Mental well-being and job satisfaction among general practitioners: a nationwide cross-sectional survey in Denmark

Danish GPs work as independent contractors for the regional health authorities. GPs working under this tax-financed public reimbursement system are organised in the Organisation of General Practitioners in Denmark (PLO).

Almost all citizens (99%) are listed with a specific general practice, which they must consult (free of charge) for medical advice. Danish GPs act as gatekeepers to the rest of the health care system (except for emergencies), and the GPs provide comprehensive primary care (including preventive maternal and child care) with high levels of continuity [28]. Listed patients on average consult the general practice nearly seven times annually, and the average list size is approximately 1600 patients per GP [21]. The GPs must provide medical care all weekdays from 8 am to 4 pm, and all acute situations must be dealt with on the same day. Many GPs are also obligated to participate in out-of-hours cooperatives.

We included only GPs who were independent contractors (owners) working with the regional health authorities (excluding locums and trainees) in practices with at least 500 listed patients. A total of 3350 GPs were eligible for inclusion (Fig. 1).

All Danish GPs listed with a valid email address at PLO in May 2016 received an email with a link to an electronically administered questionnaire. The survey was announced 1 week before by email and in a newsletter on the PLO website. Non-respondents received a reminder after 2 weeks and after 4 weeks, and the data collection terminated on 1 July 2016. The link to the questionnaire was personal; it contained a unique serial number but no personal identifiers. The PLO distributed the link, and the research group collected the survey data. The PLO provided administrative data on the GPs. The data was transferred to Statistics Denmark separately; survey data by the research group and administrative data by the PLO. The data was linked at Statistics Denmark by the unique serial number, which was subsequently deleted and combined with register data by encrypted identifiers for anonymous analysis.

The questionnaire was developed from themes identified in existing literature and interviews with experienced researchers and clinicians. The interviews included eight individual telephone interviews (with seven GPs and one social worker employed by the Danish Medical Association) and a focus-group interview involving a representative from the Collegial Network for Physicians in Denmark (an organisation offering support to physicians), an occupational psychologist employed by the Danish Medical Association and a GP.

We used validated scales whenever possible and constructed ad hoc items when validated scales were unavailable. Scales and items were discussed and agreed upon in the research group. A pilot test was conducted among 10 GPs to assess the relevance and comprehensiveness of the questionnaire battery. Only minor changes (deletion of a few ad hoc items) were made based on the pilot test.

The following scales were used in this study: the Maslach Burnout Inventory-Human Services Survey (MBI-HSS), the 10-item Danish version of Cohen’s Perceived Stress Scale (PSS-10), the Warr-Cook-Wall Job Satisfaction Scale (WCW-JSS) and the World Health Organisation (Five) Well-Being Index (WHO-5). In addition, items concerning self-rated health (from the 12-item Short Form Health Survey (SF-12)), strains in private life and work-life balance were used.

Included scales have previously been translated for research use by standardised procedures; these generally include a forward translation carried out by researchers and a linguistic expert, backward translation by a native English-speaking person who was fluent in Danish, panel discussion and pilot test [29, 30].

Additional items concerned topics such as GP demographics, practice organisation, working hours, potential job-related strains, stress management, presentism and self-rated quality of clinical work. A free-text field was also added at the end of the questionnaire.

The GPs were required to respond to all items in the specific scale in order to proceed; this setup ensured that no items were missing. Completing the questionnaire took approximately 25 min.

The administrative data on the GPs included civil registration number (CRN), age, gender, region and provider number. The number of GPs registered with each practice was categorised into: 1 GP, 2–3 GPs and >  3 GPs. Practices with one GP were defined as solo practices and practices with two or more GPs as group practices. GP age was categorised into three age groups: ≤ 45, 46–59 and ≥ 60 years.

Data on patients listed with each practice was collected from national registers. All Danish citizens are assigned a unique CRN, which allows accurate linkage of information from numerous different registers at the individual level [31].

The number of listed patients per GP was calculated as practice list size divided by the number of GPs. This was done for all listed patients, patients ≥70 years and patients with a score of ≥1 in Charlson’s Comorbidity Index (CCI). The CCI was computed based on all primary and secondary diagnoses in the Danish National Patient Register (both inpatient and outpatient hospital diagnoses) from 2006 to 2016 [32, 33].

The socio-economic burden within the practice population was measured by the Danish Deprivation Index (DADI) [25], which estimates the socio-economic burden based on eight key variables. Higher values indicate a higher proportion of deprived patients. The number of patients per GP (all patients and subgroups) and DADI scores were categorised into quartiles based on all eligible GPs and practices. Information on the practice population was collected at the end of 2015.

Burnout and engagement was measured by the MBI-HSS. This instrument is considered a gold standard for assessment of burnout. It measures three burnout dimensions: (1) emotional exhaustion (EE), which is characterised by depletion of emotional resources, (2) depersonalisation (DP), which is characterised by emotional detachment from people related to work including patients, and (3) personal accomplishment (PA), which includes perceived value of work and self-efficacy. Subscale sum scores reflect the degree of burnout on each dimension. Based on predefined cut-off values for healthcare workers, each subscale score was defined as low, moderate or high [34]. Overlap between subscale scores that were indicative of burnout (high EE, high DP and low PA) was reported (Fig. 2). Complete burnout syndrome was defined as burnout-indicative scores on all MBI subscales. The opposite positive pole (low EE, low DP and high PA) was labelled ‘engagement’ [26, 27].

Job satisfaction was measured by the WCW-JSS. Nine job satisfaction facets and the overall job satisfaction were rated on a scale from 1 (‘extreme dissatisfaction’) to 7 (‘extreme satisfaction’) and added up to a sum score [35]. As no pre-determined cut-off values exist, respondents were divided into quartiles based on their sum score. For descriptive statistics, the GPs’ own ratings of their overall job satisfaction (single item) were used: a score of ≤3 was categorised as low, 4–5 as moderate and ≥ 6 as high job satisfaction. The single item had high correlation with the full scale (Pearson’s correlation coefficient: 0.89).

Perceived stress was measured by the PSS-10 [36, 37]. This widely used instrument consists of ten items about the frequency of stress-related feelings and thoughts. Each item is rated on a scale from 0 (‘never’) to 4 (‘very often’). In accordance with previous research, a sum score of ≥18 was considered as high level of perceived stress [38].

General well-being was measured by the WHO-5, which consists of five positively phrased items about cheerfulness, calmness, energy and interest in day-to-day activities in the previous 14 days. Items are added up and multiplied by four to create a scale from 0 (worst quality of life possible) to 100 (best quality of life possible). In general populations, the mean WHO-5 score is 70. When screening for depression, a cut-off score of ≤50 is recommended [45]. We divided general well-being into three categories: ‘high’ for a score of > 70, ‘poor’ for a score of ≤50 and ‘moderate’ for a score in between.

Self-rated health was assessed by a single item from the SF-12 asking respondents to rate their general health as ‘excellent’, ‘very good’, ‘good’, ‘fair’ or ‘poor’ [39].

Work-life balance was assessed by one item: ‘Do you generally experience a good balance between work and private life?’ Responses were categorised as ‘good’ (‘to a very high degree’ and ‘to a high degree’), ‘moderate’ (‘partly’) or ‘poor’ (‘to a low degree’ and ‘to a very low degree’).

Strains in private life were assessed by one item: ‘Do you feel burdened by factors related to your private life (economic issues, family problems, health conditions or similar)?’ Responses were categorised as ‘No’, ‘Yes, but it burdens me only little’, ‘Yes, and it burdens me to some extent’, ‘Yes, and it burdens me a lot’ and ‘I do not know/I do not wish to answer’. The latter response category was classified as ‘missing’.

GPs who completed at least half of the questionnaire (including the WCW-JSS and the MBI-HSS) were classified as respondents. The response rate within subgroups and the corresponding 95% confidence interval (CI) were calculated. The response rate within subgroups was further adjusted for gender, age, GPs per practice, listed patients per GP and DADI score, and these estimates were presented as risk difference (RDs) with 95% CI.

For each scale, we computed mean sum score, standard deviation (SD), median sum score and interquartile interval (IQI). Within each category of well-being and satisfaction, the percentage of GPs was calculated with 95% CI.

Overlap of burnout symptoms (high EE, high DP and low PA) was visualised in an area-proportional Venn diagram. Likewise, we made Venn diagrams of negative mental health aspects (≥ 2 burnout symptoms, high perceived stress and poor general well-being) and of positive mental health aspects (no symptoms of burnout and high general well-being) and job satisfaction.

Associations between outcome measures and selected GP characteristics (age, gender and type of practice) were estimated as odds ratios (OR) with 95% CI using logistic regression while adjusting for the mentioned GP characteristics.

Negative mental well-being and low job satisfaction were common among Danish GPs. One in ten met the criteria for complete burnout syndrome, one in five met the criterion for high perceived stress, and one in five reported poor general well-being. These bleak aspects overlapped, and 8% of GPs experienced poor overall mental health as a combination of poor general well-being, high stress and substantial burnout.

Still, about 25% of the GPs reported no symptoms of burnout along with high job satisfaction and general well-being. Furthermore, most GPs rated their general health as good, and only few reported to feel much burdened by strains in private life. However, the majority of GPs assessed the work-life balance as only moderate or poor.

The highest frequency of low job satisfaction and complete burnout syndrome was seen among GPs aged 45–59 years and male GPs. Male GPs also had higher risk of depersonalisation and poor overall mental health than women. Yet, female GPs more often than male GPs reported a poor work-life balance. Likewise, GPs in solo practices more often than GPs in group practices reported a poor work-life balance.

The response rate was comparable to those reported in similar GP studies. In consideration of the significant size of the questionnaire, we consider this satisfactory. This study allowed for comparison of respondents and non-respondents, and response bias was present. Caring for a higher number of elderly patients, a higher number of patients with comorbidity and a more deprived patient population were all factors associated with non-response. This supports the assumption that GPs facing greater practice demands may be less likely to take out time to respond to questionnaire surveys. Thus, the proportion of GPs experiencing high levels of burnout and stress may be underestimated. The lower response rate among the oldest GPs corresponds to the findings in other GP studies [40] .

The cross-sectional design does not allow us to make any conclusions on causality. Yet, this survey provides unique opportunities for future analyses of potential causes and consequences of job satisfaction and well-being among GPs.

The number of Danish GPs who face complete burnout has increased twofold since 2012 and fourfold since 2004; this is mainly due to increased emotional exhaustion and depersonalisation [41, 42]. During the same period, a decrease in job satisfaction has been seen; we found a mean WCW sum score in this study at 48 compared with 55 in 2012 and 57 in 2004 [43]. This development is in accordance with the previously reported reciprocal relationship between burnout and job satisfaction among GPs [7, 44]. Earlier studies included GPs in only one region of Denmark. However, this is not expected to impair the comparability since job satisfaction and burnout did not vary significantly across regions in the current study. Compared with the general population, participating GPs had lower levels of general well-being, and more GPs perceived high levels of stress, especially in comparison with subpopulations with similar length of education [38, 45]. This is notable as we expect GPs in general to be more aware of the importance of mental health compared to professionals outside the healthcare system.

Burnout is considered a consequence of long-term job-related stress, and the emotional exhaustion component of burnout has been described as an individual stress experience. Thus, the finding that GPs more often report high emotional exhaustion than high levels of perceived stress might appear contradictory. However, the measures are intended to measure different concepts; the emotional exhaustion subscale of the MBI aims to measure job-related exhaustion (depletion of emotional resources), whereas the PSS-10 is constructed to measure perceived stress in life in general. Moreover, both scales measure symptom degree as a continuum, and the somewhat arbitrary cut-off points have been established differently. Hence, the two measures were expected to overlap, but only to some extent. Moreover, burnout and stress were expected to overlap with poor general well-being, and similar overlap between positive outcomes was also expected [27, 46]. The wide span in mental health and job satisfaction among GPs may relate to differences in self-care strategies, personality traits and other factors unrelated to work. However, differences in working conditions are likely to contribute as well. The changes in working conditions that we have witnessed in recent years may weigh heavier on GPs practicing in areas characterised by deprivation, many elderly patients and/or workforce shortage [25, 47].

The increasing prevalence of mental distress and job dissatisfaction among Danish GPs mirrors the findings from physician studies in other countries [3‐6]. However, the reported prevalence of distress and dissatisfaction varies substantially across countries [10].

Few other studies have measured GP burnout using the MBI-HSS. Despite marked increases in high EE (30.6%) and high DP (21.0%) among Danish GPs, most recent GP studies from elsewhere in Europe and North America have reported even higher prevalence with high EE ranging from 46% (UK) to 53% (Ireland) and high DP ranging from 32% (Ireland) to 46% (Canada) [48‐50]. Still, lower prevalence of high EE (25%) and high DP (16%) was reported in one Portuguese study [51]. Additional categorisation of burnout status based on MBI subscale scores has been approached in multiple ways, but consensus on the diagnostic criteria is lacking [26]. Only few GP studies report prevalence of GPs with high burnout scores on all MBI-HSS subscales (complete burnout syndrome). In a recent Irish study, the prevalence of GPs with complete burnout was 6.6% [48]. In earlier studies, the prevalence ranged from 3.5% in a Swiss study from 2002 to 12% (with wide inter-country variation) in a GP survey across 12 European countries from 2000 [7, 51, 52].

Low overall job satisfaction was less common in our sample (22%) than in a study among GPs in the UK from 2015 (32%) using the WCW-JSS single item [5]. In contrast, studies from Switzerland in 2009 and Norway in 2008 reported higher levels of GP job satisfaction [53, 54]. However, to our knowledge, no recent publications explore whether the job satisfaction among GPs has declined in these countries, as has been the case in Denmark.

The pronounced inter-country variation in the reported GP distress and discontentment may relate to sociodemographic and cultural differences and to different research methods and response rates [2]. However, different working conditions may also contribute substantially as the contextual factors associated with working in general practice (e.g. working hours, time per patient and type of employment) vary substantially across countries [55]. Job-related factors that are consistently reported to influence the levels of distress and dissatisfaction among GPs include lack of recognition, dissatisfaction with income, increased number of administrative tasks, long working hours, workload intensity and time pressure [13, 19]. Working conditions that undermine the GPs’ experience of providing high-quality patient care may reduce the professional fulfilment, which has been described as the satisfying inner experience of being useful and making progress and has been identified as a key motivator for physicians’ engagement in clinical work and healthcare development [56].

The finding that male GPs more often reported negative mental health and low job satisfaction is in accordance with previous research [48, 53]. Barriers to help-seeking and the propensity among physicians to ignore indicators of distress may be more pronounced in men [57]. This may contribute to increased risk of severe negative mental health among male GPs. The finding that male GPs more often reported low EE is consistent with previous research [7].

Superior mental health among GPs in the oldest age band echoes previous research [48]. This may be due to early retirement among GPs who experience low job satisfaction, workload pressure and poor health, whereas delayed retirement is seen among GPs who thrive in the job [58]. Other explanatory factors may include a cohort effect and age-related obligations unrelated to work (e.g. children living at home). This is supported by the finding that GPs of higher age reported better work-life balance. Furthermore, work-related strains may seem less intruding when retirement is approaching and you have many years of clinical experience. The high job satisfaction among the youngest GPs and the high prevalence of severe burnout symptoms among GPs in mid-career is in line with previous research [7, 52]. This increased burnout risk in middle-aged GPs compared with younger colleagues may reflect gradual development of burnout over time due to gradual depletion of resources [26].

The high prevalence of severe mental distress implicates a need for enhanced support to the affected GPs. Special attention should be paid to subgroups at high risk (e.g. male GPs and GPs in mid-career). The variation in GP responses concerning external factors calls for future research to examine the significance of differences in size and composition of patient populations.

The substantial increase in GP distress and discontentment signifies an urgent need to address the working conditions in general practice in order to maintain a sustainable GP workforce. Further exploration of the personal and environmental factors that are related to (positive and negative) well-being and satisfaction among GPs is suggested. Initiatives aiming to enhance GP well-being, career satisfaction and engagement are recommended to strengthen primary care and support the provision of high-quality patient care in a health care sector characterised by increasing workloads and looming workforce shortage.