenomenon.33

PA leadership literature suggests that the 13 leadership tasks listed in the AAPA 2018 Salary Report encompass core components of PA leadership. As frontline providers of medical care, PAs are well positioned to improve transactional aspects of healthcare delivery, such as workflow efficiency, quality improvement, and workload management of clinical and nonclinical staff.34,35 Additionally, by having an intimate understanding of their occupational model, PAs have much to contribute in the realm of regulatory and accreditation adherence, as well as performance and competency assessment of other PAs.36 As such, the tasks listed in the AAPA 2018 Salary Report are valid indicators of PA leadership responsibility.

Data collection and analysis

Ethical procedures used by the AAPA to collect the original dataset included delivery of informed consent, deidentification of study data with member profiles, and storage of data on a secure local network. Respondents also were notified of the potential for the data to be provided to secondary researchers as password-protected, deidentified datasets. The original research entity completed data cleaning. Before dataset release, members of the AAPA research department screened for and removed items with missing responses. The salary survey also is programmed at the response level for responses to fit within two standard deviations (SDs) of the mean, verifying that each response is within a real limit (email from Noël Smith, senior director of PA and industry research and analysis at the AAPA, December 12, 2019).

Descriptive, correlation, and predictive statistical analyses were completed. Means and SDs were calculated for both outcome variables and all predictor variables, including the potential mediator variable and all control variables, to verify adequate range and normal distribution. Pearson correlations were calculated between dyads of variables. Bivariate linear regression was conducted to evaluate the direct effects of experience level and leadership responsibility on professional fulfillment. Multiple linear regression with mediation was conducted to determine the effect of leadership responsibility on the association between experience level and PA professional fulfillment or burnout.

Sex and recent employer or specialty change were controlled in the regression analyses. Sex has consistently produced statistically significant differences among samples of PAs regarding various well-being indicators.15,20,37 Recent employer or specialty change was controlled because either circumstance may confound the relationship between career length and professional well-being (Table 1).

RESULTS

Descriptive statistics for each variable are listed in Table 2. Mean years of experience for sample respondents was 11.23 years (SD = 9.632, range = 0 to 48 years). Mean aggregate leadership task score, termed leader task score going forward, was 1.21 tasks (SD = 2.618, range = 0 to 13 years), with 74.1% reporting having no leadership tasks and 25.9% as having between 1 and 13 leadership tasks. Table 3 demonstrates the frequency of respondents who reported either having or not having each leadership task. Aggregate leadership task scores from respondents who indicated yes to at least one leadership task are shown in Figure 1.

TABLE 2. - Descriptive statistics of predictor, mediator, and outcome variables Range Variable type n Mean SD Minimum Maximum Predictor: Years of experience 8,761 11.23 9.632 0 48 Potential mediator: Aggregate leader score 9,906 1.21 2.618 0 13 Outcome1    PFI burnout average 7,662 1.04 .656 0 4    PFI professional fulfillment average 7,710 2.66 .874 0 4

1The criterion for high professional fulfillment was an average item score of 3 or greater; for high burnout, an average item score of 1.33.24

TABLE 3. - Pearson correlation coefficient (r) of control, predictor, mediator, and outcome variables A small effect is indicated by r = 0.1 to 0.3 or -0.1 to -0.3, a medium effect by r = 0.3 to 0.5 or -0.3 to -0.5, and a large effect by r = 0.5 to 1 or -0.5 to -1.38 Variable Results Sex Changed employer Changed specialty Years Leader score PFI burnout average PFI professional fulfillment average Sex (1 = female) Pearson r 1 P value n 8,873 Changed employer Pearson r 0.027 1 P value .017 n 7,923 7,961 Changed specialty Pearson r 0.033 0.517 1 P value .003 0 n 7,923 7,961 7,961 Years Pearson r -0.155 -0.052 -0.071 1 P value 0 0 0 n 8,714 7,961 7,961 8,761 Leader score Pearson r -0.108 -0.081 -0.06 0.232 1 P value 0 0 0 0 n 8,873 7,961 7,961 8,761 9,906 PFI burnout average Pearson r 0.052 -0.011 -0.014 -0.104 -0.062 1 P value 0 .335 .233 0 0 n 7,628 7,662 7,662 7,662 7,662 7,662 PFI professional fulfillment average Pearson r -0.083 -0.026 -0.015 0.071 0.143 -0.584 1 P value 0 .023 .196 0 0 0 n 7,676 7,710 7,710 7,710 7,710 7,652 7,710
FIGURE 1.:

Percentage of respondents by aggregate leadership task score

Results for both professional fulfillment and burnout were reported as scale averages, consistent with previously established parameters for the PFI. The criterion for professional fulfillment was an average item score of 3 or greater; for burnout, an average item score on the burnout scale of 1.33 or higher.24 The mean professional fulfillment scale item average was 2.66 (SD = 0.874, range = 0 to 4 years) and mean burnout scale item average, calculated using averages from the exhaustion and interpersonal disengagement subscales, was 1.04 (SD = 0.656, range = 0 to 4 years). Most respondents in the sample did not meet the criteria for professional fulfillment or burnout.

Pearson correlation analysis

Coefficients (r) were calculated for dyad relationships between the control, predictor, mediator, and outcome variables (Table 4). Using standard criteria for effect size, a small effect is indicated by r = 0.1 to 0.3 or -0.1 to -0.3, a medium effect size by r = 0.3 to 0.5 or -0.3 to -0.5, and a large effect size as r = 0.5 to 1 or -0.5 to -1.38 Because of their significant correlations with other variables, sex, changed employer in the past year, and changed specialty in the past year were used as covariates in the mediation analyses.

TABLE 4. - Direct effects of independent variables on PA professional fulfillment and burnout All paths controlled for sex and whether respondents had changed employer or specialty in the past year. Paths with statistically significant direct effects are in boldface. A small effect is indicated by r = 0.1 to 0.3 or -0.1 to -0.3, a medium effect by r = 0.3 to .5 or -.3 to -0.5, and a large effect by r = 0.5 to 1 or -0.5 to -1.38 Type of independent variable Direct effect path Standardized coefficients beta SE t P value r r 2 Potential predictor Years → professional fulfillment 0.059 0.001 5.121 0 0.104 0.011 Years → burnout -0.1 0.001 -8.61 0 0.112 0.013 Potential mediator Leader score → professional fulfillment 0.135 0.004 11.889 0 0.16 0.025 Leader score → burnout -0.059 0.003 -5.107 0 0.112 0.013
Bivariate linear regression

To evaluate the direct effects of the potential predictor and mediator variables on the outcome variables, bivariate linear regression was conducted, controlling for sex and for respondents who had recently changed either their employer or specialty (Table 5). Years of experience demonstrated statistically significant direct effects on both professional fulfillment (beta = 0.059, standard error [SE] = 0.001, P < .001) and burnout (beta = -0.1, SE = 0.001, P < .001). For every 1-year increase in experience, average PA professional fulfillment scores increased by 0.059 units and average PA burnout scores decreased by 0.1 units. Years of experience explained about 1% of the variability in both professional fulfillment (r = 0.104, r2 = 0.011, P < .001) and burnout (r = 0.112, r2 = 0.013