The sample was selected by 2Logical (Lisbon, Portugal). The sample was randomly selected from the company’s main database with information regarding all active physicians in Portugal, to create a final representative sample stratified by age, sex, and geographic region (n=549). A phone-based survey was first conducted with this sample to select participants willing to participate in the study. Consenting participants were invited to take a web-based survey during May 4-25, 2020, during the first wave of the COVID-19 pandemic in Portugal (English version available in Multimedia Appendix 1). At the time, 813 of the 22,749 total infected patients were receiving treatment for COVID-19 at the hospital, with 143 patients receiving intensive care [5]. Since Portugal has a total of 24,000 hospital beds available, including 430 beds in the intensive care unit, the health services were not operating at maximum capacity at the time [4,7]. However, Portugal has a low ratio of nurses per inhabitant, which may contribute to the burden of health care professionals during the pandemic [4,7]. Health care professionals from different fields (eg, internal medicine, anesthesiology, and pneumology) were assigned to intensive care services at this time [29]. Moreover, several emergency service units and COVID-19 community dedicated areas were established to treat patients [7]. Outpatient health care centers continued operating by replacing some presential appointments with telephonic or email contacts and by helping with the diagnosis, treatment, and follow-up of patients with COVID-19 [30,31].

Verbal informed consent was obtained from all subjects. All study procedures comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the tenets of the 2008 revision of the Helsinki Declaration of 1975. All procedures were approved by the Ethical Committee for Life Sciences of the University of Minho (Braga, Portugal; approval# 014/2020). Participants were not compensated for their participation in this study (Portuguese law# 21/2014).

The survey assessed information on age, sex, marital status, and geographic region of residence. The geographic region was categorized into low- and high-risk regions, based on the number of COVID-19 cases (high-risk regions were characterized as having ≥1,000 cases).

Data on living conditions were also acquired: if a participant was displaced from his/her regular habitation, type of current housing (apartment or house), house characteristics including the presence of green spaces and a balcony, the number of house habitants, and the presence of infants, children, teenagers, adults, elders, and pets in the current housing.

Additionally, the survey measured the levels of depression, anxiety, and stress with the 21-items Depression, Anxiety, and Stress Scale (DASS-21) and OC symptoms with the Obsessive-Compulsive Inventory-Revised (OCI-R). The DASS-21 scale assesses symptoms experienced in the prior week. This scale has 21 items with the following response options ranging 0=“did not apply to me at all” to 3=“applied to me very much or most of the time.” The total score varies between 0 and 63. Each subscale (depression, anxiety, and stress) has 7 items and a total score ranging 0-21. Severe depression, anxiety, and stress symptoms correspond to scores higher than 10, 7, and 12, respectively [32]. The OCI-R measures OC symptoms in the previous month. This scale has 18 items divided into 6 categories (3 items each): hoarding, checking, ordering, neutralizing, washing, and obsessing. The answer for each item ranges from 0=“not at all” to 4=“extremely.” The total score varies between 0 and 72, and scores above 20 indicate severe symptomatology [33].

Finally, participants were asked if they were working at the frontline and had direct contact with patients with COVID-19 and if they were currently in a quarantine period. Accordingly, participants were divided into two groups: participants working at the frontline during the COVID-19 pandemic (FRONT) and nonfrontline workers (NFRONT).

Statistical analyses were conducted using the JASP software (version 0.12.2.0; JASP Team, University of Amsterdam). P values under .05 were considered significant.

For scalar variables, we assessed the assumptions of normality (the Shapiro–Wilk test) and homogeneity of variances (the Levene test) in each group. Between-group differences in parametric variables were estimated with the 2-tailed independent samples t test. The Mann–Whitney U test was applied for nonparametric variables. For categorical variables, we used the chi-square test to assess differences between groups.

We used multiple linear regression models to investigate which variables explained DASS-21 depression, anxiety, and stress scores and the OCI-R total score. We analyzed the following independent variables: age, sex, group, marital status, geographic region, house type, house green space, house balcony, number of house habitants, house infants, house children, house teenagers, house adults, house elders, and house pets. Since most of the sample was not under quarantine and was not displaced from their regular habitation, these variables were not included in the regression models (Table 1). Normality, linearity, and homoscedasticity assumptions were visually assessed with Q-Q plots and residuals vs predicted plots. The Durbin-Watson value was used to assess residuals correlations, and the tolerance and variance inflation factor values were analyzed to check for multicollinearity.

In this study, we investigated whether the mental health of physicians working at the frontline during the COVID-19 pandemic is affected in comparison to other physicians. We found that frontline physicians have increased levels of stress. Furthermore, female physicians presented higher levels of stress and anxiety. Younger physicians also displayed augmented stress levels. Concurrent with these results, being female and working at the frontline are potential risk factors for stress, and having a house with green space is a potential protective factor for stress and anxiety.

These findings are consistent with those of our previous study on the general Portuguese population, which reported that being male, being older, and living in a house with green space appeared to be beneficial for mental health during the COVID-19 pandemic [10]. Moreover, previous studies have reported that women and younger individuals have a higher risk of mental disorders, including anxiety and mood disorders [34,35]. Furthermore, previous studies have reported negative associations between mental health variables (eg, stress and anxiety) and age among health care workers, indicating that older individuals might be more resilient to the psychological impact of COVID-19 [20,36,37]. Moreover, older physicians might cope better with stressful situations because they have more years of experience. On the other hand, older physicians might have been assigned to lower-exposure wards, considering the higher risk of more severe manifestations of COVID-19 with an older age [29]. Indeed, NFRONT participants were older than their FRONT counterparts in our study. Of note, age was not a significant variable in the stress regression model; this finding should be interpreted with caution. Parallel with our results, several recent studies reported that female medical workers have higher levels of anxiety and stress [13,18,20,21,36-38]. In addition to working outside of the house, women usually have more responsibilities within their household and with childcare than men. Moreover, they rely more on social support as a coping strategy for stress [39-41]. Thus, female physicians may have been exposed to more stressors during the COVID-19 pandemic because they might need to reconcile home, childcare, and work duties while having less social support from family and friends to follow social isolation guidelines, along with less support from spouses and their workplaces. Additionally, as a protective measure, schools and daycare centers for children were closed during the first wave of COVID-19 in Portugal [42], thus aggravating the lack of childcare support for women and possibly affecting their work schedule. On the other hand, in Portugal, more female physicians work in specialties including general and family practice and anesthesiology, internal medicine, and pneumology [43], which involve high contact with patients, while male physicians are more predominant in surgical fields, where several services were canceled during the pandemic [29]. This might also have contributed to the observed levels of anxiety and stress among female physicians. Lastly, having a house with green spaces, such as a garden, might help increase physical activity and provide outdoor relaxation periods, thus contributing to better mental health [44,45]. Thus, the effect observed for this variable might be a surrogate for high-quality leisure time.

Furthermore, in comparison with our previous results [10], frontline physicians presented a higher prevalence of severe OC symptoms (19.5%) than the general population (12.4%). However, we did not observe significant differences between FRONT and NFRONT physicians with respect to OC symptoms. Interestingly, the prevalence of severe OC symptoms in NFRONT physicians (16.4%) was also higher than that in the general population. Thus, COVID-19 might have an impact on OC symptoms among physicians in general, concurrent with previous reports [17,18]. The fear of self-infection or disease transmission among family members or coworkers because of working at health facilities with a high risk of infection may boost excessive washing and cleaning behaviors or contamination-related obsessive thoughts.

Furthermore, frontline workers in our study showed higher stress levels than other physicians and a higher prevalence of severe stress symptoms (11.5%) than the general population (9.3%) [10]. Previous studies have also reported increases in stress [36,37,46], distress [13], and burnout [11,12] symptoms among medical workers during the COVID-19 pandemic. These and other studies suggest that the fear of self-infection and disease transmission to family members, concern or grief for affected coworkers, social distance from family members, lack of proper COVID-19–related training and protective equipment, longer working hours, lack of sleep, and difficult life-or-death decision-making might be factors contributing to increased stress [38,46-51]. Other studies have also observed positive correlations between stress and the fear of infection and between stress and burnout symptoms among health care professionals [52]. Furthermore, previous outbreaks of other infectious diseases (eg, severe acute respiratory syndrome, Middle East respiratory syndrome, and A/H1N1 influenza) were associated with an increase in stress, distress, and burnout symptoms among health care professionals [53,54]. Long-term effects of augmented stress levels may translate to posttraumatic stress disorder and burnout, especially among frontline physicians [47,48,50,51,53,55].

Studies conducted in several countries have reported that health care professionals have elevated levels of anxiety and depression during the COVID-19 pandemic [22,23]. We did not observe evidence of increases in the symptoms of severe anxiety and depression in our sample when compared to a previous study on the general Portuguese population [10], and frontline workers were not different from other physicians in terms of these symptoms. These findings may be explained by the lower number of COVID-19 cases and the subsequent occurrence of the disease in Portugal when compared to countries that were among the first to be affected by this disease.

This study has some limitations. We used a cross-sectional design, which prevents the inference of causality. The selection of participants was not free of putative bias because the physicians who consented to participate in the study might have better mental health indicators or a reduced workload (eg, by not working at the frontline) than the ones who refused to participate. Additionally, our conclusions cannot be applied to other health care professionals such as nurses who have had direct contact with patients with COVID-19. Since few physicians in our sample were in a quarantine period, we could not study the impact of this variable on mental health. Finally, other factors that were not considered in our analysis might have had a significant impact on mental health, including workload [19,21,46,52], history of psychiatric disorders [56], and the use of coping strategies (eg, use of substances, exercise, and mindfulness) [14,46]. Furthermore, we did not determine how long these physicians have been working at the frontline during the COVID-19 pandemic, whether they switched to other wards during the pandemic, whether they received psychological or psychiatric support [57], or whether they were facing difficulties in daily functioning at work or during social activities [58,59]. Thus, our findings should be interpreted with caution in consideration of these limitations.

This is the first study to provide evidence of increased stress levels in Portuguese physicians at the frontline of during the COVID-19 pandemic. We observed that female physicians are more susceptible to stress. Prolonged exposure to COVID-19–related stressors may lead to burnout syndrome. Thus, our results potentially provide essential guidelines for future preventive actions by health care systems, namely the establishment of proper rest periods, control of excessive workloads, the supply of basic needs (eg, personal safety and childcare), development of pandemic-related training programs, and the incorporation of protective measures for mental health (eg, virtual mindfulness-based interventions). Additionally, social contact with family and friends should not be overpowered by occupational demands [47-50].