ARTIGO ARTICLE Joint effect of paid working hours and multiple job holding on work absence due to health problems among basic education teachers in Brazil: the Educatel Study Efecto conjunto de las horas de trabajo remunerado y el pluriempleo sobre la ausencia al trabajo por problemas de salud en profesores de educacuión básica en Brasil: el Estudio Educatel Lucía Rodríguez-Loureiro 1,2,3 Lucía Artazcoz 1,2,4,5 Efeito conjunto das horas de trabalho María López-Ruiz 1,2,3,6 remunerado e de múltiplos empregos sobre Ada Ávila Assunção 7 o absenteísmo por motivo de saúde entre Fernando G. Benavides 1,2,3 professores de ensino básico no Brasil: Estudo Educatel doi: 10.1590/0102-311X00081118 Abstract Correspondence L. Rodríguez-Loureiro Centro de Investigación en Salud Laboral, Universitat Pompeu The objectives were to assess the joint effect of working hours paid per week Fabra. and multiple job holding on sickness absence, by sex, among basic education Edificio PRBB, C/Dr. Aiguader, 88, Barcelona / Barcelona – teachers in Brazil. This study is based on a survey carried out over a repre- 08003, España. sentative sample of 5,116 active basic education teachers in Brazil between lrloureiro92@gmail.com 2015 and 2016 (Educatel Study). We created a dummy variable to assess the 1 Centro de Investigación en Salud Laboral, Universitat Pompeu joint effect of weekly paid working hours [standard (35-40 hours); part-time Fabra, Barcelona, España. (< 35 hours); moderately long (41-50 hours); and very long (> 50 hours)] and 2 Centro de Investigación Biomédica en Red – Epidemiología y multiple job holding (working in several schools – no/yes). Working 35-40 Salud Pública, Madrid, España. 3 Instituto Hospital del Mar de Investigaciones Médicas, hours in one school was the reference category. We conducted Poisson regres- Barcelona, España. sion models with robust variance to obtain prevalence ratios (PR) and 95% 4 Agència de Salut Pública de Barcelona, Barcelona, España. confidence intervals (95%CI) of the association with self-certified sickness 5 Institut d'Investigació Biomèdica de Sant Pau, Barcelona, absence and medically certified sickness absence. Models were adjusted for España. age, type of contract and salary, and stratified by sex. Significant associations 6 Facultad Latinoamericana de Ciencias Sociales, Salamanca, with sickness absence were only found among teachers working in more than España. 7 Faculdade de Medicina, Universidade Federal de Minas one school. Associations with self-certified sickness absence were found among Gerais, Belo Horizonte, Brasil. women with standard and men with moderately long working hours, and for both women and men working > 50 hours (PR: 1.21, 95%CI: 1.09-1.35; PR: 1.40, 95%CI: 1.18-1.66; respectively). Associations with medically certi- fied sickness absence were found among teachers working > 50 hours, among women (PR: 1.30, 95%CI: 1.03-1.63) and men (PR: 1.41, 95%CI: 1.04-1.92). Teachers working longer hours in several schools could be suffering health problems, deriving in work absence. School Teachers; Occupational Health; Absenteeism; Socioeconomic Factors 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. Cad. Saúde Pública 2019; 35 Sup 1:e00081118 2 Rodríguez-Loureiro LR et al. Introduction The relationship between long working hours and health status has been extensively assessed in the literature. A recent systematic review reported associations between working over 40 hours weekly and a range of health outcomes. Although long working hours could lead to certain health problems, such as cardiovascular disease, diabetes, anxiety, depression or disability 1,2,3,4, findings are still con- tradictory 5. Thus, despite most studies have found consistent associations with working ≥ 55 hours weekly, results for moderately long working hours are still inconsistent 3. Moreover, differences among countries have been reported 6,7. In contrast, the effect of multiple job holding on health status has been scarcely studied. Studies in the United States have observed a higher risk of injuries and less sleep among multiple job-holders 8,9, while a study conducted in Denmark found no associations of multiple job holding with long-term sickness absence 10. Also, although the effect of working hours in this association has been discussed, its joint effect with multiple job holding has not been thoroughly assessed. Additionally, gender division of labor strongly determines the time allocated to paid work: where- as men tend to work longer hours due to their breadwinner role, women frequently work part-time, since they usually assume more care responsibilities. Hence, working longer hours or moonlighting could have a stronger impact on women’s health due to the combination of paid and family work 11,12. Therefore, although most previous research was based on men-only samples or did not separate the analysis by sex 3, any analysis of the relationship between working hours or multiple job holding and health status should be separated by sex and consider gender roles. In Brazil, the intensification of teachers’ work derived from the reforms in education has gradu- ally become a matter of concern 13. In 2008, the legal regulation established the implementation of an extra-class working day, exclusive dedication of the teachers to one school and better salaries, which represent important improvement for the quality of the school as well as for working and employ- ment conditions 14. Regarding work hours in the main job among Brazilian teachers, there has been a reverse in the downward trend observed between 2002 and 2006; and the proportion in the range of 36 to 40 weekly hours, which represented 32% in 2006, increased to 39% in 2011 and to 41% in 2013 15. Although this tendency coincides in time with the implementation of the new legislation, according to recent reports, these findings are not attributable to the law reform, but to the need of salary increases or the lack of teachers in a given educational network 16. The 2013 Teaching and Learning International Survey (TALIS) showed that Brazilian teachers spent 25 weekly hours teaching, exceeding in 6 hours the average of the Organization for Economic Co- operation and Development (OECD). Brazil had also one of the larger class sizes, with more than 30 students in each class, and one of the highest student-teachers ratio 17. Regarding exclusive dedica- tion, a study conducted in seven Brazilian states showed that the proportion of basic education teach- ers working in several schools reached 62% among secondary education teachers, 49% among teach- ers of primary education and 28% among teachers of childhood education 18. It should be taken into account that the teacher’s exclusive dedication to one school improves the student’s attainment 19. As far as we know, there are no studies about the joint association of working hours and mul- tiple job holding with sickness absence. Therefore, the aim of this study was to assess the relation- ship between the combination of multiple job holding and weekly paid working hours with work absence due to health problems and to examine potential gender differences among basic education teachers in Brazil. Methods Study design This cross-sectional study uses the information from the Educatel Study, a phone survey conducted over a representative sample of basic education teachers, held between October 2015 and March 2016 in Brazil. Briefly, as it has already been described 20, the information of the survey was extended using the registries of the Research Institute of the Brazilian Ministry of Education and Culture. Prefer- Cad. Saúde Pública 2019; 35 Sup 1:e00081118 PAID WORKING HOURS, WORKING IN SEVERAL SCHOOLS AND ABSENTEEISM 3 ably, the contact with the participant was made through the telephone number of the school and the interview was carried out in the participants’ workplace. Study population and sample The survey provides information about 6,510 basic education teachers aged 18 or older who were active the year before the questionnaire administration. Only active teachers at the time of the inter- view were included in the study (n = 5,116). The survey presented an 85% response rate 20. Variables Work absence was measured by asking the respondents if they had been absent from work for at least one day in the last 12 months because of a health problem. If they answered affirmatively, the respon- dents were subsequently asked if a doctor had evaluated the health problem and certified the absence. Two dichotomous variables (yes/no) were analyzed: (i) self-certified sickness absence; (ii) medically certified sickness absence. Self-reported work absence due to health problems, measured as having lost at least one day of work due to illness in the past 12 months, is used as a valid indicator to assess the prevalence of sickness absence among countries 21. In countries with official registries, several studies have proven its comparability with administrative databases 22. In fact, in the Whitehall II Study, sickness absence was conceptualized as an integrate health indicator, regardless of whether it is medically certified or self-notified 23. Using self-reported data on work absence due to health problems is a reliable alternative when official registries are lacking. The predictor variables were working in more than one school (“no”, “yes”) and weekly paid work- ing hours (categorized into “35-40h” as standard working hours”, “< 35h” as part-time, “41-50h” as moderately long working hours, and “> 50h” as very long working hours), including the total hours working in each school. To thoroughly explore the interaction between them, the joint effect analysis proposed by Szklo & Nieto 24 was performed. A dummy variable was created to assess the combina- tion of the two variables, examining the effect of the different subgroups on the health outcomes in comparison with the reference category: working 35-40 weekly hours in one school. The adjustment variables were age (“≤ 34 years”, “35-44 years”; “45-54 years”; “≥ 55 years”), type of contract (“Public administration: Permanent/Temporary”; “Private administration: Permanent/ Temporary”) and salary (“< 2 minimum wages”, “2-3 minimum wages”, “> 3 minimum wages”). Statistical analysis First, several bivariate analyses were performed to estimate the number and percentage of each study variable by sex, and to estimate the number and percentage of each study variable according to the main explicative variable by sex. Poisson regression models with robust variance were conducted to obtain crude and adjusted prevalence ratios (PR) and their 95% confidence intervals (95%CI) regard- ing the association between the joint effect of working in more than one school and weekly paid working hours on the self-certified sickness absence and medically certified sickness absence. Analy- ses were stratified by sex. All analyses were performed with the statistical program Stata, version 13 (https://www.stata.com). Results The general description of the sample is displayed in Table 1, showing that women had a higher preva- lence of self-certified sickness absence than men (54% and 45%, respectively), although this difference decreased for medically certified sickness absence (21% and 19%, respectively). Women worked more frequently standard working hours than men (35-40 hours weekly) and in exclusive dedication to one school. Tables 2 and 3 show the distribution of the outcomes and the predictor variables according to the composite variable. Among both sexes, the highest prevalence of work absence due to health problems Cad. Saúde Pública 2019; 35 Sup 1:e00081118 4 Rodríguez-Loureiro LR et al. Table 1 General description of the sample, by sex. Educatel Study, 2015-2016. Study variables Women Men n % n % Number of schools and weekly paid working hours Working in one school (hours) 35-40 823 25.6 353 18.6 < 30 935 29.0 480 25.3 41-50 55 1.7 31 1.6 > 50 47 1.5 22 1.2 Working in more than one school (hours) 35-40 530 16.5 328 17.3 < 30 292 9.1 219 11.6 41-50 232 7.2 191 10.1 > 50 306 9.5 272 14.3 Self-certified sickness absence 1,740 54.0 848 44.7 Medically certified sickness absence 684 21.2 356 18.8 Age (years) ≤ 34 992 30.8 643 33.9 35-44 982 30.5 588 31.0 45-54 895 27.8 442 23.3 > 55 351 10.9 223 11.8 Type of contract Public school Permanent 1,388 43.1 632 33.3 Temporary 521 16.2 293 15.5 Private school Permanent 806 25.0 675 35.6 Temporary 505 15.7 296 15.6 Salary (minimum wages) * <2 1,233 38.3 593 31.3 2-3 931 28.9 427 22.6 >3 1052 32.7 872 46.1 Total 3,220 100.0 1,896 100.0 * The variable salary has 4 missing values among women (0.1%) and 4 missing values among men (0.2%). was found among teachers working more than 50 hours in more than one school, being slightly higher among women than among men (for self-certified sickness absence: 64% and 57%, respectively; for medically certified sickness absence: 28% and 25%, respectively). Teachers working in several schools were more frequently hired in a private school with a permanent contract, especially among men. Besides, the proportion of teachers earning ≥ 3 minimum wages was much higher when working in one school compared to those working the same weekly hours in different schools. The association between the composite variable and self-certified sickness absence, by sex, is shown in Table 4. Among women, working part-time was negatively associated with work absence due to health problems, although these associations were not significant after adjustment. Women working in more than one school had a higher probability of experiencing self-certified sickness absence, both if they worked weekly 35-40 hours (PRa: 1.17, 95%CI: 1.07-1.29) and > 50 hours (PRa: 1.21, 95%CI: 1.09-1.35). Among men, there was a positive and significant association for those work- ing in several schools 41-50 hours (PRa: 1.35, 95%CI: 1.12-1.63) and > 50 hours weekly (PRa: 1.40, 95%CI: 1.18-1.66). Cad. Saúde Pública 2019; 35 Sup 1:e00081118 PAID WORKING HOURS, WORKING IN SEVERAL SCHOOLS AND ABSENTEEISM 5 Table 2 Distribution of outcome and explicative variables according to the number of schools and weekly hours worked. Women. Educatel Study, 2015-2016. Study variables Working in one school Working in more than one school 35-40 hours < 35 hours 41-50 hours > 50 hours 35-40 hours < 35 hours 41-50 hours > 50 hours n % n % n % n % n % n % n % n % Self-certified sickness 437 53.1 450 48.1 28 50.9 28 59.6 326 61.5 141 48.3 135 58.2 195 63.7 absence Medically certified 181 22.0 161 17.2 11 20.0 11 23.4 121 22.8 58 19.9 55 23.7 86 28.1 sickness absence Age (years) ≤ 34 244 29.6 287 30.7 18 32.7 17 36.2 169 31.9 99 33.9 78 33.6 80 26.1 35-44 273 33.2 254 27.2 18 32.7 9 19.1 178 33.6 80 27.4 83 35.8 87 28.4 45-54 211 25.6 270 28.9 11 20.0 13 27.7 143 27.0 81 27.7 61 26.3 105 34.3 > 55 95 11.5 124 13.3 8 14.5 8 17.0 40 7.5 32 11.0 10 4.3 34 11.1 Type of contract Public school Permanent 444 53.9 383 41.0 21 38.2 27 57.4 223 42.1 81 27.7 78 33.6 131 42.8 Temporary 131 15.9 209 22.4 9 16.4 6 12.8 71 13.4 51 17.5 23 9.9 21 6.9 Private school Permanent 94 11.4 117 12.5 10 18.2 6 12.8 202 38.1 121 41.4 115 49.6 141 46.1 Temporary 154 18.7 226 24.2 15 27.3 8 17.0 34 6.4 39 13.4 16 6.9 13 4.2 Salary (minimum wages) <2 182 22.2 447 47.9 14 25.5 14 29.8 231 43.7 155 53.1 98 42.2 92 30.1 2-3 252 30.7 248 26.6 11 20.0 17 36.2 171 32.3 80 27.4 62 26.7 90 29.4 >3 387 47.1 239 25.6 30 54.5 16 34.0 127 24.0 57 19.5 72 31.0 124 40.5 Total 823 100.0 935 100.0 55 100.0 47 100.0 530 100.0 292 100.0 232 100.0 306 100.0 Finally, Table 5 shows the association between the composite variable and medically certified sickness absence, by sex. For both women and men, associations were found among those working in more than one school and > 50 weekly hours (PRa: 1.30, 95%CI: 1.03-1.63; PRa: 1.41, 95%CI: 1.04- 1.92; respectively). Discussion This study assessing the joint effect of weekly paid working hours and multiple job holding on work absence due to health problems among basic education teachers in Brazil produced three main find- ings. First, we only found an association between working hours and self-certified/medically certi- fied sickness absence among teachers working in more than one school. Second, women working standard hours and men working moderately long hours in more than one school were more likely to have self-certified sickness absence. Lastly, for both women and men, to work very long weekly hours in more than one school was positively associated with self-certified and medically certified sickness absence. Our findings suggest an increased likelihood of being absent from work due to health problems only among Brazilian teachers working in several schools. Consistent with previous research 10, there seems to be an interaction between multiple job holding and long working hours, resulting in an increased risk for work absence due to health problems. However, when considering exclusively mul- tiple job holding, the aforementioned study found no associations with long-term sickness absence. In Cad. Saúde Pública 2019; 35 Sup 1:e00081118 6 Rodríguez-Loureiro LR et al. Table 3 Distribution of outcome and explicative variables according to the number of schools and weekly hours worked. Men. Educatel Study, 2015-2016. Study variables Working in one school Working in more than one school 35-40 hours < 35 hours 41-50 hours > 50 hours 35-40 hours < 35 hours 41-50 hours > 50 hours n % n % n % n % n % n % n % n % Self-certified sickness 139 39.4 178 37.1 13 41.9 11 50.0 148 45.1 101 46.1 104 54.5 154 56.6 absence Medically certified 64 18.1 72 15.0 7 22.6 2 9.1 59 18.0 43 19.6 40 20.9 69 25.4 sickness absence Age (years) ≤ 34 118 33.4 157 32.7 9 29.0 4 18.2 118 36.0 83 37.9 74 38.7 80 29.4 35-44 108 30.6 143 29.8 9 29.0 7 31.8 91 27.7 60 27.4 61 31.9 109 40.1 45-54 84 23.8 106 22.1 11 35.5 7 31.8 85 25.9 49 22.4 44 23.0 56 20.6 > 55 43 12.2 74 15.4 2 6.5 4 18.2 34 10.4 27 12.3 12 6.3 27 9.9 Type of contract Public school Permanent 161 45.6 182 37.9 10 32.3 13 59.1 82 25.0 49 22.4 47 24.6 88 32.4 Temporary 70 19.8 112 23.3 6 19.4 2 9.1 42 12.8 27 12.3 17 8.9 17 6.3 Private school Permanent 61 17.3 68 14.2 8 25.8 5 22.7 174 53.0 105 47.9 104 54.5 150 55.1 Temporary 61 17.3 118 24.6 7 22.6 2 9.1 30 9.1 38 17.4 23 12.0 17 6.3 Salary (minimum wages) <2 46 13.1 168 35.1 8 25.8 1 4.5 122 37.2 113 51.6 59 30.9 76 28.0 2-3 68 19.3 102 21.3 2 6.5 6 27.3 93 28.4 46 21.0 52 27.2 58 21.4 >3 238 67.6 208 43.5 21 67.7 15 68.2 113 34.5 60 27.4 80 41.9 137 50.6 Total 353 100.0 480 100.0 31 100.0 22 100.0 328 100.0 219 100.0 191 100.0 272 100.0 addition, despite the limited existing literature, long working hours alone had so far been negatively associated with sickness absence in both women and men 25,26. There are several plausible interpretations for our findings. Firstly, teachers working in more than one school might have worse working and employment conditions. These teachers might have mul- tiple part-time contracts, which could lead to stressful situations when trying to get full-time work- load done or having unpredictable working schedules or lower salaries than teachers with exclusive dedication to one school 27,28. In our sample, we observed remarkable lower salaries among teachers working in more than one school and with the same working hours than teachers with exclusive dedication (Tables 2 and 3). Teaching in Brazil is marked by an expansion of daily work, absence of remuneration for extra-class work and concomitant work in several schools. These conditions might jeopardize the teacher’s well-being 29,30,31. Moreover, teachers working in several schools might experience a lack of engagement with the students and a lack of school membership, which could also predispose work absence due to health problems 18. Secondly, teachers might be forced to work in more than one school due to financial pressures. In fact, a recent report has attributed to low salaries the increase in working hours among Brazilian teachers 15. Family financial stress has been described as a relevant factor that forces teachers to work longer hours, resulting in poorer health outcomes 32,33. It is likely that factors leading to working in several jobs or longer hours are non-mutually exclusive. As already suggested for working hours, the choice (voluntary or forced) may be a relevant factor shaping the relationship between multiple job holding and health 6,34,35. Several studies have observed that long working hours could not fully explain the higher risk of injuries and less sleep of workers with multiple job holding 8,9. Another study assessing the relationship between multiple job holding and long-term sickness absence only Cad. Saúde Pública 2019; 35 Sup 1:e00081118 PAID WORKING HOURS, WORKING IN SEVERAL SCHOOLS AND ABSENTEEISM 7 Table 4 Number, prevalence and association of the joint effect of weekly paid working hours and multiple job holding on self-certified sickness absence, according to sex. Educatel Study, 2015-2016. Main independent variable, by sex n % PRc 95%CI PRa 95%CI Women Number of schools and paid working hours Working in one school (hours) 35-40 437 53.1 1.00 1.00 < 35 450 48.1 0.91 0.83-0.99 * 0.93 0.85-1.02 41-50 28 50.9 0.96 0.73-1.25 0.98 0.75-1.28 > 50 28 59.6 1.12 0.88-1.43 1.11 0.87-1.43 Working in more than one school (hours) 35-40 326 61.5 1.16 1.06-1.27 ** 1.17 1.07-1.29 *** < 35 141 48.3 0.91 0.79-1.04 0.95 0.83-1.09 41-50 135 58.2 1.10 0.97-1.24 1.11 0.98-1.27 > 50 195 63.7 1.20 1.08-1.33 *** 1.21 1.09-1.35 *** Men Number of schools and paid working hours Working in one school (hours) 35-40 139 39.4 1.00 1.00 < 35 178 37.1 0.94 0.79-1.12 0.97 0.81-1.15 41-50 13 41.9 1.06 0.69-1.64 1.08 0.70-1.66 > 50 11 50.0 1.27 0.82-1.97 1.27 0.82-1.96 Working in more than one school (hours) 35-40 148 45.1 1.15 0.96-1.37 1.13 0.95-1.36 < 35 101 46.1 1.17 0.97-1.42 1.18 0.97-1.43 41-50 104 54.5 1.38 1.15-1.66 *** 1.35 1.12-1.63 ** > 50 154 56.6 1.44 1.22-1.70 *** 1.40 1.18-1.66 *** 95%CI: 95% confidence interval; PRa: adjusted prevalence ratio (for age, type of contract and salary); PRc: crude prevalence ratio. * p-value < 0.05; ** p-value < 0.01; *** p-value < 0.001. found associations among employees with multiple jobs working longer than 37 weekly hours. However, there were no interactions with working hours among workers with second jobs as self- employed 10. A possible hypothesis for our results would be that Brazilian teachers are a highly homo- geneous group, presenting similar incentives to work in several schools. Additionally, their condition of salaried workers might reduce their flexibility to balance work and family demands. Thirdly, there might be a different management of time use among teachers working in different schools compared to those working in one school. Prior research has evinced that multiple job hold- ers seem to have longer commuting hours and less leisure and rest time. Additionally, they seemed to allocate time during the weekdays to activities such as household chores, probably trying to gain leisure time for nonworking days 36. In our sample, only slight differences were observed in commut- ing time according to multiple job holding. However, teachers working in several schools used more frequently individual motorized transportation than teachers with exclusive dedication (results not shown), which could reflect the need to save time in the displacements between schools. However, this hypothesis is speculative and should be further investigated. Fourth, the results of self-certified sickness absence differed according to gender. Among women working standard hours and men working moderately long hours in several schools, positive associa- tions were found with self-certified sickness absence. Our first hypothesis was that women would be Cad. Saúde Pública 2019; 35 Sup 1:e00081118 8 Rodríguez-Loureiro LR et al. Table 5 Number, prevalence and association of the joint effect of weekly paid working hours and multiple job holding on medically certified sickness absence, according to sex. Educatel Study, 2015-2016. Main independent variable, by sex n % PRc 95%CI PRa 95%CI Women Number of schools and paid working hours Working in one school (hours) 35-40 181 22.0 1.00 1.00 < 35 161 17.2 0.78 0.65-0.95 * 0.84 0.70-1.03 41-50 11 20.0 0.91 0.53-1.57 0.93 0.54-1.60 > 50 11 23.4 1.06 0.62-1.81 1.07 0.63-1.83 Working in more than one school (hours) 35-40 121 22.8 1.04 0.85-1.27 1.10 0.89-1.36 < 35 58 19.9 0.90 0.69-1.18 0.99 0.75-1.30 41-50 55 23.7 1.08 0.83-1.40 1.14 0.87-1.49 > 50 86 28.1 1.28 1.02-1.59 * 1.30 1.03-1.63 * Men Number of schools and paid working hours Working in one school (hours) 35-40 64 18.1 1.00 1.00 < 35 72 15.0 0.83 0.61-1.13 0.91 0.66-1.23 41-50 7 22.6 1.25 0.63-2.48 1.31 0.67-2.57 > 50 2 9.1 0.50 0.13-1.91 0.48 0.12-1.82 Working in more than one school (hours) 35-40 59 18.0 0.99 0.72-1.37 1.04 0.75-1.44 < 35 43 19.6 1.08 0.76-1.53 1.21 0.85-1.72 41-50 40 20.9 1.16 0.81-1.65 1.17 0.82-1.67 > 50 69 25.4 1.40 1.047-1.89 * 1.41 1.04-1.92 * 95%CI: 95% confidence interval; PRa: adjusted prevalence ratio (for age, type of contract and salary); PRc: crude prevalence ratio. * p-value < 0.05. more affected by multiple job holding or longer working hours due to the different time patterns of paid work among genders. However, the associations were not altered after adjusting by the variables related to volume of unpaid care work, like taking care of children, hours of domestic work or civil status (results not shown). Some studies have highlighted that the traditional variables to measure work-family conflict might not be able to capture the complexity of the burden resulting in combin- ing work and care responsibilities 37. Nevertheless, we have not been able to prove this hypothesis in our study with the available variables. We therefore suggest that women working standard hours in several schools might be suffering all the disadvantages of working in several schools, but none of the benefits, such as greater salaries. Additionally, we found that women working standard hours in sev- eral schools who experienced self-certified sickness absence reported more frequently having sleep- ing problems because of being worried than their counterparts working moderately long working hours (69% and 63%, respectively). Several studies had already highlighted the relationship between sleeping problems and self-certified sickness absence, which might be the result of a need to restore from insufficient sleep 38. However, this hypothesis is still undeveloped and should be contrasted with further studies assessing the gender differences of the relationship between the joint effect of multiple job holding and paid working hours with sickness absence. Finally, we only found an association with medically certified sickness absence among teachers working very long working hours in several schools. In this sense, it is possible that findings on self- certified sickness absence and medically certified sickness absence need different interpretations. Cad. Saúde Pública 2019; 35 Sup 1:e00081118 PAID WORKING HOURS, WORKING IN SEVERAL SCHOOLS AND ABSENTEEISM 9 It has been suggested that self-certified sickness absences and medically certified sickness absences need to be analyzed separately. Self-certified sickness absence is sensitive to subjectivity and could therefore reflect minor health problems. In fact, several studies have considered that self-certified sickness absence could serve as a coping mechanism for adverse working conditions, such as long working hours or excessive workload 39. Thus, self-certified sickness absence could reflect a specific organizational culture within a workplace on a larger extent than health problems. In contrast, it has been observed that medically certified sickness absence is a strong predictor of all-cause mortality and specific causes of early retirement 40,41. Therefore, our findings suggest that teachers working very long weekly hours in several schools might develop severe health problems, constituting a risk factor for temporary and permanent disability, and early retirement. This study has some limitations. Firstly, since it is a cross-sectional study, we cannot discard reverse causation, although it is unlikely that teachers working longer hours are those who pres- ent poorer health status. Secondly, we did not include the educational level that teachers imparted, because around half of the teachers were teaching in several levels simultaneously. Finally, this survey gathered no information on the number of sickness absence spells during the previous 12 months, which would have enriched the discussion. Likewise, the study presents several strengths. Firstly, it is based on a representative sample of all the Brazilian basic education teachers. Secondly, the separate use of self-certified and medically certified sickness absence provides information about the severity of the health problem and the organizational functioning of the schools. Lastly, this is the first study assessing the joint effect of multiple job holding and working hours on health outcomes, using a comprehensive approach for the classification of working hours. In conclusion, working longer hours in several schools is a risk factor for work absence due to health problems. The teacher’s wellbeing is a crucial factor affecting the quality of education. Sick- ness absence constitutes key indicators to assess this important social and economic problem. Future actions directed to improve the working and employment conditions of teachers are needed. Contributors Additional informations L. Rodríguez-Loureiro participated in the discus- ORCID: Lucía Rodríguez-Loureiro (0000-0003- sion of the analysis strategy, cleaned and prepared 0092-3748); Lucía Artazcoz (0000-0002-6300- the dataset for the analysis, carried out the analyses, 5111); María López-Ruiz (0000-0003-3453-0408); participated in the interpretation and discussion of Ada Ávila Assunção (0000-0003-2123-0422); Fer- the results, drafted the manuscript and completed nando G. Benavides (0000-0003-0747-2660) its final version to be published. L. Artazcoz partici- pated in the discussion of the analysis strategy, made substantial contributions in the interpretation and Acknowledgments discussion of the results, and revised and approved the final version of the manuscript to be published. To Department of School Systems Linkage of M. López-Ruiz participated in the discussion of the the Brazilian Ministry of Education; the Brazil- analysis strategy, in the interpretation and discus- ian National Institute for Educational Studies and sion of the results, and revised and approved the Research “Anísio Teixeira” (INEP); and the Brazilian final version of the manuscript to be published. A. National Research Council (CNPq) for the financial A. Assunção participated in the conception and support. design of the study, in the selection of the sample, in the interpretation and discussion of the results, and revised and approved the final version of the manuscript to be published. F. G. Benavides par- ticipated in the discussion of the analysis strategy, made substantial contributions in the interpreta- tion and discussion of the results, and revised and approved the final version of the manuscript to be published. Cad. Saúde Pública 2019; 35 Sup 1:e00081118 10 Rodríguez-Loureiro LR et al. References 1. van der Hulst M. Long workhours and health. 14. Brasil. Lei no 11.738, de 16 de julho de 2008. Scand J Work Environ Health 2003; 29:171- Regulamenta a alínea “e” do inciso III do caput 88. do art. 60 do Ato das Disposições Constitucio- 2. Virtanen M, Heikkilä K, Jokela M, Ferrie J, nais Transitórias, para instituir o piso salarial Batty G, Vahtera J, et al. Long working hours profissional nacional para os profissionais do and coronary heart disease: a systematic re- magistério público da educação básica. Diário view and meta-analysis. Am J Epidemiol 2012; Oficial da União 2008; 17 jul. 176:586-96. 15. Departamento Intersindical de Estatística e 3. Bannai A, Tamakoshi A. The association be- Estudos Socioeconômicos. Transformações tween long working hours and health: a sys- recentes no perfil do docente das escolas esta- tematic review of epidemiological evidence. duais e municipais de educação básica. http:// Scand J Work Environ Health 2014; 40:5-18. www.dieese.org.br/notatecnica/2014/nota 4. Kivimäki M, Jokela M, Nyberg S, Singh-Ma- Tec141DocentesPnadvf.pdf (accessed on 26/ noux A, Fransson E, Alfredsson L, et al. Long Oct/2018). working hours and risk of coronary heart dis- 16. Oliveira DA, Dourado LF, Cabral Neto A, ease and stroke: a systematic review and meta- Cury CRJ, Oliveira JF, Pinto JMR, et al. Por analysis of published and unpublished data for um Plano Nacional de Educação (2011-2020) 603838 individuals. Lancet 2015; 386:1739- como política de Estado. Rev Bras Educ 2011; 46. 16:483-92. 5. Ganster DC, Rosen CC, Fisher GG. Long 17. Organisation for Economic Co-operation and working hours and well-being: what we know, Development. TALIS 2013. Results: an inter- what we do not know, and what we need to national perspective on teaching and learning. know. J Bus Psychol 2016; 33:25-39. Paris: OECD Publishing; 2014. 6. Artazcoz L, Cortès I, Escribà-Agüir V, Bartoll 18. Oliveira D, Vieira L. Condições do trabalho X, Basart H, Borrell C. Long working hours docente: uma análise a partir de sete estados and health status among employees in Europe: brasileiros. In: Oliveira D, Vieira L, organiza- between-country differences. Scand J Work dores. Trabalho na educação básica: a condição Environ Health 2013; 39:369-78. docente em sete estados brasileiros. Belo Hori- 7. Artazcoz L, Cortès I, Benavides FG, Escribà- zonte: Fino Traço Editora; 2012. p. 153-90. Agüir V, Bartoll X, Vargas H, et al. Long work- 19. Matijascic M. Professores da Educação Básica ing hours and health in Europe: gender and no Brasil: condições de vida, inserção no mer- welfare state differences in a context of eco- cado de trabalho e remuneração. http://repo nomic crisis. Health Place 2016; 40:161-8. sitorio.ipea.gov.br/bitstream/11058/7929/1/ 8. Marucci-Wellman HR, Willetts JL, Lin TC, td_2304.pdf (accessed on 26/Oct/2018). Brennan MJ, Verma SK. Work in multiple jobs 20. Vieira MT, Claro RM, Assunção AA. Sam- and the risk of injury in the US working popu- ple design and participation in the Educa- lation. Am J Public Health 2014; 104:134-42. tel Study. Cad Saúde Pública 2019; 35 Suppl 9. Marucci-Wellman HR, Lombardi DA, Wil- 1:e00167217. letts JL. Working multiple jobs over a day or 21. Gimeno D, Benavides FG, Benach J, Amick a week: short-term effects on sleep duration. BC. Distribution of sickness absence in the Chronobiol Int 2016; 33:630-49. European Union countries. Occup Environ 10. Bouwhuis S, Garde AH, Geuskens GA, Boot Med 2004; 61:867-9. CRL, Bongers PM, van der Beek AJ. The longi- 22. Ferrie JE, Kivimäki M, Head J, Shipley MJ, tudinal association between multiple job hold- Vahtera J, Marmot MG. A comparison of self- ing and long-term sickness absence among reported sickness absence with absences re- Danish employees: an explorative study using corded in employers’ registers: evidence from register-based data. Int Arch Occup Environ the Whitehall II study. Occup Environ Med Health 2017; 90:799-807. 2005; 62:74-9. 11. Abramo L. Inserção das mulheres no mercado 23. Marmot M, Feeney A, Shipley M, North F, de trabalho na América Latina: uma força de Syme SL. Sickness absence as a measure of trabalho secundária? [Doctoral Dissertation]. health status and functioning: from the UK São Paulo: Faculdade de Filosofia, Letras e Whitehall II study. J Epidemiol Community Ciências Humanas, Universidade de São Paulo; Health 1995; 49:124-30. 2007. 24. Szklo M, Nieto F. Defining and assessing het- 12. Artazcoz L, Cortès I, Borrell C, Escribà-Agüir erogeneity of effects: interaction. In: Nieto F, V, Cascant L. Gender perspective in the analy- Moore M, editors. Epidemiology beyond the sis of the relationship between long work- basics. Gaithersburg: Aspen Publishers; 2000. hours, health and health-related behavior. p. 211-53. Scand J Work Environ Health 2007; 33:344- 25. Niedhammer I, Chastang JF, Sultan-Taïeb H, 50. Vermeylen G, Parent-Thirion A. Psychoso- 13. Assunção AA, Oliveira DA. Intensificação do cial work factors and sickness absence in 31 trabalho e saúde dos professores. Educação & countries in Europe. Eur J Public Health 2013; Sociedade 2009; 30:349-72. 23:622-9. Cad. Saúde Pública 2019; 35 Sup 1:e00081118 PAID WORKING HOURS, WORKING IN SEVERAL SCHOOLS AND ABSENTEEISM 11 26. Lesuffleur T, Chastang J, Sandret N, Niedham- 35. López-Ruiz M, Benavides F, Vives A, Artazcoz mer I. Psychosocial factors at work and sick- L. Informal employment, unpaid care work, ness absence: results from the French National and health status in Spanish-speaking Central SUMER Survey. Am J Ind Med 2014; 57:695- American countries: a gender-based approach. 708. Int J Public Health 2017; 62:209-18. 27. Artazcoz L, Benach J, Borrell C, Cortès I. So- 36. Marucci-Wellman HR, Lin T-C, Willetts JL, cial inequalities in the impact of flexible em- Brennan MJ, Verma SK. Differences in time ployment on different domains of psychoso- use and activity patterns when adding a sec- cial health. J Epidemiol Community Health ond job: implications for health and safety in 2005; 59:761-7. the United States. Am J Public Health 2014; 28. Koukoulaki T. New trends in work environ- 104:1488-500. ment: new effects on safety. Saf Sci 2010; 37. Nilsen W, Skipstein A, Østby KA, Mykletun 48:936-42. A. Examination of the double burden hypoth- 29. Batista JB, Carlotto MS, Coutinho AS, Pereira esis: a systematic review of work-family con- DA, Augusto LG. O ambiente que adoece: con- flict and sickness absence. Eur J Public Health dições ambientais de trabalho do professor do 2017; 27:465-71. ensino fundamental. Cad Saúde Colet (Rio J.) 38. Rahkonen O, Lallukka T, Kronholm E, Vahtera 2010; 18:234-42. J, Lahelma E, Laaksonen M. Sleep problems 30. Tabeleão VP, Tomasi E, Neves SF. Qualidade and sickness absence among middle-aged em- de vida e esgotamento profissional entre do- ployees. Scand J Work Environ Health 2012; centes da rede pública de Ensino Médio e Fun- 38:47-55. damental no Sul do Brasil. Cad Saúde Pública 39. Ala-Mursula L, Vahtera J, Kouvonen A, Vään- 2011; 27:2401-8. änen A, Linna A, Pentti J, et al. Long hours in 31. Pereira EF, Teixeira CS, Andrade RD, Silva paid and domestic work and subsequent sick- -Lopes A. O trabalho docente e a qualidade de ness absence: does control over daily work- vida dos professores na educação básica. Rev ing hours matter ? Occup Environ Med 2006; Saúde Pública 2014; 16:221-31. 63:608-17. 32. Artazcoz L, Borrell C, Benach J. Gender in- 40. Vahtera J, Pentti J, Kivimäki M. Sickness ab- equalities in health among workers: the rela- sence as a predictor of mortality among male tion with family demands. J Epidemiol Com- and female employees. J Epidemiol Commu- munity Health 2001; 55:639-47. nity Health 2004; 58:321-6. 33. Artazcoz L, Cortés I, Puig-Barrachina V, Bena- 41. Kivimäki M, Forma P, Wikström J, Halmeen- vides FG, Escribà-Agüir V, Borrell C. Combin- mäki T, Pentti J, Elovainio M, et al. Sickness ing employment and family in Europe: the role absence as a risk marker of future disability of family policies in health. Eur J Public Health pension: the 10-town study. J Epidemiol Com- 2014; 24:649-55. munity Health 2004; 58:710-1. 34. Bamberry L, Campbell I. Multiple job hold- ers in Australia: motives and personal impact. Aust Bull Labour 2012; 38:293-314. Cad. Saúde Pública 2019; 35 Sup 1:e00081118 12 Rodríguez-Loureiro LR et al. Resumen Resumo El objetivo de este estudio fue evaluar el efecto O estudo teve como objetivos avaliar o efeito con- conjunto de las horas laborales pagadas semanal- junto das horas semanais de trabalho remunerado mente y pluriempleo, en relación con las ausencias e do fato de ter vários empregos sobre o absenteís- por enfermedad, según el por sexo, entre profesores mo por motivo de doença entre professores de en- de educación básica en Brasil. Este estudio se llevó sino básico, desagregado por sexo, no Brasil. Este a acabo sobre una encuesta de una muestra repre- estudo teve como base um inquérito realizado em sentativa de 5.116 profesores activos de educación uma amostra representativa de 5.116 professores básica en Brasil, entre 2015 y 2016 (Estudio Edu- de ensino básico em atividade no Brasil entre 2015 catel). Creamos una variable dummy para evaluar e 2016 (Estudo Educatel). Criamos uma variável el efecto conjunto de las horas laborales pagadas dummy para avaliar o efeito conjunto das horas semanalmente [estándar (35-40 horas); a tiempo semanais de trabalho remunerado [padrão (35-40 parcial (< 35 horas); moderadamente largas (41- horas); tempo parcial (< 35 horas); semana de tra- 50 horas); y muy largas (> 50 horas)] y el pluriem- balho moderadamente longa (41-50 horas) e muito pleo (trabajando en varias escuelas no/sí). Estar longa (> 50 horas)] e múltiplos empregos (trabalho trabajando 35-40h en una escuela fue la categoría em mais de uma escola – não/sim). A categoria de de referencia. Se realizaron modelos de regresión referência foi o trabalho em uma única escola du- de Poisson con varianza robusta para obtener la rante 35-40 horas semanais. Realizamos modelos razón de prevalencia (RP) e intervalos de 95% de de regressão de Poisson com variância robusta pa- confianza (IC95%) de la asociación con las ausen- ra obter razões de prevalência (RP) e intervalos de cias por enfermedad justificadas personalmente y confiança de 95% (IC95%) para a associação com las ausencias por enfermedad con certificado mé- ausência no trabalho por motivo de saúde alegado dico. Los modelos fueron ajustados por edad, tipo pelo próprio indivíduo, e com atestado médico. Os de contrato y salario, y estratificados por sexo. Las modelos foram ajustados para idade, tipo de vín- asociaciones significativas con ausencias por en- culo de trabalho e salário, e estratificados por sexo. fermedad se encontraron sólo entre profesores que Os resultados só mostraram associações signifi- trabajaban en más de una escuela. Las asociacio- cativas com absenteísmo por motivo de saúde em nes con las ausencias por enfermedad justificadas professores que trabalhavam em mais de uma es- personalmente se hallaron entre mujeres con horas cola. Foram observadas associações com ausência de trabajo estándar y hombres con horas de traba- por motivo de saúde alegado pelo próprio indiví- jo moderadamente largas, y para ambos, mujeres y duo em mulheres com semana de trabalho padrão hombres trabajando > 50 horas (RP: 1,21, IC95%: e em homens com semana moderadamente longa, 1,09-1,35; RP: 1,40, IC95%: 1,18-1,66; respecti- e em mulheres e homens que trabalhavam mais de vamente). Las asociaciones con las ausencias por 50 horas por semana (RP: 1,21, IC95%: 1,09-1,35; enfermedad con certificado médico se hallaron en- RP: 1,40, IC95%: 1,18-1,66; respectivamente). Fo- tre profesores trabajando > 50 horas, entre mujeres ram encontradas associações com ausência com (RP: 1,30, IC95%: 1,03-1,63) y hombres (RP: 1,41, atestado médico entre professores que trabalha- IC95%: 1,04-1,92). Los profesores que trabajan vam mais de 50 horas por semana, em mulheres más horas en varias escuelas podrían estar su- (RP: 1,30, IC95%: 1,03-1,63) e homens (RP: 1,41, friendo problemas de salud, ocasionando ausencias IC95%: 1,04-1,92). Professores brasileiros que tra- laborales. balham horas longas em várias escolas podem so- frer problemas de saúde, levando ao absenteísmo. Maestros; Salud Laboral; Absentismo; Factores Socioeconómicos Professores Escolares; Saúde do Trabalhador; Absenteísmo; Fatores Socioeconômicos Submitted on 25/Apr/2018 Final version resubmitted on 26/Oct/2018 Approved on 05/Nov/2018 Cad. Saúde Pública 2019; 35 Sup 1:e00081118