Being satisfied at work does affect burnout among psychiatrists: A national follow-up study from New Zealand

Abstract

Background:

Burnout and job satisfaction in psychiatrists has been an area of considerable interest. Longitudinal studies on the subject are lacking, rendering it difficult to establish whether burnout changes with time or whether low job satisfaction may predict high burnout with time in psychiatrists.

Aims:

This longitudinal study of burnout and job satisfaction in a cohort of New Zealand psychiatrists was conducted to examine if initial scores on the Job Diagnostic Survey (JDS) predicted scores on the Maslach Burnout Inventory (MBI) three years later and vice versa.

Methods:

Three questionnaires (a socio-demographic questionnaire, the JDS and the MBI) were sent to all registered psychiatrists in 2008, which included all those who had participated in a study three years earlier. Scores on these three questionnaires were compared for those who had participated in both studies.

Results:

The overall number of psychiatrists reporting a high level of emotional exhaustion (EE) did not change across the two phases. The number of psychiatrists reporting high levels of depersonalization (DP) increased from 31 (13%) to 45 (20.2%); the mean DP score for the cohort significantly increased by 17.5% (p < .01). Those reporting reduced personal accomplishment (PA) increased from 90 (37.7%) to 98 (43.9%); the mean PA score for the cohort significantly reduced by 14.5% (p < .001). Low scores on skill variety, task Identity, and feedback of the JDS were significantly correlated with high EE scores three years later, whereas low scores on skill variety were significantly correlated with high scores on DP, and low scores on task significance and feedback were correlated with low scores on PA three years later.

Conclusions:

Paying attention to aspects of job satisfaction may assist us in developing specific interventions for psychiatrists who may score high on different dimensions of burnout.

Keywords

Burnout job satisfaction mental health service delivery work engagement

Introduction

Burnout among psychiatrists is an area of concern and is of increasing research interest (Firth-Cozens, 2007; Kumar, Hatcher, Dutu, Fischer & Ma’u, 2011; Snibbe, Radcliffe, Weisberger, Richards & Kelly, 1989; Thomsen, Dallender, Soares, Nolan & Arnetz, 1999). A term in common usage from 1970s, burnout is recognized to be a response to chronic emotional and interpersonal stressors at work. One of the more widely accepted definitions of burnout conceptualizes it as consisting of three dimensions: emotional exhaustion (EE); depersonalization (DP); and a decreased sense of personal accomplishment (PA) (Maslach, Schaufeli & Leiter, 2001). Burnout is reported to adversely affect work environment, retention of skilled staff (Thommasen, Lavanchy, Connelly, Berkowitz & Grzybowski, 2001), job satisfaction and involvement with the organization (Melchior, Bours, Schmitz & Wittich, 1997) to the extent that it may render the delivery of quality community mental health care difficult to sustain (Wykes, Stevens & Everitt, 1997).

We have previously reported findings on the prevalence of burnout and job satisfaction in New Zealand psychiatrists (Kumar, Fischer, Robinson, Hatcher & Bhagat, 2007) and the factors that may cause or protect against burnout in the same cohort (Kumar et al., 2011). Among other findings these studies reported a rather unusual relationship between job satisfaction and burnout in psychiatrists: despite experiencing high levels of EE, New Zealand psychiatrists reported high levels of job satisfaction (Kumar et al., 2007) similar to their counterparts in the UK (Prosser et al., 1996), the USA (Vaccaro & Clark, 1987) and Australia (Rey, Walter & Giuffrida, 2004) and in contrast to others (Bressi et al., 2009). Due to the cross-sectional nature of these studies it is difficult to conclude from them whether burnout in psychiatrists is a state that changes with time, or a trait that stays stable in the absence of any intervention. It also remains unclear whether the high levels of job satisfaction in psychiatrists who scored high on burnout measures would change with time. It is important to know whether those psychiatrists who are not satisfied at work at some point in their career would be at a greater risk of developing burnout in the future. There are no longitudinal studies that have examined the relationship between job satisfaction and burnout in psychiatrists.

Understanding the longitudinal nature of burnout and the relationship between job satisfaction and burnout is important for a variety of reasons. The temporal impact of different dimensions of burnout on productivity also differs. EE is considered to be the most important predictor of absenteeism, high levels of DP predict poor retention and low PA predicts poor client satisfaction (Schaufeli & Enzmann, 1998). Similarly each of the three dimensions of burnout is influenced by different aspects of job satisfaction and work environment over time, which may not be elucidated by cross-sectional studies. EE is often considered to arise from work stress; DP and PA are considered to arise from reduced autonomy, social support and feedback at work (Bruce, Conaglen & Conaglen, 2005; Demerouti, Bakker, Nachreiner & Schaufeli, 2001). Longitudinal studies of burnout and job satisfaction, preferably with the same cohort of psychiatrists, may inform our understanding of these complex relationships. This study was the concluding phase of a longitudinal study of burnout and job satisfaction in New Zealand psychiatrists wherein data was collected in 2005 and 2008 with the following aims and objectives:

To compare and contrast the socio-demographic variables, the Maslach Burnout Inventory (MBI; Maslach, Jackson & Leiter, 1996) score and the Job Diagnostic Survey (JDS; Hackman & Oldman, 1975) score in New Zealand psychiatrists between 2005 and 2008.

To examine if psychiatrists’ scores on the JDS predicted the score on MBI three years later.

To examine if psychiatrists’ scores on the MBI predicted the score on JDS three years later.

Study hypotheses

The mean scores on the three sub-scales of the MBI (EE, DP and PA) would not have changed from 2005 to 2008 for the cohort of psychiatrists.

The mean scores on the five core characteristic dimensions of the JDS (feedback, task significance, task variety, task identity, autonomy) would not have changed from 2005 to 2008 for the cohort of psychiatrists.

Low scores on the five core dimensions of the JDS (feedback, task significance, task variety, task identity, autonomy) in 2005 would predict a high score on the dimensions of EE and DP and low scores on the PA of the MBI in 2008 for the cohort of psychiatrists.

High scores on two dimensions of the MBI (EE and DP) and low scores on one dimension of the MBI (PA) in 2005 would predict low scores on the five core dimensions of the JDS in 2008 for the cohort of psychiatrists.

Method

Sample recruitment

Methodology and data collection for the 2005 study have been described previously (Kumar et al., 2007). The process of subject recruitment for this phase of the study is depicted in Figure 1. For this phase, a list of vocationally registered psychiatrists was obtained from the Medical Council of New Zealand database and combined with additional psychiatrists or Medical Officers of Speciality Scale working in psychiatry as identified by the Directors of Area Mental Health Service. All were required to be engaged in clinical practice in New Zealand at the time of the study. A cover letter and questionnaires containing a socio-demographic questionnaire (SDQ; available on request), the JDS (Hackman & Oldman, 1975) and the MBI (Maslach, Jackson & Leiter, 1996) were mailed to all eligible participants in November 2008. Non-respondents were sent a reminder one month later. Each psychiatrist was given the same unique identifier number as in the 2005 study. Scores on these three questionnaires were compared for those who responded to the questionnaires mailed in both 2005 and 2008. This subset of respondents compiled from both databases formed the study cohort. Responses were pooled and anonymized in accordance with the assurance given in the cover letter.

Ethics approval for the study was granted by the multi-region ethics committee of New Zealand (BOP/03/09/047).

Figure 1.

Sample selection

Statistics

The means for each dimension of the study tools (three for MBI and five for JDS) and the number of respondents scoring low, medium and high on the dimensions were calculated and compared for Phases 1 (2005) and 3 (2008) of the study. Differences between mean scores on a dimension for the two periods informed whether there was an overall improvement or decline on that dimension during the three-year study period for the cohort. The difference between the numbers of respondents scoring low, medium and high in 2005 and in 2008 informed about the changes in the size of the problem during the three-year study period.

The cohort means of each dimension of the MBI of the cohort in 2005 and 2008 were compared using Student’s t-test. Spearman’s correlation coefficient was calculated for estimating the temporal relationship between the scores on the dimensions of the JDS in 2005 and the scores on the dimensions of the MBI in 2008, and vice versa. χ² tests were performed between the scores on the five dimensions of the JDS in 2005 and the three dimensions of the MBI in 2008.

Results

Of the 485 questionnaires mailed to all registered psychiatrists in New Zealand in 2008, 262 responses were received, giving a response rate of 54%. Of these, 49 further surveys were returned but had incomplete data and thus were excluded from the analysis, giving a final response rate of 48% for Phase 3. Of the 233 respondents, Phase 1 responses on the three study questionnaires (SDQ, MBI and JDS) could be traced for 127 psychiatrists. These psychiatrists became our study cohort for the longitudinal analysis because we were able to trace data on study parameters from both phases.

Responses on study questionnaires for all the respondents in the 2008 (N = 233) and 2005 (N = 239) databases were compared to ascertain whether there were key differences between the two cohorts of psychiatrists.

Comparison of the socio-demographic variables between Phases 1 and 3

These comparisons are presented in Table 1. No statistically significant differences were found between these two groups (p > .05).

Table 1.

Comparison of socio-demographic variables of New Zealand psychiatrists who responded to a postal survey in 2005 and 2008.

Variables	2005 (N = 239)n (%)	2008 (N = 223)n (%)
Gender
Female	89 (37.23)	86 (38.6)
Male	149 (62.34)	137 (61.4)
	1 (0.41)
Ethnicity
Asian	21 (9)	18 (8.1)
European	211 (88)	183 (82.1)
Other	4 (2)	12 (5.4)
Not Specified	3 (1)	3 (1.3)
Age range
≤ 40	42 (10.04)	26 (11.7)
41–50	113 (47.2)	93 (41.7)
51–60	54 (22.5)	69 (30.9)
61–70	21 (8.7)	26 (11.7)
≥ 71	7 (2.9)	7 (3.1)
Not specified	2 (0.8)	2 (0.9)
Trained in metropolitan area
Yes	221 (92.4)	201 (90.1)
No	7 (3)	8 (3.6)
Not specified	11 (4.6)	14 (6.3)
Qualifications
FRANZCP (alone or with other qualifications)	107 (44.7)	136 (61)
MRCPsych	36 (15)	35 (15.7)
Other	95 (39.7)	51 (22.9)
Not specified	1 (0.41)	1 (0.4)
Employment status		8 (3.6)
MOSS
Consultant		201 (90.1)
Not specified		14 (6.3)
Family status
Long-term relationship, non-dependent children	53 (22.1)	50 (22.4)
Long-term relationship, dependent children	150 (62.7)	143 (64.1)
Single, dependent children	14 (5.8)	16 (7.2)
Single, non-dependent children	19 (7.9)	13 (5.8)
Other-blended family	3 (1)	1 (0.4)
Domicile
Urban	216 (90.3)	203 (91)
Non-urban	12 (5)	12 (5.3)
Not specified	11 (4.6)	8 (3.6)
Years of practice
≤ 10	42 (18)	97 (43.5)
11–20	24 (10.8)	73 (32.7)
21–30	102 (43)	19 (8.5)
31–40	66 (28)	8 (3.6)
≥ 40	22 (9)	2 (0.9)
Not specified	4 (2)	3 (1.3)

Comparison of the three sub-scales of MBI between Phases 1 and 3

The number of psychiatrists who scored low, average or high on the three dimensions of the MBI (EE, DP and PA) in 2005 and 2008 are given in Table 2. The number of psychiatrists reporting a high level of EE in 2005 and 2008 was unchanged, but those who reported a moderate level of EE decreased. However, the number of psychiatrists reporting high DP scores increased from 31 (13%) in 2005 to 45 (20.2%) in 2008, whereas those reporting low DP decreased by 12.7%. The number of psychiatrists reporting low PA increased from 90 (37.7%) to 98 (43.9%), whereas those reporting high PA decreased by 11.7%. The magnitude of change was not evenly distributed and differed between the high, medium and low groups for each of the three dimensions.

Table 2.

Number of psychiatrists scoring low, medium or high on the three dimensions of the Maslach Burnout Inventory (MBI).

Scores on MBI sub-scales	Number of respondents in 2005(N = 239)	Number of respondents in 2008(N = 223)	% change
Low emotional exhaustion (EE) (score range 0–16)	80 (33.5)	76 (34.1)	0.6
Moderate EE (score range 17–26)	80 (33.5)	64 (28.7)	−4.8
High EE (score range ≥ 27)	79 (33)	79 (35.4)	2.4
Low depersonalization (DP) (score range 0–6)	145 (60.7)	107 (48)	−12.7
Moderate DP (score range 7–12)	63 (26.3)	68 (30.5)	4.2
High DP (score range ≥ 13)	31 (13.0)	45 (20.2)	7.2
Low personal accomplishment (PA) (score range ≥ 39)	90 (37.7)	98 (43.9)	6.2
Moderate PA (score range 32–38)	92 (38.5)	93 (41.7)	3.2
High PA (score range 0–31)	57 (23.8)	27 (12.1)	−11.7

Comparison of the five core dimensions of JDS in 2005 and 2008

The numbers of psychiatrists scoring high, medium or low on the five dimensions of the JDS were compared to the responses in 2005 and 2008 (Table 3). The results suggested that scores on the dimensions of skill variety, task identification, task significance, autonomy and feedback dimensions had declined from 2005 to 2008.

Table 3.

Core dimensions of the Job Diagnostic Survey (JDS) in 2005 and 2008

Core dimensions of the JDS	Number of respondents in 2005(N = 239)	Number of respondents in 2008(N = 223)	% change
Skill variety low	39 (16.3)	55 (24.7)	8.4
Skill variety high	200 (83.7)	167 (74.9)	−8.8
Task identification low	99 (41.4)	95 (42.6)	1.2
Task identificationhigh	140 (58.6)	127 (57)	−1.6
Task significance low	57 (23.9)	73 (32.7)	8.8
Task significance high	182 (76.2)	145 (65.0)	−11.2
Autonomy low	81 (33.9)	95 (42.6)	8.7
Autonomy high	158 (66.1)	126 (56.5)	−9.6
Feedback low	123 (51.5)	151 (67.7)	16.2
Feedback high	116 (48.5)	69 (30.9)	−17.6

Comparison of psychiatrists between 2005 and 2008

The burnout scores in 2008 differed if a psychiatrist belonged to high, medium or low group in 2005. To explain these changes subsequent analysis was conducted on psychiatrists for whom socio-demographic details, and scores on the MBI and the JDS were available for both 2005 and 2008 (N = 127). Socio-demographic details of those psychiatrists were compared simply to ensure the detection of any changes in data: 107 (84%) respondents had not changed their place of work and 20 (16%) had. The number of people who had moved from rural to urban areas (n = 10) was the same as those who had moved from urban to rural areas (n = 10).

Comparison of psychiatrists’ scores on MBI in 2005 and 2008

Paired t-tests were performed to find if the mean scores of EE, DP and PA differed significantly between 2005 and 2008 for the 127 psychiatrists (Table 4).

Table 4.

Changes in the mean scores of three dimensions of the Maslach Burnout Inventory (MBI) over the three-year study period.

Dimension of the MBI	Mean score 2005	Mean score 2008	Statistic	% change
Emotional exhaustion	21.98, n = 122	21.01, n = 122	p = .321, d.f. = 121	−4.41
Depersonalization	6.27, n = 124	7.60, n = 124	p = .007, d.f. = 123	17.5
Personal accomplishment	37.37, n = 121	31.88	p = .000, d.f. = 120	−14.6

The difference between the mean scores on EE in 2005 and in 2008 was 4.4% but remained in the moderate range without reaching statistical significance (p > .05), whereas the mean DP score increased by 17.5%, which was significant (p < .01), increasing from the low (6.27) to moderate (7.6) range. Mean PA score reduced by 14.5% (from 37.37% to 31.88% – both being in the moderate range) and was significant (p < .001). Thus, the statistical significance of change in scores differed for the three dimensions of burnout and did not correspond with the percentage change between means in 2005 and 2008. The change did, however, depend on base burnout scores in 2005. Paired t-tests were performed to find if the mean scores of EE, DP and PA differed significantly between 2005 and 2008 for the 127 psychiatrists (Table 4).

Relationship between JDS in 2005 and MBI in 2008

Spearman’s correlation coefficient was calculated between the five core dimensions of the JDS in 2005 and the three of the MBI in 2008 (Table 5). Scores on task significance (r_s = .295*, p = .003, n = 124), autonomy (r_s = .295*, p < .001) and feedback (r_s = .231*, p = .010, n = 124) in 2005 were significantly positively correlated with PA score in 2008. Similarly, scores on skill variety (r_s = −.190*, p < .05), task identity (r_s = −.208*, p < .05), autonomy (r_s = −.210*, p < .05) and feedback (r_s = −.210*, p < .05) in 2005 significantly negatively correlated with scores on EE in 2008. Finally, scores on autonomy (r_s = .295*, p < .001) and feedback (r_s = .231*, p < .010) in 2005 were significantly positively correlated with DP scores in 2008.

Table 5.

Correlations between Job Diagnostic Survey (JDS) scores in 2005 and Maslach Burnout Inventory (MBI) scores in 2008.

JDS dimensions in 2005	MBI sub-scales in 2008
	Statistic	Emotional exhaustion	Depersonalization	Personal accomplishment
Skill variety	Correlation coefficient	−.190*	−.191*	.134
	Significance (two-tailed)	.035	.034	.138
	n	123	124	124
Task identity	Correlation coefficient	−.208*	−.138	.165
	Significance (two-tailed)	.021	.127	.067
	n	123	124	124
Task significance	Correlation coefficient	−.083	−.109	.295*
	Significance (two-tailed)	.368	.234	.003
	n	121	122	122
Autonomy	Correlation coefficient	−.210	.295*	.295
	Significance (two-tailed)	.020	.001	.001
	n	123	124	124
Feedback	Correlation coefficient	−.210*	.231*	.231*
	Significance (two-tailed)	.020	.010	.010
	n	123	124	124

p < .05

χ² tests were performed between the 2008 scores on the three dimensions of the MBI and the five dimensions of the JDS in 2005. A high score on task identity in 2005 was associated with a low score on PA in 2008 (χ² = 7.76, d.f. = 2, p = .021) and a low feedback score was associated with high DP score in 2008 (χ² = 8.25, d.f. = 2, p = .016) as well as a low PA score in 2008 (χ² = 8.91, d.f. = 2, p = .012).

Relationship between MBI in 2005 and JDS in 2008

Spearman’s correlation coefficients were calculated between scores on the MBI sub-scales in 2005 and the JDS scores in 2008 (Table 6).

Table 6.

Maslach Burnout Inventory (MBI) in 2005 versus Job Diagnostic Survey (JDS) in 2008.

MBI dimensions in 2005	JDS dimensions in 2008
	Statistic	Skillvariety	Task identity	Tasksignificance	Autonomy	Feedback
Emotional exhaustion	Correlation coefficient	−.011	−.310*	−.085	−.228*	−.211*
	Significance (two-tailed)	.905	.000	.348	.011	.018
	n	126	126	124	124	125
Personal accomplishment	Correlation coefficient	.134	.228	.336*	.217*	.248
	Significance (two-tailed)	.137	.011	.000	.017	.006
	n	124	124	122	122	123
Depersonalization	Correlation coefficient	−.043	−.124	−.130	−.154	−.149
	Significance (two-tailed)	.634	.166	.149	087	.099
	n	127	127	125	125	126

p < .05

The EE score in 2005 was negatively correlated with task identity in 2008 (r_s = −.310*, p = .000, n = 126) and with task autonomy (r_s = −.228*, p = .011, n = 124). The PA score in 2005 was positively correlated with task identity in 2008 (r_s = .228*, p = .011, n = 124) and with task significance (r_s = 0.336*, p = .000, n = 122) and feedback (r_s = .248*, p = .006, n = 123). The DP score in 2005 was not correlated with any of the five dimensions of the JDS in 2008.

Discussion

This was the concluding phase of a longitudinal study of a cohort of psychiatrists over three years to monitor any changes in the dimensions of burnout and job satisfaction. The cohort appeared fairly stable with regard to socio-demographic variables.

This study found that the overall proportion of psychiatrists scoring low, medium and high on the three dimensions of burnout did differ over the study period. While the mean DP score had increased significantly and mean PA score had reduced significantly over the three-year study period, the EE mean score did not change. This finding suggests that among the three dimensions EE may be the most stable and least likely to change – a finding reported by previous authors (Boersma & Lindblom, 2009). This trend underscores the importance of reporting mean scores and the number of people scoring across various cut-off points in the three dimensions of the MBI. Longitudinal studies of burnout should therefore report changes in scores of all three dimensions of burnout and should investigate their impact on such variables.

Let us examine the impact of initial scores on the burnout dimensions of the JDS. The EE score in 2005 was negatively correlated with task identity in 2008 (r_s = −.310, p = .000, n = 126) and with task autonomy (r_s = −.228, p = .011, n = 124). A correlation of -0.35 between the general job satisfaction scale of the JDS and EE was reported by Maslach and Jackson (1981), which was considered as evidence of discriminant validity between the two constructs. If a single trait (such as EE) correlates at a low level with a conceptually different trait (such as job satisfaction) measured via a similar method, then discriminant validity is generally agreed to exist (Meier, 1984). The study found that as the scores on DP and PA increased, scores on the five dimensions of the JDS decreased over the three years, suggesting an inverse relationship between burnout and job satisfaction. In many cross-sectional surveys, psychiatrists as a group have been noted to report high levels of job satisfaction despite experiencing high burnout (Kumar et al., 2007; Prosser et al., 1996; Rey et al., 2004; Vaccaro & Clark, 1987). One recent cross-sectional study has reported that low levels of job satisfaction predicted high levels of burnout (Bressi et al., 2009). Being valued, doing a variety of tasks and having support in the psychiatrist role are reported to have contributed to job satisfaction (Clark & Vaccaro, 1987). This longitudinal study looked at the bidirectional relationship between job satisfaction and burnout. Job satisfaction scores in 2005 affected burnout scores in 2008 and burnout scores in 2005 affected job satisfaction in 2008. These relationships are worth examining in detail.

All the five dimensions of the JDS affected scores on one or more MBI dimensions. Low scores on skill variety in 2005 were significantly correlated with high EE and DP scores in 2008. Having variety in one’s work has been long identified as a significant contributor to improved job satisfaction among psychiatrists (Clark & Vaccaro, 1987). Psychiatrists can introduce variety in their work by splitting their time across clinical, administrative, educational and research activities (Sullivan, Jinnett, Mukherjee & Henderson, 2003). In particular, adding academic commitments to psychiatrists’ work may be effective in increasing a sense of professional accomplishment (Rutter, Herzberg & Paice, 2002), reducing DP, EE and overall stress (Agius, Blenkin, Deary, Zeally & Wood, 1996).

Low scores on task identity in 2005 were correlated with high scores on EE in 2008 (p = .035) and high scores on DP (p = .034). Task identity has been defined in the JDS as an employee’s ability to ‘do a job from the beginning to its end’ (Hackman & Oldham, 1975, p. 161). In a psychiatric context task identity would be reflected in a clinician’s ability to follow their patients through from the point of entry to exit in the clinical pathway, which has been associated with lower levels of work stress in psychiatrists (Isett, Ellis, Topping & Morrissey, 2009; Rathod, Mistry, Ibbotson & Kingdon, 2011).

Task significance was positively significantly correlated with the PA score. This finding could imply that psychiatrists derive a sense of worth and accomplishment from having a high degree of value attached to their role and profession. The possibility that task significance and PA could be surrogate measures of the same thing (i.e. passion for one’s work) needs to be considered. If that were the case, any association between the two could be an iterative effect. Task significance is defined as ‘the degree to which the job has a substantial impact on the lives or work of other people’ (Hackman & Oldham, 1975, p. 161) and hence its association with PA may be easily understood. This explanation becomes weaker when one examines the fact that the correlation between PA and task significance did not appear in 2005 but only on follow-up three years later. It has been suggested that factors that tend to undermine a sense of task significance also lead to job dissatisfaction among psychiatrists (Clark & Vaccaro, 1987). These factors include a lack of administrative support and validation, low pay, responsibility without authority, and pressure to sign documents related to patients unknown to them.

Low scores on autonomy in 2005 were significantly correlated with high DP scores in 2008 (r_s = −.210, p = .020, n = 123). This is consistent with previous findings suggesting that cynicism and reduced PA can arise from organizational factors such as lack of autonomy, social support and feedback at work (Bruce et al., 2005; Demerouti et al., 2001).

A low feedback score in 2005 was associated with high DP score (χ² = 8.25, d.f. = 2, p = .016) and low PA score in 2008 (χ² = 8.91, d.f. = 2, p = .012). One dimension of the JDS (feedback) affected scores on all three dimensions of the MBI (EE, DP and PA) after three years. Taking the definition of feedback ‘the degree to which carrying out the work activities required by a job results in the individual’s obtaining direct and clear information about the effectiveness of his or her performance’ (Hackman & Oldham, 1975, p. 162), it is reasonable to speculate that providing direct and clear feedback about psychiatrists’ performance may not only reduce cynicism and exhaustion, but also improve their sense of accomplishment. A positive relationship between feedback and PA has been reported in a cross-sectional study (Kumar et al., 2007). Anecdotal reports also identify built-in feedback mechanisms as protective against burnout (Holloway, Szmukler & Carson, 2000). This study offers both empirical and longitudinal support to the relationship between feedback and burnout.

Such a significant impact of the different dimensions of the JDS on burnout has implications for intervention. Paying attention to aspects of job satisfaction may assist us in developing specific interventions for psychiatrists who may score high on different dimensions of burnout. For instance, psychiatrists scoring high on EE may be encouraged to introduce variety into their clinical practice (e.g. teaching, administration and clinical work), be aligned to integrated models of service delivery and be given adequate opportunity for feedback about their performance. The impact of such interventions on burnout will need to be tested in well-designed studies and was beyond the scope of this study.

Strengths and limitations

This study followed up a relatively stable cohort of psychiatrists practising in New Zealand over a three-year period. It is the first longitudinal study that has looked at the relationship between dimensions of burnout and job satisfaction among psychiatrists. Findings of the study also rendered support to the view that EE may be the most stable dimension of burnout over a period of time. They show a bidirectional relationship between job satisfaction and burnout in psychiatrists.

The study had some limitations, which may make generalization of the findings difficult. The sample size of the final study was relatively small (N = 127). We were only able to follow up approximately a quarter (127/485) of the psychiatrists over the three-year period. It must be acknowledged that attrition rates are high among mental health professionals (Wykes et al., 1997). Finally, the three-year period of follow-up may only reflect a relatively short period of the professional lives of the psychiatrists, given the fact that 80% of them had more than two decades of post-qualification experience. Longer periods of follow-up may be better but could also potentially suffer from even greater attrition rate.

Study implications

Certain findings of the study may have significant implications for managers and leaders of mental health services. This study showed that there is a bidirectional relationship between burnout and job satisfaction in psychiatrists. Service providers may therefore wish to raise awareness that over a period of time burned-out psychiatrists may become dissatisfied with their jobs and dissatisfied psychiatrists may be prone to burnout. Improving mechanisms of constructive feedback may increase feelings of accomplishment and reduce cynicism in psychiatrists. EE, on the other hand, can be reduced by bringing skill variety and clarity in roles (task identity) into psychiatrists’ work.

Conclusions

This study found that EE may be the most stable of the three dimensions of burnout and that DP and PA appear to change with time. This trend implies that these two dimensions may serve as indices of worsening burnout, with potential implications for service delivery and patient care. Furthermore, DP and PA are influenced by two dimensions of job satisfaction, namely task significance and feedback, which can be addressed by paying attention to the environment in which psychiatrists work. Service planners and administrators may wish to note that burnout does affect psychiatrists’ job satisfaction longitudinally.

References

Agius

Blenkin

Deary

Zeally

Wood

(1996). Survey of perceived stress and work demands of consultant doctors. Occupational and Environmental Medicine, 53, 217–224.

Boersma

Lindblom

(2009). Stability and change in burnout profiles over time: A prospective study in the working population. Work & Stress, 23, 264–283.

Bressi

Porcellana

Gambini

Madia

Muffatti

Peirone

Altamura

(2009). Burnout among psychiatrists in Milan: A multicenter survey. Psychiatric Services, 60, 985–988.

Bruce

Conaglen

(2005). Burnout in physicians: A case for peer-support. Internal Medicine Journal, 35, 272–278.

Clark

Vaccaro

(1987). Burnout among CMHC psychiatrists and the struggle to survive. Hospital & Community Psychiatry, 38, 843–847.

Demerouti

Bakker

Nachreiner

Schaufeli

(2001). The job demands–resources model of burnout. Journal of Applied Psychology, 86, 499–512.

Firth-Cozens

(2007). Improving the health of psychiatrists. Advances in Psychiatric Treatment, 13, 161–168.

Hackman

Oldham

(1975). Development of Job Diagnostic Survey. Journal of Applied Psychology, 60, 159–170.

Holloway

Szmukler

Carson

(2000). Support systems. Advances in Psychiatric Treatment, 6, 226–237.

10.

Isett

K. R.

Ellis

A. R.

Topping

Morrissey

J. P.

(2009). Managed care and provider satisfaction in mental health settings. Community Mental Health Journal, 45, 209–221.

11.

Kumar

Hatcher

Dutu

Fischer

Ma’u

(2011). Stresses experienced by psychiatrists and their role in burnout: A national follow-up study. International Journal of Social Psychiatry, 57, 166–179.

12.

Kumar

Fischer

Robinson

Hatcher

Bhagat

(2007). Burnout and job satisfaction in New Zealandpsychiatrists: A national study. International Journal of Social Psychiatry, 53, 306–316.

13.

Maslach

Jackson

(1981). The measurement of experienced burnout. Journal of Occupational Behaviour, 2, 99–113.

14.

Maslach

Jackson

Leiter

(1996). Maslach Burnout Inventory manual, 3rd edn. Palo Alto, CA: Consulting Psychologist Press.

15.

Maslach

Schaufeli

Leiter

(2001). Job burnout. Annual Review of Psychology, 52, 397–422.

16.

Meier

(1984). The construct validity of burnout. Journal of Occupational Psychology, 57, 211–219.

17.

Melchior

Bours

Schmitz

Wittich

(1997). Burnout in psychiatric nursing: A meta-analysis of related variables. Journal of Psychiatric and Mental Health Nursing, 4, 193–201.

18.

Prosser

Johnson

Kuipers

Szmuckler

Bebbington

Thornicroft

(1996). Mental Health, ‘burnout’ and job satisfaction among hospital and community-based mental health staff. British Journal of Psychiatry, 69, 334–337.

19.

Rathod

Mistry

Ibbotson

Kingdon

(2011). Stress in psychiatrists: Coping with a decade of rapid change. The Psychiatrist, 35, 130–134.

20.

Rey

Walter

Giuffrida

(2004). Australian psychiatrists today: Proud of their profession but stressed and apprehensive about the future. Australian and New Zealand Journal of Psychiatry, 38, 105–110.

21.

Rutter

Herzberg

Paice

(2002). Stress in doctors and dentists who teach. Medical Education, 36, 543–549.

22.

Schaufeli

Enzmann

(1998). The burnout companion to study and practice: A critical analysis. London: Taylor and Francis.

23.

Snibbe

Radcliffe

Weisberger

Richards

Kelly

(1989). Burnout among primary care physicians and mental health professionals in a managed health care setting. Psychological Reports, 65, 775–780.

24.

Sullivan

Jinnett

Mukherjee

Henderson

(2003). How mental health providers spend their time: A survey of 10 Veterans Health Administration mental health services. Journal of Mental Health Policy and Economics, 6, 89–97.

25.

Thommasen

Lavanchy

Connelly

Berkowitz

Grzybowski

(2001). Mental health, job satisfaction, and intention to relocate: Opinions of physicians in rural British Columbia. Canadian Family Physician, 47, 737–744.

26.

Thomsen

Dallender

Soares

Nolan

Arnetz

(1998). Predictors of a healthy workplace for Swedish and English psychiatrists. British Journal of Psychiatry, 173, 80–84.

27.

Vaccaro

Clark

Jr. (1987). A profile of communitymental health centre psychiatrists: Results of a nationalsurvey. Community Mental Health Journal, 23, 282–289.

28.

Wykes

Stevens

Everitt

(1997). Stress in community care teams: Will it affect the sustainability of community care? Social Psychiatry & Psychiatric Epidemiology, 32, 398–407.