Residency is a stressful stage of physician training during which residents work for long hours while their responsibility towards the lives of patients simultaneously increases, ideally along with their knowledge and experience. However, this is not always the case: accumulation of responsibilities disproportional to residents' knowledge and experience can be rather common and, combined with physical exhaustion, can set the scene for burnout 123. Resident burnout has been associated with worse health status, substance abuse, tendency towards depression, higher anxiety and suicidal thoughts 124567. Nevertheless, two extensive reviews of the literature by Thomas (2004) 1 and by Prins et al. (2007) 2 demonstrated the surprising scarcity of published data on resident burnout. Particularly among European states, very few reports on resident burnout have thus far been published 58910111213. The syndrome's repercussions do not solely affect the giving end of the health care process (health workers) but the receiving end as well; particularly, the users of health services: patients and society as a whole. Therefore, high levels of burnout in doctors can lead to poor performance, medical errors, decreased quality of medical care and deterioration of confidence in health services 1241214151617.

The Greek situation may be even more complex as a consequence of the peculiarities of the Greek National Health System (NHS) 18192021. More than twenty-five years after its foundation in 1983, the Greek NHS faces a manifold crisis (financial and organizational) and is in need of urgent reforms 202223242526. The medical education system is outdated 2728293031, the European Work Time Directive (EWTD) is still not implemented and Greek junior doctors face the risk of unemployment due to the overwhelming inflation of medical graduates 32. We have previously shown that such discouraging conditions prompt an alarming percentage (70.3%) of Greek medical students to consider entering a medical residency abroad 33.

The organizational difficulties particular to the Greek NHS 181920212223242526 may result in increased work stress, role ambiguity and higher uncertainty about the future prospects of residents. The underdevelopment of Greek integrated primary health care 22 intensifies the workload of residents working in secondary institutions. Furthermore, the widespread use of informal payments within the Greek NHS 23, the growing imbalance between the Greek public and private health sector 25 as well as the perceived inefficiency and inequity of Greek health care 18202425 result in increased patient dissatisfaction which may aggravate conflicts between patients and residents. Higher work-related stress, work role problems, anxiety about the future, work overload and confrontations with patients have previously been associated with burnout 12343536 and may thus render those residents who do choose to train in Greece vulnerable to this syndrome.

The EWTD regulates the working time schedule, the annual paid leave, the characteristics of night shift duties and the time to rest after on-call duty. European Member States were thus obliged to gradually reduce the working hours of residents to no more than 48 hours per week by 1 August 2009 37. This directive was incorporated in Greek Legislation by the presidential edict 76/05 (2005). However, at the time of writing and the latest revision of this manuscript (June 2010), the new working hour limits have yet to be implemented in Greece.

It should be noted that concerns have been expressed regarding the EWTD effects on resident training, especially on surgical residents who may experience reductions in the number of didactic outpatient and operative sessions, as well as discontinuity of presence during clinical and postoperative care 3839404142. Such problems may be more pronounced in Greece due to the notable limitations of structured resident training 27282930. Particularly in the surgical residencies, log-books have not been formally adopted by all training centers, the residency curricula among the various clinics have not been systematized and significant heterogeneity in quality of the various training programs exists 2931. The Greek educational system is thus largely based on the informal practical training of younger residents by those most experienced. Decreasing the active working hours may reduce clinical and operative experience as well as significantly limit the time available for senior residents or specialists to teach younger residents 29.

The overall aim of this study is to report the prevalence of burnout syndrome among Greek medical residents. In addition, Greek residents' satisfaction and opinion on aspects of their training are investigated. This study also reports on Greek residents' attitudes towards the EWTD as well as the demographic and work-related factors that are associated with these opinions. Furthermore, the present study aims to delineate the relationship of burnout scores with Greek residents' demographic characteristics, residency field, future outlook, self-reported training satisfaction and subjective perceptions towards the EWTD.

Eight major Greek hospitals, all with a capacity of over 200 beds, representing three large Greek cities (Athens/Piraeus, Heraklion and Volos) and covering 30 different specialties were selected for this cross-sectional study. The Greek system does not allow official residency training in private hospitals and therefore all institutions surveyed in this study were public. Every resident was given a number according to the official enrolment lists that had been kindly provided by each of the hospitals. Out of 1407 eligible residents in all hospitals, 400 (50 from each hospital) were randomly selected by a computerized method to participate in the sample, a number that represents approximately 4% of the total number of residents training in Greece 43. Attendance was elective and all participants were informed that responses would be anonymous and were blinded to the scope and purpose of the study. The residents were asked to return the completed questionnaires to a sealed box provided in each hospital.

Our survey used a two-stage cluster sampling design and a design effect value of 2.0 was thus adopted. There are currently approximately 10 300 residents training in Greece 43. A minimum sample size of 296 was thus required for this survey to represent the population of Greek residents at the 95% confidence level and a confidence interval of 8. The questionnaire was administered to 400 residents and 311 completed questionnaires were returned and included in this study.

The first 9 questions queried the socio-demographic profile, specialty choices and the respondents' opinion on their future employment. We combined residencies according to their conceptual and occupational relations and formed four different medical 'fields': a. internal medicine (n = 108; paediatrics and neurology were also included in this group) b. surgical specialties (n = 135) c. diagnostic and laboratory specialties (n = 45) and d. general practice (n = 23). The second part of the questionnaire was devoted to medical training quality. These questions were developed based on a list of factors related to residents' satisfaction with the quality of their training programs that were identified by a review of the relevant literature 444546. Residents were asked to rate their opinion of seven statements regarding their residency training on a 5-point Likert scale ranging from "strongly disagree" (score of 1) to "strongly agree" (score of 5). The third part consisted of three multiple choice questions ("Yes", "No" or "I do not know/do not have an opinion") querying residents' opinion on the EWTD and its effects on their personal and social life and their medical training.

The last part of the survey assessed resident burnout using the Maslach Burnout Inventory (MBI) 47 which is the most established and widely used measure of burnout among physicians 1. The MBI is a 22-item questionnaire that evaluates the 3 domains of burnout: emotional exhaustion (measured by 9 items), depersonalization (measured by 5 items) and lack of personal accomplishment (measured by 8 items) 47. Respondents rate the frequency of burnout symptoms on a 7-point Likert-scale ranging from 0 (never) to 6 (daily) and independent scores are calculated for each of the 3 components of burnout. Each component score was categorized as "low", "moderate" or "high" based on previous results on 1104 medical professionals 47. Thus, scores on emotional exhaustion of ≥ 27 were considered high, of 19-26 moderate, and of ≤ 18 low. Depersonalization scores of ≥ 10 were considered high, of 6-9 moderate and of ≤ 5 low. Personal accomplishment scores of ≤ 33 were considered low, of 34-39 moderate and of ≥ 40 high. A high classification on the emotional exhaustion or depersonalization subscales relates to higher professional burnout, while the personal accomplishment subscale has an inverse relationship to burnout with lower personal accomplishment scores indicating higher professional burnout 47. According to previously established criteria 48 considered to be the most accurate in obtaining a dichotomous burnout score 248, resident responses with a high score on the emotional exhaustion subscale combined with a high score on depersonalization or a low score on personal accomplishment were classified as having burnout. The Greek translation of the MBI has been previously validated 49 and used among Greek health care professionals 505152.

Variables were maintained as continuous or categorical according to their original form in the questionnaire. The Kolmogorov-Smirnov test was applied for analysis of variance in all continuous variables. The choice of methods for statistical testing of continuous variables was based on whether the data permitted parametric or non-parametric analysis. Categorical variables were compared using Pearson's chi-square tests. Multivariate analysis was performed using linear regression analysis. Reliability of the three MBI subscales was determined using Cronbach's alpha. A minimum alpha value of 0.70 is generally desirable 53. A P-value of < 0.05 was considered statistically significant. P-values of post hoc paired comparisons, following Kruskal Wallis or chi-square tests, were adjusted with the Bonferroni method.

Table 1 lists the respondents' demographic profile and specialty. A total of 311 completed questionnaires were returned (response rate 77.8%, 172 males, 139 females, median age 32 years). Respondents' demographic characteristics were similar across the 8 hospitals and the four medical fields surveyed (Kruskal-Wallis one-way analyses of variance and Pearson's chi-square tests, all P-values were > 0.1).

Reliability analysis of each MBI subscale showed acceptable alpha values: 0.85 for emotional exhaustion; 0.81 for depersonalization and 0.78 for personal accomplishment. Table 2 shows the distribution of residents in each MBI subscale category. Notably, 154 residents (49.5%) met the criteria for overall burnout and 99 of the respondents (31.8%) satisfied all three subscale scores for high burnout (high emotional exhaustion, high depersonalization and low personal accomplishment).

A considerable portion of respondents (46.6%) were unoptimistic about their future employment prospects as physicians and 29.9% were oblivious to what their future might bring (Table 2). These proportions were not significantly affected by the respondents' medical field or demographic variables (Kruskal-Wallis one-way analyses of variance and Pearson's chi-square tests, P-values > 0.1) with the exception of hospital of residency. More specifically, 32 respondents training at one General Hospital of Crete (66.7%) believed that it will be hard to find or keep a job as a physician in the future, a percentage that is significantly higher compared to responses from all other hospitals (Pearson's chi-square tests, P < 0.05 following adjustment with the Bonferroni method).

As shown in Table 2, a substantial proportion of residents believed that their supervisors are indifferent towards their training and were dissatisfied with the overall quality of training in their clinic and with their continuous medical education (CME) activities. Furthermore, approximately 3 in every 5 Greek residents asserted that their daily residency schedule is hampered by too much 'scutwork' (menial tasks). On the other hand, a sizable proportion of residents (44.7%) did feel that they dedicate a considerable amount of time to their patients' care, yet a notable fraction of these residents (22.3%) were dissatisfied with the patient care responsibilities that they are administered, and only 43.1% felt that the time they invested in patient care was well spent. On a more positive note, more than half of the residents (56.9%) conveyed their satisfaction with the level of support and cooperation with their peers (Table 2). Residents' gender, marital status and parenthood did not significantly modify any of the satisfaction scores (Kruskal-Wallis one-way analyses of variance, P-values > 0.05) while age significantly correlated (Spearman rank correlation), albeit weakly, with only two variables, i.e. older residents were more likely to be dissatisfied with peer interactions (r = -0.121, P = 0.033) and less likely to feel that they are doing too much 'scutwork' (r = -0.142, P = 0.012).

Residents' opinions on their supervisors' interest towards their training, on the merit of their CME activities, on the responsibilities and time given to them for patient care and on the overall quality of training in their clinic were similar between the four medical fields (Kruskal-Wallis one-way analyses of variance, P-values > 0.1). However, surgical residents were less likely to report satisfaction with peer interactions compared to the other groups (Mann-Whitney U tests, P-values were < 0.01 following adjustment with the Bonferroni method). Furthermore, those belonging in the diagnostic and laboratory specialties group were less likely to report excessive 'scutwork' during their residency training compared to those training in the internal medicine, surgical or general practice fields (Mann-Whitney U tests; P-values were < 0.05 following adjustment with the Bonferroni method).

The predominant opinion among medical residents training in Greece is that implementing the EWTD is a necessary step (74.9%) and that the EWTD will be beneficial to their personal and social life (74.3%). In comparison, the percentage of residents who feel that the EWTD will have positive effects on their quality of training was lower (55.9%) but still represented more than half of the total responses (Table 2). Only 32 residents (10.3%) stated that the EWTD will not have beneficial effects on any aspect of their life. Pearson's chi-square tests did not reveal any significant differences (P-values > 0.1) on EWTD attitudes between the four medical fields. Likewise, residents' age, gender, marital status, parenthood and hospital of training did not significantly modify any of the EWTD responses (Pearson's chi-square tests and Kruskal-Wallis one-way analyses of variance, all P-values > 0.05).

Respondents' demographics, residency field and opinions on the EWTD did not correlate significantly with any of the MBI subscales (Spearman rank correlations, Kruskal-Wallis one-way analyses of variance and Mann-Whitney U tests, all P-values > 0.05). Due to Greek hospitals varying in specialties for residency training they offer, these two variables are not independent of each other and may thus confound interpretations of their potential relationship to burnout. Multivariate linear regression analyses were performed to adjust for this potential confounding effect. The linear regression models showed that neither residents' hospital of training nor their residency field affected the MBI subscale scores (P-values > 0.05). Residents' opinion on their future prospects did not affect their depersonalization and personal accomplishments scores (Kruskal-Wallis one-way analyses of variance, all P-values > 0.1). However, residents who believed that it will be difficult to find employment as physicians in the future demonstrated significantly higher emotional exhaustion (Mann-Whitney U tests, P-values were < 0.05 following adjustment with the Bonferroni method), compared to those who were unsure or those who did not have such concerns.

The correlation coefficients (Spearman rank correlations) between each MBI subscale and residents' opinions on aspects of their residency training are presented in Table 3. The strongest correlations noted were those between personal accomplishment and satisfaction with colleague support/teamwork (r = +0.323, P < 0.001) as well as between personal accomplishment and time spent on patient care (r = +0.275, P < 0.001). Interestingly, more time in patient-care activities also resulted in increased emotional exhaustion of residents (r = +0.116, P = 0.041). Despite being statistically significant, the correlation coefficients were weak indicating that the variable pairs do not fully represent the same construct.

The present study's limitations include the cross-sectional design which prevented determination of causality. Furthermore, training satisfaction was documented by questions constructed by the authors based on a review of the relevant literature 444546 and not with a validated instrument. While the response rate was considerably higher than what is typical of physicians 75, the possibility of response biases cannot be excluded. In order to fully protect residents' anonymity, we were unable to collect any further data on non-respondents. Residents with burnout symptoms may have been less willing to complete and return the questionnaire. Thus, the prevalence of overall burnout could range from 39.5% if all 89 non-respondents were not burnt out, to 60.8% if all non-respondents were burnt out. However, even if the first scenario is true, a burnout prevalence of 39.5% is approximately two-fold higher compared to medical residents from other European Union countries 5854.

Our study also has a number of strengths. It is a large, multi-institutional study in a wide range of residency programs, conducted in three geographically distant Greek cities, although it must be noted that our population was biased towards large centres, each a capacity of over 200 beds. Moreover, we applied a validated and widely used instrument to measure resident burnout, which allowed direct comparisons with different populations and with other samples of medical residents and physicians. Our burnout data are consistent with previous reports, and the present study is the first to document Greek residents' training satisfaction, opinions on the EWTD and job insecurity, as well as being the first to correlate these factors with burnout dimensions.