In the general working population, burnout is considered a significant social and economic problem (Collier, 2017). In contrast to many other professions, relatively little is known about the causes of burnout among general practitioners (GPs).

Burnout is a psychological syndrome characterized by the three core symptoms of emotional exhaustion, depersonalization, and diminished personal effectiveness (Freudenberger and Richelson, 1980; Maslach et al., 2001). Since the introduction of the concept of burnout in the 1970s by Herbert Freudenberger, many studies have passed, attempting to explain this intriguing phenomenon (Freudenberger, 1974). Indeed, after introducing the Maslach Burnout Inventory (MBI) (Maslach and Jackson, 1981), a psychological assessment tool to measure the three dimensions of burnout, the research on this subject exploded (Maslach et al., 1996). Not surprisingly, considering the background of Freudenberger, the initial researchers found burnout exclusively related to human services, such as social work, healthcare, and teaching (Maslach and Schaufeli, 1993). However, the historical imperfection of exclusively relating burnout to human services, has been corrected abundantly in the past two decennia given a large amount of literature outside the human service sectors (Maslach and Schaufeli, 1993). Moreover, there was no theoretical argument to limit burnout to the human service professions (Maslach and Leiter, 1997). Moreover, there is enough evidence that shows that stressors leading to burnout in human services can be found in other non-human service professions as well (Buunk et al., 1998). Therefore, there appears to be sufficient theoretical arguments for releasing the initial conventional definition, restricting burnout to the human service professions. To meet this shortcoming of the Maslach Burnout Inventory-Human Services Survey (MBI-HSS), the Maslach Burnout Inventory–General Survey (MBI-GS) was developed, which was suitable for all workers, including those without intensive contact with recipients of services. One of the main differences between the two versions of the MBI was that the depersonalization scale from the MBI-HSS was transformed into the cynicism scale in the MBI-GS. Both scales reflect the social perception of distancing, distancing from people in the case of depersonalization and distancing from things in the case of cynicism (Schaufeli and Leiter, 1996). Simultaneously, the Oldenburg Burnout Inventory (OLBI) was developed, which consists of two dimensions: emotional exhaustion and disengagement from work (Demerouti, 1999). Despite their strengths and weaknesses, the MBI-HSS, the MBI-GS, and the OLBI are all widely practised.

According to Schaufeli and Taris (2005), burnout should be conceptualized, primarily as a work-related syndrome with at least emotional exhaustion and depersonalization as dimensions. One of the criticisms of the MBI is that depersonalization is a coping strategy under certain circumstances and should be better studied along with other coping strategies (Tipa et al., 2019). Furthermore, diminished personal efficiency, one of the three burnout dimensions of the MBI, should be perceived as one of the many consequences of long-term stress (Kristensen et al., 2005). These and other objections are amply and dignifiedly refuted by Schaufeli and Taris (2005). To address these concerns, the OLBI (Demerouti et al., 2003) was developed. One of the limitations of the OLBI is its limited construct validity (Halbesleben, 2003). However, a detailed discussion of this debate is beyond the scope of this study. Despite all the criticism, both the OLBI and the MBI are still widely used measurement tools today.

It is remarkable and as yet not easy to explain why the MBI is still so popular, despite all the criticism.

In response to the many well-documented objections to the MBI, a new Burnout Assessment Tool (BAT) has been developed (Schaufeli et al., 2019, 2020). The BAT comprises four core dimensions: exhaustion, mental distance, and emotional and cognitive impairment. Moreover, three secondary dimensions emerged: depressed mood, psychological distress, and psychosomatic complaints.

To classify the determinants of burnout globally, this study distinguished between job demands, job resources, and personal resources. Job demands are defined as those physical, social or organizational aspects of the job that require sustained physical or mental effort and are therefore associated with maintaining physiological and psychological costs (Demerouti et al., 2001). Examples of generic job demands are work overload and interpersonal conflict. Job resources were defined as those physical, social or organizational aspects of the job that may do any of the following: (a) be functional in achieving work goals; (b) reduce job demands and the associated physiological and psychological costs; (c) stimulate personal growth and development (Demerouti et al., 2001). Examples of generic job resources are feedback and social support. Personal resources are defined as the psychological characteristics or aspects of the self that are generally associated with resiliency and that refer to the ability to control and impact one’s environment successfully (Schaufeli and Taris, 2014). Examples of personal resources are self-efficacy and optimism. Extensive overviews of generic job demands, generic job resources and personal resources have been published by, among others, Lee and Ashforth (1996) and Schaufeli and Taris (2014).

Common theories in this regard are the job demands resources (JDR) theory and the conservation of resources (COR) theory. The JDR theory states that a balance between all possible work demands and all possible resources leads to the health and wellbeing of the worker (Demerouti et al., 2001; Schaufeli and Taris, 2014). The JDR model distinguishes two underlying processes, namely the health-limiting or energetic process and the motivational process (Demerouti et al., 2001). The health-limiting process arises at work demands and is mainly focused on exhaustion. The motivational process arises at work resources and is mainly focused on depersonalization. The JDR theory explains that there is an interaction between job demands and work resources, but the theory does not explain why, this requires additional psychological theories, such as the COR theory (Hobfoll, 2002). The central assumption of the COR theory is that people strive to preserve and protect the things they value, the resources. According to the COR theory, stress then arises when critical resources are threatened with loss, when they are lost, or when they are insufficiently regained after intensive effort (Hobfoll, 2002).

To empirically assess occupational stress, researchers usually study the effect of work demands and work resources on, for example, burnout. This research is usually done through questionnaires with generic job demands and job resources (Brough et al., 2009). The use of generic job demands and job resources is primarily driven by the wide range of occupations to which they apply so that the outcomes of studies can be compared between themselves but also with normative data (Brough et al., 2009).

The inclusion of occupation-specific job demands and job resources focuses on the assessment of the specific work context. This not only increases the ecological validity of the assessment but also improves the description of the work environment (e.g., Brough and Frame, 2004). The assessment of occupation-specific job demands and job resources is also beneficial for the development of targeted interventions (e.g., O’Driscoll et al., 2009). Finally, it is often overlooked by researchers that examining occupation-specific job demands and job resources simultaneously, yields additional explained variance over the generic job demands and job resources (e.g., Kop et al., 1999).

Occupational job demands and job resources have now been examined for various professions, for example, police officers (Juniper et al., 2010), teachers (Van der Doef and Maes, 2002; Bakker et al., 2007), firefighters (Tuckey and Hayward, 2011), prison staff (Brough and Biggs, 2015), nurses (Sundin et al., 2011), dentists (Hakanen et al., 2005), and GPs (Bakker et al., 2000a; Cathebras et al., 2004).

Traditionally, stressors or job demands of GPs have been related to the emotionally charged doctor-patient relationship. The demanding nature of the doctor-patient relationship has once been considered the root cause of burnout (Maslach, 1978). In the meantime, other than emotionally bound occupation-specific job demands for GPs have been identified, for example, high workload, and administrative job demands but also conflicts with social work (Cathebras et al., 2004).

Given the above-described added value of research into occupation-specific job demands and job resources, and the importance for a society of a well-functioning and healthy primary healthcare system, in particular for GPs, it is surprising that so little is known about occupation-specific job demands and job resources of GPs. The amount of academic literature on the determinants of burnout is impressive. However, the amount of literature on the determinants of burnout among GPs is much less. The amount of literature on occupation-specific determinants of burnout among GPs is very limited.

Therefore, the present study aims to gain insight into the existing academic literature on generic and occupation-specific determinants of burnout among GPs. We aim to contribute insight into and knowledge of the determinants that govern the mental energy household of GPs and that often lead to burnout. Moreover, by emphasizing occupation-specific determinants, we want to contribute to the ecological validity of this study.

We propose the following research question for the present study:

In the present study, we used the systematic quantitative literature review (Pickering and Byrne, 2014). The method is systematic because of how studies are initially identified for inclusion, i.e., explicit and reproducible. The method quantifies where there is research but also where there are gaps. The technique can be used for all types of studies, qualitative, quantitative and mixed-method. The method bridges the gap between traditional narrative review methods and meta-analyses (Pickering and Byrne, 2014).

The present study was conducted in four phases. First, the literature search was conducted over the period 1970–2021. In the first phase, existing literature on determinants of burnout among GPs was collected. The beginning of the search marks a period when the term burnout was introduced in the scientific literature by Freudenberger (1974) and Maslach (1976). The search was performed with the search terms “burnout,” “general practice,” “GP,” “family practitioner,” “family doctor,” “family physician,” “primary care,” “primary care practitioner,” “primary care doctor,” “primary care physician,” and combinations of these search terms. The following search engines were used: Open University digital library interface, including Web of Science, Embase, PsycINFO, Google Scholar, PubMed, and Mendeley.

The following inclusion criteria were used in the second phase, the screening phase, to narrow the natural result. An abstract of an identified study must be available for screening purposes, and the language of the study must be Dutch or English. Type of study means any study that addresses the nature of a determinant and its relationship with burnout, i.e., no intervention studies. In 2016, a comparative study on GPs was published by Schäfer et al. (2016), which included 34 nationalities because they have comparable and well-documented primary care. It concerns 31 European countries, Canada, Australia, and New Zealand. The inclusion criterion “Nationality” in the present study is based on the said study by Schäfer et al. (2016), i.e., we used the 34 nationalities as a “Nationality” inclusion criterion. A separate inclusion criterion, the “Language” criterion, which is separate from the “Nationality” criterion, is the language in which the study is published and must be English or Dutch.

The USA is a rich source of scientific literature on primary care. This literature has provided evidence that a well-developed primary healthcare system offers benefits in terms of better coordination, continuity of care and possibilities for cost control. However, part of this research has limited relevance to the European situation (Kringos et al., 2010). Therefore, in the current study, we have limited ourselves to the European situation, and a comparable situation, i.e., the 34 nationalities mentioned in the study by Schäfer et al. (2016).

After identifying duplicate studies and screening on title and abstract with the above inclusion criteria, phase three followed.

In the third phase, the full text of the remaining studies was reviewed and assessed against the same inclusion criteria as in the previous screening phase. Naturally, the full text of the study had to be available. The remaining studies were used for further analysis in the next phase.

In the fourth and final phase, the remaining studies were coded according to a range of variables: authors, year of publication, source journal, area of focus of the journal, study location, type of study, psychological theories used, and outcomes of the studies. Finally, the results of the studies were ranked according to generic and occupation-specific job demands and job resources, job characteristics, socio-demographic data, personal characteristics and a group of miscellaneous. The miscellaneous group consists of determinants that cannot be classified into other categories.

At several points in the procedure, decisions have to be made that are open to discussion. Making those decisions by only one reviewer increases the risk of selection bias. For example concerning the inclusion criteria or the distinction between generic and occupation-specific work demands and resources. Concerning the latter, the distinction between generic and occupation-specific work demands and resources is made by several authors who have used a profession-specific measurement scale, for example, Bakker et al. (2007) concerning teachers and Hakanen et al. (2005) concerning dentists. Therefore, in the present study, we have taken the measurement scale for job-specific work demands and resources from our previous study (Verhoef et al., 2021) as a starting point and the categories in the current study have been adjusted accordingly.

To better compare and interpret the results, two effect sizes were calculated, namely, the frequency effect size (FES) and the intensity effect size (IES) (Onwuegbuzie, 2003; Sandelowski et al., 2007). The FES is used to determine the relative magnitude of the abstracted findings and is calculated by dividing the number of studies that found a result by the total number of studies. The IES represents the impact of a study and is calculated by dividing the number of findings in that study by the total number of results in all studies.

The outcomes of the four screening phases are indicated in the process diagram in Figure 1. The first rough search identified 9,508 studies, of which 60 studies ultimately remained for further analysis. The bibliographical details of these studies (authors and journals) are presented in Table 1. The methodological details of these studies are presented in Table 1. The outcomes of the quality assessments of all studies are presented in Table 1 and are denoted as strong, moderate or weak.

The findings study is explained in detail, including the FES, in Tables 2–8. Table 1 lists the locations of the 60 studies reviewed. Of the 22 different nationalities, the Netherlands and England produced the highest number of studies, namely 9. The psychological theories used for the theoretical underpinning in the studies examined indicate a study’s solidity. Note, that the limited space in a journal can also cause the absence of a theory! Several methodological parameters have been identified and included in Table 1. Of the 24 distinct psychological theories, the COR theory (Hobfoll, 2002) and the JDR theory (Demerouti et al., 2001) have the highest frequencies with 7 (29.1%) and 6 (25%), respectively. Of the 60 studies examined, 65% do not appear to use a supporting psychological theory (see Table 1). Furthermore, of the 60 studies examined, only 10% appear to have formulated and tested a hypothesis and 55% a research question that guides the research.

The results of an additional critical appraisal of the quality of the studies are presented as part of Table 1. The impact of a study is represented by its IES and is shown in Table 8 for all 60 studies examined. The mean IES = 1.99 (range 0.35–13.99). This parameter shows that the meta-analysis of Lee and Ashforth (1996) showed an IES = 13.99 and had thus the greatest impact on the current study. In addition, the qualitative study by Clough et al. (2020) (IES = 5.59) has a major impact on the results of the current study.

Finally, a Fisher exact test was conducted to test the null hypothesis that there is no difference between the FES in the strong studies (n = 30) and the moderate studies (n = 30) (see Tables 2–6) (Freeman and Julious, 2007). Tables 2–7 demonstrate that of the 75 items in total, only two had a significant p-value. Those items were Stressful events, Table 2, Job demands, had a p-value of 0.0001, and the other item was Patient care, Table 3, Job resources, had a p-value of 0.0007.

It is essential to gain a good understanding of the nature and strength of generic work requirements and resources. However, to increase ecological validity, it is also important to identify and further explore occupationally specific work demands and resources (Sundin et al., 2011; Brough and Biggs, 2015).

In the present study, we identified a substantial number of generic and some occupation-specific job requirements and resources for GPs (see Tables 2, 3). To better evaluate the results of a systematic review and the studies used for it, we used two effect measures. The FES of a finding, for example, a job requirement, is considered a measure of the strength of evidence for that finding (Sandelowski et al., 2007). The IES is regarded as a measure of the influence of the study in question on the final result of the ongoing study (see Table 8). For example, a study with an IES of 10 had much more influence on the final result of the ongoing study than a study with an IES of 2. In the current study, we selected 4 studies with relatively high IES, from Table 8, that is, Abrahams et al. (2013) (IES = 5.59), Clough et al. (2020) (IES = 5.59), Croxson et al. (2017) (IES = 4.90), and Cheshire et al. (2017) (IES = 4.90).

The quantitative study by Abrahams et al. (2013) aimed to investigate sex and gender differences in the prevalence and determinants of burnout among 349 Dutch GPs. The authors conclude that there is a relationship between gender and burnout that is mediated by work pressure, social support from the partner and a depressive response pattern. While these are highly recognizable work requirements and resources, they are all generic.

The mixed methods study by Clough et al. (2020) aimed to compare stressors leading to burnout and protective factors among 252 Australian GPs. The authors found 12 stressors (e.g., workload and time management) and nine protective factors (e.g., clinical interest). Except for this last resource, clinical interest, the other stressors and protective factors (resources) mentioned are generic.

The qualitative study by Croxson et al. (2017) aimed to investigate perceptions and attitudes toward workload among 34 UK GPs. The reasons cited for an increased workload are increased needs and expectations of patients, a changed relationship between primary and secondary healthcare and bureaucracy. It is primarily the workload balance within the practice that erodes the resource continuity of care. They, therefore, conclude that management of patient expectations and reduction of bureaucracy should be a high priority. It is the increased expectations of patients and the changed relationship between primary and secondary healthcare that can be regarded as occupation-specific work demands and continuity of care as a profession-specific resource.

Finally, the qualitative study by Cheshire et al. (2017) among 22 British GPs aimed to explore experiences with workplace challenges, stressors and coping. It was experienced as very stressful by the participants to feel bound by the moral implications of good doctors, resulting in anxiety, sleep disorders and stress. As individuals, they felt powerless to do anything about it. Various generic work requirements are mentioned (e.g., administration, workload, complexity of the work) as well as occupation-specific work requirements (e.g., demanding patients). Continuity of care is mentioned as an important occupational resource.

All in all, it is mainly generic work demands and resources that are mentioned in the results of the above four studies and relatively few profession-specific work demands and resources. Moreover, the generic job demands with an FES_avg of 7.33 seem more important than the occupation-specific job demands with an FES_avg of 5.27 (see Table 2). Further exploration of job-specific job demands and resources among GPs seems highly desirable. The situation is reversed for the resources, the generic resources with an FES_avg = 5.38 seem much less important than the profession-specific resources with a much higher FES_avg = 11.90 (see Table 3).

The fact that the share of England with nine studies and the Netherlands with nine studies have the largest share in the current study may mean that primary healthcare is changing in both countries.

Finally, Table 1, containing several methodological parameters, raises concerns about the methodological quality of studies. That only 35% of the studies use a supportive psychological theory is low. That only 10% of the studies generate a hypothesis is too low. That 55% of the studies use a directional research question seems average. However, to overcome this concern, we paid additional attention to the presence of the research question, supporting theory and hypotheses in the included studies and whether the quality assessments accounted for these items. Subsequently, we performed Fisher exact tests. From the results, we learned that only two out of 75 p-values were significant. We concluded that there was insufficient evidence to reject the null hypothesis. In other words, the quality of the studies did not affect the outcomes of the present study. We consider this finding a strength of this study.

GPs are independent healthcare professionals in small-scale working-groups outside the hospital. As a medical specialist, the GP has a broad medical knowledge of many diseases and disorders and is the first contact for people with various physical or mental complaints (Verstappen and Hobma, 2002). GPs are considered gatekeepers for hospital access (Van der Zee et al., 2004). The GP can also offer a solution if you have problems at home, school, or work. At least as necessary is GP care for the elderly in an aging society (Verlee et al., 2017; Grol et al., 2022). It is essential that a GP knows their patients and has time for a good conversation (van Ballegooijen et al., 2021). Fewer referrals to hospitals and less medication are also crucial for better health, to which the GP can make an excellent contribution (Gillam, 2022). During the current COVID-19 pandemic, the importance of good primary care has become apparent, as we can also learn from UK experiences (Mitchell et al., 2021). Some GPs vaccinate but also motivate people to get vaccinated. By taking over hospital care by the GP, patients can be discharged earlier and thus reduce the pressure on hospitals (Kloos, 2020; Cals et al., 2021).

If GPs are so important and valuable and have such a high risk of burnout (Bakker et al., 2000b; Soler et al., 2008), why is research into the causes and consequences of burnout among GPs so limited in size? The answer to this question, while partly obscure, is likely to be pluralistic. For example, recruitment problems for participants in research projects in primary healthcare have been reported (Bower et al., 2009; Graffy et al., 2009). But also, understaffed academic institutes for primary healthcare, with too few researchers, contribute to the limited literature (Campbell et al., 2015). Therefore, the urgent call in the current study to stimulate and initiate more research into burnout among GPs requires an ultimate effort that is worth it.

The original contributions presented in this study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

RB supervised this project. NV wrote the main body of the manuscript and carried out selection procedures. Both authors carried out extraction and synthesis procedures, read and approved the final manuscript.