Assessing the Short-Term Effects of Air Pollution on Morbidity; Use of a Subject-Specific Morbidity Index to Analyze Three Years of Pollution and Daily Physician-Visits of 40,000 Subjects

Abstract

P-051 Background: Previous time-series studies, on the short-term effect of air pollution were mostly ecological and relied mainly on hospital visits, which represent only a small fraction of total morbidity. Our study uses physician visits for a respiratory or cardiovascular indication, in a large population, as a health outcome, to estimate earlier pollutions effects using a subject-level design. Methods: Subjects were toddlers and elderly, in Tel Aviv area, who visited a primary doctor during 1999–2001. Daily air pollution measurements were collected from monitoring stations besides the daily visits. The relationship between visits and exposure, accounting for seasonality and meteorology was performed at the community-level by a Poisson regression; and at the subject-level by a logistic regression that included a morbidity index that we developed. Results: In both age groups, a positive association (OR > 1) was found between all the pollutants and the likelihood of a physician visit, with the exception of Ozone. In the elderly age group, an inter-quartile change in exposure level led to a small, but statistically significant, increase in the probability of a physician visit of 4% (SO2, PM2.5), 2% (NOX, CO), 0.8% (PM10). Point estimates were slightly higher in the subject-level results compared to community-level results. In the toddlers, the subject-level model estimated a positive relationship between exposure to PM2.5 and morbidity whereas the community-level analysis estimated a negative effect (OR = 1.006, and 0.992 respectively). The effect of the morbidity index was highly significant and the precision in estimating the risk, as expected, was higher in the subject-level models.