[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
December 2002

Association of Air Pollution With School Absenteeism Due to Illness

Author Affiliations

Copyright 2002 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2002

Arch Pediatr Adolesc Med. 2002;156(12):1235-1239. doi:10.1001/archpedi.156.12.1235

Objective  To assess the association of air pollution and school absenteeism among elementary students.

Design  Time-series analysis of air pollution and school absenteeism data with controlling for long-term trends, seasonality, day of the week, and holiday as well as meteorologic variables.

Setting  School absenteeism data for the period from March 2, 1996, to December 22, 1999, were collected from student attendance reports of one elementary school in Seoul, Korea.

Main Outcome Measures  The number of daily illness-related absences was analyzed against the daily levels of air pollution by generalized additive Poisson regression. The relative risks of absenteeism for air pollution exposure of interquartile ranges (the range from the lowest 25% of the value to the lowest 75% of the value) on absenteeism were calculated on the same day.

Results  Exposure to air pollutants such as particulate matter of 10 µm or less in aerodynamic diameter (PM10), sulfur dioxide, and ozone, was associated with illness-related absenteeism. The estimated relative risks were 1.06 (95% confidence interval, 1.04-1.09) per 42.1-µg/m3 increase in PM10, 1.09 (95% confidence interval, 1.07-1.12) per 5.68–part per billion increase in sulfur dioxide, and 1.08 (95% confidence interval, 1.06-1.11) per 15.94–part per billion increase in ozone. There was no significant relationship between nitrogen dioxide level and illness-related absenteeism.

Conclusion  Air pollution is associated with illness-related absences among elementary students.

EXPOSURE TO ambient air pollution is one of the main causes of respiratory disease among children. Air pollutants, including ozone, nitrogen dioxide, and particulate matter, aggravate respiratory symptoms and illnesses and increase visits to clinics or emergency departments.1-4 For children, short-term variations in level of air pollutants are associated with hospital admission due to respiratory symptoms.5 Nitrogen dioxide is associated with respiratory diseases in children younger than 2 years.6,7 Charpin et al8 reported that moderate daily changes in sulfur dioxide levels induce a significant but transient increase in the prevalence of respiratory symptoms in children. Particulate matter of 10 µm or less in aerodynamic diameter (PM10) was also found to be a pollutant associated with respiratory symptoms and lung function changes.9 Among elementary students, elevated PM10 concentration was associated with a statistically significant decline in peak expiratory flow, and it increased the incidence of respiratory diseases as well as the use of asthma medication.9

Such effects of air pollution on children's health may increase school absences. Diette et al10 reported that school attendance may be affected by frequent awakening at night in children with asthma aggravated by air pollution. Children with asthma were reported to be absent 2 more days than the children without asthma in a school year.11 Therefore, we hypothesized that air pollution leads to an increase in the number of illness-related absences.

Absenteeism is an important index that relates not only to morbidity but also directly to daily activities. School absences may lead to a student's decline in school performance and overall education received. However, a number of studies that have dealt with absenteeism associated with air pollution produced inconsistent results.12-16 This study assesses the effects of air pollution on school absenteeism among elementary students after controlling for long-term trends, seasonality, day of the week, holiday, and meteorologic influences.


The protocol was approved by the institutional review board on human subjects at Ewha Woman's University, Seoul, Korea.

Absence data

School absentee data in the attendance reports of the students of one elementary school in Seoul were collected. This school is located in Kangnam Ku, where traffic is particularly heavy. The grades at this school ranged from the first to sixth grade, with 5 classes assigned per grade, and typically 45 students were in each class. Average enrollment count was 1264 (671 boys and 593 girls) in 1996. The study period extended from March 2, 1996, through December 22, 1999. An absence was defined as a day when a student did not attend school while school was in session. An absent day as a result of changing schools was not counted as an absence in this study. The numbers of illness-related and non–illness-related absences were counted separately. A teacher recorded an illness-related absence, by parent's report, in the attendance record for each class.

Air pollution data

The Ministry of Environment provided data on air pollution for Seoul. The pollution status of Seoul was evaluated by measuring pollution exposure during the study period at 27 monitoring sites. Levels of PM10 (β-ray absorption method), carbon monoxide (nondispersive infrared photometry), nitrogen dioxide (chemiluminescence), sulfur dioxide (ultraviolet photometry), and ozone (ultraviolet photometry) were measured hourly. Twenty-four–hour pollutant concentration averages were constructed. For ozone, an 8-hour daytime average was used instead of a 24-hour average.

Meteorologic information from a station in the central part of Seoul was obtained from the National Meteorological Office, including 24-hour average temperature and relative humidity. Except for occasionally missing or excluded observations, data for air pollution and meteorologic measures were available for the period from March 1, 1996, to December 22, 1999.

Statistical analysis

The absence data and exposure data were merged to create an analysis file with days. To allow for nonparametric relations between absence and predictor variables with respect to time and weather, we applied LOESS, a moving regression smoother, into a time-series analysis of the counts by means of a generalized additive model17 with a log link and Poisson error allowing overdispersion. We also used a locally weighted running-line smoother18 to check graphically the linearity assumption between air pollution and school absenteeism.

A model that best fit the data was selected with the use of Akaike's information criteria.19 Because the effects of pollutant exposure were likely to occur over several days, 1-day lag and 7-day moving average of the pollutant level were considered as well as the same-day concentrations. Because the same-day pollution best fit the model, we used the same-day concentration as the air pollution exposure in the results. To compare the effect of pollutants on school absence, we calculated the relative risks of illness-related absence in units of interquartile ranges (range from the lowest 25% of the value to the lowest 75% of the value) of each air pollutant. We used 2-pollutant models to estimate the effects of a pollutant after controlling for another pollutant. All analysis was done with S-PLUS 2000 (MathSoft Engineering & Education, Inc, Cambridge Mass).


The average number of daily absences was 5.89 (SD, 4.04) and the number due to illness was 4.20 (SD, 3.50). On the basis of the enrollment in 1996, the absence rate was 4.66 (95% confidence interval [CI], 0.94-9.38) per 1000 student-days and the illness-related absence rate was 3.69 (95% CI, 0.43-7.95) per 1000 student-days.

Table 1 summarizes the statistics on air pollution and weather in Seoul during the study period. The 24-hour average concentrations of PM10, nitrogen dioxide, sulfur dioxide, and carbon monoxide were 68.11 µg/m3, 33.12 parts per billion (ppb), 9.19 ppb, and 1.11 ppm, respectively. The 8-hour average concentration of ozone was 22.86 ppb.

Table 1. 
Summary Statistics for Concentrations of PM10, Nitrogen Dioxide, Ozone, Sulfur Dioxide, and Carbon Monoxide in Seoul, South Korea, 1996-1999*
Summary Statistics for Concentrations of PM10, Nitrogen Dioxide, Ozone, Sulfur Dioxide, and Carbon Monoxide in Seoul, South Korea, 1996-1999*

Table 2 shows the correlation matrix for the air pollutant concentrations during the study period. The concentrations of carbon monoxide, nitrogen dioxide, sulfur dioxide, and PM10 positively correlated with each other (0.56≤r≤0.76). However, the concentration of ozone negatively correlated with that of the other pollutants.

Table 2. 
Pearson Correlation Coefficients for Pollution Variables in Seoul, South Korea, 1996-1999*
Pearson Correlation Coefficients for Pollution Variables in Seoul, South Korea, 1996-1999*

We compared relative risks of non–illness-related and illness-related absence for the interquartile range increase of air pollutant levels (Table 3). Illness-related absences occurred more frequently when levels of PM10, sulfur dioxide, and ozone increased. Relative risks of the absences for the interquartile range increase of the pollutants were 1.06 (95% CI, 1.04-1.09), 1.09 (95% CI, 1.07-1.12), and 1.08 (95% CI, 1.06-1.11), respectively. Figure 1 shows plots of the relationship between these air pollutants and illness-related absenteeism. There was clearly a relatively linear positive relationship between the number of daily illness-related absences and concentrations of PM10, ozone, and sulfur dioxide. The effects on illness-related absenteeism changed little when 2 pollutants were included in the model together (Table 4). However, a statistically significant association between nitrogen dioxide and absenteeism among elementary school students was not observed.

Table 3. 
Relative Risk and 95% Confidence Interval in Absenteeism Due to an Interquartile Range Increase in PM10, Nitrogen Dioxide, Ozone, Sulfur Dioxide, and Carbon Monoxide, 1996-1999*
Relative Risk and 95% Confidence Interval in Absenteeism Due to an Interquartile Range Increase in PM10, Nitrogen Dioxide, Ozone, Sulfur Dioxide, and Carbon Monoxide, 1996-1999*
Relationship between concentrations of particulate matter of 10 µm or less in aerodynamic diameter (PM10) (A), sulfur dioxide (B), and ozone (C) and illness-related absenteeism; ppb indicates parts per billion.

Relationship between concentrations of particulate matter of 10 µm or less in aerodynamic diameter (PM10) (A), sulfur dioxide (B), and ozone (C) and illness-related absenteeism; ppb indicates parts per billion.

Table 4. 
Relative Risk (RR) and 95% Confidence Interval (CI) in Illness-Related Absenteeism Due to an Interquartile Range Increase in PM10 (42.1 µg/m3), Ozone (15.94 ppb), and Sulfur Dioxide (5.68 ppb) Using 2-Pollutant Models*
Relative Risk (RR) and 95% Confidence Interval (CI) in Illness-Related Absenteeism Due to an Interquartile Range Increase in PM10 (42.1 µg/m3), Ozone (15.94 ppb), and Sulfur Dioxide (5.68 ppb) Using 2-Pollutant Models*

For non–illness-related absenteeism, air pollutants were not associated statistically significantly with absences, except for ozone and sulfur dioxide. Concentrations of ozone and sulfur dioxide were negatively associated with non–illness-related absences.


Absenteeism is considered the index for showing the effect of air pollutants on daily activity before severe adverse events, such as hospitalization, occur. Work absences or restricted activity days have been reported to be associated with air pollution.20-22 However, the number of studies on the relationship between school absenteeism and air pollutants has been limited. A study that was done in Finland12 showed a significant association between levels of sulfur dioxide and absenteeism from day care centers, schools, and workplaces. Ransom and Pope13 found an association between PM10 and overall absenteeism among kindergarten children in a study done in Utah Valley, Utah. They observed that a 100-µg/m3 increase in the 28-day moving average of PM10 was associated with a 40% increase in overall absences. This association was observed even at PM10 levels below 150 µg/m3. Chen et al14 found that carbon monoxide and ozone levels were significant predictors of daily absenteeism in elementary schools. For every 1.0-ppm and 50-ppb increase in carbon monoxide and ozone levels, the absence rate increased 3.79% and 13.01%, respectively. However, PM10 concentration was negatively correlated with school absenteeism.

Although most absences are associated with illness, there could be other reasons for the absences. Therefore, illness-related absences rather than overall absences are more important for evaluating the effects of air pollutants on health. In a study done in Mexico, ozone was associated with absenteeism related to respiratory diseases among preschool children.23 However, other studies have not shown an association between air pollutants and absenteeism due to illness. Wayne and Wehrle24 failed to find a risk of air pollutants in respiratory disease related to school absenteeism in California. In a recent study, Gilliland et al16 reported an association between PM10 and all absences except illness-related ones. Increased daily PM10 exposure was associated with increases only in non–illness-related absences. A change of 10 µg/m3 in PM10 was associated with a 22.8% increase in all types of school absences and a 97.7% increase in non–illness-related absences, but a 5.7% increase in illness-related absences. However, an increase in ozone levels was associated with increases in respiratory illness–related school absences in children. Nitrogen dioxide had only a weak association with school absenteeism.

In our study, absenteeism was classified as illness-related and non–illness-related absenteeism. Increased levels of air pollutants such as PM10, sulfur dioxide, and ozone were associated with the number of illness-related absences. Acute effect of nitrogen dioxide on school absenteeism was not shown in this study. Although nitrogen dioxide exposure may be associated with respiratory symptoms, we could not find any evidence that symptoms from nitrogen dioxide exposure resulted in school absences.16 Gilliland et al16 also reported that acute effects of nitrogen dioxide on school absenteeism were not observed.

Our results show that air pollutants such as ozone and sulfur dioxide are protective for the non–illness-related absences. Other pollutants also tended to decrease the number of non–illness-related absences. Because the causes of non–illness-related absences were not ascertained in our study, it is hard to explain why air pollutants decrease the number of such absences. One possible explanation is that air pollution decreases outdoor activities such as family tours or visits, thereby increasing attendance at school. Further studies including detailed causes of absences are needed.

In this study, the effect of air pollution on absenteeism, both illness related and non–illness related, was investigated separately. Because attendance records were used to obtain data, there may be no ascertainment bias. On the other hand, our study had some limitations. The specific reasons for non–illness-related absences could not be explored because we depended on the past records. The reasons for non–illness-related absences vary. Students might have such absences when they go away to visit relatives or have troubles with their teacher or friends. There was no detailed information about the reason for these absences in the attendance report. Teachers distinguish between an absence related to illness and that not related to illness by relying on the parent's report. Since absence during elementary school does not affect performance grade and entrance to a higher level of school, there is no reason not to believe the parent's report about the reason for absence.

In addition to the limitations, we did not consider individual-level covariates in the model. Even though including individual-level covariates in the analysis would give greater statistical power than the time-series model,25 our results clearly showed the significant effect of pollutants on illness-related absences.

In conclusion, air pollutants were associated with illness-related absences, which suggests that air pollution is closely linked to daily activities, particularly of school-aged children.


Accepted for publication August 6, 2002.

This study was funded by grant number 02-PJ1-PG1-CH03-0001 from the Korea Health 21 R&D Project, Ministry of Health & Welfare, Gwacheon, Korea.

Corresponding author and reprints: Hyesook Park, MD, PhD, Department of Preventive Medicine, College of Medicine, Ewha Woman's University, 911-1, Mok-6-dong, Yangcheon-ku, Seoul, 158-710, Korea (e-mail: hpark@ewha.ac.kr).

What This Study Adds

It has been known air pollution leads to respiratory illness among children. A few studies have shown that air pollution is related to school absence, but they have produced inconsistent results. Illness-related school absences are considered a comprehensive index for evaluating health-related effects of air pollution on children.

This study suggests that air pollution increases illness-related school absences. This result provides an important basis for establishing policy for ambient environment.

Braga  AFSaldiva  PHNPereira  LA  et al.  Health effects of air pollution exposure on children and adolescents in São Paulo, Brazil.  Pediatr Pulmonol. 2001;31106- 113Google ScholarCrossref
Ostro  BDEskeland  GSSanchez  JMFeyzioglu  T Air pollution and health effects: a study of medical visits among children in Santiago, Chile.  Environ Health Perspect. 1999;10769- 73Google ScholarCrossref
Gouveia  NFletcher  T Respiratory diseases in children and outdoor air pollution in São Paulo, Brazil: a time series analysis.  Occup Environ Med. 2000;57477- 483Google ScholarCrossref
Pikhart  HBobak  MKriz  B  et al.  Outdoor air concentrations of nitrogen dioxide and sulfur dioxide and prevalence of wheezing in school children.  Epidemiology. 2000;11153- 160Google ScholarCrossref
Schwartz  JSpix  CWichmann  HEMalin  E Air pollution and acute respiratory illness in five German communities.  Environ Res. 1991;561- 14Google ScholarCrossref
Speizer  FEFerris Jr  BBishop  YMSpengler  J Respiratory disease rates and pulmonary function in children associated with NO2 exposure.  Am Rev Respir Dis. 1980;1213- 10Google Scholar
Shima  MAdachi  M Effect of outdoor and indoor nitrogen dioxide on respiratory symptoms in schoolchildren.  Int J Epidemiol. 2000;29862- 870Google ScholarCrossref
Charpin  DKleisbauer  JPFondarai  JGraland  BViala  AGouezo  F Respiratory symptoms and air pollution changes in children: the Gardanne Coal-Basin Study.  Arch Environ Health. 1988;4322- 27Google ScholarCrossref
Pope III  CA Respiratory hospital admissions associated with PM10 pollution in Utah, Salt Lake, and Cache Valleys.  Arch Environ Health. 1991;4690- 97Google ScholarCrossref
Diette  GBMarkson  LSkinner  EANguyen  TTAlgatt-Bergstrom  PWu  AW Nocturnal asthma in children affects school attendance, school performance, and parents' work attendance.  Arch Pediatr Adolesc Med. 2000;154923- 928Google ScholarCrossref
Silverstein  MDMair  JEKatusic  SKWollan  PCO'Connell  EJYunginger  JW School attendance and school performance: a population-based study of children with asthma.  J Pediatr. 2001;139278- 283Google ScholarCrossref
Ponka  A Absenteeism and respiratory disease among children and adults in Helsinki in relation to low-level air pollution and temperature.  Environ Res. 1990;5234- 46Google ScholarCrossref
Ransom  MRPope III  CA Elementary school absences and PM10 pollution in Utah Valley.  Environ Res. 1992;58204- 219Google ScholarCrossref
Chen  LJennison  BLYang  WOmaye  ST Elementary school absenteeism and air pollution.  Inhal Toxicol. 2000;12997- 1016Google ScholarCrossref
Makino  K Association of school absence with air pollution in areas around arterial roads.  J Epidemiol. 2000;10292- 299Google ScholarCrossref
Gilliland  FDBerhane  KRappaport  EB  et al.  The effects of ambient air pollution on school absenteeism due to respiratory illnesses.  Epidemiology. 2001;1243- 54Google ScholarCrossref
Hastie  TJTibshirani  RJ Generalized Additive Models.  London, England Chapman & Hall1990;
Cleveland  WSDevin  SJ Robust locally-weighted regression and smoothing scatterplots.  J Am Stat Assoc. 1979;74829- 836Google ScholarCrossref
Akaike  H Statistical predictor identification.  Ann Inst Stat Math. 1970;22203- 217Google ScholarCrossref
Ostro  BD The effects of air pollution on work loss and morbidity.  J Environ Econ Manage. 1983;10371- 382Google ScholarCrossref
Ostro  BD Air pollution and morbidity revisited: a specification test.  J Environ Econ Manage. 1987;1487- 98Google ScholarCrossref
Portney  PMullahy  J Urban air quality and acute respiratory illness.  J Urban Econ. 1986;2021- 38Google ScholarCrossref
Romieu  ILugo  MCVelasco  SRSánchez  SMenses  FHernández  M Air pollution and school absenteeism among children in Mexico City.  Am J Epidemiol. 1992;1361524- 1531Google Scholar
Wayne  WSWehrle  PF Oxidant air pollution and school absenteeism.  Arch Environ Health. 1969;19315- 322Google ScholarCrossref
Hwang  JSChen  YJWang  JDLai  YMYang  CYChan  CC Subject-domain approach to the study of air pollution effects on schoolchildren's illness absence.  Am J Epidemiol. 2000;15267- 74Google ScholarCrossref