Confirmatory factorial analysis of the Maslach Burnout Inventory – Human Services Survey in health professionals in emergency services*

Objective: to confirm the factorial validity of the Maslach Burnout Inventory – Human Services Survey version in a sample of health professionals from the emergency services. Method: a quantitative, exploratory, descriptive and analytical study. Two hundred and eighty-two health professionals participated in the study. For data collection, a sociodemographic questionnaire and the Maslach Burnout Inventory were used. The psychometric sensitivity for the MBI-HSS items was estimated by measures of central tendency, variability and the distribution shape. Internal consistency was estimated using Cronbach’s alpha coefficient and the adequacy of the sample was verified using the Kaiser-Meyer-Olkin (KMO) index. As indexes for assessing the quality of fit of the model, the chi-square ratio by the degrees of freedom (χ2/DoF), the comparative fit index (CFI), the goodness of fit index (GFI), the Tucker Lewis index (TLI) and the root mean square error of approximation (RMSEA) were considered. To test data fit, the maximum likelihood method was used. Results: the three-factor structure of the Maslach Burnout Inventory was confirmed. Items 9, 12, 15 and 16 had a factorial weight below what was considered appropriate and were removed from the model. The second order hierarchical model with the aforementioned modifications presented an adequate adjustment to the data and can be considered the best and most parsimonious model tested according to the information theory indexes. The internal consistency of the instrument’s factors was recalculated considering the exclusion of the items and the three factors were considered adequate. Conclusion: the results obtained show that the Maslach Burnout Inventory is a reliable and factorially valid instrument for measuring the burnout syndrome in emergency service professionals in Brazil.


Introduction
The Maslach Burnout Inventory -Human Service Survey (MBI-HSS) stands out as the most used instrument to assess the burnout syndrome and its configuration in three dimensions has been confirmed, worldwide, in different populations. However, there is no Brazilian study that has investigated its validity in health teams from emergency services. Furthermore, it is relevant to develop analyses that prove that the items on a scale measure exactly what they propose. This is a psychological syndrome that develops in response to chronic interpersonal stressors in the workplace (1)(2)(3) . It is characterized by emotional exhaustion (feelings of extreme tiredness related to excessive physical and emotional effort); depersonalization (negative attitudes in interpersonal relationships, marked by cynicism and disinterest) and low personal fulfillment (negative self-assessment of oneself, work ability, and to deal with other people) (1,(4)(5)(6)(7) .
The Maslach Burnout Inventory was developed in 1981, originally to be applied to human service professionals. The three dimensions that make up the scale emerged from exploratory items collected from interviews with health care professionals, with the aim of reflecting on the experiences related to the phenomenon (1) . With increasing interest in the burnout syndrome, other versions of this instrument have been developed (3) .
There are currently three versions of the Maslach Burnout Inventory: the Human Services Survey (MBI-HSS), used for the health services; the Educators Survey (MBI-ES), used in the educational area and the General Survey (MBI-GS), used for workers in general (8) . There are other assessment instruments, however, the MBI is the most used by the national and international scientific community, showing high reliability regardless of the sample (4,7) .
The MBI-HSS has 22 statements that comprise the frequency of feelings and attitudes towards clients and work. These statements are divided into three dimensions: emotional exhaustion (made up of nine items), depersonalization (made up of five items) and personal fulfillment (made up of eight items). The answers follow a five-point Likert scale ranging from 1 to 5 (from never to every day). There is the burnout syndrome in the manifestation of high emotional exhaustion, high depersonalization and low personal fulfillment (1)(2)(3)(4) .  (1)(2)(3)(4)(9)(10) .
Confirmatory structural analysis is widely used to assess the relationship between items and factors in an instrument and in the international scientific literature there are studies that indicate from one to six factors for MBI-HSS (4,(11)(12)(13) . Most of them indicate the original composition, with three factors/dimensions, as the most suitable (11,13) .
The emotional exhaustion dimension stands out as the least vulnerable to cultural issues (26) . This result may indicate a more universal relationship between the statements that make up the emotional exhaustion dimension and the feeling of weariness due to work that occurs in individuals (relationship between the dimension and the construct).
The low internal consistency of the depersonalization dimension, in comparison with the other two dimensions of emotional exhaustion and personal fulfillment, is often found in other studies (3)(4)11,25) . This result may be related to the small number of items that make up this dimension (11) .
In addition, social demands can be a burden for the health professionals, whose main objective is to care for others.
Admitting certain level of depersonalization can pose a psychological threat and interfere with their levels of self-esteem and perceived self-efficacy, as feelings of depersonalization refer to certain distancing from the service recipients.
Despite the lower internal consistency of the factors that make up the depersonalization dimension and the lack of personal fulfillment in the face of emotional exhaustion, the latter is considered, according to the scale's own authors, the most important and the one that really reflects emotional strain (8) . Also according to the authors, depersonalization can act as a coping strategy in a situation of exhaustion, while the lack of personal fulfillment would be a consequence (3) . Therefore, maintaining it in the instrument is relevant and was adopted in this study.
It was possible to obtain an adequate factorial solution, with satisfactory adjustment levels, paying attention to the psychometric sensitivity of the items in the sample. All the items that make up the MBI-HSS showed adequate asymmetry and kurtosis, allowing asserting that, in general, they fit well to a normal distribution being able to, adequately, discriminate different levels of evaluation. There is no consensus regarding the items that should be excluded from the scale. However, some items are known to be more ambiguous, as is the case of items 12 and 16, as well as items 2, 6 and 20 (4,11,19,21) .
The scale's authors themselves suggested not considering items 12 and 16 in confirmatory factorial studies (22,27) .
In this sense, different studies have shown that, due to their inadequate factor weight, some items end up being removed. In an Argentinian study conducted with mental health professionals, it was observed that the adjustment of the model improved satisfactorily, confirming the three-factor structure after removing item 12. This item obtained a satisfactory factorial load in the three dimensions, behaving in an ambiguous way; the same was verified in the original study (19) . In Córdoba, a study conducted with professionals from different work services tested six models that varied from one to four factors.  behavior to the matrix, model eight presented the best fit (28) . Another Brazilian study with a sample of nursing assistants, divided between groups with major depressive disorder and absence of the disorder, found that the acceptable rates for RMSEA and CFI occurred in the group with no disorder. And in the general sample of the study, the three-factor model proved to be acceptable, but concluding that the two-factor model was better adjusted in this sample (29) .
In the same sense, a number of European studies showed the relevance of confirmatory analysis studies for the scale in different samples. A Hungarian study conducted with elementary and high school teachers tested eight models, with the best fit obtained with the bi-factorial model (30) . A Spanish investigation, conducted with a sample of social workers, supported the superior three-factor model compared to alternative models of one or two factors. In addition, items 12, 13 and 16 were excluded, which favored better adjustment and internal consistency to MBI-HSS in the referred sample (31) .
Another study, conducted with Australian lawyers, concluded that the five-factor structure better explains the multi-factorial nature of the burnout syndrome (18) . In a study conducted in Thailand with Medicine post-graduate students, identified that the three-factor model obtained values considered acceptable, after data adjustment (23) .
In turn, with respect to items 9 and 15 excluded

Conclusion
The confirmatory factor analysis of the MBI-HSS