A Three-year Cohort Study of the Relationships between Coping, Job Stress and Burnout after a Counselling Intervention for Help-seeking Physicians

Karin E Isaksson Ro; Reidar Tyssen; Asle Hoffart; Harold Sexton; Olaf G Aasland; Tore Gude


BMC Public Health. 2010;10(213) 

In This Article


Study Design and Sample

The consecutively participating physicians in a counselling intervention at The Resource Centre for Health Personnel in Norway, from August 2003 through July 2005, were eligible for inclusion in the study. Participants signed an informed written consent. The cohort comprised 227 physicians at baseline (94% of 242 eligible) (see Figure 1). Self-report instruments were completed before the intervention (baseline) and were mailed to participants approximately one and three years after the intervention (two reminders sent). Three-year follow-up was completed 36.9 months (SD 1.9, range 34–44.5) after baseline.

Figure 1.

Flow chart over participation in the study.


The Resource Centre is available for all Norwegian physicians. It is funded by the Norwegian Medical Association and is located at a psychiatric facility, Modum Bad.

The interventions are based on an integrative approach incorporating psychodynamic, cognitive and educational theories.[26]

Physicians chose to participate in one of two different interventions. The first was a single day, six to seven hour counselling session for one physician with a psychiatrist or a specialist in occupational medicine (MD). A "non-treatment" setting without medical records and with absolute confidentiality was ensured. After being invited to describe his or her situation the physician was asked to map both work-related and private contextual factors contributing to stress. Coping strategies, often related to sources of identity, self-esteem and self-reliance in the individual, were identified, acknowledged, and challenged. The physician's present needs in both a short and a longer perspective were identified, and it was usually recommended that the doctor actively should deal with these needs (for example by stress reduction or obtaining treatment, such as psychotherapy).

The second type of intervention was a five day, group based course for eight participants, boarding at the Centre, and led by one of the same counsellors in collaboration with an occupational therapist. Daily lectures, group discussions, and physical activity were offered as well as an individual counselling session during the week. The intervention is described in more detail elsewhere.[26,33]


Burnout In this study we used the subscale emotional exhaustion (10 items, Cronbach's α = .92) of Maslach's Burnout Inventory.[34] As in previous studies of Norwegian physicians, a five-point scale scoring perceived fit (1-does not fit, 5-fits very well), with reference to the last two weeks at work, was used to score the MBI,[26,29] contrary to the seven-point scale scoring frequency that is used in most studies internationally. The seven-point scale has been criticized for having categories that are not mutually exclusive.[35]

Personality Eysenck's abbreviated personality questionnaire with six items for neuroticism explaining 82% variance of the original scale was used (α = .71).[36] Items were scored dichotomously (1-yes or 0-no), and a sum score from 1–6 was obtained, in which higher scores designate more neuroticism.

Perceived Job Stress 17 items from a modified version of the Cooper Job Stress Questionnaire, used in the Norwegian student/doctor cohort, were selected using Principal Component Analysis. In addition nine clinically prompted items from the questionnaire were included, as previously described.[26] Job stress with 26 items (α = .92), as reported in this paper, consists of three subscales – emotional stress (10 items, α = .85), social stress, including work-home interface stress and time stress (10 items, α = .83) and fear of litigation which also covers fear of complaints or criticism (6 items, α = .86). The correlations between the subscales ranged from .48 to .64. Scores were given on a five-point scale (1 = no stress, 5 = very much stress), with reference to the two last weeks at work.

Coping Strategies Eighteen of 42 items in Vitaliano and colleagues' Ways of Coping Checklist were selected by a Principal Component Analysis of data from the Norwegian student/doctor cohort.[6,7] Vitaliano defines two main dimensions of coping strategies – the more adaptive ways of coping and the potentially maladaptive ways of coping.[6] Among the adaptive strategies two subscales are "problem-focused coping" and "seeking social support." In this study they are described with four and five items respectively, together designated as "active coping strategies" (α = .81). Among the potentially maladaptive strategies two subscales are "wishful thinking" and "blaming self." In this study they are described with seven and two items respectively, together designated as "emotion-focused coping strategies" (α = .82). (For a list of items, see Additional file 1). Scores were given on a five-point scale (1 = does not fit at all and 5 = fits completely).

Demographic Data Gender, age, marital status, having children less than 16 years of age (dichotomous variable).

Specialist Status Categorized into internal medical specialties, surgical specialties, psychiatric specialties, general practice (in Norway general practice or family medicine is an approved specialty, and about half of the general practitioners are specialists), public health and laboratory medicine, non-specialist (usually specialists in training).[33]

Work hours Sum of hours per week used in direct patient contact, meetings, paperwork, on the telephone etc., research, and "other work activities".[26]

Psychotherapy Attending psychotherapy during the first year after baseline intervention (0-no and 1-yes).

Sick Leave Number of weeks on full time/part time sick leave/rehabilitation/disability during the preceding year and current sick leave.


Continuous, repeated parameters were tested with repeated measures ANOVA (repeated contrast) with time (baseline, one-year and three-year), and with interactions between time and psychotherapy. Variables were normally distributed. The condition of sphericity was examined with Mauchly's test. In case of violation of the assumption of sphericity, degrees of freedom were corrected using Greenhouse-Geisser estimates of sphericity (ε = .91).

Two repeated measures in the same cohort were tested with paired t-tests for parametric data, Wilcoxon's rank test for continuous, non-parametric data, and McNemar's test for dichotomous variables. T-tests and Chi-square tests were used respectively for comparison between different groups.

Effect sizes using pooled SD were calculated according to the method of Cohen (the mean difference/sum of standard deviation for the two measures/2), defining values of <0.20 as indicating no effect, 0.20–0.49 indicating small and 0.50–0.79 indicating moderate effect.[37]

The sequential relationships between change in emotional exhaustion relative to changes in job stress, coping and neuroticism respectively were examined using the structural modelling program EQS 6.1, beta version, in a series of cross-lagged and synchronous panel models. Cross-lagged models examine if the baseline value of one parameter influences change in the other parameter. The synchronous or co-temporal model examines if change in one parameter appears to influence change in the other parameter,[[38] pp 22–37]. (See Figure 2). As the time span from the intervention to follow-up (to both one-year and three-years) was relatively long, we would expect that synchronous or co-temporal panel models, rather than cross-lagged, would be likely to detect temporal relationships that might be present between these variables,[[38] p32]. Cross-lagged paths were examined first. The synchronous relationships were studied if the cross-lagged paths had an inadequate model fit or did not show significant cross-lagged relationships (Figure 2).

Figure 2.

Cross-lagged and synchronous/co-temporal panel models.

The variables were allowed to freely correlate at baseline. The model fit criterion of the confirmatory fit index, CFI, is a frequently used measure of the adequacy with which the structural model represents the observed data and was used to determine the adequacy, or fit, of the model. A comparative fit index (CFI) >= 0.95 in combination with the standardized root mean squared residual (SRMR) <= 0.08 (which is independent of sample size) were used to determine model fit.[39] A critical ratio (parameter/standard error) of 1.96 or greater was used to determine whether or not a path was significant at the 0.05-level.

The sample size was adequate for maximum likelihood estimation of models with a small number of parameters to be estimated, but not for the use of latent variables.[39] The relationships were studied from baseline to both one- and three-year follow-up. Cronbach's α-values from 0.71–0.92 indicate satisfactory internal consistency of the variables. The lack of excessive kurtosis or skewness in the variables indicated a sufficiently normal distribution of the data. The correlations at baseline, one-year and three-year follow-up were between emotional exhaustion and (i) job stress 0.70, 0.64 and 0.67 (ii) active coping -0.23, -0.21 and -0.23 (iii) emotion-focused coping 0.47, 0.59 and 0.62 (iv) neuroticism 0.53, 0.61 and 0.55 with significance levels ranging between <0.01 and <0.001.

The level of significance was set to p < 0.05 in general. To avoid type I-error due to number of tests a statistical correction was performed in relation to the ANOVAs and to the panel modelling. This correction implied that for the four overall ANOVAs a p-value of <0.05/4 = 0.0125 should be required. The overall ANOVAs all had a significance of <0.001, indicating that the results were not incidental. For the panel models, involving fourteen tests, a p-value of <0.05/14 = 0.0036 should be required. This corresponds to a Critical ratio CR = parameter/standard error (distributed as z) of >2.93, indicating that, except for stress due to fear of litigation, the significant relationships found at three-year follow-up stay significant also after correction.

Missing Data

One or a few missing items in instruments measuring coping strategies and neuroticism were replaced by the mean score of completed items. The instruments measuring job stress and emotional exhaustion included items that were not relevant for all respondents due to differences in working conditions (not working directly with patients as in laboratory work, leadership, research). Mean score of relevant items for each individual was used.

Zero to four/184 (0–2.2%) of the instruments on burnout, job stress, coping and neuroticism at baseline, and 0–18/184 (0–9.8%) at three year follow-up were insufficiently completed. N was adjusted accordingly in the individual analyses.


Participants signed an informed, written consent. The study has been approved by the Data Inspectorate through the Norwegian Social Science Data Services. The Regional Ethical Research Committee in the South of Norway did not find special consent necessary for this study.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: