Hypotheses , design , and instruments in the Educatel Study , Brazil , 2015 / 2016

The Educatel Study 2015/2016 was designed to evaluate health and work conditions in a representative sample of the 2,220,000 schoolteachers working in Basic Education in Brazil. The article aimed to describe the telephone survey’s basis and design, using a questionnaire consisting of 54 short, simple questions, most of which with multiple-choice answers (closed questions) addressing diseases, accidents, absenteeism, frequency of healthy behaviors, physical and psychosocial environment, and employment characteristics. In the pilot stage, the multi-theme questionnaire was assessed in order to verify the effects of the terminology, the format of the questions and the multiple-choice answers, the questionnaire’s internal organization, production of the answers, and duration of the interview. The interviewers’ training and follow-up and listening of the calls in real time aimed to identify communications problems. The teachers were interviewed at school after prior contact with the school assistant to set appointments. The advantages and risks of biases related to the telephone interview modality should be considered to interpret the results. The results on the teachers’ profile, illnesses, and school environment will provide inputs for elaborating inter-sector measures to improve the target group’s health, which is related to Brazil’s school system indicators based on the concepts presented here. Methods; Health Surveys; School Teachers; Occupational Health Correspondence A. A. Assunção Faculdade de Medicina, Universidade Federal de Minas Gerais. Av. Alfredo Balena 190, Belo Horizonte, MG 30310-450, Brasil. adavila@medicina.ufmg.br 1 Faculdade de Medicina, Universidade Federal de Minas Gerais, Belo Horizonte, Brasil. 2 Escola de Enfermagem, Universidade Federal de Minas Gerais, Belo Horizonte, Brasil. 3 Instituto de Ciências Exatas, Universidade Federal de Juiz de Fora, Juiz de Fora, Brasil. doi: 10.1590/0102-311X00108618 Cad. Saúde Pública 2019; 35 Sup 1:e00108618 QUESTÕES METODOLÓGICAS METHODOLOGICAL ISSUES This article is published in Open Access under the Creative Commons Attribution license, which allows use, distribution, and reproduction in any medium, without restrictions, as long as the original work is correctly cited. Assunção AA et al. 2 Cad. Saúde Pública 2019; 35 Sup 1:e00108618 Introduction A survey is a concepts-based activity developed with methodological resources planned to collect and analyze data in a given population. Survey procedures are well-defined and accompanied by an exercise of data processing, treatment, and compilation 1. Health surveys have been conducted in industrialized countries since the 1960s as valid resources for formulating and evaluating public policies. They are generally developed on the basis of the need for information when there are no data, or when data are available but insufficient to define the health situation in a population or in specific groups, aimed at assessing access to services or to identify risks of certain diseases 2. Meanwhile, occupational surveys are intended to study groups – fractions of the overall population – who have in common the fact that they are connected to the same employment under particular work conditions. This type of study is widely used in the workers’ health field, given the advantage of simultaneously examining the health situation and work conditions possibly associated with the target phenomena 3. In short, the strategy of conducting surveys in occupational groups allows examining the hypothesis concerning the contribution of work conditions to adult workers’ health 4,5. Survey results are descriptive but allow obtaining analytical results via adjustments to statistical models 6. It is thus possible to evaluate associations between the respondents’ characteristics (e.g., teacher’s age), context (characteristics of the school and surrounding area), and target outcomes (voice disorders, etc.). They do not refer either to objective measurements in the physical environment (dust and noise) or measurements of vital data or objective approximations of the individuals’ behaviors (verification, at the worker’s home, of sleep duration and hours), although such measurements are feasible, as attested by recent surveys 7,8. The Educatel Study – a telephone survey on health, work conditions, and work absences in schools in Basic Education in Brazil – is part of the activities developed in the debate and development of the Brazilian National Educational System (SNE). The prevalence of illnesses and their consequences for teachers’ well-being and work attendance clash with widely accepted guidelines on the relevance of valuing teachers for achieving educational objectives, as indicated in targets 15, 16, 17, and 18 of the Brazilian National Plan for Education 9. These specific targets are considered strategic for achieving the others 10. Valuing teachers is related to both classroom teaching conditions and the interfaces with external dimensions of teaching work per se: training, career plans, wage policies, and social recognition 11. Two components, namely teachers’ workday and salaries, are key health issues on the agenda of teachers’ unions 12 and are frequently mentioned by specialists in education 10,13. Most local governments in Brazil fail to ensure time in the formal workday for teachers to perform non-classroom activities or guarantee exclusive and fulltime employment for teachers in the same school, both of which are formally provided for under the country’s legislation 14. From 1981 to 2009 there was a mean increase of four hours in the workweek of teachers that answered the Brazilian Institute of Geography and Statistics (IBGE) survey on workweeks in the person’s primary employment. During this same period, 10 to 20% of teachers reported having more than one job, or a 5% increase in this contingent in recent years 13. According to Oliveira & Vieira 15, three-fourths of teachers interviewed in seven states of Brazil routinely took work home to finish after hours. Among these, more than half were not paid for this non-classroom overtime work. In 2016, ten Brazilian states and more than half the municipalities (counties) 16 fell short of complying with the law establishing a nationwide minimum wage for schoolteachers 17. Teachers received lower salaries than other professionals with similar educational levels. This likely explains why 18% of Brazilian teachers hold two or more jobs 18. One can conjecture on the relationship between such data on teachers’ health indicators and issues of quality of education 19. First, how can a teacher with such a long workday manage to study and keep up with the dynamics of the cultural and scientific world? Besides, these characteristics of employment in Basic Education may limit teachers’ access to health services, leisure-time activities, and decent transportation 20,21. These conditions are stressful, reduce quality of life, and produce effects that result in illness 22. Second, how can the system expand the students’ school day, induce greater teacher retention in the same school, and open new learning opportunities for pupils if teachers are discouraged, plagued with health problems, and facing financial dilemmas in their lives? DESIGN OF THE EDUCATEL STUDY 3 Cad. Saúde Pública 2019; 35 Sup 1:e00108618 In the attempt to provide inputs to operationalize the targets for valuing teachers, Educatel 2015/2016 was designed to directly measure the prevalence of diseases and accidents and absenteeism and its factors, in addition to identifying employment status, work attendance, and healthy behaviors in a representative sample of the 2,220,000 teachers working in preschool, elementary, and secondary school or in special education, which jointly comprise Basic Education in Brazil. The article describes the basis and techniques used in the representative nationwide survey of Brazilian


Introduction
A survey is a concepts-based activity developed with methodological resources planned to collect and analyze data in a given population.Survey procedures are well-defined and accompanied by an exercise of data processing, treatment, and compilation 1 .
Health surveys have been conducted in industrialized countries since the 1960s as valid resources for formulating and evaluating public policies.They are generally developed on the basis of the need for information when there are no data, or when data are available but insufficient to define the health situation in a population or in specific groups, aimed at assessing access to services or to identify risks of certain diseases 2 .
Meanwhile, occupational surveys are intended to study groups -fractions of the overall population -who have in common the fact that they are connected to the same employment under particular work conditions.This type of study is widely used in the workers' health field, given the advantage of simultaneously examining the health situation and work conditions possibly associated with the target phenomena 3 .In short, the strategy of conducting surveys in occupational groups allows examining the hypothesis concerning the contribution of work conditions to adult workers' health 4,5 .
Survey results are descriptive but allow obtaining analytical results via adjustments to statistical models 6 .It is thus possible to evaluate associations between the respondents' characteristics (e.g., teacher's age), context (characteristics of the school and surrounding area), and target outcomes (voice disorders, etc.).They do not refer either to objective measurements in the physical environment (dust and noise) or measurements of vital data or objective approximations of the individuals' behaviors (verification, at the worker's home, of sleep duration and hours), although such measurements are feasible, as attested by recent surveys 7,8 .
The Educatel Study -a telephone survey on health, work conditions, and work absences in schools in Basic Education in Brazil -is part of the activities developed in the debate and development of the Brazilian National Educational System (SNE).The prevalence of illnesses and their consequences for teachers' well-being and work attendance clash with widely accepted guidelines on the relevance of valuing teachers for achieving educational objectives, as indicated in targets 15, 16, 17, and 18 of the Brazilian National Plan for Education 9 .These specific targets are considered strategic for achieving the others 10 .Valuing teachers is related to both classroom teaching conditions and the interfaces with external dimensions of teaching work per se: training, career plans, wage policies, and social recognition 11 .Two components, namely teachers' workday and salaries, are key health issues on the agenda of teachers' unions 12 and are frequently mentioned by specialists in education 10,13 .
Most local governments in Brazil fail to ensure time in the formal workday for teachers to perform non-classroom activities or guarantee exclusive and fulltime employment for teachers in the same school, both of which are formally provided for under the country's legislation 14 .From 1981 to 2009 there was a mean increase of four hours in the workweek of teachers that answered the Brazilian Institute of Geography and Statistics (IBGE) survey on workweeks in the person's primary employment.During this same period, 10 to 20% of teachers reported having more than one job, or a 5% increase in this contingent in recent years 13 .According to Oliveira & Vieira 15 , three-fourths of teachers interviewed in seven states of Brazil routinely took work home to finish after hours.Among these, more than half were not paid for this non-classroom overtime work.In 2016, ten Brazilian states and more than half the municipalities (counties) 16 fell short of complying with the law establishing a nationwide minimum wage for schoolteachers 17 .Teachers received lower salaries than other professionals with similar educational levels.This likely explains why 18% of Brazilian teachers hold two or more jobs 18 .
One can conjecture on the relationship between such data on teachers' health indicators and issues of quality of education 19 .First, how can a teacher with such a long workday manage to study and keep up with the dynamics of the cultural and scientific world?Besides, these characteristics of employment in Basic Education may limit teachers' access to health services, leisure-time activities, and decent transportation 20,21 .These conditions are stressful, reduce quality of life, and produce effects that result in illness 22 .Second, how can the system expand the students' school day, induce greater teacher retention in the same school, and open new learning opportunities for pupils if teachers are discouraged, plagued with health problems, and facing financial dilemmas in their lives?
In the attempt to provide inputs to operationalize the targets for valuing teachers, Educatel 2015/2016 was designed to directly measure the prevalence of diseases and accidents and absenteeism and its factors, in addition to identifying employment status, work attendance, and healthy behaviors in a representative sample of the 2,220,000 teachers working in preschool, elementary, and secondary school or in special education, which jointly comprise Basic Education in Brazil.The article describes the basis and techniques used in the representative nationwide survey of Brazilian schoolteachers.

Preliminary stage, prior to data collection
Since 2001, this group of researchers has focused on teachers' health in Basic Education.In addition to qualitative (clinical and ergonomic) studies, in 2004 a survey was held in the municipal schools in Belo Horizonte, Minas Gerais State, providing the main references adopted in the Educatel Study 23 .
The first results of the microdata available at the Brazilian National Institute for Educational Studies and Research "Anísio Teixeira" (INEP) 24 , covering the entire universe of Brazilian schoolteachers, revealed differences when the schools were compared according to the country's five major geographic regions and census areas (rural vs. urban).The parameters for the sampling plan were thus prepared to reflect this diversity: teachers in relation to their own characteristics and their employment and the schools where they worked as to geographic location, teaching modality, and physical and psychosocial environments.Absenteeism for health reasons was considered a relevant event and was incorporated as the principal outcome both in the sampling plan and in calculating the study's size.Additional details have been published in a previous study 25 .
A "face-to-face" survey would have been unfeasible, because it would have been too expensive, given the objective of obtaining information on a national scale.To achieve a response rate consistent with the desired representativeness, we opted for a telephone interview with a computer-assisted questionnaire, based on speed in obtaining and processing data with this survey modality 26 .Telephone interviews are a valuable method for avoiding the costs and difficulties involved in moving the interviewer from site to site 27 .This strategy's consistency had been proven in previous studies that identified similar results when comparing telephone and face-to-face interviews 28 .The most recent and robust Brazilian experience -VIGITEL 29 -was corroborated by results that identified similarity in the prevalence of chronic diseases when comparing home interviews with a telephone survey 26 .
Concerning the location for the interview, there is a known advantage to approaching workers during their normal workday, which facilitates their recalling workplace conditions and related characteristics 30 .We thus used the INEP website 24 to identify telephone numbers for the schools where the selected teachers worked, in order for them to answer the questionnaire.
While barriers related to efficiency are overcome with the telephone survey modality, there are well-known disadvantages.By telephone, the interviewer perceives the interviewee's embarrassment with the content of some questions.Sensitive questions become less embarrassing with a self-applied questionnaire than with a questionnaire applied by the interviewer by telephone or a face-to-face interview (in that order).This is an example of the modality.On the one hand, in a face-to-face interview, when the interviewer and the participant meet in person, the odds are lower for misunderstandings or for the participant to decide not to participate.The face-to-face modality also allows recording nonverbal communication, which is useful for expanding the understanding of what occurs when the participant interacts with the questions' content.
Finally, the survey was made possible thanks to authorization by INEP 24 to access the information available in the institute's national registry of schoolteachers.

Developing the questionnaire
The study design's underlying concepts were reproduced as empirically testable elements through questionnaires developed to meet the specificities of a telephone survey 31 .
Cad. Saúde Pública 2019; 35 Sup 1:e00108618 The questionnaire took considerable time to develop, since it was done in step with the development of the theoretical framework and exploration of the field to characterize the study population and the scenario where the teachers work, plan the sample, and collect elements for the hypotheses based on the notion of the health/disease process, as appropriate in the field of workers' health.
In the Educatel Study, selection and measurement biases were minimized in order to ensure the interpretations' consistency and the study's credibility.To offset any selection bias, we adopted a complex sampling plan to obtain estimates with preestablished precision measurements.Having concluded the data collection, we identified the participation rates (for eligible teachers) according to major geographic regions of Brazil, school location (rural vs. urban), age bracket, gender, school's administrative status (public vs. private), and teacher's employment status.Sampling weights adjusted for non-response and estimation procedures were adopted.Further methodological details have been published elsewhere 25 .
The questionnaire was developed in careful detail, as we shall see, and the interviewers were adequately trained to avoid the measurement bias expected in epidemiological surveys, regardless of the interview modality (telephone or face-to-face).The variables and measurement format were based on careful study of the literature to identify validated and widely used formats, as described below.In addition to this care, the questions on the workplace and behaviors, commonly criticized in society (difficulty in missing work when sick, for example) were drafted in keeping with consistent recommendations in the literature.Box 1 lists the references for each scale or sensitive question used.
The questions formulated to obtain information from teachers on their health status and work conditions aimed to produce original data to fill gaps in this area.Knowing that the question constructs the object and that the answers depend on the question's format 32 , tests and retests were performed to avoid measurement bias 33 .Essentially psychological questions are not always faced naturally by the respondent, besides requiring more concentration 33 , so some questions and choices of answers were adapted.The questionnaire's approval depended on the validation of these questions and answers that were adapted for the study's purposes.
During the development of the questionnaire, we examined the known and validated scales for multidimensional events (e.g., violence, social support, autonomy).Cross-cultural adaptation was the criterion for adopting both the scales and the key questions that had been developed outside of Brazil and in other languages 34 .We also examined the adequacy and sufficiency of the questionnaire's psychometric performance.
Teachers' exposure to interpersonal conflicts in the development of their activities at school 15,22,35 , including with the students themselves, motivated the inclusion of two specific questions: "In the last 12 months, have you suffered any VERBAL violence from students?" and "In the last 12 months, have you suffered any PHYSICAL violence from students?".The choices for answers were: "never", "once", and "twice or more".The two questions were drafted and validated in the Educatel Study.
The social support dimension was investigated with pertinent questions taken from the Job Stress Scale ( JSS), which has been adapted and validated in Brazil 36 .The response categories for these questions are "strongly disagree", "disagree", "agree", and "strongly agree".In order to facilitate the telephone interview, which excludes other forms of communication between the interviewer and interviewee (e.g., facial expressions), the JSS answers were adapted to "frequently", "occasionally", "rarely", and "never or almost never".We did not identify any refusals or attempts by teachers to give at evasive answers or similar verbalizations that might denote embarrassment on the part of the interviewee.
As for autonomy, the question used previously was reproduced 35 (Box 1): Does the school give teachers the opportunity to participate actively in decisions?("frequently", "occasionally", "rarely", "almost never", or "never").
The physical activity scale adopted in the VIGITEL survey 29 was used in full.Key outcomes in the study were defined with the same strategy used for the scales, whether reproducing consensuses for the question's format and content, as in the case of absenteeism 37 , or adapting the format to the modality of choice (Box 1).We did not perform linguistic adaptations to local and regional specificities, considering the sample's relatively homogeneous schooling level.
Testing of the multi-thematic questionnaire aimed to verify the effects of its terminology, the questions' format (open or closed), the choices of answers, the questions' organization within the

Box 1
Thematic blocks, variables, and reference source or scale in the Educatel Study, 2015.questionnaire, the production of answers, and duration of the interview.This stage incorporated nine volunteers, including medical and nursing students who were currently teaching in Basic Education.The questionnaire's adequacy was confirmed after assessment of each question as to understanding of the wording and its objective.During the pilot stage, special attention was given to occasional impasses on the interviewee's part when asked about a specific event.Reports of failure to recall a certain event, doubts with the questions or choices for answers, and expressions of embarrassment, as described, were taken into account to adjust the questionnaire's structure and content.As for recall bias, it is possible to minimize its interference when the questions refer to more recent time frames for the target event.However, in some cases we maintained the 12-month time frame for asking about work absences due to illness, for example.The justification for adopting longer time frames in these and other situations was the need to follow consensuses in the literature in order to ensure the results' comparability with those from other studies.The answers' validity was verified via tests to measure internal consistency.For example, we attempted to identify any inconsistency between the answer concerning the reason for the work absence and reports of symptoms related to that reason.Next, a new assessment of the questionnaire was performed with other participants.The questionnaire's structure, content, and basis were addressed in the Educatel instructions manual 38 .

Thematic blocks
Following weeks of work, the questionnaire was approved, consisting of 54 short and simple questions, mostly with multiple-choice answers (closed questions).Relevant information such as the teacher's gender, age (date of birth), employment status, area of residence, and schooling was obtained from the 2014 School Census 24 in order to save time in the interview.The same source was used for the following data on the school: location, schooling level, administrative system, size according to number of teachers, census area, safe water supply, electricity, sewage disposal, garbage disposal, and the school's equipment and installations.

Telephone interview
The telephone interviews were conducted from October 2015 to March 2016.The team included 30 interviewers, two supervisors, and a general supervisor.They all received prior training and were accompanied by the study coordinators.Interviewees tend to have less patience answering questions by telephone when compared to face-to-face interviews.To avoid this unwanted effect, which might have interfered in the information's quality, we opted for a short interview.The team used a maximum of 8 minutes for the interview.This was a limiting factor, since some questions from validated scales had to be eliminated (Box 1), as mentioned.
Teachers were contacted first via a call to the landline telephone at the school where they worked.After confirming with the school assistant that the teacher worked there (eligibility criterion), the interview started right then if the teacher was able and agreed to answer.Some interventions were performed to adjust the approach to the school assistant that answered the first call looking for the selected teacher in order to schedule the interview per se or to interview the teacher right then whenever possible.If the teacher was not at the school or unable to participate in the interview at the time of the call, we attempted to obtain another telephone number or set another time according to the teacher's convenience and comfort.In case of impediments, further contacts were made on different days of the week and at different times until the interview was actually performed or the teacher effectively declined to participate.The number of attempts varied from one, in case of a successful call to initiate and conclude the interview, to fifteen, in cases when the interview was not initiated or had to be interrupted.During the first contact, the teacher was informed of the survey's website to access a 3-minute video explaining the objectives, ethical aspects, and institutional responsibility 39 .If the teacher was interested, he or she could also receive the video via WhatsApp.
The data were entered in real time thanks to the electronic system.In other words, the questions were read on a computer screen by an applier that directly and immediately recorded the answers digitally.The tool developed especially for this purpose allowed scheduling the interviews, automatically skipping questions that were not applicable due to previous answers and real time critique of the nonvalid answers, in addition to directly and continuously feeding the system's databank.Twenty percent of the interviews were randomly selected and supervised.In addition, 10% of the calls not converted into interviews were audited.The supervisor monitored the quality of the interviews by listening to the recordings and identifying tendencies, lapses, etc., which (when identified) were corrected immediately to decrease the measurement bias.Selection and training of the interviewers before the data collection began emphasized the importance of empathy, listening skills, capacity to answer the teacher's doubts without generating insecurity in communicating the survey's objectives 40 .

Final remarks
Progress in the field of workers' health requires empirically confirming or refuting plausible hypotheses.Despite its limitations, the telephone survey is a means to know how subjects perceive their job situation and to allow researchers to formulate hypotheses on the target phenomena.Such results Cad. Saúde Pública 2019; 35 Sup 1:e00108618 are also powerful for orienting measures to transform workplace conditions, backed by social and community constructs that are more robust than those generated through spreadsheets applied by occupational risk managers.
The self-report method in relation to the physical work environment, for example, or on the subject's reactions to job circumstances (satisfaction, for example) received some criticism in the 1990s.However, other techniques (direct measurement of ambient sound levels, for example) are also subject to distortions 41 , since there is variation in exposure to specific factors over the course of the day, the week, and the months, besides seasonal problems that can influence a factor's concentration in the microenvironment in drier or hotter seasons compared to other times of the year.The above-mentioned limitation is compensated for by the advantage of considering self-perception of health-related facts according to the subjects' own ideas concerning such facts.This is known as representation, which is related to the individual's constructs within a given social milieu, and whose health is not disconnected from these ideas.
Finally, the area where the selected teacher was located may not have been covered by the landline telephone system, as already mentioned and addressed by researchers using this strategy 42 .Post-stratification statistical adjustment procedures allow mitigating the effects of the bias related to telephone coverage 43 .Details on these procedures have been published in another article 25 .
The basis for the Educatel Study featured proper attention to the consensuses in the literature when elaborating the questions and drawing on validated and adapted scales 3,4,5 , as well as by the adoption of the telephone interview modality.Despite the limits of a survey for identifying relations between health and work, the results of the study analyzing a probabilistic and representative sample of Brazilian schoolteachers 25 will provide useful input for programs in the school system to implement the Brazilian National Plan for Education 9 .This methodological strategy was inspired by Brazilian national surveys on noncommunicable diseases and was an innovative initiative in the field of workers' health.

GHQ- 12 :
General Health Questionnaire; IBGE: Brazilian Institute of Geography and Statistics; JSS: Job Stress Scale; PNAD: Brazilian National Household Sample Survey; TALIS: Teaching and Learning International Survey; VIGITEL: Risk and Prospective Factors Surveillance System for Chronic Non-Communicable Diseases through Telephone Interview; V-RQOL: Voice-Related Quality of Life.Cad.Saúde Pública 2019; 35 Sup 1:e00108618 Contributors A. A. Assunção analyzed the literature and organized and wrote the article.A. M. Medeiros revised the literature and participated in writing the article.R. M. Claro and M. T. Vieira collaborated in writing the article.E. G. Maia collaborated in the data analysis and participated in writing the article.J. M. Andrade collaborated in the data analysis and interpretation and participated in writing the article.