Environmental Engineering Reference
In-Depth Information
All residents of NYS were included in the analysis. Health outcomes included
hospitalizations for respiratory disease, years 1997-2006. Specifically, respiratory
diseases included the following principal diagnoses based on the International
Classification of Disease, 9th Revision (ICD-9): bronchitis (491), emphysema
(492), asthma (493), and chronic obstructive pulmonary disease (496). In addition,
for children ages 0-4 years, a diagnosis of acute bronchitis and bronchiolitis (466)
and bronchitis, not specified as acute or chronic (490) were also included because
they are common respiratory diseases among very young children and the symptoms
are difficult to distinguish from asthma at this age. The control admissions included
hospitalizations for gastrointestinal diseases (009) and accidental falls (E880-E888).
Monthly, seasonal, and yearly respiratory hospital admission rates during the
study period were calculated. The rates of seasonal and yearly hospital admissions
for combined disease categories were compared across three different time periods.
Based on the actual timing of the regulation, these periods were defined as follows:
(1) 1997-2000 (pre-NOx regulation or baseline); (2) 2001-2003 (partial-NOx SIP
implementation); and (3) 2004-2006 (post-NOx SIP or post implementation).
Periodic trends in respiratory hospitalizations during the three NOx time periods
(baseline, partial-implementation and post-implementation) were compared by
geographic region, race/ethnicity, gender, age, and disease sub-groups. We then
calculated the percent change in average daily respiratory admissions partial- and
post- implementation compared to the average daily admissions during the baseline
period.
2.2. Statistical methods
Time series models using an intervention analysis approach were used to investigate
the temporal trends for the three NOx study periods. ARIMA models were used to
examine the change in average daily ozone levels, respiratory hospitalizations, and
their relationship during the summer months (June-August) between each of the
time periods, after adjusting for day-of-the-week, temperature, secular trend, and
other temporal/seasonal effects. The temporal/seasonal effects included a long
term trend indicator variable for capturing the baseline, partial implementation and
post implementation effect, as well as a separate indicator variable capturing the
seasonal trend of summertime effect during these periods. In addition, the respiratory
analysis included terms for holidays and day-of-the-week to control for their potential
confounding effects. The autocorrelation of the residuals were investigated at
various lags. The final model was determined when it included sufficient orders of
autocorrelations and moving averages such that the model residuals contained no
signal beyond random variation. Following passage of these autocorrelation tests,
statewide and region-specific time series models were estimated separately
for ambient ozone concentrations and respiratory admissions. To examine the
intervention effect, the baseline time period was compared against the partial and
post implementation periods. Analyses were stratified by eight NYS regions as
defined by the NYS Department of Environmental Conservation (DEC),
including: Long Island, New York City (NYC) Metro, Lower Hudson, Upper
