One Big Happy Family? Interdisciplinary Variation in Job Satisfaction among Hospice Providers

Abstract

Background:

Job satisfaction is particularly important in the hospice industry, given the emotional and interpersonal challenges that hospice staff face in providing care to patients near the end of life and their families. However, little is known about the job satisfaction of hospice providers, or about variation in satisfaction among disciplines.

Methods:

Staff at participating hospices completed the Survey of Team Attitudes and Relationships (STAR) using an online user interface. The STAR has 6 domains that comprise 45 items.

Results:

Results were submitted for 8,495 staff from 177 hospices in 41 states. The mean total score was 28 on a 0–100 scale (range, 0–100; interquartile range, 8–45) and hospice-level scores ranged from 15 to 44. Nonclinical staff (n = 3260) and clinical staff (n = 5235) had similar total scores (28 for both). Among clinical staff, in a mixed effects model adjusting for individual and hospice characteristics, physicians had the highest total scores (adjusted mean 42; 95% confidence interval: 35–46) compared to chaplains (30; 28–33), bereavement coordinators (27; 24–30), nurses' aides (29; 27–33); nurses (26; 28–33), and social workers (25; 23–26).

Conclusions:

There is significant variation in job satisfaction both among hospices and disciplines.

Introduction

The Medicare Hospice Benefit was established in 1982 to ensure that patients in the last 6 months of life have access to high-quality palliative care provided by a physician, visiting nurse, social worker, chaplain, home health aide, and volunteers.¹ In large part because of Medicare funding, the hospice industry has grown rapidly since then. In 2009, approximately 1,560,000 people used hospice, compared to 1,200,000 in 2005. The number of hospice programs has also increased, and in 2009 there were approximately 5000 hospices in the United States, compared to 4160 in 2005.¹

However, little is known about the experience of the staff who provide this care. For instance, it is not known whether hospice employees are satisfied with their jobs, or whether job satisfaction varies across organizations. Nor is it known whether some disciplines have higher rates of job satisfaction than others.

It is essential to better understand the way that hospice employees view their work and work environment because previous studies in other settings have found that job satisfaction is highly associated with employee morale, productivity, and inversely associated with burnout.^2–5 Satisfaction in other health care sectors is also inversely associated with job turnover.^6,7

Particularly in light of growing patient volumes, short lengths of stay, and increasing competition among hospices, it is important to understand the work experience of hospice staff. Hospice staff face a work environment that is intellectually and emotionally challenging, and upcoming changes to hospice regulations and reimbursement are likely to increase those challenges significantly. As the hospice industry continues to grow, and as hospices increasingly compete for trained staff, it is essential that we understand the variation in job satisfaction among hospice providers in order to develop models for the hospice work environment that will be sustainable. Therefore, the goal of this study was to measure job satisfaction in a diverse sample of hospice staff from multiple disciplines.

Methods

The Survey of Team Attitudes and Relationships (STAR) comprises 45 items in 6 domains: daily work (e.g., work hours), teamwork (e.g., interdisciplinary collaboration), management and oversight (e.g., support from management), organizational structure (e.g., organization's commitment to patient care), rewards of work (e.g., meaningfulness of work), and global job satisfaction.⁸ Each item has a 5-point Likert scale ranging from “strongly disagree” to “strongly agree.” Previous testing has demonstrated good psychometric characteristics, with acceptable floor and ceiling effects. The overall STAR has a Cronbach α of 0.93, indicating good homogeneity, and all domains have similar alphas (range, 0.74–0.84).⁸

The STAR was administered through a set of hospices that voluntarily participate in the National Hospice and Palliative Care Organization's (NHPCO's) data collection and reporting initiatives. Each hospice shared an Internet link with all staff, who were given a defined time period (typically 2–4 weeks) to complete the survey online.

Hospices were allowed to categorize their surveys by office or location and so some hospices are defined by one provider identification number, whereas others are defined by several. Thus the total number of reporting sites is larger than the number of participating hospices as they are defined by their NHPCO membership provider numbers.

Some hospices began using the STAR in 2008 but others began in 2009 or 2010. Some use it annually, while others have skipped a year. Therefore, in analyzing these results, we restricted the study sample to the most recent survey administration for each hospice, ensuring that each hospice and respondent is represented only once.

Each STAR item is scored using “top box” coding that distinguishes the best possible response versus all others, in order to make the results more meaningful to hospices than mean scores would be. For each item, we calculated the proportion of respondents who gave the best response (either “strongly disagree” or “strongly agree”) and used the mean to calculate domain scores and total scores. Finally, we multiplied each proportion by 100 to give a possible range from 0–100. Thus, each score reflects the mean percentage of respondents who gave the best possible response.

In order to define variation in job satisfaction across organizations and disciplines, we analyzed STAR scores in three ways. First, we examined associations between scores and respondent characteristics (e.g., discipline, education) using appropriate non-parametric bivariate tests (e.g., Wilcoxon rank sum test, Spearman ρ), followed by linear regression analysis. Second, we performed a similar analysis using hospice-level characteristics (e.g., profit status, average daily census). Third, we developed a mixed effects regression model that included respondent characteristics as fixed effects and hospice characteristics as random effects. Stata software (version 10; StataCorp, College Station, TX) was used for all statistical analysis.

We estimated that a sample of at least 8000 surveys would provide adequate power (1-β > 0.80) to detect even small differences in total STAR scores between clinical and nonclinical staff, assuming at least 20% were nonclinical (e.g., 1.5 points on the STAR scale of 0–100; α = 0.05). Furthermore, this sample size would also be adequate to detect similar differences (e.g., 2 points on a 0–100 scale) between clinical and nonclinical staff for each of the 6 domains (α < 0.008, adjusted for multiple comparisons).

These data were collected by NHPCO and reported to hospices as part of their operations and performance improvement efforts. Therefore, Institutional Review Board (IRB) approval was not obtained for survey administration. However, the University of Pennsylvania's IRB granted an exemption for secondary analysis of existing data.

Results

Results were submitted for 8495 staff from 177 hospices in 41 states, divided into 191 sites. Of these, 1665 surveys were submitted in 2008, 1614 in 2009, and 5216 in 2010. Sites returned a mean of 35 surveys (range, 2–389; interquartile range, 34–123). Because of changes in staffing and job turnover, it was impossible to accurately determine the total potential staff size at participating hospices at the time that each survey was administered. However, based on data obtained from each hospice during this period, we estimate that these hospices employed a total of approximately 18,198 staff who could have completed a survey (response rate = 47%). The characteristics of participating staff and organizations are summarized in Table 1.

Table 1.

Characteristics of Respondents (n = 8495)

Female	7433 (89%)
Ethnicity: Hispanic	341 (4%)
Race
White	7201 (86%)
African American	604 (7%)
Asian	177 (2%)
Pacific Islander	35 (0.5%)
Native American	76 (1%)
Other	279 (3%
Missing	172 (2%)
Discipline
Physician	65 (0.77%)
Nurse (includes advance practice nurse, LVN, LPN)	2362 (27.8%)
Chaplain	308 (3.6%)
Social Worker	727 (8.6%)
Nurses' aide	1043 (12.3%)
Bereavement counselor	194 (2.3%)
Other clinical staff^a	536 (6.3%)
Nonclinical staff	3260 (38.7%)
Education
Less than high school	51 (0.6%)
High school or GED	791 (9.5%)
Some college	1486 (17.9%)
2-year college degree (Associate)	1731 (20.8%)
4-year college degree (BA/BS)	2000 (24.1%)
Master's degree	1528 (18.4%)
Professional degree (MD/JD)	113 (1.4%)
Doctoral degree	81 (1.0%)
Other	527 (6.3%)
Years worked for the organization
<1	1262 (15%
1–2	1922 (23%)
3–5	2578 (31%)
6–10	1614 (19%)
11–20	812 (10%)
>20	145 (2%)
Hospice characteristics
For-profit: n (%)	720 (8.5%)
Rural	852 (10.3%)
Average daily Census
<50	1278 (18.2%)
51–100	1104 (15.7%)
101–200	2545 (36.2%)
>200	2093 (28.8%)

Other clinical staff either filled several clinical roles, or had largely nonclinical responsibilities.

LVN, licensed vocational nurse; LPN, licensed practical nurse.

The mean total STAR score was 28 on a 0–100 scale (range, 0–100; interquartile range, 8–45). That is, on average hospice staff gave the best possible answer for 28% of items. The score for the global domain assessing overall satisfaction was 32. Specific domain scores were lowest for satisfaction with daily work (20), organizational structure (23), and management and oversight (25). Scores were highest for satisfaction with teamwork (34) and rewards of work (35).

Staff characteristics and STAR scores

Nonclinical staff (n = 3260) and clinical staff (n = 5235) had similar total scores (28 for both), but global scores were higher for clinical staff (35 versus 29; rank sum test p < 0.001). Nonclinical staff also rated management and oversight slightly higher than clinical staff did (26 versus 24, rank sum p < 0.001). However, using a p-value adjusted for multiple comparisons (0.008), clinical staff and nonclinical staff gave similar ratings of teamwork (35 versus 33; rank sum test p = 0.01), daily work (21 versus. 18; rank sum p = 0.589), rewards of work (34 versus 35; rank sum test p = 0.027) and organization and management (22 for both; rank sum test p = 0.57).

Among clinical staff, there were significant differences in scores across disciplines for all domains and for the total score (Table 2). In particular, physicians consistently had the highest satisfaction for all domains and for the total score. Chaplains also had scores that were higher than those of many other disciplines. In contrast, nurses and social workers generally had scores that were lower than those of other disciplines (Table 2).

Table 2.

STAR Scores by Discipline

	Work rewards	Teamwork	Management and oversight	Organization	Daily work	Global	Total score
Bivariaate analysis of STAR domain scores, by discipline ^a
Physician	51	49	31	37	32	52	42
Nurse (RN, LVN, LPN)	33	35	23	20	16	33	26
Chaplain	38	39	28	26	20	37	30
Social work	34	31	21	20	18	28	25
Nurses' aide	32	33	22	23	21	32	28
Bereavement	39	32	22	23	18	33	27
Adjusted STAR scores and 95% confidence intervals from mixed effects models, adjusted for staff and hospice characteristics ^b
Physician	50 (36–39)	47 (39–55)	30 (26–35)	36 (30–42)	31 (25–37)	51 (37–41)	42 (35–46)
Nurse (RN, LVN, LPN)	33 (31–34)	35 (33–36)	23 (22–24)	20 (19–22)	16 (15–17)	33 (31–34)	26 (25–27)
Chaplain	38 (35–42)	39 (35–42)	27 (25–30)	26 (23–29)	20 (17–23)	37 (33–41)	30 (28–33)
Social work	33 (32–36)	31 (28–33)	21 (20–23)	20 (18–22)	18 (16–20)	28 (25–31)	25 (23–26)
Nurses' aide	33 (28–36)	32 (28–36)	23 (20–26)	21 (19–23)	19 (17–21)	29 (27–31)	29 (27–33)
Bereavement	40 (36–44)	32 (27–37)	22 (20–25)	23 (19–27)	18 (14–21)	33 (28–38)	27 (24–30)

Kruskal Wallis test p < 0.001 for all domains and total score.

Results adjusted for staff characteristics (education, gender, race, time at the organization) and hospice characteristics (average daily census and profit status).

STAR, Survey of Team Attitudes and Relationships; RN, registered nurse; LVN, licensed vocational nurse; LPN, licensed practical nurse.

There was no association between total STAR score and race (28 for both white and non-white; p = 0.477), ethnicity (28 for both Hispanic and non-Hispanic; p = 0.587), or education (Spearman ρ 0.013; p = 0.135). Men had slightly higher total scores than women did (30 versus 28; p = 0.028). Of those respondents (n = 5865) who reported their salaries, there was a very small association between total scores and salary (Spearman ρ 0.03; p = 0.002). There was also a very small negative association between time worked for the organization and total score (Spearman rho 0.04; p = 0.002). In a series of multivariable regression models, only discipline retained a significant association with the total STAR score (p < 0.001).

Hospice characteristics and STAR scores

In order to better understand the variation in scores across hospices, we also examined the association between scores and hospice characteristics. We restricted the sample to 187 sites (of 191) with at least 20 surveys, resulting in a sample of sites ranging in size from 21 to 414 surveys. The mean site-level total score was 28, and site means ranged from 15 to 44 (median, 27; interquartile range, 24–30).

Compared to government and not-for-profit hospices, for-profit hospices had higher total scores (33 versus 27; rank sum p < 0.001). Staff at rural sites had slightly higher scores compared to those in urban and mixed sites (28 versus 27; p < 0.001). However, there was no association between total score and hospice census (Spearman ρ 0.03; p = 0.275).

These results were used to build two mixed effects models that included both fixed effects at the individual level, and random effects at the hospice level. For both models, we used the subsample of those hospices with at least 20 completed surveys (n = 187). We included education, gender, race, and time at the organization as fixed effects, and profit status and location (rural/urban) as random effects.

In the first model, we compared clinical versus non-clinical staff. This model showed no significant differences in the total scores between clinical (27; 95% confidence interval [CI] 26–28) and nonclinical staff (28; 95% CI 27–29). Clinical staff did have significantly higher global scores (34; 33–35) compared to nonclinical staff (29; 28–30) p < 0.001) and higher teamwork (35, 34–36 versus 33, 32–34; p = 0.010). However, clinical staff had lower scores for management and oversight (24, 23–25 versus 26, 26–27; p < 0.001) and daily work (18, 17–19 versus 21, 20–22; p < 0.001). There was no difference in satisfaction with work rewards (35; 34–36 for both) and organizational structure (23; 22–24 for both).

In the second model, we examined differences among clinical disciplines using the same fixed and random effects variables described above. In this model, physicians had the highest adjusted total scores (42; 95% CI 35–46), followed by chaplains (30; 28–33), bereavement professionals (27; 24–30), nurses (26; 25–27) and social workers (25; 23–26). Analysis of discipline scores for domains, presented in Table 2, are consistent with the results of bivariate analysis, indicating the highest scores for physicians and the lowest scores for nurses and social workers.

Discussion

Despite rapid growth in the hospice industry, with concomitant increases in the hospice workforce, little is known about how satisfied hospice staff are with their jobs. This study, which is the first organized effort to describe the job satisfaction of a national sample of hospice providers, begins to shed much-needed light on the work experience of hospice providers. In particular, there are two results of this study that should be important to the growing hospice industry.

First, this study identified significant differences in job satisfaction among disciplines. In particular, we found that the physicians in this sample were more satisfied with their jobs than most other disciplines were. This seems to be true for the total STAR scores and for most domain scores. Furthermore, this difference persists after adjustment for respondent- and hospice-level characteristics.

Second, we found substantial variation in satisfaction scores among other disciplines as well. For instance, the scores of nurses and social workers were generally among the lowest measured. In particular, nurses' satisfaction with respect to daily work was among the lowest scores across all domains and disciplines.

These results are important because extensive research has found that job satisfaction is related to other workplace measures of well-being, including a sense of empowerment and burnout.^2,4,9,10 In addition, job satisfaction is also inversely associated with staff turnover.^3,5,6 Therefore, findings that describe disparities in job satisfaction should prompt consideration of ways in which hospice organizations can better support hospice staff, and particularly nurses and social workers.

Fortunately, growing experience in nursing administration in particular has identified numerous ways in which job satisfaction can be improved. Some interventions are broad in scale, and incorporate regulatory changes to nurse staffing ratios that may be difficult to generalize to the hospice industry.¹¹ But other interventions, such as a “job preview,” may help prospective staff to make more informed choices about whether hospice is the right work environment for them.¹² Furthermore, changes to the local work environment that focus on additional education, support, and flexibility have been valuable in improving satisfaction in other health care settings, and should be considered in hospice as well.^2–4,13

This study has three main limitations that should be noted. First, as with any voluntary survey, there is the possibility of non-response bias. In light of the small number of physician respondents, the results for this group may be particularly susceptible to bias. Without detailed records of every employee at every participating hospice, it is not possible to evaluate the response rate precisely. Nor is it possible to estimate the effect of non-response on the results reported here. Nevertheless, the response rate for this study is typical of other surveys of health care providers including nurses^14,15 and social workers,¹⁶ making it unlikely that this study suffers from unique problems of selection bias.

Second, although the sample size is large, this study represents the views of hospice staff from a relatively small proportion of hospices. In particular, the low proportion of for-profit hospices is not representative of the hospice population as a whole.¹ Therefore, further research is needed to determine whether the results reported here are typical of the experience of hospice staff as a whole. In particular, it will be important to better understand the range of job satisfaction results of the disciplines in this sample that were particularly high (e.g., physicians) or particularly low (e.g., nurses and social workers) to determine if these results are representative.

Third, these surveys reflect only a single year's results for each participating hospice. It is possible that the impact of staff satisfaction initiatives or secular trends could not be detected. However, these results reflect the last results from each hospice, and thus offer the most current picture of staff satisfaction at that site.

Although these results are preliminary, they offer a valuable first glimpse of the experience of health care providers who work in the hospice industry. As the hospice industry continues to grow, and as this workforce increases, it will be important to better understand both how staff experience their jobs. Most importantly, it will be essential to understand how hospices can ensure that their staff has a supportive and rewarding work environment.

Footnotes

Acknowledgments

Financial support for data collection and analysis was provided by the National Hospice and Palliative Care Organization.

Author Disclosure Statement

Three of the authors (D.C., C.S., M.H.) are employees or consultants to the National Hospice and Palliative Care Organization, which is responsible for survey data collection and reporting.

References

National Hospice and Palliative Care Organization Facts and Figures. Hospice Care in America, 2010. www.nhpco.org/files/public/Statistics_Research/Hospice_Facts_Figures_Oct-2010.pdf. 2011 June 7.

Kovner

, Brewer

, Wu

Y-W

, Cheng

, Suzuki

. Factors associated with work satisfaction of registered nurses. J Nurs Scholarsh, 2006; 38:71–79.

Ulrich

, Buerhaus

, Donelan

, Norman

, Dittus

. How RNs view the work environment: Results of a national survey of registered nurses. J Nurs Admin, 2005; 35:389–396.

Jourdain

, Eanevert

. Job demands-resources, burnout and intention to leave the nursing profession: A questionnaire survey. Int J Nurs Stud, 2010; 47:709–722.

Fitzpatrick

, Campo

, Graham

, Lavandero

. Certification, empowerment, and intent to leave current position and the profession among critical care nurses. Am J Crit Care, 2010; 19:218–226.

Ellenbecker

, Porell

, Samia

, Byleckie

, Milburn

. Predictors of home healthcare nurse retention. J Nurs Scholarsh, 2008; 40:151–160.

Estryn-Behar

, van der Heijden

, Fry

, Hasselhorn

. Longitudinal analysis of personal and work-related factors associated with turnover among nurses. Nurs Res, 2010; 59:166–177.

Qaseem

, Shea

, Connor

, Casarett

. How well are we supporting hospice staff? Initial results of the Survey of Team Attitudes and Relationships (STAR) validation study. J Pain Symptom Manage, 2007; 34:350–358.

Newton

, Waters

. Community palliative care clinical nurse specialists' descriptions of stress in their work. Int J Palliat Nurs, 2001; 7:531–540.

10.

Zeytinoglu

, Denton

, Davies

, Baumann

, Blythe

, Boos

. Retaining nurses in their employing hospitals and in the profession: Effects of job preference, unpaid overtime, importance of earnings and stress. Health Policy, 2006; 79:57–72.

11.

Spetz

. Nurse satisfaction and the implementation of minimum nurse staffing regulations. Policy Politics Nurs Pract, 2008; 9:15–21.

12.

Crow

, Hartman

, McLendon

. The realistic job preview as a partial remedy for nursing attrition and shortages: The role of nursing schools. J Cont Educ Nurs, 2009; 40:317–323.

13.

Hyrk

, Appelqvist-Schmidlechner

, Haataja

. Efficacy of clinical supervision: Influence on job satisfaction, burnout and quality of care. J Adv Nurs, 2006; 55:521–535.

14.

Sorensen

, Seebeck

, Scherb

, Specht

, Loes

. The relationship between RN job satisfaction and accountability. West J Nurs Res, 2009; 31:872–888.

15.

McDonald

, Tullai-McGuinness

, Madigan

, Shively

. Relationship between staff nurse involvement in organizational structures and perception of empowerment. Crit Care Nurs Q, 2010; 33:148–162.

16.

Evans

, Huxley

. Factors associated with the recruitment and retention of social workers in Wales: Employer and employee perspectives. Health Soc Care Commun, 2009; 17:254–266.