Doctor, are you healthy? A cross-sectional investigation of oncologist burnout, depression, and anxiety and an investigation of their associated factors

The Barretos Cancer Hospital (BCH) is located in Barretos – SP (Brazil), a city with approximately 110,000 inhabitants. It is currently one of the largest cancer hospitals in Brazil, providing free care for 10,000 new cancer cases annually. BCH is an assistential, teaching, and research institution.

The Research Ethics Committee of HCB approved the study protocol (CEP/HCB no. 1.091.484/2015). Participants indicated their agreement to participate in the study via the electronic informed consent included in the survey form.

A cross-sectional study design with online surveys was employed.

During the research period, 323 physicians (i.e., staff, residents, and fellows) worked in the hospital and were invited to participate in the study.

Estimates suggest that approximately 60% of the population of physicians who work in hospitals present with burnout [12, 13]. Accepting an estimate of absolute precision (i.e., how close the estimate is to the true value) of 10% and a level of significance of 1%, the minimum estimated sample size was 160 physicians [14].

The MBI is composed of 22 questions answered using a seven-point Likert scale ranging from 0 (never) to 6 (every day). Of the 22 questions, nine evaluated EE, five evaluated DP, and eight evaluated PF. The classification for each dimension is given by the sum of their respective affirmations, making it possible to highlight them in low, moderate, or high levels. The PF dimension was reverse scored [15, 16]. The following cutoff values were used: EE (low level, ≤ 18; intermediate level, 19–26; high level, ≥ 27); DP (low level, ≤ 5; intermediate level, 6–9; high level, ≥ 10); and PF (low level, ≥ 40; intermediate level 39–34; high level ≤ 33). Burnout was diagnosed when high levels of EE, DP, or both were found [17]. We paid for the rights to use the MBI, and its use was duly authorized (Mind Garden, Inc., http://​www.​mindgarden.​com/).

The HADS [18] is composed of seven items related to anxiety symptoms and seven related to the depression symptoms, totaling 14 items. All of the items are answered using a four-point Likert scale. For each dimension (anxiety/depression), cut-off scores are established for “possible cases” or “probable cases”. These scores are calculated based on the sum of the questions used to evaluate the anxiety and depression domains. In this study, scores ≥11 were considered positive for both the anxiety and depression domains.

The physician characteristics potentially related to burnout, anxiety, and depression were included in a questionnaire developed specifically for the present study. This questionnaire was created after meetings among the authors and a review of the literature. These characteristics were grouped into four main categories: (1) personal characteristics of physicians that are innate or difficult to modify; (2) physical or objective workplace characteristics; (3) the work-related stressors perceived by physicians; and (4) outside-of-work characteristics such as leisure, religiosity/spirituality, physical activity, and family relationships (Additional file 1). Content validation was conducted with 10 physicians from different departments using cognitive debriefing and think-aloud method. All the questions were adequately understood. Two items were modified after suggestions; both suggestions were to include examples between parenthesis (regarding leisure and physical activity). An Expert Committee was formed to analyze the development of the questionnaire and the results of the pilot testing. In addition to the authors of the study, the committee was composed of one physician (MD, PhD), two experts in questionnaire validation and a psychologist. They analyzed all responses and considered the questionnaire valid to be used in the present survey. The translated English version of the questionnaire is shown in Additional file 2.

To apply the instruments, the web-based program SurveyMonkey® (https://​pt.​surveymonkey.​com) was used. The doctors received an e-mail explaining the research and a link that provided access to the informed consent and the survey. They received three reminder e-mails about the survey each week for 3 weeks. Unanswered e-mails were considered refusal to participate.

Variables were individually compared based on diagnoses of burnout (yes/no), depression (yes/no), and anxiety (yes/no). Categorical and continuous variables were analyzed using Fisher’s exact test and Mann-Whitney U test, respectively. Variables associated with p < 0.10 in the univariate analyses were included in blocks in a hierarchical binary logistic regression model to identify the predictors of burnout, depression, and anxiety. The blocks were divided a priori into (1) personal characteristics of the physicians; (2) workplace characteristics; (3) workplace-related stressors perceived by the physicians; and (4) outside-of-work characteristics. The difference in the measurement of error (− 2 log likelihood) between different blocks is the block χ². Statistical tests for the estimated models (model χ²) and block within each model were described. In addition, we assessed the amount of variance (%) explained by the logistic models using Nagelkerke’s R² parameter [19]. A two-tailed p-value of < 0.05 was considered significant. All statistical analyses were conducted using SPSS for Windows version 21 (SPSS, Inc., Chicago, IL, USA).

A total of 323 physicians were invited to participate in this study; 237 accessed the research link, and four chose not to participate after reading the electronic informed consent. Of the 233 responders, the data for seven were not analyzed because those doctors did not complete at least the MBI. Thus, the final sample was composed of 227 physicians (response rate = 70.5%). A CONSORT flow diagram is described in Additional file 3.

The median age of the physicians was 34 years old (25th percentile [p25]–p75 = 30–40 years old). Most physicians (n = 123, 54.2%) received more than BRL 20,000 annually, were married (n = 140, 61.7%), and had no children (n = 130, 57.3%). A total of 139 (61.2%) were staff members, and 88 (38.8%) were residents or fellows. In total, 88 (38.8%) were clinicians, 63 (27.8%) were surgeons, 57 (25.1%) worked in the diagnostic sector, 11 (4.5%) were emergency or Intensive Care Unit (ICU) physicians, and eight (3.5%) were anesthesiologists (Additional file 4).

In total, 132 (58.1%, 95% CIs = 51.5–64.3%) physicians were identified as having burnout (i.e., EE, high DP, or both), 28 (12.3%, 95% CIs = 8.0–17.2%) had depression (HADS-D ≥ 11), and 44 (19.4%, 95% CIs = 14.6–24.8%) had anxiety (HADS-A ≥ 11). Regarding the MBI domains, 95 (41.9%, 95% CIs = 35.4–48.7%), 85 (37.6%, 95% CIs = 31.0–44.2%), and 115 (50.9%, 95% CIs = 44.2–57.1%) physicians were considered as having high EE, high DP, and low PF, respectively (Fig. 1).

The physicians were further categorized according to the median time of practice in the hospital (≤2 years vs. > 2 years). Physicians with less than 2 years of practice in the hospital presented a higher number of burnout compared to physicians with longer practice times (64.6% vs. 51.3%, p = 0.045). When analyzed by the burnout domain scores, an statistical significant difference was observed only in relation to EE (≤2 years, high EE = 49.1% vs. > 2 years, high EE = 34.2%, p = 0.012). Additional file 4: Table S2 describes the analysis of burnout in function of years of practice in the present hospital.

All variables with p-values < 0.10 in the univariate analyses (Table 1) were included in hierarchical binary logistic regression models for each analyzed outcome. The variables were included in blocks according to an a priori defined model. The final burnout regression model explained 43.6% of the total variance; blocks 1, 2, 3, and 4 individually explained 10.5%, 7.6%, 22.4%, and 3.1%, respectively. The chance of being diagnosed with burnout was higher among physicians who reported being pessimistic (OR = 7.601, p = 0.036), working in ICU/emergency department [ED] (OR = 6.456, p = 0.063), perceiving a lack of hospital recognition (OR = 2.605, p = 0.018), and excess work (OR = 4.735, p < 0.001). On the other hand, physicians who reported a higher frequency of leisure activities were less likely to be diagnosed with burnout (moderate: OR = 0.321, p = 0.009; frequent/very frequent: OR = 0.362, p = 0.065; Table 2).

The final depression regression model explained 58.4% of the total variance; blocks 1, 2, 3, and 4 individually explained 25.2%, 12.1%, 7.7%, and 13.4%, respectively. Physicians who reported being pessimistic were approximately 10 times more likely to have depression than those who reported being optimistic (OR = 10.729, p = 0.021). Physicians who practiced regular physical activity more than twice per week were less depressed (OR = 0.049, p = 0.006) than those who did not receive regular physical activity (Table 3).

The final anxiety regression model explained 41.2% of the total variance; blocks 1, 2, 3, and 4 individually explained 22.9%, 4.3%, 10.6%, and 3.4%, respectively. Male doctors presented with lower anxiety rates compared with female doctors (OR = 0.410, p = 0.052). The presence of a psychological/psychiatric history of illness (OR = 8.188, p = 0.017) was associated with higher anxiety rates. Regarding the work stressors perceived by physicians, relationship problems with other health professionals (OR = 3.218, p = 0.023) and excess work (OR = 2.396, p = 0.074) were associated with more anxiety (Table 4).