Between November 2020 and February 2021 of the COVID-19 pandemic, a cross-sectional study was carried out with dentists from the state of Bahia, Brazil. All active dentists who accepted the invitation to participate in the study and signed an informed consent form were included. Convenience virtual snowball sampling was employed. A link to a Google Form questionnaire was made available via WhatsApp, email, and Facebook dentist groups. Dentists who responded to the questionnaire were asked to forward the link to other colleagues in order to maximize participation.

Information was collected about a) sociodemographic aspects: sex, age, marital status, skin colour, monthly family income (1 USD = 5.187 Brazilian Reals), amount of time worked as a dentist, and employer economic sector (public, private or both); b) aspects related to COVID-19 and their occupation: diagnosed with COVID-19, afraid of catching COVID-19, avoids seeing patients suspected of having COVID-19, increased alcohol consumption during the pandemic, much higher work demands due to COVID-19, accomplishes all work tasks, feels safe when using personal protective equipment at work, has support from fellow dentists, has support from other health professionals, feels qualified to provide care to COVID-19 cases, and has put their dentistry practice on hold because of the pandemic; c) Quality of life, evaluated by the WHOQoL-BREF questionnaire. This 26-question instrument has four domains: physical, psychological, social relations, and environmental, with scores varying from 4 to 20, where higher scores represent better quality of life [13]. The WHOQoL-BREF has been translated and validated in Brazil [14]. In a sample of 3,574 university staff from a Brazilian university, the questionnaire showed acceptable levels of internal consistency and test-retest reliability [15]; d) Anxiety, measured by the Beck Anxiety Inventory (BAI). The 21 items of this questionnaire are classified into a 4-point Likert scale, with options varying from 0 to 3. Total scores can vary from 0 to 63, with higher total scores indicating more severe anxiety symptoms [16]. This questionnaire has been translated into Brazilian Portuguese and validated, presenting good psychometric properties [17, 18]; and e) Burnout, measured by the Maslach Burnout Inventory (MBI), along three dimensions: Emotional Exhaustion, Depersonalization, and Personal Accomplishment. The 22 self-reported items are classified in Likert-type scales that can vary from 0 to 6. The maximum scores for Emotional Exhaustion, Depersonalization, and Personal Accomplishment are 54, 30, and 40, respectively. Dimension scoring was performed in line with the MBI developers [6].

Dependent variable: Burnout; the scores for each of the three MBI dimensions - Emotional Exhaustion (EE), Depersonalization (DP), and Personal Accomplishment (PA) - were taken as continuous variables. The proportion of dentists with high EE: (score ≥ 30); high DP: (score ≥ 12); and low PA (score ≤ 33) was classified according to published normative cut-off levels [19].

Independent variables: The questions about COVID-19 were originally collected from a Likert-type scale with five options that were subsequently codified as: Frequently and Sometimes = Yes; Neutral, Rarely, and Never = No.

The correlation between age and monthly family income and MBI domain scores was evaluated by calculating the Pearson linear correlation coefficient. Differences between the means of an MBI domain according to each independent variable were measured by independent t-tests. Variables that reached a P level of < 0.20 were selected to compose three multiple linear regression models, where Emotional Exhaustion, Depersonalization, and Personal Accomplishment were the respective dependent variables, using the backward method. Only those variables that reached a P level of < 0.05 remained in the final adjusted models.

In the studentized residuals analysis, cases presenting ± 3.000 standard deviations were classified as outliers and excluded from the model. Collinearity between independent variables was evaluated through tolerance statistics. Values close to zero denote strong collinearity and indicate that the independent variable has a linear performance similar to another variable or variables in the model.

This study used convenience, non-probabilistic sampling, which prevents the use of statistical inference based on P-values or confidence intervals. P-values obtained from t-tests were simply used for selecting variables for the multiple linear regression. The multiple linear regression technique was used to adjust the raw and standardized regression coefficients (also known as the BETA coefficients) obtained for the predictive variables in each model without making statistical inferences. BETA coefficients enable direct comparisons between the variables in the model since they are measured in Z-score scales.

The reliability of the MBI, WHOQoL-BREF, and BAI questionnaires was evaluated through the internal consistency of the domains using Cronbach’s alpha: 0 to 1, where 0 to 0.21 indicates little consistency; 0.21 to 0.40, average; 0.41 to 0.60, moderate; 0.61 to 0.80, substantial; and 0.81 to 1.0, nearly perfect consistency [20].

Statistical data processing was performed using the Statistical Package for the Social Sciences™ - SPSS, version 25 (IBM Corp, Armonk, NY, USA).

All three burnout domains - Emotional Exhaustion, Depersonalization, and Personal Accomplishment, presented satisfactory internal consistency, as revealed by substantial or nearly perfect Cronbach’s alpha values.

In this study, Brazilian dentists had higher means for EE, DP, and PA than dentists from Turkey [7] and lower means for EE and DP than those from Scotland [8].

Some of the studies about burnout among dentists during the COVID-19 pandemic only applied certain items from the complete set recommended by the MBI manual [6] in order to define EE [21] and PA [22]. Unfortunately, the procedures used in these studies preclude comparison with our findings.

Studies of other health professionals used comparable methodologies to measure burnout during the COVID-19 pandemic in the state of Bahia, Brazil. One study evaluated 223 physicians working on the COVID-19 frontline and reported mean levels of EE = 27.5 ± 11.6; DP = 6.4 ± 5.5; low PA = 35.6 ± 7.9 [23]. Another study evaluated 752 multi-professional residents during the early days of the pandemic and reported mean levels of EE = 29.6 ± 11.8; DP = 7.0 ± 5.6; low PA = 30.9 ± 7.3 [24]. The mean levels of burnout reported in these two studies were similar to those found in the dentists in this study: EE = 28.9 ± 8.9; DP = 8.2 ± 6.3; and PA = 28.0 ± 6.2.

Anxiety was the only variable in this study that was strongly associated with all three burnout dimensions. Following this, the association between burnout and the other variables investigated, which are strictly related to chronic workplace stress, need to be discussed.

In this study, the mean scores for several quality of life domains were strongly associated with the mean scores of all three burnout dimensions. Such findings were expected, given the broad and complex facets encompassed by the quality of life domains which, in turn, are intrinsically linked to the chronic workplace stress that defines burnout. A study of 185 physicians without anxiety, working on the COVID-19 frontline in the state of Bahia, Brazil, presented a moderate to high degree of correlation between the four WHOQol domains and EE (-0.37 < r < 0-0.56), DP (-0.28 < r <0.41), and PA (0.30 < r < 0.51) [23]. Another study of 442 Brazilian multi-professional residents without anxiety, working during the COVID-19 pandemic, reported moderate correlations between EE and WHOQol Physical (r = -0.37), and between EE and WHOQoL Psychological (r = -0.33) [24]. In both of these studies, “without anxiety” was defined by a score <16 according to the Beck Anxiety Inventory [25].

Feeling safe when using personal protective equipment at work was associated with lower Emotional Exhaustion (-0.149 points) and Depersonalization (-0.116 points) scores. The sensation of safety at work is essential for coping with the fear of contracting the virus and transmitting it to patients and loved ones, decreasing stress and consequent burnout, particularly among professionals on the frontline of a new and unpredictable pandemic. Feeling safe at work when using personal protective equipment was referred to by 62.3% of dentists, [26] 78.5% of physicians [24], and 79.5% of nursing professionals [27] working on the frontline against the COVID-19 pandemic in Brazil.

Dentists with six or more years of professional practice presented less Depersonalization. Age can be taken as a proxy for time in the profession, corroborating the finding that being young is an important factor associated with burnout [9].

Increased alcohol consumption by dentists during the pandemic was associated with a higher mean level of Depersonalization. The Emotional Exhaustion and Depersonalization burnout dimensions were negatively related to the regular use of alcohol among Lithuanian dentists [28]. In Brazil, increased alcohol consumption during the pandemic was mentioned by 36.5% of dentists [26], 37.2% of physicians [23], 47.8% of nursing professionals [27], and 51.1% of multi-professional residents [24]. Increased alcohol consumption could represent both adaptive and maladaptive methods for coping with the stress of the pandemic [4]. Alcohol consumption merits special attention because it is a potentially modifiable factor in the prevention of burnout among dentists during this pandemic.

In this study, dentists who referred to support from their fellow dentists scored higher for Personal Accomplishment. Personal Accomplishment was also inversely associated with anxiety and positively associated with the psychological dimension of quality of life. Furthermore, these findings must be interpreted in a context in which 81.3% of dentists can be classified as demonstrating low Personal Accomplishment, according to the cut-off level [19].

A recent study of dentists engaged in primary dental care in Scotland reported that the proportions of dentists with high “recognized published cut-off values” for EE and DP were 55.6% and 32.2%, respectively. Unfortunately, the study did not specify these cut-off values [8]. A study of 260 Canadian dental hygienists as they returned to work following service interruption due to the COVID-19 pandemic found EE (≥ 27) = 65%, DP (≥ 10) = 34%, and PA (≤ 33) = 24% [29]. Another study of dental hygienists from California carried out prior to the pandemic, reported EE (>27) = 30.0%, DP (>10) = 11.3%, and PA (<33) = 41.1% [30]. Our study of Brazilian dentists reported the following proportions of high EE (≥ 30) = 43.4%; high DP (≥ 12) = 26.3%; and low PA (≤ 33) = 81.3%. Other studies have reported different or non-specified cut-off levels, precluding valid comparisons with our findings. Despite this limitation, the very high proportion of Brazilian dentists with low PA remains noteworthy. Summing up the evidence, this study concludes that the dentists we investigated showed moderate levels of Emotional Exhaustion and Depersonalization but very low levels of Personal Accomplishment. The prevention and mitigation of burnout syndrome among dentists during a pandemic must take into account of several factors, such as quality of life, time spent in the profession, use of personal protective equipment, support from fellow dentists, and alcohol consumption.

This study has certain important limitations: a small number of individuals was investigated; selection bias could derive from the convenience virtual snowball sampling used; in the context of a cross-sectional study such as this, it is impossible to determine the directional relationship between risk factors and burnout. Finally, other factors related to the prevalence of burnout, stress disorders, and depression were not investigated in our model.