Association between ozone and emergency department visits: Application of geostatistics and geographic information systems (GIS)

Background. Ozone has been recognized as the most widespread and persistent urban pollution problem with demonstrated effects on health outcomes. However, there have been a limited number of studies using both temporal and spatial characteristics to study these relationships.Purpose. Using traditional statistics and geostatistics in combination with GIS, the association between ozone concentration and emergency department (ED) admissions for cardiovascular and respiratory conditions were examined at the ZIP code level.Methods. This study used multiple large datasets representing data for Maryland in the summer, 2002: (1) ozone data from the EPA, (2) weather factors from the National Climate Data Center, (3) socioeconomic status (SES) data from the Census 2000, and (4) ED visit data from the Maryland Health Services Cost Review Commission. Block kriging, a geostatistical technique, was applied to estimate daily ozone and weather factors (i.e., temperature and relative humidity) by ZIP-day level. These data were linked to aggregated ED visits. Poisson regression was used.Results. For 30 days of the 92-day study period, block kriged ozone concentration exceeded the EPA's national air quality standards for 8-hr ozone concentration, 85 parts per billion (ppb). A 10 ppb increment of 8-hr ozone as a 3-day average was associated with increased respiratory ED visits by 3.1% (95% CI: 1.8, 4.5) even after adjusting for weather factors, SES, and day of week. For cardiovascular ED visits, a 10 ppb increment of 1-hr ozone as a 5-day average increased by 4.7% (95% CI: 3.1, 6.3).Conclusion/Implication. Findings suggest that respiratory and cardiovascular ED visits increased even at lower ozone concentration than the EPA's air quality standards. As few studies examined the relationship between ozone and cardiovascular ED visits at a small area level, this study showed a statistical significance of ozone effects on increasing cardiovascular ED visits. Moreover, a limited number of studies have controlled for SES in examining relationship between air pollution and health, whereas this study adjusted for SES as a confounding factor of ozone effects on health outcomes. This study suggests that the integration of large datasets from discrete sources and the use of geostatistics and GIS can facilitate population health assessment. Future studies can use a similar approach to investigate other environmental risk factors and patient health outcomes at the population level.