The present study is, to the best of our knowledge, the first to explore the psychosocial burden of GP trainees in three different domains; testing for depression, stress burden and burnout. More than one GP trainee out of ten presented with symptoms of moderate to severe depression. One out of five GP trainees showed a high level of perceived stress (PSQ-20). Few GP trainees showed moderate rates of burnout scoring high in all three dimensions of the MBI score, with the older participants in particular reporting less emotional exhaustion and less depersonalisation.
Mental health of GP trainees in comparison to general population
In our own sample the mean PHQ-9 score was rather high, especially when compared to general population (see Fig. 1). Additionally, the GP trainees had a higher level of perceived stress than the general population: According to Fliege et al. healthy adults show PSQ-scores of 0.33 (SD 0.17) [25]. Kocalevent et al. showed in a more recent study [20] in a representative sample of the German general population (n = 2.552; households and target persons were selected at random) a mean PSQ-score of 0.3 (SD 0.15). In our own sample we found a mean PSQ-20 score of 0.4 (SD 0.18), while n = 40 (19.8%) of GP trainees showed PSQ-20 sum scores ranging between 0.45 and 0.59 being indicative of a moderate level of distress. A total of n = 42 (20.8%) of the trainees even showed a high level of perceived stress with sum scores higher than 0.59. It is worth noticing that Fliege et al. used the original 30-items version to establish the aforementioned PSQ-mean values [25]. However, the shorter version (PSQ-20), which we used for this study, was validated after the reduction from 30 to 20 items [19] and the results of both instruments can be compared after linear transformation of the resulting total score between 0 and 1.
In contrast, it is difficult to compare the mean MBI-scores of our study sample to the general population, as we were using the MBI-HSS, which was specially developed to assess burnout in the human services and not in general population [24]. To detect signs of burnout in other organisational contexts, the MBI-GS was developed [26], which is difficult to compare with the MBI-HSS due to the different number of items. We looked at a study on the experience of burnout in healthcare professionals from the private hospitals in Delhi, India [27], as the authors used the MBI-HSS also for their support staff (i.e. heterogeneous group of staff not working as physicians or nurses, e.g. security, pharmacy, front office, housekeeping, etc.). This subgroup can therefore be considered a “broader” population, although we are aware of the existence of possible regional characteristics. The support staff showed a mean MBI-EE score of 16.62 (SD 3.78), a mean MBI-DP score of 7.82 (SD 1.59) and a mean MBI-PA score of 34.69 (SD 3.67). Thus our study sample shows higher burden in the burnout-dimensions of emotional exhaustion and depersonalisation and only seems to be less burdened in terms of personal accomplishment.
Mental health of GP trainees in comparison to other physicians
As far as we know, there is no study of high quality using the PHQ-9 in a bigger sample of fully-trained GPs. However, Schwenk et al. screened randomly selected practicing physicians, i.e. those being professionally active and not in training, from Michigan for depression using the PHQ-9 [28]. They received a total of n = 1.152 usable responses with those active in primary care being the biggest subgroup. Of this study population 11.3% scored positive for moderate to severe depression. In our sample of GP trainees an almost identical share of 10.9% showed signs of moderate to severe depression.
We are not aware of a study using the PSQ in a bigger sample of (fully-trained) GPs. Bernburg et al. [29] used the PSQ to measure perceived stress in residents from various specialties in German hospitals with an average work experience of 4 years (SD 2 years). Bernburg’s study participants displayed a mean score of 0.48 (SD 0.18) which is a little higher than our mean PSQ-20 score, while the total prevalence of perceived stress at a moderate level was 39.5% and at a high level 17.1%, which is comparable to our own data.
A number of studies [5,6,7] have used the MBI to describe burnout in GPs. Due to its big sample size and its multi-centered approach the so-called EGPRN study is especially well-known [7]. In the EGPRN study almost 3500 questionnaires featuring the MBI-HSS were distributed in 12 European countries with a response rate of 41%. Participants of the EGPRN study had a mean time of 19.2 years since graduation (SD 8.5). The reported mean MBI-EE score in the EGPRN study was 24 (SD 16) as opposed to a slightly lower mean score of 22.3 (SD 9.8) in our own study sample. 33.5% of GP trainees in our study scored high in the MBI-EE dimension (vs. 43% of participants in the EGPRN study). In terms of depersonalisation (MBI-DP) fully-trained GPs showed a mean score of 7 (SD 7). In this dimension the German GP trainees from our own study seemed more burdened than their European colleagues – they had a mean MBI-DP score of 8.1 (SD 5.2). 35.2% of our participants scored high in this dimension (vs. 35.3% in the EGPRN study). The GP trainees in our study showed a mean MBI-PA score of 39.7 (SD 5.6), which was a little higher (= healthier due to the inverse scale) than the mean MBI-PA score reported. Accordingly, only 14.6% of our participants scored high in this last subdimension of the MBI, whereas 32% of their older colleagues from the EGPRN study had a high MBI-PA score. The EGPRN study further showed that 65% of European GPs have at least one high score for burnout and 12% had three high scores. In our study only n = 5 (2.5%) of the GP trainees scored high in all three dimensions of the MBI, whereas n = 96 (47.1%) participants reached no high score at all.
Association of sociodemographic characteristics and mental health
Linear regression models revealed that being a female GP trainee led to a higher PHQ-9 sum score (p < .05), which goes in line with depression statistics showing a gender gap with women being almost twice as likely as men to develop depression during their lifetime [30, 31]. There was no association between sociodemographic characteristics and PSQ-20 sum score, MBI-EE or MBI-PA. Higher age was associated with less depersonalisation in the MBI (p < .05) – another well-known phenomenon that was first described by Maslach and Jackson in 1981 in their MBI validation study with participants from human service occupations [32]. A second regression analysis (without the variable “rotation”) also revealed a negative association between age and MBI-EE. According to the well-known demands-control model by Karasek [33], job control is expected to moderate the relationship between job demands and psychological strain, which could be a possible explanation for these observations, assuming that the higher biological age is accompanied by more work experience, which in turn leads to more control over the work environment. However, a higher age does not necessarily go hand in hand with longer professional experience and it is just as conceivable that the greater life experience (with its greater arsenal of coping strategies) is responsible for this association.
Final thoughts
It becomes very obvious that GP trainees in Germany are more burdened in terms of depression, perceived stress and possibly even burnout than the general population. Alarmingly, they seem to be almost as burdened as fully-trained GPs or hospital doctors of other disciplines, although they are usually protected from the pressures of a fully qualified GP workload. Our results are comparable to a study from Galam et al. (2013), who studied burnout in French GP trainees and demonstrated that it was frequent. We are not aware of any comparable studies from Germany.
Although this descriptive study cannot answer the question of the causes of this strain, it should be mentioned that GP trainees suffer from those stressors specific to their level of training (e.g. being held accountable for their clinical decisions for the very first time, fear of showing imperfection and new level of personal involvement) as well as from those stressors, which typically occur in general practice regardless of the level of work experience (e.g. closeness between GPs and their patients with many challenging situations, feelings of isolation, bureaucratic demands and time pressures) [2].
But what are the implications of the finding that GP trainees are almost as burdened as fully-trained GPs? We believe that the most important things are to be attentive, to train GP trainees in self-perception and the ability to talk about own problems and to increase the recognition in health policies. Before specific (political) measures such as regulations on obligatory adherence to working hours or a reduction in total working hours can be taken, we must improve our understanding of the exact causes of stress in GP trainees. However, there will only be increased efforts to reach this target group if a “cultural change” takes place at the same time, i.e. if the majority of GP educators also recognise that it would be desirable to protect and maintain the psychosocial health of GP trainees in the best possible way. We know that perfectionism which is typical for physicians (not wanting to make mistakes, not showing weakness, not revealing a knowledge gap, etc.), can also contribute to the development of stress and burnout [34], it could for example be an important learning goal of the GP training that true perfection is not possible and that mistakes are part of learning. Professional mentoring as a compulsory part of GP training could make a decisive contribution to the early detection of psychosocial stress in GP trainees, and interventions to promote their physical and mental health are necessary to ensure healthy GPs in the future [2].
Strengths and limitations
We understand this study as a thorough cross-sectional analysis of the psychosocial burden of GP trainees. Our results might therefore be highly useful to all readers enrolled in GP training programmes and those who train young GPs in another setting. However, it is important to be aware of the predictive limitations of all cross-sectional studies: Without future longitudinal data, it is not possible to get a real idea where the detected stress burden comes from, i.e. to establish a true cause and effect relationship. Secondly, the response rate was exceptionally high and quality of data is good. All data was generated from only one GP training programme within one region of Germany and therefore should be handled with care if translated into another context, as prevalence of stress or psychosocial morbidity may vary by country of training.