Working conditions and Work-Family Conflict in German hospital physicians: psychosocial and organisational predictors and consequences

Work-Family Conflict (WFC)

Work-Family Conflict (WFC) has been commonly defined as a "form of interrole conflict in which the role pressures from work and family are mutually incompatible in some respect" [1, 2]. Its theoretical background is a scarcity hypothesis, which describes that individuals have a certain, limited amount of energy. When involved in multiple roles, these roles tend to drain them and cause stress or interrole conflict [3, 4] (in [5]). As conflicts may arise not only between work and family some authors extended the view on other obligations in personal life [6, 7].

Individuals identify themselves through social roles. For many people, work and family roles are the most important and self-relevant life roles. As demands and expectations within the family and work domains are not always compatible, conflicts between family and work life can arise. This conflict between work and family is bidirectional in nature; there is work interfering with family (WIF) and family interfering with work (FIW) [8]. WIF occurs when demands and obligations of the work role are deleterious to family life. FIW describes the conflict arising from family obligations that disturb one's work. Family and work as social roles have different grades of permeability: family roles tend to be less structured and more permeable than job roles [9]. This may explain the higher reported frequency of WIF compared to FIW [10, 11].

The relationship between work and family life has been investigated over the past years. Different aspects of Work-Family Conflict, its antecedents and consequences as well as gender differences have been examined by various measures and in different samples. As the number of dual-career and single-parent families in Germany has increased over the last decades [12], Work-Family Conflict as a phenomenon in the workforce gains more and more attention.

In the medical sector, Germany currently experiences a situation of major physician attrition which has been attributed to the exit or non-entrance into medical care of young physicians in particular [13]. The introduction of the DRG-based financing system in 2004 has led to major changes in work organisation and condensation of workload for physicians. Additionally in the clinical sector, the number of viable and employed physicians diminished drastically over the last few years [13]. So in 2003, already 21.7% of physicians were without medical occupation [14]. This was mainly due to decrease in medical students and graduates, increased attrition of medical students during their studies and loss of physicians to public health systems of other countries as well as into non-clinical occupations, e.g., health management, pharmaceutical industry, occupational health and safety, research, consulting, or journalism [15, 16, 14]. In 2004, the German Ministry for Health and Social Security sponsored a study on major reasons for the German physicians' exit from curative medical occupation [14]. 44,4% of the physicians in non-clinical occupations would theoretically like to continue working in hospitals. Major barriers for their return were the following: overwork (52% of male, 54% of female physicians), incompatibility of family and work obligations (37% of male, 54% of female physicians) and insufficient income (51% of male, 24% of female physicians). 82% of physicians working abroad would not like to return to Germany; major reasons for not coming back were good adaptation to the foreign country (70% of male, 72% of female physicians), professional perspectives (69% of male, 65% of female physicians), compatibility between work and family (51% of male, 64% of female physicians) and income (64% of male, 53% of female physicians). Antecedents and sources of Work-Family Conflict among physicians were not investigated in [14].

Already 92,2% of the German medical students expect the conflict between job and family demands to be very high, especially for female physicians [17]. JURKAT ET AL. described the highest expected strains on personal life to be related to childcare and finding time for other activities.

As the ratio of female medical graduates has increased to over 50% in the last 20 years in Germany, Work-Family Conflict could have serious impact on future doctor shortage in Germany leading to a serious threat of the overall quality of medical care [13, 14]. Not only in Germany but in most Western countries, the share of female physicians has increased, so that some researchers are talking of a future "pink-collar medicine" [18]. To prevent extreme physician attrition by improving working conditions in particular for women, gender differences of WFC have to be investigated intensively in future.

Physicians hold a profession that is traditionally connected to very high work commitment. Residency, the time of medical specialisation, has been described as the most strenuous and work demanding period during a career as a physician [19]. In this period, physicians mostly experience high work demands with low job control, which may by itself contribute to high job stress [19]. At the same time, residency as a career stage directly follows graduation from medical school, and often coincides with the family-founding life stage of young physicians. Young physicians are more likely to have young children and consequently experience high family or parenting demands, and hence might experience high levels of Work-Family Conflict. On the other hand, female and male physicians who endured the hardship of the time of medical education and residency training [18] may have already given precedence to work over personal activities [20] (in [21]). Data examining the Work-Family Conflict within medical professions is rare in general. Some research has been conducted concerning WFC among nurses [22, 23]. Data on WFC among physicians and gender difference in particular is very sparse in general, although DUMELOW observed higher dissatisfaction with work-life balance among female physicians [24]. Physicians' health is underinvestigated in Germany [25]. We are not aware of previous research on details of WFC among German hospital physicians. Our study tries to contribute to a more detailed knowledge of different aspects on WFC.

From the occupational health point of view, in our study we will only examine work interfering with family conflict (WIF) thus focussing on a parameter which may possibly be modified by changes of work-related strains, e.g., work organisation.

• Key Question I a : What is the prevalence of work interfering with family conflict (WIF) among German hospital physicians?

• Key Question I b: Does the prevalence of WIF among German hospital physicians differ from that of the general German population?

• Key Question I c: Which sociodemographic covariates (e.g., age) significantly predict WIF?

According to gender role theory, women are more likely to see their family role as part of their social identity than men do [26, 27]. Consequently, WIF is threatening a central social role (and self-identity) in particular for women. Overall Europe, women's roles in the workplace have increased over the last years [12, 28], but at the same time expectations within their family roles have not diminished: working mothers still invest more time in family than working men do [29, 21, 30]. Also in the medical profession this pattern is to be found [31, 32] (in [18]). Nevertheless, the situation of women in medicine is somewhat peculiar: for female physicians, it is possible to combine a medical career with child rearing, via the availability of part-time working and the ability to afford high-quality child care [33, 34] (in [18]). It still remains unclear if female physicians would report higher levels of WIF. With our study, we want to fill this gap in literature.

Findings of previous research concerning gender difference in WIF and FIW are inconsistent: some studies report higher WIF among women compared to men [29, 10, 35]. On the other hand, WIF and FIW were found to be similar for men and women by several researchers [26, 9, 36, 3, 11, 27].

Our study addresses the issue of gender difference in order to bring greater clarity to WIF among German hospital physicians. Details on the prevalence, sources, and outcomes of WIF for female and male physicians carry important implications and high potential for future interventions.

• Key Question I d: Is there a gender difference in the prevalence of WIF among German hospital physicians?

Research design

The survey was designed as an explorative study, no primary and secondary outcomes were determined beforehand. Significant differences between different groups of physicians, as shown by statistical analysis, helped to generate hypothesis. The study was conducted as a cross-sectional survey using a standardised questionnaire to assess sociodemographic data and the psychosocial work environment, including work organisation and general and mental health of hospital physicians in Germany.

The final questionnaire was distributed together with an informative letter of invitation and a postage-paid return envelope. The cover letter explained that participation in the study was voluntary and anonymous. We did not ask for written consent of the participants in particular, but judged their voluntary participation after having read the informative letter as "informed consent". The respondents were not compensated for participation. Departments with a response rate under 35% were personally contacted by the first author during a department team meeting. The study protocol was approved by the Ethical Committee of the University Witten/Herdecke (no. 15/2005). The ethical aspects were in full agreement with the Helsinki declaration.

Instruments

In order to secure a hospital-specific design of the survey, unstructured interviews were conducted with experienced hospital physicians at different stages of career (resident, attending, head of department, academic medicine) as well as related medical professions (nurse, nurse scientist). The aim of these interviews was to obtain insight into physicians' work environment in order to construct survey questions regarding hospital-specific aspects. During the development of the standardised questionnaire, two pre-tests were performed. The first pre-test was conducted to eliminate possible inconsistencies or unclear wording of the survey questions (participants: non-medical, university degree (n = 11)). The items on hospital-specific issues were reviewed in a second pre-test (participants: German hospital physicians (n = 8)). In case a pre-test participant found a survey item unclear, the item was reworded.

Statistical analyses

All categorical items described above related to workload, strain, and outcomes derived from the German COPSOQ version were transformed to a scale ranging from 0 (minimum value, e.g., "do not agree at all") to 100 points (maximum value, e.g., "fully agree"). Item non-response and the category "does not apply" were coded as missing values. Scale scores were computed as the mean of the values of the single aspects if at least half of the single items had valid answers. Thus, as all of the single items, all scales therefore have a theoretical range from 0 to 100. Since the transformation is linear, the relation between all items remains unaffected. This transformation is a standard procedure following the recommendation for the German COPSOQ version [6] and allows comparison of our findings with other studies using the same approach.

Data analysis included descriptive statistics, parametric and non-parametric correlation analyses, explorative and confirmatory factor analyses, simple and multiple regression analyses, and reliability analyses. During validation of the original German COPSOQ version [6] in addition to Cronbach's alpha and intra-class-correlation (ICC) for the total population, G-coefficients in a single-facet design [52] were determined in the analysis of the scale reliability to verify the generalisability of the item information [53]. Pearson's chi² test, correlation analysis (Pearson's r), analysis of variance (ANOVA), and linear bivariate and multiple regression analysis were performed. The comparisons between our sample of hospital physicians (n = 296) to all German hospital physicians in 2005 [15] (n = 146511) was performed through chi² tests. Significant differences of our sample to the sample of the general German population [6, 7] (n = 2561) including another sample of German physicians (n = 77) were tested through ANOVA. Gender differences were tested also by ANOVA.

All p-values given were two-tailed. A p-value of less than .05 was considered significant. Values are given as mean and standard deviation (SD). Data were analysed using SPSS V14.0.

Sample

A survey on Psychosocial work hazards and strains of German hospital physicians was conducted during the period of April to July 2005. The study subjects were physicians currently working in two different university teaching hospitals in the region of Northrhine in Germany. The hospitals were almost equal in size. Hospital A has been privately owned since January 2003 whereas hospital B was a new public holding consisting of three formerly independent public hospitals.

General data about physicians working in the participating hospitals were obtained from the human resources department as well the departments' heads. Yet, the representativeness of responders was difficult to determine because there was no complete demographic data available from the hospitals. Gender distribution (chi² = .005, DF 1, n.s. p > .9) and cultural background (chi² = .39, DF 1, n.s. p > .5) in our sample (n = 296) were similar to the distribution among German hospital physicians in 2005 (n = 146511) [15]. However, responders in our study were slightly younger (mean = 38.3 yrs, SD 8.9 vs. mean = 40.9 yrs, SD not available). Certain medical disciplines were overrepresented in our sample.

Work-Family Conflict

COPSOQ data

Besides WIF, psychosocial working conditions, general and mental health outcomes were assessed by the German COPSOQ version, see Additional file 2. Thus possible sources of WIF in the psychosocial work environment and general and mental health consequences could be examined (key questions II a – II c, key questions III a – III c).

Predictors for Work-Family Conflict: bivariate and multivariate analysis

For the identification of significant predictors for WIF (key question II a), we first performed a bivariate analysis (Pearson's correlation) with all sociodemographic variables, the parameters of the psychosocial work environment (COPSOQ scales), and the hospital-specific items depicting work organisation. Secondly, all variables that reached a significance level of p < .2 in bivariate analysis were included as potential predictors in the multivariate linear regression model. A step-wise selection procedure with the commonly used including criteria p < .05 and the excluding criteria p > .10 was applied. A pairwise exclusion of missing values was executed. The procedure followed the suggestions by HOSMER [54]. The estimated values of the final model were compared to empirical means in the survey.

Bivariate analysis revealed several significant correlations between WIF and some of the COPSOQ scales, see Additional file 3. For example higher levels of WIF correlated with higher quantitative demands (p < .01) and role conflict (p < .01). Negative correlations were found between influence at work (p < .01), sense of community (p < .01), influence and possibilities for development (p < .01), social relations and leadership (p < .05), and quality of leadership (p < .05) and WIF. The scales commitment to workplace and relational justice did not show significant correlations to WIF.

For all the predictors, the performed estimations of WIF means were satisfying and confirming the final regression model.

Outcome factors

We tested the relationship between WIF and job outcomes as well as outcomes measuring physical and mental well-being by bivariate analysis (key question III a).

Prevalence of work interfering with family (WIF) conflict

The WIF conflict scale (0–100) showed a high mean among German hospital physicians (mean = 74.2). Our sample of physicians had a significantly (p < .01) higher level of WIF compared to the German general population [6, 7] (mean = 44.8) as well as to the subgroup of German physicians in that earlier study (n = 77, mean = 59.0, SD 28.3, p < .01). This finding underlines the relevance of negative work-life balance as an existing stressor for physicians [41, 58]. As our study had a cross-sectional design, it has to remain unclear why the present sample showed higher values on the WIF scale than the physicians investigated earlier. It has to be borne in mind that the broad public discussion of physicians' work-life balance in Germany in recent years might have led to a higher awareness with regard to this issue.

In accordance with other research comparing male and female physicians [33], the present study did not find a gender difference for the level of WIF perceived. As a higher WIF level for female physicians was described by DUMELOW [24] and our sample was relatively small, this question should be addressed in future investigations with bigger sample size. In contrast to the results reported in [65, 27], we did not observe any effects of cohabitation with a child on WIF which again may be due to the rather small sample size.

Predictors for WIF

Several sources for WIF conflict could be identified in our study. Consonant with the other research findings [8], our study shows that quantitative workload was a significant predictor for WIF. This finding confirms the assumption that work role overload through quantitative demands leads to job stress [66, 63, 23], which may by itself lead to conflict between job and personal life. When comparing the predictors of WIF with data of the German general population [6, 7], significantly higher values for the quantitative demands in our sample (p < .01) were found. In the hospital setting this result is not surprising since condensation of workload in the course of re-structuring of the German health system is one of the most direct effects on the employees, e.g., the physicians.

To date, empirical results suggest that one of the time-based sources of WIF lie in work schedule irregularities [5, 23]. We examined frequency, consequences, and strains arising from unpredictable changes of the duty roster among hospital physicians. The frequency of the need to postpone planned vacations due to work schedule irregularities was identified as a significantly positive predictor for WIF: the more often a physician had to change vacation plans, the higher was the perceived WIF. Obviously, the demand to give up an important personal "reward" that was planned on long hand creates high conflict between work and personal life. The short-time changes of the work schedule may be rooted in organisational deficits, which could be easily and directly be altered.

The number of days the physicians went to work despite own health problems or illness significantly predicted WIF: the more often the respondents ignored their own physical well-being and went to work, the higher was the perceived WIF. This factor has not been described as a source of WIF in prior literature so far. It might reflect a neglect of own physical needs among physicians. Consciously ignoring one's own health might be a national peculiarity, as postponing one's recovery from illness to weekends or holidays seems to be common in the German workforce [67]. On the other hand, the low sickness absence rates of physicians [68] may reflect high work commitment in the medical profession. It has been described elsewhere that the "service ethic" connected to physicians' work demands long hours and the placing of patients' welfare above personal needs [20] (in [21]) [18]. A similar factor, excessive identification with the job, has been described as a predictor for job-related family problems elsewhere, too [8] (in [69]).

In concordance with previous findings [23], age was identified as a negative predictor for WIF: physicians under 35 years of age scored much higher on the WIF scale (mean = 79) than physicians aged 55 years and older (mean = 56). A possible explanation could be the general high stress level for young physicians in their first postgraduate years. This may be caused by task-inherent stressors such as dealing with sickness and death, high work demands, little control over assigned duties, and the absence of a work routine. However this could also be attributed simply to the worse working conditions for physicians in lower job positions as described by JURKAT ET AL. [70]. In addition, the time of job entry of young physicians often coincides with the family-founding phase thus leading to conflicting interests in both work- and family-related domains. Furthermore, advanced age could be a protective factor for WIF due to better coping strategies with job stress based on longer job experience [3]. As described by NÜBLING ET AL., different interpretations of our findings could be verified only in longitudinal studies – but remain speculative in cross-sectional surveys [71].

Similar to the results of studies conducted in Germany and the U.S. [72, 41, 25, 29], sense of community was identified as the only protective factor for WIF: high sense of community predicted low values of WIF. Our analyses revealed that women rated the sense of community significantly better (mean = 78) than men did (mean = 72).

Relationship between WIF and outcomes

Our study offers several interesting insights into the possible consequences of WIF conflict: significant relationships to all outcomes investigated were observed. These outcome parameters covered physical and mental health, general well-being, and job-related outcomes. Figure 1 displays the mean values of five outcome parameters according to four quartils covering the WIF scale.

In accordance with the results by HILL [29], who reported a direct relationship between WFC and individual stress, our study showed a positive relationship between WIF and behavioural and cognitive stress symptoms. Furthermore, WIF was significantly positively correlated to personal burnout. The relationship to emotional exhaustion has been described earlier [45]. In our sample, female physicians had a significantly higher level of burnout than their male counterparts. Previous findings reported the same [73]. According to our results, high WIF was related to high intention to leave the job (p < .01). This finding is consistent with earlier research, which identified the intention to leave the job as well as high staff turnover as consequences of job-related strains for the family or private life [69, 45, 47, 42, 44, 5, 74]. In a longitudinal study, GRANDEY described chronic WFC to drain resources over time, leading to high intention to leave the job [3]. Furthermore, our result underlines the importance of WIF for the non-entry of – respectively the exit from – the curative medical occupation in German hospital physicians [14].

Similar as in [26, 38, 42–44, 29, 23, 74], we found a significantly negative relationship to job satisfaction: High values of WIF were associated with low job satisfaction. This finding strengthens the assumption that a threat towards a self-relevant role, in this case towards the family or personal life, creates negative attitudes towards the threat, in this case the work [26, 42]. To date, empirical results have suggested that women and men have similar levels of job satisfaction [29]. Although SWANSON described a lower job satisfaction for male physicians, in our study female physicians reported a significantly lower job satisfaction than men. In accordance with several researchers [5, 38, 42, 44, 29, 23], a negative relationship between WIF and satisfaction with life could be observed: the higher the WIF, the lower was the satisfaction with life in general.

In our sample, low values of WIF were associated (p < .01) with better self-evaluation of the own health status and work ability (p < .01). REIMER ET AL. found that female physicians felt healthier than men [75]. Our findings clearly contradict this view. We found that women felt significantly less able to work than men did.

Almost all results on mental and physical outcomes differed significantly from the general population: physicians in our sample reached significantly higher values for behavioural stress symptoms and personal burnout. Compared to another sample of physicians [6, 7], our sample had a significantly higher mean for behavioural stress symptoms (p < .01). The higher burnout level was also observed in [63], although HERSCHBACH described only an average burnout-index for physicians [76]. The intention to leave the job was significantly higher for our sample than in the general population. This finding might reflect the general climate among physicians, that leaving the medical curative occupation in hospitals is a realistic and attractive opportunity which more and more physicians in Germany consider [15]. Interestingly, despite an identified negative relationship between WIF and health status respectively work ability (see above) – which would suggest low self-reported health status and work ability in our sample – physicians reported above-average scores on these scales compared to the general population. The mean of general health status was 80 in our sample compared to a mean of 72 in the general population. The work ability (mean = 80.1 vs. mean = 75.5, p < .01) was significantly better in our sample, too. The high scores on these scales could be a result of high social expectations and the existing role model of a "hard-working, but healthy" physician [77]. Other explanations could be over-commitment to the workplace [72], or a very poor perception of own stress and health [77]. Some authors suspect that physicians may have – due to their daily work with very sick and dying people – a distorted relationship to their own body and health, ignoring all signs of own illness [77].

Our findings underline the dissatisfaction of German physicians with their job: our sample had significantly lower job satisfaction compared to the general population. Prior research discusses job satisfaction among physicians controversially: results for above-average values [76, 57, 78] coexist next to those showing below-average values [79] compared to other professions.

A major finding in our study was the above-average value of the work-related stressor quantitative work demands and its strong correlation with WIF. The alerting above-average level of all outcome variables points out the current status of physicians' health in Germany. Interestingly, own health concerns, such as going to work despite own illness, significantly predicted WIF, and at the same time physicians reported above-average levels of self-reported health status and work ability.