Online Program

Economic and Atmospheric Conditions Related to Emergency Department Visits
Leonard Bieolry, UMDNJ 
David Dickey, NC State 
Van Dunn, NYCHHC 
Raymond Gregory, NYCHHC 
Shunsuke Ito, NYCHHC 
Caroline Jacobs, NYCHHC 
David Chester Low, IMF, NYU 
*Ronald Bruce Low, NYCHHC, NYU 

Keywords: Emergency Department, pollution, economy, allergens, weather, unemployment

The current health care debate includes discussions about what drives patients to emergency departments; understanding these drivers might help improve efficiencies and reduce costs. We studied the relationship between the number of ED visits (EDV) to New York City Public Hospitals (HHC), and time, weather, air pollution, airborne allergens, employment and other economic indicators. Methods: Databases from the following sources were mined: HHC billing, author’s aerobiology lab, NWS, EPA, BLS and IMF. An ARIMA model of EDV/day was developed. Model inclusion criteria: alpha=.05, SBC statistic minimized, coefficients for known exacerbating factors must be positive. Results: From 1998-2008, there were 10,561,562 EDV (2628+369 visits/day). There several time effects: yearly effects were seen as AR 364, 365, MA 364 365 (all p<.0001); visits peaked during the winter months. We saw weekly effects as MA 8 (p=.0005)and MA 28 and AR7, 21,22,28,29, and a separate DOW dummy variable (all p<.0001). The DOW dummy variable reflects a monotonically decreasing (~36 EDV/day) number of EDV during the week, with a peak on Mondays and nadir on Sunday, We also modeled 159 fewer EDV on federally declared holidays(p<.0001). There is a positive relationship to: same day increasing air temperature (10 EDV/degree C) and decreasing relative humidity(7 EDV/1% humidity), incidence of upper respiratory infection (URI) (0.6 EDV/URI patient), and a DOW*URI interaction (all p<.0001), log of NYC population size (p=.0073), same day CO level (16 EDV/ppm, p=.0307), pollen count (8 EDV/1000 grains, p=.0025), and the U2 measure of unemployment, which looks only at previously employed workers (85 EDV/1 % workforce formerly employed, p=.0009). Our model shows a decrease of 349 EDV on 9/11/2001. Conclusion: We found a modest increase in EDV with increasing unemployment of previously employed workers; we did not find any other independent effects of “economic health” on EDV. ED use decreased during weekends/holidays; it peaked on Mondays and dropped steadily during the rest of the week. ED use increased during winter, and with cold air temperature, decreasing relative humidity, URI outbreaks, and following increasing levels of pollen and CO pollution.