Factors associated with perceived stress among professors at a federal public university.

Background
Within the world of work, stress has negative socioeconomic impacts for employers, employees and the government, including the cost of medical treatments, leaves of absence, disability retirement and loss of individual productivity.


Objective
To identify the main factors that contribute to increase the level of stress of professors at Federal University of Viçosa (UFV), Brazil.


Methods
A questionnaire including Perceived Stress Scale (PSS)-14 and items to investigate personal and occupational characteristics was administered to a sample composed of 222 UFV professors. The study involved quantitative research techniques, descriptive statistics including the t-test to compare means, correlation and multiple linear regression analysis.


Results
Strength of association was highest for variables weekend work (R=0.45), physical activity (R=-0.40), administrative and teaching activities (R=0.29), scientific production (R=0.18), temporary administrative position (R=0.15) and graduate level teaching (R=0.14).


Conclusion
University professors accumulate many tasks, including teaching, research, outreach and administrative activities. Their regular working hours seem not to be enough, but they are compelled to take work home and do not have time for leisure, physical activity and family life, with consequent increase of their level of stress and risk for illness. We suggest formulating public policies to organize the teaching career with consideration of the incidence of stress.


INTRODUCTION
Stress is currently a part of the everyday life of a large part of global population.Indeed, the number of affected individuals considerably increased in the past two decades, to the point it is now considered as a public health problem 1 .Within the world of work, stress has negative socioeconomic impacts for employers, employees and the government, including the cost of medical treatments, leaves of absence, disability retirement, and certainly also loss of individual productivity 2,3 .Given its considerable impact on health, the interest in occupational stress is growing among researchers 4 .
The academic milieu is considered a source of stress for university professors as a function of psychosocial and organizational factors of work, such as poor working conditions, undervaluation of the professional image, low salaries, high exposure to hazards, shortage of human and material resources and physical exhaustion derived from increased pace and intensity of work 5,6 .These conditions contribute to the biopsychological exhaustion of professors, which makes them more prone to lose motivation and develop apathy, anguish, phobias and health problems such as arterial hypertension and coronary artery disease, mental disorders, stress and cancer, among others [7][8][9][10] .The relationship between stress and disease is, however, not simple and direct, but is influenced by coping strategies and individual and collective psychological and organizational factors which make teaching a source of pleasure 1 .
The Brazilian federal universities seem to represent a paradigmatic example of a stressful work environment.This situation is the outcome of changes in higher education policies, currently oriented by the so-called "Anglo-Saxon" model by virtue of which universities are no longer seen as social institutions, but as neo-professionalized, heteronomous, operational, business-minded and competitive social organizations 10 .Such dramatic deterioration of the working conditions of professors led to changes in their activity and social function 11 .As a result, teaching is now seen one of the most stressful occupations 12,13 .
In Brazil, teaching is influenced by countless factors which contribute to enhance job dissatisfaction.In addition to proper teaching, professors are also required to perform several other tasks, such as writing reports and reviews, seek grants and improve their and their department's visibility 14 .
Also, the time professors need to accomplish their tasks was affected.According to Leite 15 , while several tasks were facilitated by the introduction of new technologies, university professors are now required to spend more time in professional activities in the workplace, and also outside it, such as planning, among others.
The aforementioned considerations indicate that the organization of teaching work has peculiar characteristics described in the literature as causes of stress.To contribute to the studies on the teaching career, we sought to answer the following question: what factors related to the teaching career influence the level of stress of professors?
Therefore, the aim of the present study was to investigate factors related to the teaching career that influence the level of stress of professors.Such study is justified by the need to organize a profession characterized by excessive pressure to accomplish tasks and maintain and increase excellence in teaching, research and outreach -and eventually also perform additional administrative tasks -with consequent emotional, physical and social exhaustion and stress.

STUDY DESIGN
The focus of the present original cross-sectional, observational and analytic study was on analyzing factors that determine the level of stress of professors at Universidade Federal de Viçosa (UFV), Brazil.

Sample
The inclusion criteria were: tenured professors with a 40-hour weekly work regime and having worked at the institution for one year at least -since professors hired more recently were not exposed to all the factors analyzed in the present study.The exclusion criteria were: substitute, visiting or non-tenured professors, high school teachers, weekly working time of less than 40 hours and having worked less than one year at the institution.
Participants were selected by means of simple random sampling, since we intended to analyze the characteristics of the full population of professors.The study was conducted at UFV Viçosa campus in December and January 2016.All the participants responded all the questions.
There was a total of 966 professors allocated to the Viçosa campus, distributed as follows: 221 to the Agrarian Sciences Center, 247 to the Biological Sciences Center, 279 to the Exact Sciences Center and 219 to the Human Sciences Center.The calculated sample size for a total of 902 eligible subjects, with 5% of error and e=0.5 (since the odds of selection ought to be the same for all the subjects) was 229 participants.Therefore, 229 professors were selected by means of the lottery method to respond an electronic questionnaire.Seven forms were wrongly filled and were excluded from analysis.As a result, the final sample comprised 222 participants.

Data collection techniques
Data collection began by a presentation and explanation of the study to the participants, followed by the delivery of an informed consent form.The study fully complied with the ethical principles described in the National Health Council Resolution no.466/2012.The study was approved by UFV committee of ethics in research involving human beings, Certificate of Presentation for Ethical Appraisal (CAAE) no.45243915.1.0000.5153,ruling no.1,116,358.Data collection was performed by means of an ad hoc socioeconomic questionnaire and the Perceived Stress Scale (PSS).

CATEGORIES AND VARIABLES FOR ANALYSIS
The t-test for independent samples was used to establish whether there was significant difference in mean stress level among professors with different characteristics.
The influence of predictor variables on the variation of the stress level among the participants was investigated by means of Pearson's correlation and multiple linear regression analysis based on the ordinary least squares (OLS) method and using software Statistical Package for the Social Science (SPSS, version 20.0).The variables considered were: Stress level (PSS): set as outcome variable.It represented the participants' stress level as measured by PSS-14.PSS was developed by Cohen et al. 16 , and according to Machado et al. 17 it is the most widely used instrument to analyze perceived stress, having been validated in more than 20 countries.PSS assesses how unpredictable and uncontrollable respondents find their lives have been in the past month.In addition to providing a subjective assessment of stress, this scale is remarkable for its brevity, which favors its administration together with other techniques.

RESULTS
Table 1 describes the participants' socioeconomic and demographic characteristics.Males (55.4%) and professors with a PhD (79.7%) predominated.In regard to their Stress among professors at a federal public university professional activity, most participants were involved in graduate level teaching (59.4%), taught more than 11 hours/week of undergraduate level courses (56.8%), performed administrative tasks (63.51%), developed outreach projects (60.4%), had low scientific production (68.4%) and did not hold any temporary administrative position (82.0%).
We sought to establish whether the participants exhibited diseases likely derived from work overload.The results are described in Table 2.Only 18.92% of the participants did not report any health problem.The most prevalent ones were backache (38.7%), voice disorders (26.1%), frequent headaches (22.1%), hypertension (18.5%) and depression (10.4%).
We also sought to establish groups in which stress was most frequent and whether there was difference in stress level between males and females, professors with or without children and with master's or doctoral degrees, as we did not include these variables in the regression model.As Table 3 shows, the level of stress was higher among the professors with children (p=0.05) and a doctoral degree p=0.01).Difference was not found as a function of sex.
As a part of the attempt to explain the variation in stress level among UFV professors we fitted a linear regression model with the variables described in section Methods.Pearson's simple correlation test was used to investigate linear correlations between the analyzed variables and stress level.The results are described in Table 4.The variables with From eight variables analyzed, only participation in outreach programs did not exhibit significant correlation with stress at a significance level of 5%, for which reason it was removed from the model.
The regression model was fitted by means of the stepwise method.Maroco 18 observes that the advantage of procedures for variable selection is that based on exact criteria they point to the variables with strongest relationship with the dependent variable and thus are better when fitting the definitive model.
We removed variable temporary administrative position from the model due to its multicollinearity with other variables.As a result, the variables included in the multiple linear regression model were: weekend work, physical activity, administrative activities, scientific production, graduate level teaching and teaching hours (Table 5).The joint strength of association of the model thus fitted was 41% (R=0.65;R 2 =0.41).
Finally, Table 5 also shows the value of the beta coefficient for the variables included in the multiple regression model.It should be observed that the t-test allows rejecting the hypothesis that the coefficient value is zero at a significance level of 5%.
All statistical assumptions for regression were tested and confirmed the model validity.

DISCUSSION
The results of the present study show that most of the participants were male, married and with children.Thus, they In addition to personal, we also analyzed occupational characteristics of the participants.About 60% of the sample taught graduate level courses, 60.36% developed outreach projects, more than half taught more than 10 hours/week and 63.51% performed administrative activities -although without any specific paid position (additional pay, board position).Most participants held a doctoral degree, which finding agrees with that reported in the study by Souza et al. 22 , in which 58.7% of the participant held a PhD.
The participants reported to take work home, and thus their actual weekly working time was over 40 hours.About 60% of the participants reported to take work home on all or almost all weekends, while only 4.05% stated they did no work on weekends.These findings are suggestive of high levels of stress, since the participants spend working a part of their time for physical activity, leisure and family life, with consequent increase of physical and mental tiredness, as well as of the risk of illness.These results evidence the work overload to which university professors are exposed, since most perform several other tasks in addition to teaching, which contributes to increase their level of stress 23 .
The participants reported diseases likely to result from work overload, while only 18.92% did not report any health problem.In their study of the state of health, voice and working conditions of university professors, Servilha and Pereira 24 detected cases of hypertension, voice disorders and back problems, as also us in the present study.These findings might be accounted for by the fact that stress might cause serious problems of variable nature.Stress weakens the body defense system, activates mechanisms that trigger inflammation or alternatively deactivates the mechanisms that inhibit inflammation.An impaired body defense might result in high blood pressure, respiratory disorders and joint pain, which become more serious when under stressful conditions 25,26 .
The mean score on PSS-14 was 25.89.For the purpose of comparison, we surveyed studies which also had resource to PSS-14 to measure levels of stress among other categories of workers and groups.The mean score obtained by the participants in our study was higher than that reported for most of the analyzed occupations.The single exception was represented by master's and doctoral candidates, who also belong with the academic milieu and are exposed to the same pressures for productivity and meeting deadlines as professors, therefore also they have high potential to develop high levels of stress.It is worth noticing that the mean score obtained by the university professors was higher than that of physicians (20.38) 27 , nurses (21.73) 27 and military police officers (22.48) 28 all which categories are subjected to high levels of pressure.
We did not find difference in stress level as a function of sex.This finding agrees with that reported by Gonçalves et al. 29 , but disagree from the results of the studies by Kafrouni 30 and Souza et al. 22 who found higher levels of stress among females.
The results of the present study indicate that having children might increase stress, as the mean level was higher for the participants with children.This finding agrees with the results of the studies by Hyeda and Handar 31  et al. 32 in which the participants with children were found to be more prone to emotional exhaustion.The level of stress was higher among the participants with a PhD compared to those with a master's degree, with a margin of error of mere 1.3%.The explanation might be that professors with just a master's degree are not allowed to teach graduate level courses, have limited access to research resources and a more restricted scope of activities compared to those with a PhD.As a result, they are less overloaded by research activities and student supervision.
On the multiple linear regression model, the strength of association of stress level with variables weekend work, frequency of physical activity, administrative activities, scientific production, graduate level teaching and teaching hours was 64.8%.The value of the coefficient of determination (R 2 ) was 0.419, which indicates that 41.9% of the variation of stress level was explained by the joint variation of the included variables.
Variable "frequency of weekend work" explained the largest proportion of outcome stress level (R=0.45).Working on weekends might interfere with the personal and family life of individuals, as they spend a part of their leisure time working, precisely at a time when their priority should be resting and socializing.According to Borsoi 33 -based on other studies 10,34,35 -when facing an intensification of tasks and work overload, university professors simply keep on working without setting any time limit, with consequent impact on the time they should allot to their individual needs.
Variable "physical activity" ranked second in explanatory power (R=-0.40)and exhibited inverse relationship with stress levels.In other words, the higher the frequency of physical activity, the lower the level of stress.However, by taking work home university professors reduce the time available for physical activity.Our findings corroborate the results reported by Souza et al. 22 and Camargo et al. 21who also found negative correlation between physical activity and stress level.Also variable "administrative activities" exhibited strong correlation with stress level (R=0.29).The reason is that this type of activities increase the workload of university professors, shorten the time they have for teaching, research and outreach and often demand skills beyond the scope of university professors, who thus need to spend extra time learning how to perform them, with consequent increase of stress at work.Our results are similar to those reported by Ayres et al. 36 who found that professors who perform administrative tasks exhibit higher levels of stress.
Similarly, also variable "teaching hours" influenced the level of stress (R=0.29).The reason is that professors have increasingly less time for their tasks, and thus the longer the teaching hours, the shorter the time available for other tasks, with consequent increase of their workload and stress level.
"Scientific production" (R=0.18) is an aspect that demands much effort from investigators, often requires gathering large teams, which increases the number of students to be supervised, and compels professors to take work home.The pressure for productivity creates a vicious circle that keeps professors chained to it, increases their workload, impairs their personal and family life, causes physical and mental tiredness and increases their level of stress.According to Lima and Lima-Filho 37 the almost "frantic" efforts to increase production make university professors compete against another, the result being tiredness, stress and often frustration.
Also "graduate level teaching" had impact on stress level (R=0.14) as a function of the high demands for scientific productivity, supervision of master's and doctoral candidates and teaching new subjects, which increase the participants' workload.According to Borsoi 33 the demand to maintain a high level of "scientific production" to keep teaching graduate level courses exposes professors to high levels of stress.
Finally, we calculated the regression equation coefficients and all the values were statistically significant as per the t-test results.All the assumptions were measured and converged toward the validity of the analyzed model.

CONCLUSION
The aim of the present study was to investigate the main factors that increase the level of stress of UFV professors.
The results show that teaching is a source of stress and might be associated with occurrence of several health problems, particularly backache, depression, voice disorders and hypertension.
Stress among professors at a federal public university University professors accumulate many tasks, including teaching, research, outreach and administrative activities.Their regular working time is seemingly never enough, but they are compelled to take work home.Thus, they cannot allot the due time to leisure, physical activity and family life, with consequent increase of their level of stress and risk of illness.
Work on weekends, lack of physical activity, amount of administrative activities, scientific production, graduate level teaching and teaching hours were the main factors associated with the participants' level of stress.Therefore, university professors should organize themselves in a way they do not accumulate tasks after hours and do not cut the time needed for physical and leisure activities.
In this regard, public policies to organize the teaching career are needed and institutions should concern themselves with the incidence of stress and other disorders among this population of workers.
We suggest for future studies to include variables related with family life, quality of life and leisure activities, as in the present study we mainly focused on career-related variables, which have just reasonable explanatory power.We also suggest reproducing the present study at other institutions and with larger samples to increase the explanatory power of the tested model.

Table 4 .
21arson's correlation between predictor variables and stress, Viçosa, 2016 (n=222).Stress among professors at a federal public university agree with findings reported by Moraes and Moreira19and Aguiar20relative to number of children and marital status, respectively.Differently, 50.1% of the sample analyzed by Camargo et al.21corresponded to females.

Table 5 .
and Silva Beta coefficient value for the variables included in the model, Viçosa, 2016 (n=222).