Predicting nurse burnout from demands and resources in three acute care hospitals under different forms of ownership: A cross-sectional questionnaire survey

1.1. Burnout 

Burnout is a form of stress reaction in terms of mental exhaustion as a consequence of mental overload. The concept was introduced in 1974 by the psychiatrist Freudenberger, who noticed that health care personnel often suffered from chronic physical fatigue, emotional exhaustion and increased distancing from their patients (Freudenberger, 1974). According to a common definition, burnout is a state of physical, emotional and mental exhaustion as a reaction to emotionally demanding working conditions over a long period of time (Schaufeli and Greenglass, 2001). Although burnout was originally regarded as consisting of three components (Maslach and Jackson, 1981), today it is increasingly seen as consisting of two core components, exhaustion and cynicism (Bakker et al., 2004, Maslach et al., 2001). More recent studies have also shown that the symptoms are generally applicable to other occupational groups besides health care employees (see e.g. Demerouti et al., 2001). As opposed to other health-related problems, such as depression, burnout has turned out to be linked to one’s work situation since the explanations to the syndrome can be found in work relationships rather than internal predispositions (Bakker et al., 2000).

Burnout has several negative consequences for the individual as well as the organization, and in the long run also for society. Besides the personal suffering burnout inflicts in the form of decreased cognitive ability (Bakker et al., 2004), it is linked to diminished job satisfaction, impaired organizational commitment and an increased turnover intention (Lee and Ashforth, 1996). Nurses, whose work tasks include a great deal of stress and emotionally demanding contact with patients, have proven to be an occupational group especially prone to burnout (Demerouti et al., 2000). This will in turn affect their relationship with patients, in the form of less contact and a higher risk of incorrect medical treatment (West et al., 2006).

Despite certain potential negative consequences for the work environment that accompany privatization, a meta-analysis of data at a corporate level from a number of countries indicates performance improvements through more efficient resource use and improved incentive structures within organizations that have been privatized (D’Souza and Megginson, 1999). The same meta-analysis also showed that organizations in industrialized countries that go through privatization normally reduce their number of employees. This could lead to an increased stress level among the staff, which can eventually lead to burnout. It has been shown that burnout among personnel occurs more frequently in connection with restructuring and cutbacks (Burke and Greenglass, 2001, Greenglass and Burke, 2000, Greenglass et al., 2003). There are also indications in Sweden that levels of burnout (exhaustion and cynicism) are higher in private hospitals than in publicly administered ones (Hallberg and Sverke, 2004). Although very little research has been conducted on the differences in burnout levels between the various ownership types, existing studies point to an increased health risk for employees of private for-profit and private non-profit hospitals. Against this background, we formulate the following hypothesis:Hypothesis 1.

Burnout levels in personnel are higher the more commercial the hospital ownership type is.

1.2. Job demands and resources in health care 

The concept of psychosocial work environment refers to the interplay between an employee and the surrounding environment, and has an effect on how the employee feels. Many theories attempt to describe the content of the psychosocial work environment, often emphasizing that the work environment is central to the individual’s motivation, well-being and behavior at work (Hackman and Oldham, 1976, James and Sells, 1981). For the past two decades, Karasek and Theorell’s (1990) Demand–Control Model has been dominant, citing workload and time pressure as the greatest work-related stressors, and control as a protective resource. The model is limited, however, as it begins by specifying what is considered the greatest stressor and the most important defense against stress. The Job Demands–Resources Model is a newer version of Karasek and Theorell’s model, taking it a step further, which has received increased attention of late and has been used a great deal within health care research (Demerouti et al., 2001). Like its predecessor, this model states that stress occurs when there is an imbalance between high job demands and insufficient job resources, but what constitutes the most important job demands and resources depends on the type of work and task, which makes this model more flexible than the former one. According to this theory, demands and resources trigger various processes in an individual. While high job demands over time deplete an employee’s energy, sufficient resources at work lead to a motivation-increasing process (Schaufeli and Bakker, 2004).

Job demands refer to physical, psychological, social or organizational aspects of the work to be performed, which demand physical or mental effort over time (Demerouti et al., 2001). Although this does not automatically imply something negative for an employee, it can often lead to negative reactions like depression, worry or burnout (Schaufeli and Bakker, 2004). Examples of demands at work are workload, role conflict and job insecurity. In an OECD (2005) report, nurses are mentioned as a risk group for which job demands have become greater, which is assumed to affect health negatively.

Resources, on the other hand, refer to the health-protecting factors that have a positive effect on health. Job resources can thus be defined as physical, psychological, social or organizational aspects of work that are functional in achieving work goals, reduce the effect of job demands and their accompanying physical or psychological strain, and stimulate personal development (Demerouti et al., 2001). Common examples of resources at work are job autonomy, goal clarity, work group support, supervisor support, job challenge and feedback. Nurses form a vulnerable group, which despite high job demands and great responsibility has relatively limited power and authority (Bakker et al., 2005), which can indicate a lack of resources.

Although there is empirical support for the assertion that job demands and control over work affect the stress level as well as physical and mental health of health care personnel (Michie et al., 2004, Williams et al., 1998), little knowledge exists about the significance of ownership type for job demands and resources. A Canadian study, however, found that employees within for-profit health care institutions experienced a significantly higher stress level and workload than did employees within non-profit ones (Miles-Tapping, 1992). D’Souza and Megginson’s (1999) meta-analysis showed that productivity seemed to increase with privatization, which implies that job demands should also increase.

Meanwhile, the private for-profit and non-profit types also allow for extended opportunities. Compared to publicly administered hospitals, private hospitals present different possibilities for independent prioritization and formulation of goals, which can reduce the uncertainty and lack of clarity that otherwise characterize health care (Öhrming and Sverke, 2003). Previous research on hospitals as well as institutions within other sectors suggests that private organizations have a more defined leadership and a more genuine commitment to employees and their development (Cunha, 2000, Cunha and Cooper, 2002, Hellgren et al., 2005). Taken together, these differences may create prerequisites for better organizational and social job resources for both the private for-profit and the private non-profit hospital. These arguments led us to propose:Hypothesis 2a.

Levels of job demands are higher the more commercial the hospital is.

Levels of job resources are higher the more commercial the hospital is.

1.3. The significance of the psychosocial work environment for burnout 

Previous research has shown that job demands as well as resources seem to be significant for burnout in employees. The model constructed by Demerouti et al. (2001) is based on the assumption that an individual being subjected to growing demands at work along with a lack of resources will experience stress which, if it is long-lasting, can develop into burnout. More specifically, the model assumes that high job demands have a stronger relation with emotional exhaustion while lack of job resources is more related to cynicism (Demerouti et al., 2001). According to the Job Demands–Resources Model, burnout can be described as too-high demands over a longer period of time creating the depletion of an employee’s energy reserve, which ultimately leads to exhaustion. If the job resources are too limited, this leads to a decrease in motivation which makes it more difficult to handle job demands efficiently, with mental withdrawal and indifference as result (Schaufeli and Bakker, 2004).

This model has received empirical support from a number of studies (e.g., Bakker et al., 2004, Demerouti et al., 2001). In a cross-sectional study of nurses, it was shown that workload and resources in the work environment were strongly related to health and well-being, in that subjects’ health declined with high job demands and improved with sufficient job resources (Shamian et al., 2002). Job demands have been shown to play a more important role than job resources in the development of burnout (Lee and Ashforth, 1996, Schaufeli and Bakker, 2004); thus, the following hypotheses can be formulated:Hypothesis 3a.

High job demands are linked to high levels of burnout.

The presence of job resources is linked to low levels of burnout.

The Job Demands–Resources Model describes universal processes involved in the development of burnout. Although different ownership types within health care can be expected to be associated with different levels of demands and resources, there is a lack of research investigating whether the significance of demands and resources varies depending on type of ownership. Thus, the existing knowledge in this area does not allow for specific hypotheses on differences between private for-profit, private non-profit and public health care as regards possible effects of demands and resources on burnout. Rather, we examine whether the significance of demands and resources for burnout differs between these various ownership types.

2.1. Samples and setting 

The study is based on cross-sectional data collected in 2001/2002 in Sweden, a country with highly developed work procedures and favorable collective bargaining agreements. A questionnaire was sent to all registered nurses at three acute care hospitals in the Stockholm region, each with its own ownership type: private for-profit, private non-profit and publicly administered, respectively. The hospitals in the present study have been chosen because of their many similarities. First of all, they all are situated in the same region. Secondly, while certain differences in medical specialties exist, the hospitals specialize in the same areas (internal medicine, surgery and orthopedics). Despite these similarities, there are also some differences among the hospitals. For instance, as can be seen from the aggregated productivity data reported in Table 1, the two privatized hospitals, and particularly the for-profit one, are characterized by a somewhat higher productivity and work intensity as compared to the public administration unit, although these statistics do not take into consideration the possibility for so-called case-mix (i.e., the diagnoses that the patients are being treated for).

Table 1. Ratio comparison between the hospital types regarding productivity and workload (hospital total data from 2002, based only on inpatient, somatic care)

		Private for-profit	Private non-profit	Publicly administered
	Bed productivity (number of patients/number of beds)	82.74	72.90	71.40
	Average treatment time (treatment days/number of patients)	3.79	3.97	4.17
	Utilization of inpatient care (number of days of inpatient care/[number of beds×365])	0.86	0.79	0.82
	Staff per patient ([number of completed full-time job equivalents/number of patients]×1000)	43.78	56.30	75.62
	Staff turnover (number of resignations/number of employees)	0.29	0.21	0.19

2.2. Questionnaire 

The variables included in the study are presented in Table 2. The choice of demands (workload, role conflict and job insecurity) and resources (job autonomy, goal clarity, work group support, supervisor support, job challenge and feedback) follows as closely as possible the results of Lee and Ashforth’s (1996) meta-analysis in which demands and resources was found to be most strongly related to the burnout dimensions emotional exhaustion and cynicism. All variables were measured using established scales, each containing multiple items. The response alternatives ranged from 1 (strongly disagree) to 5 (strongly agree). In the study, we have also taken the background variables organizational tenure and gender into consideration. Table 3 presents descriptive statistics (means and S.D.), reliability estimates (Cronbach’s alpha) and the correlations between all included variables.

Table 2. List of variables used in the study

	Variables	Number of items	Example of item	References
	Job demands
	Workload	3	I often have too much to do in my job	Beehr et al. (1976)
	Role conflict	5	I have to do things that should be done differently	Translated and slightly modified version of Rizzo et al. (1970)
	Job insecurity	10	You may be laid off permanently	Ashford et al. (1989)
	Job resources
	Job autonomy	4	I can make my own decisions on how to organize my work	Sverke and Sjöberg (1994), based on Hackman and Oldham (1975) and Walsh et al. (1980)
	Goal clarity	4	I know exactly what is expected of me	Combination of items from Rizzo et al. (1970) and Caplan (1971)
	Work group support	4	In my work group we help and support each other in the job	Combinations of items from Nystedt (1992) and a slightly modified version of Taylor and Bowers (1972)
	Supervisor support	3	My boss is considerate	Ekvall and Arvonen (1994)
	Job challenge	3	I’m learning new things all the time in my job	Hellgren et al. (1997)
	Feedback	3	I usually know whether or not my work is satisfactory on this job	Hackman and Oldham (1975)
	Burnout
	Emotional exhaustion	9	I feel used up at the end of the workday	Maslach et al. (1996)
	Cynicism	5	I worry that this job is hardening me emotionally	Maslach et al. (1996)

Table 3. Correlations, descriptive statistics and reliability estimates (Cronbach’s alpha; n=1102 for the total sample)

	Variables	1	2	3	4	5	6	7	8	9	10	11	12	13	M	S.D.	Alpha
	Demographic variables
	1. Organizational tenure	–													9.16	9.38	–
	2. Gender (women)	.08	–												0.95	0.22	–
	Job demands
	3. Workload	−.02	.01	–											3.66	0.90	0.76
	4. Role conflict	−.10	−.07	.42	–										2.21	0.78	0.71
	5. Job insecurity	.12	−.01	.14	.35	–									1.60	0.57	0.74
	Job resources
	6. Job autonomy	.15	.05	−.31	−.39	−.12	–								3.32	0.75	0.72
	7. Goal clarity	.11	.06	−.23	−.53	−.22	.39	–							4.11	0.71	0.73
	8. Work group support	.08	.07	−.19	−.40	−.15	.48	.42	–						3.64	0.75	0.80
	9. Supervisor support	.03	.07	−.14	−.31	−.14	.31	.29	.37	–					3.35	0.71	0.85
	10. Job challenge	.06	.07	.14	−.07	−.02	.30	.17	.23	.12	–				3.97	0.74	0.64
	11. Feedback	.11	.04	−.24	−.43	−.19	.49	.54	.44	.46	.19	–			3.21	1.00	0.79
	Burnout
	12. Emotional exhaustion	−.06	−.01	.48	.46	.21	−.38	−.35	−.32	−.19	−.06	−.34	–		2.23	0.73	0.84
	13. Cynicism	−.13	−.13	.18	.35	.13	−.17	−.22	−.15	−.13	−.16	−.17	.46	–	1.64	0.69	0.74

–, not applicable. The scales range from 1 to 5 for all variables with the exception of tenure (in years) and gender (0=man, 1=woman). For r>.04, p<.05.

3.1. Differences in levels of burnout 

To examine whether there were differences in burnout between the three types of ownership, a multivariate analysis of covariance (MANCOVA) was performed. In order not to overestimate the effect of hospital type, we included two covariates (organizational tenure and gender). Table 4 shows the mean values of emotional exhaustion and cynicism, univariate F-tests and contrast tests between the hospitals. There was a significant multivariate difference in burnout between the different hospitals (F4, 2194=3.89, p=.004). However, while the follow-up univariate test for emotional exhaustion did not indicate any differences between the hospitals, the contrast tests showed that emotional exhaustion was significantly higher at the private for-profit hospital as well as the non-profit hospital compared to the publicly administered one. For cynicism, the univariate test indicated a significant difference, where the contrast tests showed that the private for-profit hospital had higher levels of cynicism than both the publicly administered and private non-profit hospitals. In accordance with Hypothesis 1, these results thus suggest that burnout was highest at the private for-profit hospital and lowest at the publicly administered one. However, it was not possible to statistically determine whether the burnout levels at the private non-profit hospital were significantly different from those of the private for-profit or the publicly administered ones.

Table 4. Burnout: means, univariate tests, contrast tests and covariate F-tests between the hospital types

		Private for-profit (1)	Private non-profit (2)	Publicly administered (3)	Univariate F (d.f.=2, 1102)	Contrast test	Covariates F-tests (d.f.=1, 1097)
							Tenure	Gender
	Emotional exhaustion	2.28	2.24	2.13	2.74	3<1, 2	2.96	0.00
	Cynicism	1.79	1.60	1.58	5.29**	1>2, 3	13.05**	13.54**

**p<.01.

3.2. Differences in psychosocial work environment 

To examine whether there were differences in work environment between the three ownership types, two additional MANCOVAs were performed, one for job demands and one for job resources. Table 5 shows the mean values of demands and resources for each hospital as well as the follow-up univariate F-tests and contrast tests.

Table 5. Psychosocial work environment: means, univariate tests, contrast tests and covariate F-tests between the hospital types

		Private for-profit (1)	Private non-profit (2)	Publicly administered (3)	Univariate F (d.f.=2, 1097)	Contrast testsa	Covariates F-tests (d.f.=1, 1097)
							Tenure	Gender
	Job demands
	Workload	3.68	3.78	3.37	18.20**	3<1, 2	0.11	0.23
	Role conflict	2.23	2.22	2.18	0.18		9.84**	4.75*
	Job insecurity	1.49	1.62	1.67	5.03**	1<2, 3	12.17**	0.78
	Job resources
	Job autonomy	3.24	3.29	3.47	4.56**	3>1, 2	20.18**	1.54
	Goal clarity	4.12	4.12	4.09	0.80		12.90**	3.48
	Work group support	3.62	3.65	3.64	0.12		6.07**	4.61*
	Supervisor support	3.38	3.40	3.23	6.20**	3<1, 2	1.72	5.08*
	Job challenge	3.85	4.00	4.03	3.40*	1<2, 3	1.62	3.21
	Feedback	3.17	3.24	3.18	0.47		13.06**	0.93

*p<.05; **p<.01.

a Contrast test for differences between hospitals (p<.05).

As regards job demands, a significant multivariate difference was detected between the three hospitals (F6, 2192=8.74, p<.001). The follow-up tests showed that workload was significantly lower at the publicly administered hospital compared to the private for-profit and the private non-profit ones, while the degree of job insecurity was lower at the private for-profit hospital than at the two others. Taken together, this finding provides only partial support for Hypothesis 2a, which predicted that demands would be highest at the private for-profit hospital and lowest at the publicly administered one.

There was also a multivariate difference for job resources between the different ownership types (F12, 2186=3.44, p<.001), but the follow-up tests were contradictory in this case as well. The hospitals did not differ regarding goal clarity, work group support or feedback, but the nurses at the publicly administered hospital reported more job autonomy but less supervisor support than did their colleagues at the two other hospitals. Job challenge was significantly lower at the private for-profit hospital than at the other two. Taken together, the results do not offer unambiguous support for Hypothesis 2b, which stated that resources should be better at a private hospital and worse at a publicly administered one.

3.3. Effects of job demands and resources on burnout 

Multiple regression analysis was used to study the degree to which the two dimensions of burnout (emotional exhaustion and cynicism) can be explained by nurses’ demands and resources. To examine the differences in the effects of job demands and resources between the three hospitals, separate analyses were performed for each hospital. The variables were arranged in three hierarchical steps. In a first step, we controlled for the background variables organizational tenure and gender. The second step contained the job demands (workload, role conflict and job insecurity), while the final step consisted of the job resources (job autonomy, goal clarity, work group support, supervisor support, job challenge and feedback). We also tested whether the effects of demands and resources differed between the hospitals (χ2-difference tests from multiple group analyses in Lisrel 8; Jöreskog and Sörbom, 2001). The results of these analyses are shown in Table 6.

Table 6. Results from hierarchical multiple regressions predicting emotional exhaustion and cynicism (standardized regression coefficients from the last step in the regression)

	Predictors	Emotional exhaustion				Cynicism
		Private for-profit	Private non-profit	Publicly administered	Differencea	Private for-profit	Private non-profit	Publicly administered	Differencea
	Step 1: Demographic variables
	Organizational tenure	−.03	−.05	.11*	*	−.05	−.14***	−.02
	Gender (women)	−.00	−.00	.14**	*	−.06	−.09*	−.07
	ΔR2	.01	.01*	.02*		.02	.05***	.01
	Step 2: Job demands
	Workload	.29***	.36***	.32***		.08	.13**	.01
	Role conflict	.18**	.15***	.23***		.42***	.18***	.31***	*
	Job insecurity	.07	.07*	.03		.08	.03	.04
	ΔR2	.30***	.34***	.27***		.21***	.09***	.07***
	Step 3: Job resources
	Job autonomy	−.10	−.13**	−.09		.03	−.01	.09
	Goal clarity	−.05	−.13**	.06		.02	−.13**	.08	*
	Work group support	−.10	−.06	−.03		.11	.01	.00
	Supervisor support	.03	−.02	.08		−.02	−.04	.01
	Job challenge	−.05	.00	−.04		−.15**	−.17***	−.11
	Feedback	−.06	−.01	−.12		−.02	.10	−.02
	ΔR2	.04**	.05***	.03		.03	.04***	.02
	Model R2	.35***	.40***	.32***		.25***	.18***	.09**

*p<.05; **p<.01; ***p<.001.

a Test of differences in regression coefficient between hospital types (χ2-test with two degrees of freedom).

The first analysis concerns emotional exhaustion. The background factors explained a marginal share of the variance at the publicly administered (2%) and the private non-profit hospital (1%), whereas they were unrelated to exhaustion at the private for-profit hospital. The second step, containing job demand factors, helped explain an additional 27–34%. Workload and role conflict demonstrated relatively strong positive associations with emotional exhaustion at all three hospitals. At the private non-profit hospital, job insecurity was positively related to the criterion, indicating that the more job insecurity the nurses felt, the higher levels of emotional exhaustion they reported. The final step, dealing with job resources, showed no association with exhaustion at the publicly administered hospital, but explained a further 4% and 5% of the variance in emotional exhaustion at the private for-profit and private non-profit hospital, respectively. However, only two resources (job autonomy and goal clarity) showed a significant negative association for nurses at the private non-profit hospital indicating that emotional exhaustion increased when nurses experienced a lower degree of job autonomy and a lower degree of goal clarity. The follow-up χ2-tests showed a significant difference between the hospitals only regarding the magnitude of the effects of organizational tenure and gender. Women and those with longer organizational tenure reported more exhaustion at the publicly administered hospital. In total, the model explained 32–40% of the variance in emotional exhaustion.

As regards cynicism, the background factors explained 5% of the variance at the private non-profit hospital, but were unrelated to cynicism at the other hospitals. The second step, job demands, helped explain 7–21% in addition to the first step. Role conflict had relatively strong positive associations with cynicism at all three hospitals. Additionally, workload related positively to cynicism at the private for-profit hospital, which indicates that cynicism levels were higher among nurses who experienced a higher degree of workload. The final step, concerning job resources, did not add to the explained variance in cynicism at either the private for-profit or the publicly administered hospital, but did explain a further 4% at the private non-profit one. However, only two resources (job challenge and goal clarity) showed significant negative effects in nurses at this hospital. This means that the nurses’ cynicism increased when they experienced a lower degree of job challenge and a lower degree of goal clarity. Finally, job challenge also related negatively to cynicism at the private for-profit hospital, indicating that nurses who did not experience their job as challenging reported higher levels in cynicism. In total, the model explained 12–29% of the variance in cynicism.

Taken together, the results provide partial support for Hypothesis 3a that high job demands are related to burnout, though not for all variables. Hypothesis 3b, that job resources are linked to lower burnout, also received partial support. However, this did not apply to all variables either.

As regards the explorative aim concerning possible patterns in the predictors that can be associated with type of ownership, the results did not generally show any differences in the significance of job demands and resources for burnout, with two exceptions. Among the demands, role conflict had a greater significance for high levels of cynicism at the private for-profit hospital than at the others. Among the resources, clear goals had a greater significance for cynicism at the private non-profit hospital than at the others.

4.1. Methodological considerations 

A limitation of this study is its cross-sectional character, which makes it impossible to make statements about direction of causality. In fact, the causal direction could be the reverse, in that a burned-out employee may experience higher job demands and fewer resources (cf. Zapf et al., 1996). Previous research has also shown that nurses with burned-out colleagues experience more burnout themselves (Bakker et al., 2005). To partially cope with this problem in future studies, a longitudinal design, with several measurements over time, is a way to study the direction of the relation. Another limitation is the difficulty establishing that the differences in burnout, demands and resources that did exist were due specifically to type of ownership. Regarding the regression analysis, we have controlled for organizational tenure and gender; however, there are naturally other factors and differences between hospitals that we have not controlled for and that could have contributed to the results. For example, even though the hospitals were recruited from the Stockholm region, it is possible that the our findings are affected by differences in geographical areas and medical specialties which makes it necessary for additional studies comparing ownership types to control for such possible differences. In this study, we have chosen to focus only on registered nurses, as previous research has shown that this group may be the most vulnerable during an ownership change (Falkenberg et al., in press). However, the tests should be replicated in other groups of health care personnel, and the association between job demands, job resources and burnout should be studied within other business sectors that have experienced ownership change.