Physician Burnout in Primary Care during the COVID-19 Pandemic: A Cross-Sectional Study in Portugal

Background Primary care physicians have been present on the frontline during the ongoing pandemic, adding new tasks to already high workloads. Our aim was to evaluate burnout in primary care physicians during the COVID-19 pandemic, as well as associated contributing factors. Methods Cross-sectional study with an online questionnaire disseminated through social media, applying the snowball technique. The target population was primary care physicians working in Portugal during the first outbreak of the COVID-19 pandemic. In addition to sociodemographic data, the questionnaire collected responses to the Copenhagen Burnout Inventory (CBI), the Resilience Scale and the Depression, Anxiety, and Stress Scales (DASS-21). Data were collected from May 9 to June 8, 2020, a period comprising the declaration of a national calamity and then state of emergency, and the subsequent ease of lockdown measures. Levels of burnout in 3 different dimensions (personal, work, and patient-related), resilience, stress, depression, and anxiety were assessed. Logistic regression analyses were conducted to identify factors associated with burnout levels. Results Among the 214 physician respondents, burnout levels were high in the 3 dimensions. A strong association was found between gender, years of professional experience, depression and anxiety, and burnout levels. Conclusions Physician burnout in primary care is high and has increased during the pandemic. More studies are needed in the long term to provide a comprehensive assessment of COVID-19’simpact on burnout levels and how to best approach and mitigate it during such unprecedented times.

institutions. Burnout was already a problem in Portugal prior to COVID-19, with 43.6% of Portuguese physicians reporting high levels of burnout. 3 As the COVID-19 pandemic increases pressure on healthcare systems worldwide, physician burnout levels are also expected to rise. With a total of 32 500 confirmed cases and 1410 deaths from COVID-19 as of May 31, 4 the pandemic has brought new stressors possibly contributing to physician burnout, such as fears of becoming infected or infecting a relative, lack of appropriate personal protective equipment, lack of access to up-to-date information and communication, limited time with family and friends, reductions in economic revenue, and increased demands from childcare and household tasks. [5][6][7] Around the world, general practitioners and family doctors were on the frontline fighting the spread of infection. In the Portuguese health care system, the family doctor acts as a gatekeeper; while maintaining this role during the pandemic, they had to adjust and add new activities to their daily schedules (such as the diagnosis and monitoring of infected patients, prescribing their respective sick leaves, and adapting to the use of telemedicine for most consultations) while keeping up-to-date with new evidence and guidelines on the novel coronavirus, all in addition to their normal preventive and curative medical activity. [8][9][10] In a study to understand the impact of COVID-19 around the world that included a total of 2707 participants from 60 countries, 51% of HCWs reported burnout. 11 In a recent study of 2008 HCWs in Portugal during the pandemic, where burnout was measured by the validated Portuguese version of the Copenhagen Burnout Inventory (CBI), 12 high burnout levels were reported by 52.5% of participants for personal burnout, 53.1% for work-related burnout, and 35.4% for patient-related burnout. 13 Another work focusing on the levels of burnout in HCW in Italy during the COVID-19 pandemic used the Maslach Burnout Inventory (MBI), whose subscales assess levels of emotional exhaustion, depersonalization and personal accomplishment. This study revealed high levels of emotional exhaustion in 41% and high levels of depersonalization in 27%. 14 Wu et al. specifically compared levels of burnout in frontline vs. other HCWs in Wuhan, China to find that the first had significantly lower levels of burnout and were less worried about becoming ill compared to those in the "usual ward" group, 15 highlighting that all HCWs should be considered when designing well-being policies, irrespective of their position.
Kristensen et al. developed the CBI based in the concept that burnout is not just fatigue or exhaustion, it is the attribution of these feelings to specific domains of one's life, as a result CBI is a tool with 3 subscales: personal, work, and client-related burnout. The personal burnout subscale measures feelings of physical, emotional, and mental fatigue and exhaustion. The work-related burnout subscale assesses the symptoms that respondents attribute to work. The client-related burnout subscale describes the aforementioned feelings that respondents attribute to their work with clients. 16 To our knowledge, no study to date has focused on assessing primary care physicians' burnout levels in addition to stress, depression, anxiety, and resilience during the COVID-19 outbreak, nor the associated contributing factors.

Study Design
Cross-sectional study with an online questionnaire disseminated through social networks, spread using the snowball technique.

Participants
The study population was Portuguese-speaking primary care physicians working in Portugal during the COVID-19 pandemic.

Data Collection and Variables
Data were collected from May 9 to June 8, 2020, a period comprising the declaration of a national calamity, the subsequent state of emergency, and the ease of lockdown measures that followed. The questionnaire, developed using Google ® Forms, was spread as a web link through social networks and institutional mailing lists. All participants gave their informed consent to participate in this study and anonymity was assured. The first section of the questionnaire addressed sociodemographic data. Psychological variables collected in the following sections included the Copenhagen Burnout Inventory (CBI), the Resilience Scale and the Depression, Anxiety, and Stress Scale (DASS-21).
The validated Portuguese version of the CBI was used to measure burnout. 12 This is a 19-item scale with 3 subscales: personal (6 items), work (7 items), and patient-related burnout (6 items). Each item is answered on a 5-point Likert scale (always/to a very high degree = 100, often/to a high degree = 75, sometimes/somewhat = 50, seldom/to a low degree = 25, and never/almost never/to a very low degree = 0). The score for each subscale is the average of item scores within the subscale, ranging from 0 to 100. A score of 50 or higher in any of the subscales was considered high-level burnout. 12,16 These subscales demonstrated high internal consistency (original version: α = .84; Portuguese version: α = .86, where α is the Cronbach's alpha). 12,16 In the current study, α was .91, .89, and .89 for personal burnout, work-related burnout, and patient-related burnout, respectively.
The Resilience Scale is composed of 25 items, each with a 7-point Likert response scale ranging from 1 (strongly disagree) to 7 (strongly agree). 17 The total score corresponds to the sum of the 25 items, thus ranging from 25 to 175. A score below 121 is considered "low resilience," 121 to 145 points "moderate resilience" and higher than 145 "high resilience." The validated Portuguese version presented high internal consistency, with α = .89 18 and α = .94 in a medical sample 19 and α = .95 in our study.

Data Analysis
Data from the survey were exported to a Microsoft Excel ® 2016 (USA) spreadsheet and statistical analyses were performed using SPSS ® Statistics (version 26.0; SPSS Inc., Chicago, Illinois, USA).
Categorical variables were described using absolute and relative frequencies, n (%). Quantitative variables with a normal distribution were described with mean ( x ), standard deviation (SD), minimum (min), and maximum (max) values. Those without a normal distribution were described using medians (med) and interquartile intervals [Q1; Q3] (in these cases, means and standard deviations were also included for comparison purposes). Normal distribution was assessed with visual inspection of the histograms.
The internal consistency of each subscale was assessed using Cronbach's alpha (α), where a value higher than .7 was considered acceptable. 22 Separate multiple logistic regressions were performed for personal, work-related, and patient-related burnout. The independent variables to include in each multiple logistic regression were chosen by conducting simple logistic regressions with each variable. All variables correlating with the outcomes with P ≤ .20 in the simple logistic regression were included in the multiple logistic regression analyses. Only the significant variables were maintained in the final multivariate models for each burnout dimension. To convey the results of logistic regressions odds ratios (OR), 95% confidence intervals [95% CI], and P-values are presented. The final models were evaluated using the Hosmer & Lemeshow test of adequate fit. P ≤ .05 was considered significant.

Sample Size
Considering there are 8000 Portuguese physicians in a Primary Healthcare specialty (either Family Medicine or Public Health) we planned a sample size between 192 and 367 respondents, considering the most conservative scenario (a proportion of 50%), for a level of confidence of 95% and an error margin between 5% and 7%.

Results
Participants 225 primary care physicians answered the questionnaires, but 11 were excluded from the analysis for being either retired or absent from work due to disease or leave. Therefore, a total of 214 respondents were considered (80.8% female, mean age 38.6 years old). Of these 214 respondents, 95.3% were working on the frontline (defined as those indicating they worked face to face, full-time, or part-time) during the pandemic. The sociodemographic characteristics of the participants are described in Table 1.

Burnout Dimensions, Resilience, Depression, Anxiety, and Stress
High levels of burnout were found for the 3 dimensions: 65.9% for personal burnout, 68.7% for work-related burnout, and 54.7% for patient-related burnout ( Figure 1 and Table 2). Resilience was high or moderate in 77.1% of the sample.

Logistic Regression Analysis
Results for the multiple logistic regression analyses for personal, work-related, and patient-related burnout are presented in Tables 3 to 5. Being a female was significantly associated with higher odds of patient-related burnout (OR = 2.57;95% CI = [1.17; 5.65]; P = .019). Having worked for 6 to15 years was also significantly associated with higher odds of patientrelated burnout (OR = 3.12;95% CI = [1.53; 6.34]; P = .002) compared to those with 5 or fewer years of practice. A reduction in monthly income was significantly and inversely correlated with patient-related burnout.

Summary of Main Findings
This is to our knowledge the first study assessing burnout in primary care physicians during the ongoing COVID-19 pandemic. Our findings demonstrate a significant burden of burnout, anxiety, depression, and stress during these unprecedented times in the context of primary care. A strong association was found between gender, years of professional experience, depression and anxiety, and burnout levels.
Of note, an inverse correlation was found between patient-related burnout and reduction in monthly income; we hypothesize this may be due to less contact with patients and a lesser workload in the cases where a reduction in salary occurred.

Comparison with Existing Literature
We found a higher prevalence of burnout compared to pre-COVID-19 levels, which is in line with other recent similar studies. 23 In a 2016 Portuguese study of burnout among healthcare professionals, 43.6% of doctors were found to have high burnout, although comparison of this figure with our 55% to 69% prevalence rate should be made with caution, as the former study measured burnout with the Maslach Burnout Inventory. 3 In a multinational, cross-sectional study of 3537 healthcare workers, 67% screened positive for burnout; gender, anxiety, and depression were significant determinants for burnout, in line with our results. 24 A subgroup analysis in a systematic review and meta-analysis of depression, anxiety, and insomnia among healthcare workers during the COVID-19 pandemic revealed gender differences; according to our findings, female workers reported higher levels of affective symptoms. 25 A 2008 study assessing burnout in family doctors found that approximately two thirds scored high for burnout in any dimension, although direct comparison with our findings is not possible due to the use of different scales. 26 Another study assessing burnout during the pandemic found that younger age and being female were independent determinants of burnout, similar to our results. 27 In Portugal, 6 to 15 years of experience in primary care coincides with the first years of practice as a specialist and, frequently, taking on a new role as parent; therefore, we can hypothesize that those factors contribute to the observed differences, although having children ≤12 years was not significantly associated with burnout in our study. Additionally, a previous Portuguese study found that healthcare professionals with more years on the job were less affected by burnout. 3 Our findings are also consistent with the results of a study evaluating the prevalence of burnout and its associations with the work environment among hospital physicians in  Lithuania, which used the CBI scale and found a significative inverse relationship between work-and patient-related burnout and length of employment and job control (assessed using the Job Content Questionnaire). 28

Strengths and Limitations
The use of the CBI scale to assess burnout is one of the strengths of our study: in addition to its excellent psychometric properties and validation for the Portuguese population, we found this scale to be particularly appropriate for the population being studied, since the CBI is composed of 3 different domains that also reflect primary care physicians' daily activities.
Having used a convenience sample, our results may not be generalized to other populations and contexts. Additionally, we cannot rule out that the physicians who decided to answer our survey might be different from those who did not respond, nor can a self-reporting bias be excluded. Our sample consisted of 80.8% female physicians, which could suggest a gender-biased response, which may reflect the female predominance in family ; . χ HL P ; . χ HL P 2 5 8701 122 ; . medicine in Portugal: in 2019, 62.2% of family doctors were female, with 73.2% females in the 36 to 40 years age group. 29 The questionnaire may not address other important contributors to burnout. Notably, it is not possible to establish a causal relationship between the pandemic and our findings.

Implications for Clinical Practice and Research
Burnout has negative impacts on physicians, patients, and healthcare organizations. 30 Our findings reinforce that strategies to counteract physician burnout during a pandemic need to be further investigated. Our workgroup suggests next steps should include, at an organizational level, involving physicians in designing guidelines and contingency plans and also in implementing physician's access to feedback channels. To feel better prepared, they should receive rapid basic training and have the opportunity to talk to experts. A supportive network should be created, including childcare, transportation, and lodging. 7 Emotion management strategies and self-care should also be endorsed, encompassing rest, work breaks, sleep, shift work, fatigue, and healthy lifestyle behaviors. 31 Physicians with depressive and anxious symptoms are at increased risk of personal and work-related burnout; therefore, specific interventions should be aimed at identifying and helping this group. Specific programs to prevent burnout should also be implemented for physicians just starting their careers, such as coping and self-care strategies for medical residents. More studies are needed in the long term to provide a comprehensive assessment of the impact of the COVID-19 pandemic and to further evaluate it as a contributing cause for burnout, for instance, in a study using a sequential mix-method approach.

Author Contributions
SB, AT, LC, MC, CS, AR, and ID designed the study. CS, AR, and ID performed the data collection. AT and LC analyzed the data. SB interpreted the data supported by AT, LC, MC, CS, and ID. SB and MC drafted the manuscript and SB wrote the final manuscript. All authors approved the final manuscript after revising it critically.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. ; . χ HL P 2 7 7820 349 ( ) = = . ; .