Background
Health problems related to poor indoor air quality (IAQ), sometimes referred to as sick-building syndrome or building-related symptoms, include headache, nausea, eye, nose and throat problems, chest tightness or shortness of breath, fatigue, chills and fever, dizziness, dry skin, or even clusters of serious health problems [
1]. In Finnish office worker population (n = 11 154), the most common symptoms of poor indoor air were irritated, stuffy, or runny nose (20%), itching, burning, or irritation of the eyes (17%), and fatigue (16%) [
2]. These symptoms are often relatively minor, non-specific, and common amongst the general population. Despite their minor nature, these symptoms may have a great effect on public health and incur costs to the economy through widespread absenteeism and lowered productivity among the affected workers [
3,
4].
Possible sources of indoor air pollutants are building materials, furniture, office equipment, heating, ventilation, air conditioning, detergents used for cleaning, and pollutants originating from people and their activities [
5]. Increasing outdoor ventilation, removing or managing sources of contaminants, and increasing air cleaning or filtration have shown beneficial health effects [
6].
The quality of IAQ and ventilation are often assessed using objective measures such as CO
2 or NO
2 concentrations [
3,
7‐
9]. Although these measures are reliable and valid, they measure the IAQ from a narrow point of view and large-scale studies are expensive and time-consuming. Subjective reports of IAQ, i.e., perceived IAQ, provide a wider scope of potential problems and subjective experience of symptoms. Good perceived IAQ has been associated with cool and dry air, [
10] low pollution loads in the office [
11], good ventilation systems [
12‐
14], the use of air purifiers [
5], and low levels of volatile organic compounds (VOCs) [
15].
However, with regard to the health effects of poor perceived IAQ, IAQ reports and the measured outcomes (health symptoms) usually come from the same source, leaving the results open to common-method bias. Perceived IAQ had only weak effects on health at follow-up in prospective, cross-lagged design. The results however indicated a reversed effect of health problems leading to increased IAQ complaints, which led the authors to conclude that having pre-existing health problems may lead to increased IAQ complaints even when adjusting for the overall level of complaining [
16].
It has been proposed that psychosocial stress, i.e. pathophysiological changes related to psychological responses to the social environment [
17], possibly mediates environmental symptoms by altering the body’s sensitivity to physical demands and toxic exposures. Research has linked psychosocial factors quite strongly with building-related symptoms, especially among women [
18,
19], but not on psychosocial factors alone: most of the studies have supported the conclusion that these symptoms are of multifactorial origin related to chemical, physical, biological and psychosocial factors [
18‐
23].
IAQ problems, such as poor ventilation, are common in schools worldwide [
6,
24‐
26]. Also pupil-related psychosocial problems are rather common, and they have shown to negatively associate with teachers’ health and subsequent sick leave [
27,
28]. Low pupil cohort socioeconomic composition at school [
29] and lack of school resources as measured by pupil-teacher ratio (PTR) [
30] have also been associated with teacher sick leave. Moreover, school location, i.e., school urbanicity, low pupil socioeconomic composition, and high PTR have been associated with poor IAQ [
26].
In this study, we used pupil evaluations of IAQ in school as a predictor of the subsequent health problems of their teachers, as indicated by sick leave. Specifically, we aimed to test whether the level of or a change in pupil-evaluated IAQ was predictive of teacher sick leave when prior sick leaves and changes in the pupil-related psychosocial factors, pupil school satisfaction [
27] and bullying [
28], were controlled for. Furthermore, we controlled for changes in pupil cohort socioeconomic composition and in the PTR at school.
Results
Table
1 presents the descriptive statistics. The PTR was high at both times in 77% of schools where a negative change in IAQ had occurred. However, also the pupil cohort socioeconomic composition was high in 67% of those schools. Schools in which the perceived IAQ was good at both times had fewer negative changes in pupil school satisfaction. Negative changes IAQ seemed to somewhat co-occur with negative changes in bullying.
Table 1
Descriptive statistics of study sample by the level of or change in indoor air quality (IAQ) as evaluated by pupils
Mean age, years | 47.9 | 47.6 | 48.1 | 48.4 | 47.6 | 0.57 |
Percentage of male teachers | 26.5 | 25.0 | 29.0 | 21.5 | 25.5 | 0.20 |
Percentage of teachers with fixed-term job contracts | 14.0 | 14.6 | 13.6 | 20.7 | 11.8 | 0.10 |
Percentage of special education teachers | 5.5 | 6.2 | 5.7 | 1.7 | 5.6 | 0.58 |
Percentage of teachers in schools with
|
high PTR* | 46.8 | 50.5 | 36.3 | 76.9 | 52.0 | <0.001 |
high pupil cohort socioeconomic composition** | 48.1 | 36.7 | 47.6 | 66.9 | 57.6 | <0.001 |
Negative change***
|
school satisfaction | 13.8 | 20.8 | 4.3 | 14.1 | 22.5 | <0.001 |
pupils being bullied | 20.9 | 28.0 | 20.6 | 25.6 | 10.5 | <0.001 |
pupils bullying others | 17.1 | 16.5 | 15.2 | 36.4 | 15.3 | <0.001 |
School location/follow-up time****
|
Metropolitan area | 66.5 | 77.9 | 53.7 | 100 | 64.9 | <0.001 |
Other | 33.5 | 22.1 | 46.3 | 0 | 35.1 | |
The school location (metropolitan area vs. other) indicated also the follow-up time. In the schools located in the Helsinki metropolitan area (with a 2 year follow-up), the perceived IAQ was more often poor at follow-up than in schools located in other towns (with a 4 year follow-up). The negative changes in the IAQ all occurred in schools located in the Helsinki metropolitan area despite the shorter follow-up time. (Table
1) Moreover, the PTR was high at both times in 82% of the metropolitan area schools, when the corresponding percentage in other areas was 47%. The pupil socioeconomic composition was high at both times in 59% of the metropolitan area schools, when the corresponding percentage in other areas was 26%. (Data not shown in tables.)
We tested whether the school location/follow-up time acted as an effect modifier in the association between IAQ and teacher sick leave. The interaction effect (PTR*IAQ) was however nonsignificant (p > .60). Thus, school location/follow-up time was used as a covariate in further analyses.
At baseline (in 2001–02), 58% (n = 971) had no sick leaves, 25% (n = 416) had short-term sick leaves, and 17% (n = 291) had long-term sick leaves (with or without short-term sick leaves). Both short- and long-term sick leaves increased during the follow-up: at follow-up (in 2004–05), of the teachers in the study, 37% (n = 618) had no sick leaves; 35% (n = 581) had short-term sick leaves only; and 28% (n = 479) had long-term sick leaves (with or without short-term sick leaves) in 2004–05. In the empty model (Table
2), variance in teachers’ sick leaves between schools or municipalities was nonsignificant, i.e., no schools or municipalities had either a particularly large or particularly small number of teacher sick leaves.
Table 2
Summary of random effects of associations between changing school environment and teacher sick leave
Short-term
|
Empty | 0.056 | 0.043 | 0.10 | 0 | - | - |
with IAQ | 0.019 | 0.037 | 0.20 | 0 | - | - |
Model I* | 0.031 | 0.042 | 0.23 | 0.039 | 0.036 | 0.14 |
Model II** | <0.001 | - | - | 0.028 | 0.035 | 0.21 |
Long-term
|
Empty | 0.038 | 0.049 | 0.22 | 0.016 | 0.021 | 0.21 |
with IAQ | 0.029 | 0.047 | 0.27 | 0.017 | 0.021 | 0.21 |
Model I* | <0.001 | - | - | 0.121 | 0.083 | 0.07 |
Model II** | <0.001 | - | - | 0.084 | 0.076 | 0.13 |
School environment as predictor of teacher sick leave
The pupil-reported IAQ improved slightly during the follow-up (t = 14.8[DF 1677], p < 0.001). The mean level of perceived IAQ was 2.7 (SD = 0.2) at baseline and 2.6 (SD = 0.2) at follow-up. At follow-up, 1072 (64%) teachers were in schools with above median perceived IAQ, and 606 (36%) were below the baseline determined median of perceived IAQ. Both the unadjusted and adjusted associations between the level of or change in perceived IAQ and teachers’ short-term sick leave are presented in the first part of Table
3, and the associations with long-term sick leaves are in the second part of Table
3 (Additional file
1: Table S1).
Table 3
Change in or the level of pupil-reported school environment as predictor of teachers’ short-term sick leave (vs. no sick leaves) in 2004–05
Women vs. men | | | 1.41 (1.08-1.86) | 0.01 | 1.45 (1.10-1.91) | 0.009 |
Age/10 years | | | 0.66 (0.57-0.76) | <0.001 | 0.67 (0.58-0.77) | <0.001 |
Special vs. general education (2004–05) | | | 0.82 (0.49-1.38) | 0.46 | 0.90 (0.53-1.50) | 0.68 |
Fixed-term vs. permanent job (2004–05) | | | 0.88 (0.62-1.26) | 0.49 | 0.89 (0.62-1.27) | 0.52 |
Short-term sick leaves in 01–02: yes vs. no | | | 3.05 (2.22-4.19) | <0.001 | 3.14 (2.28-4.32) | <0.001 |
School characteristics
|
School location/follow-up time: 2 vs. 4 years | | | 1.20 (0.79-1.83) | 0.39 | 1.22 (0.78-1.91) | 0.38 |
High vs. small PTR at school*** | | | | | 0.84 (0.54-1.29) | 0.41 |
Low vs. high pupil socioeconomic composition**** | | | | | 0.63 (0.43-0.90) | 0.01 |
Pupil school satisfaction
|
1 Poor at both times, 22 schools (n = 414) | | | | | 1.00 = Referent | |
2 Good at both times, 36 schools (n = 619) | | | | | 1.00 (0.69-1.44) | 0.99 |
3 Negative change; from good to poor, 13 schools (n = 232) | | | | | 1.78 (1.13-2.81) | 0.01 |
4 Positive change; from poor to good, 22 schools (n = 413) | | | | | 1.44 (0.98-2.11) | 0.06 |
Pupils being bullied
|
1 Poor at both times, 27 schools (n = 501) | | | | | Referent | |
2 Good at both times, 29 schools (n = 488) | | | | | 1.15 (0.79-1.69) | 0.47 |
3 Negative change; from good to poor, 18 schools (n = 350) | | | | | 0.97 (0.68-1.40) | 0.89 |
4 Positive change; from poor to good, 19 schools (n = 339) | | | | | 0.90 (0.60-1.36) | 0.61 |
Pupils bullying others
|
1 Poor at both times, 28 schools (n = 464) | | | | | Referent | |
2 Good at both times, 28 schools (n = 549) | | | | | 0.96 (0.66-1.40) | 0.83 |
3 Negative change; from good to poor, 16 schools (n = 287) | | | | | 1.51 (1.01-2.25) | 0.04 |
4 Positive change; from poor to good, 21 schools (n = 378) | | | | | 1.42 (0.97-2.08) | 0.07 |
Indoor air quality
|
1 Poor at both times, 28 schools (n = 485) | 1.00 = Referent | | Referent | | Referent | |
2 Good at both times, 37 schools (n = 699) | 0.60 (0.45-0.79) | <0.001 | 0.63 (0.46-0.86) | 0.003 | 0.61 (0.45-0.85) | 0.003 |
3 Negative change; from good to poor, 7 schools (n = 121) | 0.80 (0.50-1.30) | 0.37 | 0.78 (0.50-1.30) | 0.34 | 0.80 (0.47-1.35) | 0.40 |
4 Positive change; from poor to good, 21 schools (n = 373) | 0.59 (0.42-0.83) | 0.002 | 0.64 (0.45-0.92) | 0.02 | 0.60 (0.41-0.88) | 0.009 |
In the fully adjusted model, teachers in schools with good perceived IAQ at both times were at a 0.61 -fold (95% CI: 0.45-0.85) decreased risk of short-term sick leave compared to teachers in schools in which the perceived IAQ was poor at both times. The corresponding OR was 0.60 (95% CI: 0.41-0.88) when we compared these teachers to teachers in schools in which a beneficial change in perceived IAQ had occurred. A negative change in pupil school satisfaction (OR = 1.78, 95% CI: 1.13-2.81) and in pupils bullying others (OR = 1.51, 95% CI: 1.01-2.25) increased the risk of teachers’ short-term sick leaves compared to schools in which pupil satisfaction/pupil bullying was at poor, but stable, at both times.
The level of or change in pupil cohort socioeconomic composition was also associated with teacher short-term sick leaves. Teachers in schools with lower pupil socioeconomic composition at both times were at a decreased risk of short-term sick leaves compared to teachers in schools with higher pupil socioeconomic composition at both times (OR = 0.63, 95% CI: 0.43-0.90). Other factors predicting teacher short-term sick leaves were prior sick leaves, female gender, and younger age. School location/follow-up time was not associated with teacher sick leaves.
The level of or change in school environment factors were not associated with teachers’ long-term sick leaves. Factors associated with long-term sick leaves were: prior sick leaves, female gender, younger age, and permanent job contract. (Additional file
1: Table S1, continued from Table
3).
The relationship of IAQ and pupil-related psychosocial factors in predicting teacher sick leave
We found no evidence of mediation of pupil-related psychosocial factors between IAQ and sick leaves, i.e., the odds ratios were similar in the unadjusted and adjusted models (Table
3). We also tested for whether there was an interaction (effect modification) of change in IAQ and change in school satisfaction/pupils bullying others in explaining teacher sick leaves. The interaction effects were however nonsignificant (p > .50). Neither was there an interaction of change in IAQ and change in PTR (p>. 30), or change in pupil socioeconomic composition (p > .90).
Discussion
In this prospective study of 1678 teachers in 93 Finnish lower secondary schools, we examined whether school environment, as indicated by pupil-reported IAQ and pupil-related psychosocial factors predicted teacher sick leave. We found that good perceived IAQ at school was associated with decreased the risk of teachers’ short-term sick leave even after adjustment for baseline sick leaves, level of or a change in pupil cohort socioeconomic composition, the PTR at school, and pupil-related psychosocial factors. Moreover, improved IAQ at school also decreased the risk for short-term sick leaves.
The study is unique; we are not aware of any published studies regarding this issue. The results are plausible as short-term sick leaves are an indicator of minor or short-term health problems [
31], and poor IAQ has been linked to minor health problems [
3]. Along with poor perceived IAQ, negative changes in the pupil-related psychosocial factors seemed to increase the risk of teachers’ short-term sick leaves.
Based on this study, it seems that lower pupil cohort socioeconomic composition, pupil-related psychosocial problems, and prolonged problems in IAQ were somewhat clustered in the same schools, even though their effects on teacher sick leaves were different and independent from each other. Contrary to our expectations, there were more teachers’ short-term sick leaves in schools with higher pupil cohort socioeconomic composition than in schools with lower pupil cohort socioeconomic composition. This was however explained by the school location: short-term sick leaves were more prevalent in schools located in the metropolitan area, where the socioeconomic composition of pupils was higher than in other areas. Consequently, also IAQ problems were more prevalent in the metropolitan schools. However, the school location did not moderate the effect of IAQ on short-term leaves: when the IAQ was constantly poor, the risk for short-term leaves increased despite of school location.
Previous studies have suggested that the psychosocial work environment might contribute to the link between IAQ and health [
3,
18‐
23]. Comparison of models with and without adjustment for pupil-related psychosocial factors did not however indicate mediation or effect modification.
The major strengths of our study were that we were able to use exposure and outcome variables derived from independent data sources, objective measurement of sick leave from records with good coverage, and multilevel modeling taking into account the hierarchical nature of the data. We were also able to control for a large number of covariates on both the individual and school level, and use a prospective study design. As our measures came from independent sources, common method bias cannot explain our findings on the detrimental effects of constantly poor IAQ and pupil-related psychosocial problems on teacher health. Moreover, the survey on IAQ was conducted on the same month each year. Thus, the perceived changes in IAQ cannot be due to seasonal changes.
However, although the use of subjective evaluations of school IAQ is considered cost-effective, and these evaluations have shown associations with objective measures of IAQ [
10‐
15], the subjective measurement of IAQ is also a limitation of the study. Pupil evaluations on whether poor ventilation or indoor air at school hinders one’s school work may reflect, in addition to actual IAQ, pupils’ awareness of air quality factors, susceptibility to air quality factors, or psychological or socioeconomic factors indicating propensity to attribute school work performance to external (environmental) factors. Moreover, based on these subjective evaluations, we cannot evaluate what is specifically wrong with indoor air or ventilation at these schools. The schools may be either too warm or too cold; the air may be either stuffy or drafty. Along with poorer perceived IAQ, feelings of thermal discomfort (thermally warm) have been associated with poorer performance and physiological symptoms [
32], Previous research has also confirmed that of the indoor air pollutants, asbestos, carbon monoxide and carbon dioxide, nitrogen dioxide, formaldehyde, radon, VOCs, indoor biological allergens, fungi, bacteria, and viruses are associated with ill-health and poorer functioning [
3]. Moreover, outdoor pollutants may also contribute to indoor air quality [
33].
Another limitation of this study may relate to the healthy worker effect [
34]. It is possible that those that remained in the study group during the follow-up were healthier than those that dropped out, i.e., those who were no longer employed at follow-up, and thus were excluded from our analyses. This would suggest that the effects of IAQ on health would be underestimates.
Since lower secondary school in Finland lasts for three years (grades 7-9), the pupils evaluating IAQ and the psychosocial school environment in 2001/02, and those evaluating them in 2004/05 were largely different. This can be considered either a strength or a weakness; evaluations from different pupils decrease common method bias, but may also mean that the changes were not necessarily in the variables investigated but in pupil composition. Pupil composition, with regard to follow-up time and the level of or change in pupil cohort socioeconomic composition was, however, controlled for in our analysis.
Conclusion
To conclude, problems in the school environment, as indicated by poor IAQ, as well as pupil school dissatisfaction and bullying, are associated with increased risk of short-term sick leaves among teachers. The fact that sick leave levels decreased among teachers from schools in which perceptions of IAQ improved, suggests that such improvements have potentially positive effects on the health and well-being of teachers and on the costs incurred by health care and absenteeism, as well as on the well-being of pupils. Finally, constantly poor IAQ, psychosocial problems and low pupil socioeconomic composition seem to cluster in certain schools often located in metropolises.
Acknowledgements
The study was supported by the Academy of Finland (grant numbers 133535, 124322, 124271 and 129262); the New and Emerging Risks in Occupational Safety and Health (New OSH ERA) Research Programme, European Union; and the BUPA Foundation, UK. The funders had no role in design, in the collection, analysis, or interpretations of data, in the manuscript preparation, or in the decision to submit the manuscript for publication.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JE was the principal author of the paper and carried out the data analyses with help from JP and MV, who supervised the data analyses and participated in designing the study, and writing the paper. MK ja JV are the directors of the Finnish Public Sector Study; they helped in writing the paper and interpreting the results. RP is the director of the School Health Promotion Study, and participated in interpreting the results. TP, IK, SVS, and TO contributed to the interpretation of the results and edited the manuscript. All authors read and approved the final manuscript.