Physicians as clinical directors: working conditions, psychosocial resources and self-rated health

Transcription

1 Occupational Medicine 2004;54: DOI: /occmed/kqh021 Physicians as clinical directors: working conditions, psychosocial resources and self-rated health Maud Lindholm 1,3, Elisabeth Dejin-Karlsson 1, Jan Westin 2, Bo Hagström 3 and Giggi Udén 1 Introduction Background Physicians in clinical directors positions fulfil their commitments in demanding work environments characterized by organizational changes and economic cutbacks. Little is known about the self-rated health of this group. Aim Methods Results Conclusion Key words Received 24 June 2003 To investigate whether self-rated health was associated with psychosocial working conditions, professional networks, job support, social networks and social support, sick leave and salary in Swedish physicians working as clinical directors. A self-reported questionnaire was sent to 373 clinical directors. Odds ratios (ORs) were used for estimating the bivariate association between self-rated health and psychosocial resources. A total of 274 clinical directors agreed to participate in the study. The response rate was 73%. The clinical directors exposed to high job demands had a significantly higher probability of low self-rated health [OR = 3.4 and 95% confidence interval (CI) = ] than those who were not in this situation. Furthermore, participants who were exposed to high job demands had an increased risk of low self-rated health (OR = 3.8 and 95% CI = ) irrespective of available social support inside or outside work. High average working hours more than doubled the risk of low self-rated health (OR = 2.2 and 95% CI = ). The job demands on physicians in clinical directors positions may exceed ordinary means of support with consequent adverse effects on self-rated health. More research is needed to investigate the interaction between job demands and support systems in this group of health care workers. Clinical director; job demand; professional network; self-rated health; social support; working hours. Revised 16 October 2003 Accepted 24 November 2003 There has been considerable reform and reorganization within the Swedish health care system, with potential consequences for the work environment and conditions 1 Department of Nursing, Malmö University, Malmö, Sweden. 2 Department of Medicine, Lund University, Lund, Sweden. 3 National Institute for Working Life, Malmö, Sweden. Correspondence to: Maud Lindholm, School of Health and Society, Department of Nursing, Malmö University, SE Malmö, Sweden. Tel: ; fax: ; maud.lindholm@hs.mah.se for management and leadership at all levels [cf. 1,2]. Methods derived from industry designed for enhancing efficacy have often been applied uncritically to the health care sector in order to ensure higher medical output with fewer resources [3]. It is the responsibility of the Swedish Government to ensure that, through local government, the health care system develops efficiently according to its overall purpose, which is based on overall goals, the constraints of social welfare and macro-economic factors. There is a legal requirement to have clinical directors in departments who represent and are responsible to the health Occupational Medicine, Vol. 54 No. 3 Society of Occupational Medicine 2004; all rights reserved 182

2 care provider [4]. The physicians in clinical directors positions fulfil their commitments in demanding work environments characterized by political decisions, organizational changes and financial constraints. Karasek and Theorell [5] found psychological job demands along with time pressure and conflicts to be significant sources of risk for stress-related illness. Individual control over the work process was one resource available to workers when confronting job demands. Social job support might, according to Johnson [6], function as an important coping resource, which may potentially modify the impact of social environmental stress. Karasek and Theorell [5] stated that social support at work referred to the overall levels of helpful social interaction available on the job from both co-workers and supervisors. Berkman [7] described social networks and social support as two different concepts. Social networks were defined as webs of social relationships that surround an individual and the characteristics of those linkages, while social support was defined as the emotional, material or financial aid that was obtained from an individual s social network. Low levels of psychosocial resources such as weak social networks and low social support have emerged as risk factors in health-related research. Different studies have shown associations between different aspects of social networks and/or social support within various groups and health [8 10], stress and psychiatric disorder in health care professions and hospital staff [11,12], resiliency and burnout in human resource managers [13] and the psychosocial work environment in municipal middle managers [14]. Yukl [15] described social (professional) networks in an organizational perspective as major sources about what was happening inside or outside the organization. They could be used for obtaining assistance and political support for plans and proposals, for helping implement changes and innovations, for obtaining advice from fellow professionals and for facilitating coordination with people in other parts of the organization. The exchange of information within the networks was built on trust [16]. A person s health can be assessed in two fundamentally different ways: on one hand by an expert, such as a doctor and on the other by the person him/herself. It has often been found that medical examinations are poor predictors of illness behaviour, such as absence from work or use of a health system. The main reason is that people s illness behaviour seems to be largely influenced by their own perceptions of their health, in other words self-rated health [17]. The aim of the study was to investigate whether self-related health was associated with psychosocial working conditions, professional networks, job support, social networks and social support, sick leave and salary in Swedish physicians in clinical directors positions. M. LINDHOLM ET AL.: PHYSICIANSASCLINICALDIRECTORS 183 Subjects and methods The sample consisted of 373 clinical directors within hospital and community health care in Sweden. The criterion for participation was that the physicians were active in a management position. Formally, clinical directors hold a basic position, e.g. as a senior physician, together with managerial responsibility. It is up to the individual clinical director to determine the proportion of their time spent either clinically or managerially. Selection of the sample was made by systematically selecting every fifth physician recorded in a clinical director s position in the Register of Swedish Hospitals 2001 and Register of Swedish Community Care The study was conducted using a postal self-administered questionnaire. Questionnaires were originally mailed in January 2002, with two reminders mailed to non-respondents at intervals of 3 weeks. Thirty-six (10%) of the returned questionnaires were not completed because the physician contacted did not currently hold a managerial position and they were therefore replaced in the study. Nineteen (5%) of the clinical directors abstained from participation and 81 (22%) did not answer. No notable differences were found with regard to gender, age and employment area in the data analysis of clinical directors who chose not to participate and those who did not answer compared to the participants. In the end 274 clinical directors (73%) participated in the study, 70 (26%) of whom were women. The questionnaire contained items related to psychosocial working conditions. It also inquired about demographic, socio-economic and work-related conditions, such as age, cohabiting status, time and type of employment, educational level, average working hours per week, sick leave, income and participation in professional networks, psychosocial characteristics outside the workplace and self-reported health. Definitions Socio-economic variables Age was dichotomized close to the median into above or below 50 years. Cohabiting status was dichotomized as living either in the same domicile with a husband/partner or being the only adult in the household. Participation in a leadership programme was classified as either no participation in such a programme, participation at a local or regional level or participation at university level and dichotomized into participation in any programme or not. Years of work experience was based on total work experience and dichotomized close to the median into above or below 19 years. Years in a chief position was based on managerial

3 184 OCCUPATIONAL MEDICINE experience and dichotomized close to the median at the level of 4 years. Working hours per week denoted the average working time and was dichotomized close to the median at the level of 50 h/week. The salary of the participants (before taxation) was dichotomized close to the median at the level of 5945 Euros per month. Professional network denoted participation in networks consisting of chief colleagues, with at least two meetings per year and focusing on management questions and was divided into affiliation or no network affiliation (Table 1). Quality of a professional network denoted perceived support and was classified into high and low network support (Table 2). Sick leave was classified as being absent or not from work owing to illness. Psychosocial variables Work-related psychosocial factors were assessed by means of a Swedish version of Karasek s [5] demand/control instrument. The participants were asked to rate their current work position in terms of demands and decision latitude. The dimension of job demand, which consisted of five items, covered the degree of psychological demands such as hectic work, work overload and role ambiguity. The dimension of job control covered the degree of decision latitude experienced in the work situation and the opportunity for learning new things related to work: it consisted of six items. A model developed by Hanson and Östergren [18] was used for obtaining social network and social support indices. Social network, which was regarded as a structural concept, was divided into two indices: social stability and social participation. Social stability, which consisted of five items, described the degree to which a person belongs to and is anchored within formal and informal groups such as the family, neighbourhood and friends and the feeling of membership in these groups. Social participation described how actively a person takes part in the social activities of formal and informal groups in society. The instrument assessing social participation was from the Swedish National Survey on Living Conditions and was assessed by 14 items [19]. Social support, which was regarded as a function of a person s interaction with his/her social network, was divided into two indices: emotional support and material support. Emotional support reflected a person s experience of receiving encouragement of personal value and feelings of confidence and trust from relatives, friends, neighbours and colleagues: it consisted of three items. Material support consisted of one item measuring a person s access to advice, information and practical service. Job support described the atmosphere and conditions of the workplace, which included the understanding and Table 1. Background characteristics of the clinical directors investigated (n = 274) Background characteristics n % Age (years) Missing Gender Male Female Cohabiting status Cohabiting Not cohabiting Participation in a leadership programme Yes No Years of work experience Years in chief position Type of organization Hospital Community care Missing Professional network Affiliation No affiliation Missing Average work hours per week Missing Salary (Euros) > Sick leave (previous year) Yes No Missing support received from superiors and workmates defined by Johnson [20] and assessed by an instrument modified by Theorell et al. [21] and was measured by six items. Self-rated health, which was regarded as a person s non-comparative global assessment of his/her physical /psychological health and well-being at the present time, consisted of one item ranging from 1 (very bad, could not be worse) to 7 (excellent, could not be better). Low self-rated health was defined as a score of 1 3.

4 M. LINDHOLM ET AL.: PHYSICIANSASCLINICALDIRECTORS 185 Table 2. Association between psychosocial characteristics, salary and sick leave and low self-rated health in clinical directors presented as crude ORs, adjusted ORs and 95% CIs (n = 274) Psychosocial characteristics Number Low self-rated health Crude OR 95% CI Adjusted OR a 95% CI Demand Low 147/ High 119/ * * Control High 146/ Low 123/ Professional network Affiliation 133/ No affiliation 139/ Professional network High support 45/ Low support 88/ Job support High 157/ Low 107/ Social stability High 190/ Low 68/ * * Social participation High 183/ Low 88/ * * Instrumental support High 114/ Low 157/ Emotional support High 139/ Low 130/ * * Average working hours per week 50 h 136/ h 135/ * * Salary (Euros) / / * * Sick leave No 241/ Yes 29/ *P < a Adjusted for age, gender and leadership education. The distribution of the score of each item of the scales assessing social network and social support outside work, job support, job demand and job control was first dichotomized as close as possible to the median. The exposed category was assigned 1 point whereas the unexposed one was scored 0 points. The points for the items of each index were added into their respective total index score, which resulted in a new distribution. In the second step a score below the tertile of the total index was defined as low and the score above as high. All low scores were considered a potential risk factor. Dichotomizations were made as near the tertile as possible when the distributions were narrow. Statistical analysis Odds ratios (OR) were used for estimating the bivariate association between work-related variables, socioeconomic variables, social network/social support variables and level of self-rated health. Logistic regression analysis was used for the multivariate analysis and the variables age, gender and post-basic education were regarded as potential confounding factors. In order to test for a possible effect modification of professional network, job support and emotional support on the relationship between job demand and self-rated health, a Synergy Index score was calculated as proposed by Rothman [22]. This was performed, in the first case, by combining the variables assessing job demand and professional network into dummy variables so that participants who had experience of low job demand and affiliation to a professional network constituted the reference category. Furthermore, participants with low

5 186 OCCUPATIONAL MEDICINE job demand and no affiliation to a professional network, participants experiencing high job demand and affiliation to a professional network and, finally, participants who were exposed to both high job demands and no professional network affiliation made up the other categories. Age, gender and post-basic education were regarded as potential confounding factors. ORs were used as estimates of the relative risks. In order to assess the potential confounding effect regarding the relationship between job demand and low self-rated health in a logistic regression analysis, the variables professional network, job support and emotional support were added to the analysis one by one. Statistical significance was determined by means of confidence intervals (CIs) at the 95% level. The Statistical Package for the Social Sciences (SPSS 11.0) was used for all statistical purposes in this study. The Ethic Committee at the Medical Faculty of Lund University approved the study. Results Table 1 presents the socio-demographic characteristics of the participating clinical directors. When examining average working hours per week we found that 61% of clinical directors in hospitals and 30% in community care reported working more than 50 h/week. Table 2 shows the associations between psychosocial characteristics, salary level, previous sick leave and self-rated health in clinical directors, which are presented as crude and adjusted ORs. The group of clinical directors who reported low self-rated health consisted of 41 persons (15%), 30 (11%) of whom worked in hospitals. We found that participants who reported high job demands had a significantly higher probability of low self-rated health than those who were not in this situation. In relation to social network and social support outside work, clinical directors who reported low social stability, low social participation and/or low emotional support increased their probability of low self-rated health many times. Furthermore, in this study it was also found that high average working hours per week and a low salary level more than doubled the odds for low self-rated health in the clinical directors, while no significant association was found between sick leave and self-rated health. In order to test for effect modification, the associations between low self-rated health in the clinical directors and double exposure to psychosocial job demands combined with either professional network, job support or emotional support were presented as adjusted ORs in Table 3. The adjustment was made for age, gender and post-basic education as potential confounding factors. When testing for professional networks we found that the risk was increased for low self-rated health in clinical directors who were exposed to high job demands, regardless of whether or not they were affiliated to professional networks. Furthermore, the association with low self-rated health remained increased in combinations with high job demands and also when the variable affiliation or not to a professional network was replaced with the variable level of job support. The risk of low self-rated health turned out to be three times higher when high job demands were related to high job support and four times higher when related to low job support. The Synergy Index was calculated as 1.6, which indicates a weak synergistic effect. In the further examination we found a statistically significant association between low Table 3. Effect modification of combined exposure to demands and professional network/job support/emotional support with regard to low self-rated health in clinical directors presented as adjusted ORs with 95% CIs and Synergy Index (n = 274) Variables n Adjusted OR a 95% CI Synergy Index Demands and professional networks: affiliation/no affiliation Low demand and affiliation 78/ Low demand and no affiliation 69/ High demand and affiliation 53/ * High demand and no affiliation 66/ * 0.77 Demands and job support Low demand and high job support 93/ Low demand and low job support 51/ High demand and high job support 61/ * High demand and low job support 57/ * 1.6 Demands and emotional support Low demand and high emotional support 74/ Low demand and low emotional support 74/ * High demand and high emotional support 61/ * High demand and low emotional support 56/ * 2.0 *P < a Adjusted for age, gender and leadership education.

6 M. LINDHOLM ET AL.: PHYSICIANSASCLINICALDIRECTORS 187 self-rated health and low job demands in combination with low emotional support. By changing the low job demands into high job demands and keeping the low level of emotional support, the probability of low self-rated health increased more than three times for the exposed clinical directors. The Synergy Index was calculated as 2.0, thus indicating a synergistic effect (Table 3). Finally, in the last analysis focusing on the association between low self-rated health and high job demands, three of the psychosocial characteristics regarded as supportive resources in this study, namely professional network, job support and emotional support, were tested as potential confounders in a multivariate analysis. Here we found that the OR for a high level of job demands in relation to low self-rated health remained statistically significant when the three variables were entered into the analysis one by one (Table 4). When clinical directors working in hospitals were compared with clinical directors working in community care, significantly higher odds were found for low job support (OR = 1.7 and 95% CI = ) in those working in hospitals, while significantly higher odds were found for exposure to high job demands (OR = 1.7 and 95% CI = ) in those working in community care (not shown in Table 4). Discussion This study revealed that physicians in clinical directors positions who were exposed to high job demands had a significantly higher probability of low self-rated health than those who were not in this situation. It was also found that those who were exposed to high job demands had an increased risk of low self-rated health irrespective of available social support inside and outside work. High average working hours more than doubled the risk for low self-rated health. A total of 274 clinical directors participated. The researchers decision to replace clinical directors in the sample who did not currently hold a senior position was made in order to ensure an acceptable level of participation. An excluded participant was replaced by the closest name recorded in the same register used. This mode of selection could mean a threat to the internal validity if differences affect the dependent variable in ways extraneous to the effect of the independent variables. However, since the interest of this study was in investigating a possible association between self-rated health in physicians in a clinical director s position related to their experience of working conditions and psychosocial resources, this mode of selection was not considered inappropriate. Another possible threat to the internal validity could be misclassification in relation to clinical directors professional network affiliation, as there is no accepted description of what really constitutes a professional network. The criteria chosen in this study were chosen in order to match an earlier study focusing on nurse managers self-rated health [10]. There is, of course, a risk of bias in using a wide description, since there could be personal apprehensions about whether a group of colleagues is to be considered a professional network or not. The instrument chosen for assessing social networks and social support has been used in a number of previous studies and has documented good validity and reliability [23]. Karasek s [5,24] demand/control model, which was used for assessing psychosocial working conditions, is a well-recognized instrument of good validity and reliability. Good reliability was found of a measure of self-rated health in a test retest, with a κ coefficient of 0.72 [25]. Confounding might be another risk of bias. There is always a possibility of residual confounding owing to the existence of unmeasured variables. In this study three confounding factors were adjusted for, namely age, gender and leadership education, but these factors changed the crude estimates only marginally. External validity is obtained when the results can confidently be generalized outside the specific setting [26]. The physicians in clinical directors positions participating in this study embraced all Sweden and represented hospitals as well as community care, which might make it possible to generalize the findings to Table 4. Association between high demands at work and low self-rated health at work in clinical directors presented as crude ORs and 95% CIs (n = 274) Psychosocial characteristics Model 1 Model 2 Model 3 Model 4 Crude OR 95% CI Crude OR 95% CI Crude OR 95% CI Crude OR 95% CI Demand (high/low) * * * * Professional network (no affiliation/affiliation) Job support (low/high) Emotional support (low/high) * *P < 0.05.

7 188 OCCUPATIONAL MEDICINE physicians at corresponding managerial levels within work environments characterized by changing processes. The results indicated that job demands on physicians in a clinical director s position might have become too high to be buffered within ordinary supporting limits related to the work situation. These results were supported by a study by Lindholm et al. [10] who investigated the psychosocial work environment in 205 nurse managers. The study showed that those who were exposed to high job demands had elevated odds for low self-rated health, regardless of the level of psychosocial resources within or outside work. Similar results were found by Westerberg and Armelius [14] in a study focusing on municipal middle managers, where a high level of control and a satisfactory level of social support did not seem to compensate for the high levels of demands in terms of psychosomatic reactions owing to work. Low emotional support in combination with high job demands turned out to be a serious threat to those clinical directors who reported low self-rated health. Almost half of the participants reported low emotional support. In this respect it seems noteworthy that 90% of all participants reported themselves as cohabiting. Although the single question of cohabiting status was quantitative and no further questions were asked about cohabiting frequency or length, previous work situation with high job demands on clinical directors might not be neglected as a contributing factor in the low emotional support experienced. Thulin [27], who investigated work conditions in 23 physicians in clinical directors positions at surgical clinics, found that 18 of the clinical directors answered that they often or always neglected their family in favour of their work. In a study focusing on fatigue among the major occupational groups of health care workers, the highest level of fatigue and the subjective sensation of tiredness were experienced by physicians (particularly women physicians), in professions allied to medicine and managers. The differences in general fatigue between men and women and between professional groups could be accounted for by high work demands and lack of role clarity [28]. These findings altogether might indicate that high job demands could also influence the quality of clinical directors private lives. The importance of high job demands became even more obvious in this study when available support, such as professional network, job support and emotional support, altogether did not seem to ease the experiences of workload and low self-rated health in the participating clinical directors. One way of handling an increased amount of work is to work more than 40 h/week voluntarily and regularly, which in Sweden is the limit for normal working hours. It was revealed that half of the participants worked 51 h or more per week. This group of clinical directors turned out to have more than a 2-fold higher risk for low self-rated health than those who worked 50 h or less. Feldman [29], who presented a multilevel framework for understanding when and why managers will engage in long working hours, showed that the propensity to work long hours was by no means synonymous with greater organizational productivity or higher quality of life outside the working place. These findings were supported by Atkinson [30] who demonstrated that many individuals working long working days had increased fatigue and a decrease in the acuity of their decision making. For physicians in management positions, who usually and by tradition try to uphold both clinical work and a management position at the same time, these findings point at an increased risk of negative impact on their health and well-being. As stressed by Westin [31], the decision to accept a managerial position means a breaking up and a choice: a choice away from heavier clinical activities and deeper research engagement, but also a choice towards something new and untried. This study shows the effects of high job demands and deficient support systems related to self-rated health in physicians in clinical directors positions. One proposal for strengthening the support system in this group is to develop national and local network systems consisting of senior colleagues through further research. Apart from efforts at decreasing job demands it also seems necessary to explore the kind of support this group of clinical directors considers to be most appropriate in their present work situation. Acknowledgements The National Institute for Working Life (Arbetslivsinstitutet Syd), Malmö, Sweden and Swedish Foundation for Health Care Sciences and Allergy Research (Vårdalstiftelsen), Stockholm, Sweden, supported this study. The authors are most grateful to the physicians in chief positions for their cooperation during the data collection. References 1. Lindholm M, Råstam L, Udén G. Management from four perspectives. J Nurs Manage 1999;7: Lindholm M, Udén G. Nurse managers management direction and role over time. Nurs Admin Q 2001;25(4): Arnetz B. Psychosocial challenges facing physicians of today. Soc Sci Med 2001;52: SOSFS. Socialstyrelsens Allmänna råd om Verksamhetschef inom Hälso-och Sjukvård, Vol8(General Advice for Chief Executives Within the Health Care Sector). Stockholm: Socialstyrelsen, 1997 [in Swedish]. 5. Karasek R, Theorell T. Healthy Work. Stress, Productivity, and the Reconstruction of Working Life. New York: Basic Books, 1990.

8 M. LINDHOLM ET AL.: PHYSICIANSASCLINICALDIRECTORS Johnson JV. The significance of the social and collective dimensions of the work environment for human health and well-being. In: Enander A, Gustavsson B, Karlsson J, Starrin B, eds. Work and Welfare. Papers from the Second Karlstad Symposium on Work. Karlstad: University of Karlstad, Berkman L. Assessing social networks and social support in epidemic studies. In: Lykke Mortensen E, Egsgaard J, eds. Social Networks and Health. Copenhagen: Hans Jörgensens Bogtrykkeri/Offset, 1995; Östergren P-O. Psychosocial resources and health: with special reference to social network, social support and cardiovascular disease. Thesis, Studentlitteratur, Lund, Hanson BS, Östergren P-O. Social network and health: theories and empirical evidence from Malmö, Sweden. In: Lykke Mortensen E, Egsgaard J, eds. Social Networks and Health. Copenhagen: Hans Jörgensens Bogtrykkeri/Offset, 1995; Lindholm M, Dejin-Karlsson E, Östergren P-O, Udén G. Nurse managers professional networks, psychosocial resources and self-rated health. J Adv Nurs 2003;42: Sheikh A, Hurwitz B. Psychological morbidity in general practice managers: a descriptive and explanatory study. Br J Gen Pract 2000;50: Weinberg A, Creed F. Stress and psychiatric disorder in health care professionals and hospital staff. Lancet 2000;355: Zunz SJ. Resiliency and burnout: protective factors for human service managers. Admin Soc Work 1998;22(3): Westerberg K, Armelius K. Municipal middle managers: psychosocial work environment in a gender-based division of labour. Scand J Manage 2000;16: Yukl G. Leadership in Organizations, 5th edn. Upper Saddle River, NJ: Prentice-Hall, Inc., Joint P, Warner M. Managing Across Cultures: Issues and Perspectives. London: International Thomson Business Press, Bjorner JB, Sondergaard Kristensen T, Orth-Gomér K, Tibblin G, Sullivan M, Westerholm P. Self-rated Health A Useful Concept in Research, Prevention and Clinical Medicine. Stockholm: Forskningsrådsnämnden (the Swedish Council for Planning and Coordination of Research), Hanson BS, Östergren P-O. Different social network and social support characteristics, nervous problems and insomnia: theoretical and methodological aspects of some results from the population study Men Born in 1904, Malmö, Sweden. Soc Sci Med 1987;25: National Central Bureau of Statistics. Living Conditions. Isolation and Togetherness An Outlook on Social Participation Report No. 18. Stockholm: The National Central Bureau of Statistics, Johnson JV. The impact of place of work social support, job demands and work control upon cardiovascular disease in Sweden. Thesis, Report No. 1, Department of Psychology, University of Stockholm, Stockholm, Theorell T, Harms-Ringdahl K, Ahlberg-Hultén G, Westin B. Psychosocial job factors and symptoms from the locomotor system a multicausal analysis. Scand J Rehab Med 1991;23: Rothman KJ. Modern Epidemiology. Boston: Little, Brown & Co., Hanson BS, Östergren P-O, Elmståhl S, Isacsson S-O, Ranstam J. Reliability and validity of measures of social networks, social support and control results from the Malmö shoulder and neck study. Scand J Soc Med 1997;4: Karasek R. Job demands, job decision latitude, and mental strain. Admin Sci Q 1979;24: Lundberg O, Manderbacka K. Assessing reliability of a measure of self-rated health. Scand J Soc Med 1996;24: Polit DF, Hungler BP. Nursing Research: Principles and Methods. Philadelphia: Lippincott, Thulin L. Occupational conditions of chief surgeons. Does a heavy work schedule interfere with sleep and family life? Läkartidningen 1993;90: [in Swedish]. 28. Hardy GE, Shapiro DA, Borrill CS. Fatigue in the workforce of National Health Service trusts: levels of symptomatology and links with minor psychiatric disorder, demographic, occupational and work role factors. J Psychosom Res 1997;43: Feldman DD. Managers propensity to work longer hours. A multilevel analysis. Human Resource Manage Rev 2002;12: Atkinson W. Wake up! Fighting fatigue in the workplace. Risk Manage 1999;46: Westin J. The Physician as Chief and Leader. Facts, Advices and Encouragement to the Physician in Chief Role. Lund: Studentlitteratur, 2002 [in Swedish].