Prevalence and risk factors of common mental disorders among teachers

This cross-sectional study was conducted with a calculated random sample of the population of 1,250 teachers distributed in all 37 elementary schools located in a large city in the metropolitan area of Porto Alegre (in the state of Rio Grande do Sul, Brazil). The parameters for sample size calculation were the 50.3% prevalence ( Gasparini et al., 2006 ), 5% error, 80% effect power, and 20% potential losses. The final sample consisted of 713 teachers. The collected sample consisted of 679 teachers, with a loss of 34 participants.

Most participants were women (91.8%), with a partner (60.8%) and children (68.6%). Their average age was 42 years ( SD = 9) and their salaries were higher than three minimum wages, a reference in Brazil (51.2%). Most participants have a post-graduate degree (61.8%). The majority of the professionals work exclusively at the school under investigation (74.2%). Participants have an average of 17 years of professional experience (range = 1-47, SD = 8.9, Mode = 15) and 8.8 years of professional experience at the school ( SD = 7.2). The workload varies between 16 and 57 hours a week, with an average of 34 hours a week (SD = 11.6, Mode = 40). The amount of students they work with per day ranges between 7 and 500, with an average of 77 students per day ( SD = 74, Mode = 60).

Data were collected using a questionnaire that was specifically designed to investigate some sociodemographic factors (gender, age, marital status, children) and working variables (education, weekly workload at the institution, wages, time working at the institution, work in other activities, jobs at other schools). Additionally, the following self-report instruments were used:

Self-Reporting Questionnaire (SRQ-20). This scale has been designed to assess the presence of common mental disorders. It has been validated for use in Brazil by Mari and Williams (1986) . It comprises 20 yes/no questions about physical and psychological symptoms, for the detection of minor psychiatric disorders (anxiety, depression, psychosomatic reactions, irritability, and mental fatigue). Each affirmative answer corresponds to a score of 1 in the final score (sum of all scores). The scores obtained indicate the probability of the presence of non-psychotic disorders, and may range from 0 (in probability) to 20 (extreme probability). The instrument has a Cronbach's alpha of .86 and a cutoff point of 7/8, regardless of gender ( Gonçalves, Stein, & Kapczinski, 2008 ). In the current sample the Cronbach's alpha was .88.

Battery of Psychosocial Risk Assessment (Unidad de Investigación Psicosocial de la Conducta Organizacional- UNIPSICO), by Gil-Monte (2005) , evaluating: Role Ambiguity (5 items, ? = .78), Social Support (5 items, ? = .75), Self-efficacy (8 items, ? = .79), Autonomy (5 items, ? = .76), Role Conflict (5 items, ? = .76), and Overload (6 items, ? = .79). All items are assessed on a five-point Likert scale of frequency, ranging from 0 ( never) to 4 (every day ). The Cronbach's alpha obtained in the present study was: Role Ambiguity (? = .77), Social Support (? = .84), Self-efficacy (? = .89), Autonomy (? = .76), Role Conflict (? = .76), and Overload (? = .71).

First, we contacted the City's Educational Department and presented the object of the study in order to obtain authorization and support to implement the instruments. Teachers answered the instruments at their workplace. The instruments were collected after they were filled out. The application occurred from September to November 2013. The first author of this study collected the data. Research ethics committees of the Pontifical Catholic University of Rio Grande do Sul/Brazil (PUCRS) approved the study.

The statistical software PASW, version 17 (SPSS/PASW, Inc., Chicago, IL), was applied to conduct the data analysis. Deive statistics were used to calculate frequencies, mean scores, and standard deviations. The data analysis began with bivariate analyses between the outcome and the study factors. The multivariate logistic regression d the variables that originally presented p-values < .20. The logistic regression was performed according to a three-stage hierarchical model. The demographic variables were introduced in the first stage, and only those with p < .10 were kept in the model. The working variables were introduced in the second stage, using the same criterion to keep them in the regression. The third stage d psychosocial variables. Significant associations were defined as those with p-values < .05 in the stage in which they were introduced in the model.

The findings of this study showed that out of the Brazilian teachers of the sample 236 (34.8%) met with criteria for CMDs. This prevalence rate is within the 7%-26% range for the prevalence reported in different studies with Western populations ( WHO, 2003 ). When compared to the same al category, namely, teachers, the percentage identified in this study is lower than that in a study conducted by Gasparini et al. (2006) , which found a 50.3% prevalence.

Findings from the block models that demographic factors and working variables were not associated with CMDs. Psychosocial variables, such as role ambiguity, overload, low level of social support and low perceived self-efficacy were significantly associated with the increase in the odds ratio for CMDs. All these factors had a moderate impact by increasing the odds twofold, except for social support. Those with low levels of social support had 1.74 times higher odds ratio for CMDs compared to those with high levels.

Psychosocial risks have been associated with numerous mental and physical health problems. Common mental disorders are among them ( Leka & Jain, 2010 ). Work-related stress, depression, and anxiety can be directly associated with the exposure to psychosocial hazards at work ( Cox, Griffiths, & Leka, 2005 ). Studies on workers’ health have shown similar results regarding psychosocial stressors, such as role ambiguity ( Jamal, 1990; Sobrinho, Carvalho, Bonfim, Cirino, & Ferreira, 2006 ), overload (Sobrinho et al., 2006 ), low level of social support ( Lakey & Orehek, 2011; Stansfeld, Clark, Caldwell, Rodgers, & Power, 2008 ), and low perceived self-efficacy ( Luszczynska, Gutiérrez-Do?a, & Schwarzer, 2005 ).

Role ambiguity and overload have been identified as important causes of work-related stress ( Vanishree, 2014 ). Ambiguity arises in those situations when expectations in relation to the role the individual will play are not clear ( Jones, 1993 ). Overload happens when there is too much work to be developed in a short period of time and the individual has scarce resources to do it. According to Maslach and Leiter (1997) , this is one of the most notable indications of imbalance between people and their work.

Social support involves empathy, identification, cooperation, and harmonious interaction with others ( Crandall, 1984 ). Thus, dividing individual difficulties with coworkers and supervisors, within an environment of respect and emotional comfort, helps professionals to distance themselves from problems directly related to their target audience, that is, their students. Self-efficacy, in the perspective of the social cognitive theory, is defined as the set of beliefs that people have about their ability to organize their cognitive, affective, and motivational resources, in order to implement the d actions to achieve certain goals, perform certain tasks, or manage specific situations ( Bandura, 1997).

In a study with teachers, conducted by Brouwers, Tomic, and Boluijt (2011) , support from coworkers had a moderating effect in the relation between work ments and professional fulfillment, while support from leaders had a moderating effect in self-efficacy belief, work ments, and professional fulfillment. In this regard, the impact of the lack of social support and low self-efficacy represents significant risk factors for the health conditions of teaching professionals. A successful adaptation to stressful demands contributes to better levels of health ( Brouwers et al., 2011; Schwerdtfeger, Konermann, & Schönhofen, 2008 ).

It is important to highlight that out of the three blocks of variables being studied only the psychosocial variables presented risk factors for CMDs. One might think that in the context of illness at work, psychosocial variables represent the dynamics of the organization of the teaching work. These results revealed a risk profile formed by psychosocial variables, demonstrating the importance of environmental/contextual factors for the development of CMDs, as Goldberg and Huxley (1992) state. Therefore, the study suggests the need to deepen the relation of CMDs according to the social conditions in which the work is performed. Identifying the profile of illness in populations is important to plan and develop public policies ( Borrell & Artazcoz, 2008).

One of the strengths of the present study is that data about teachers was ed from a probabilistic sample. In what concerns the size of the sample, this was sufficient to provide power to the significant effect size in statistical analysis ( Cohen, 1988 ). Additionally, this study used reliable and valid instruments to collect data.

The results of this study should be considered in light of some limitations, such as the cross-sectional design. Another limitation of the study is the utilization of self-reported measures, which may increase the possibility of response bias.

Overall, this study contributes to the development of theoretical knowledge in the area of CMDs among teachers, since this professional category has been poorly investigated concerning this type of harm. Furthermore, the findings have important implications for occupational psychology, applied practitioners, and principals and school administrators.

Psychosocial risks are present in all occupational contexts, representing a constant threat to workers’ health. In the school context these threats have increased considerably in recent decades ( Carlotto, 2012 ). Acknowledging this is essential to develop actions to prevent occupational diseases ( Leka & Jain, 2010).

Thereby, the results should be used to shed light on possible applied interventions among teachers, such as increasing self-efficacy and social support, resizing activities aimed at reducing the workload, and defining professional assignments in order to reduce role ambiguity.