Subject-domain approach to the study of air pollution effects on schoolchildren's illness absence

Jing Shiang Hwang, Yi Ju Chen, Jung-Der Wang, Yu Min Lai, Chun Yuh Yang, Chang Chuan Chan

Research output: Contribution to journalArticlepeer-review

15 Citations (Scopus)

Abstract

In this paper, the authors propose a new statistical modeling technique, the subject-domain approach, which is theoretically proven to be equivalent to the time-domain approach in detecting an association between exposure and response with time trends. The authors use an empirical data set from a school absence monitoring study conducted during the 1994-1995 school year in Taiwan to demonstrate this subject-domain approach's application to environmental epidemiologic studies. Because the subject-domain models can control the influential personal confounding factors in the models, they show greater statistical power than the traditional time-domain approaches in determining the relation between air pollution and illness absences. The authors' models found that the schoolchildren's risks of illness absence were significantly related to acute exposures to nitrogen dioxide and nitrogen oxides with a 1-day lag (p < 0.01) at levels below the World Health Organization's guidelines. By contrast, the authors could not detect significant associations between air pollution and schoolchildren's absenteeism using time-domain approaches. Such findings imply that the models built on subject domain may be a general solution to the problem of the ecologic fallacy, which is commonly encountered in environmental and social epidemiologic studies.

Original languageEnglish
Pages (from-to)67-74
Number of pages8
JournalAmerican Journal of Epidemiology
Volume152
Issue number1
DOIs
Publication statusPublished - 2000 Jul 1

All Science Journal Classification (ASJC) codes

  • Epidemiology

Fingerprint Dive into the research topics of 'Subject-domain approach to the study of air pollution effects on schoolchildren's illness absence'. Together they form a unique fingerprint.

Cite this