Impact/Purpose:

We will develop models to predict acute respiratory morbidity (including asthma exacerbation) using near-real time weather, ambient air pollution, and respiratory emergency department (ED) visits in New York City (NYC). We will take advantage of a unique syndromic surveillance system that monitors ED visits daily by the NYCDOHMH. We will systematically characterize the sequence of events among weather conditions, air pollution buildup, and health effects indicators. Using sub-area analysis, we will also determine spatial and neighborhood/socio-economic factors that influence the prediction power and efficiency of the models. We will estimate model uncertainties by computing prediction errors of candidate models in a series of real-time validation tests. Overall, this proposed project will create a framework to model, in near real time, acute health outcome indicators of environmental exposures in a large metropolitan area.

We will develop models to predict acute respiratory morbidity (including asthma exacerbation) using near-real time weather, ambient air pollution, and respiratory emergency department (ED) visits in New York City (NYC). We will take advantage of a unique syndromic surveillance system that monitors ED visits daily by the NYCDOHMH. We will systematically characterize the sequence of events among weather conditions, air pollution buildup, and health effects indicators. Using sub-area analysis, we will also determine spatial and neighborhood/socio-economic factors that influence the prediction power and efficiency of the models. We will estimate model uncertainties by computing prediction errors of candidate models in a series of real-time validation tests. Overall, this proposed project will create a framework to model, in near real time, acute health outcome indicators of environmental exposures in a large metropolitan area.

Description:

Impact of heat wave on mortality: We evaluated model performance for an exhaustive set of alternative weather models, using both parametric and non-parametric time-series Poisson models, to predict the heat wave effects on natural mortality. The result is useful in advising the NYC Office of Emergency Management (OEM) in determining the appropriate threshold of temperature index for heat wave advisory/warnings. We found that including individual lags 0 through 3 days of heat index (HI) best predicted impacts of heat waves on mortality in both parametric (which included linear, quadratic, and cubic terms) and non-parametric models. The fitted HI/mortality relationship showed a non-linear relationship, with a steeper slope above a 100 degrees HI. The impact of consecutive days of 95 degrees HI on mortality was approximately equivalent to a single day of 100 degrees HI. These findings have been communicated to the OEM, and subsequently DOHMH formally recommended that these agencies modify the threshold for heat advisories and activating a response. Thus, our research project contributed to the improvement of a public health program in New York City. The manuscript of analysis was published in 2010 (Metzger, Ito, and Matte, 2010).

Real-time surveillance of heat-related morbidity: relation to heat-related excess mortality:

The impact of heat on mortality is well documented, but deaths tend to lag extreme heat and mortality data are generally not available for timely surveillance during heat waves. In this study, we explored the associations among weather, indicators of heat-related ambulance calls and emergency department visits and excess natural cause mortality in NYC. Data analyzed included daily weather conditions, emergency medical system (EMS) calls flagged as heat-related by EMS dispatchers, emergency departments (ED) visits classified as heat-related based on chief complaint text, and natural cause deaths. EMS and ED data were obtained from data reported daily to the city health department for syndromic surveillance. We fit generalized linear models to assess the relationships of daily counts of heat-related EMS and ED visits to natural cause deaths after adjustment for weather conditions during the months of May-September between 1999 and 2008. Controlling for temporal trends, an 11% (95% confidence interval (CI): 5-18) and 7% (95% CI: 4-9) increase in natural cause mortality was associated with an increase from the 50th percentile to 99th percentile of same-day and one-day lagged heat-related EMS calls and ED visits, respectively. After controlling for both temporal trends and weather, we observed a 10% (95% CI: 4-16) increase in natural cause mortality associated with one-day lagged heat-related EMS calls and a 5% mortality increase with one-day lagged ED visits (95% CI: 2-8). Heat-related illness can be tracked during heat waves using EMS and ED data, which are indicators of heat associated excess natural cause mortality during the warm weather season. Thus, heat-related morbidity indicators can predict mortality above and beyond weather variables. 

Evaluations of the relationships among asthma ED syndrome, asthma ED visits (physician diagnosed), and asthma hospitalizations: For the near-real time surveillance of asthma exacerbations, the main health outcome of interest in this project was asthma ED syndrome (i.e., available the day after the ED visits). However, this health outcome indicator is based on subjects’ chief complaints, and it was of interest to determine how well this indicator was correlated to the physician-diagnosed asthma ED visits and asthma hospitalizations. Thus, we analyzed these outcomes for the years 2005-2008, when all of these outcomes data were available. We examined temporal correlations of these series for four age groups: (1) 0-4; (2) 5-17; (3) 18-44; and (4) 45-64 (note that “asthma” at older age group was considered to overlap with COPD cases and thus was not analyzed). The asthma ED syndrome data were highly temporally correlated with the physician diagnosed asthma ED visits data for all the age groups (Pearson correlation [r]: 0.91, 0.97, 0.90, and 0.83, respectively) and moderately to highly correlated with asthma hospitalizations data (r: 0.74, 0.82, 0.54, and 0.48, respectively). Thus, the asthma ED syndrome data is a good indicator of physician-diagnosed asthma ED visits and a reasonable indicator of physician-diagnosed asthma hospitalizations. 

Within-city effect modifiers of short-term air pollution effects: Asthma morbidity is temporally associated with daily variation in central site air pollution measures and spatially associated with socioeconomic factors such as poverty and environmental factors such as residential proximity to traffic. We used NYCDOHMH near-real-time syndromic illness surveillance data on daily ED visits to evaluate the extent to which neighborhood characteristics modify the temporal relation of asthma to ambient air quality. Daily asthma ED syndromic illness counts for children (age 5 – 17) during the years 2002-2006 were analyzed for their associations with fine particles (PM_2.5), nitrogen dioxide, and ozone. Zip code-specific Poisson regression models adjusted for smooth functions of temporal trends, immediate and delayed temperature, and day of week. The second-stage random effects meta-regression model included available zip code-level census data (socio-economic status and race) and estimated traffic density as vehicle miles per unit area. All of the pollutants were positively significantly associated with asthma ED counts. The combined percent excess PM_2.5 risk for 115 zip codes was 3.8% (95%CI: 1.5, 6.2) per 10 µg/m³ increase in the average of 0- and 1-day PM_2.5. In the second stage model, traffic density was the most significant effect modifier of the PM_2.5-asthma association, increasing the excess risk estimate by 44% per one standard deviation of traffic density. Traffic density modified PM_2.5-asthma association. This examination of the within-city variation of air pollution effects and the role of effect modifiers is one of our major goals of this project, and is a new and relevant contribution to this field. This line of research also is relevant to EPA’s mission to identify at-risk population of environmental pollution. A manuscript for this analysis is in preparation by extending the study period from 2002-2006 to 2002-2009.

Assessing syndromic surveillance of cardiovascular outcomes from emergency department chief complaint data in New York City: Prospective syndromic surveillance of emergency department visits has been used for near-real time tracking of communicable diseases to detect outbreaks or other unexpected disease clusters. The utility of syndromic surveillance for tracking cardiovascular events, which may be influenced by environmental factors and influenza, has not been evaluated. We developed and evaluated a method for tracking cardiovascular events using emergency department free-text chief complaints. There were three phases to our analysis. First, we applied text processing algorithms based on sensitivity, specificity and positive predictive value to chief complaint data reported by 11 NYC emergency departments for which ICD-9 discharge diagnosis codes were available. Second, the same algorithms were applied to data reported by a larger sample of 50 NYC emergency departments for which discharge diagnosis was unavailable. From this more complete data, we evaluated the consistency of temporal variation of cardiovascular syndromic events and hospitalizations from 76 NYC hospitals. Finally, we examined associations between PM_2.5, syndromic events and hospitalizations. Sensitivity and positive predictive value were low for syndromic events, while specificity was high. Utilizing the larger sample of emergency departments, a strong day of week pattern and weak seasonal trend were observed for syndromic events and hospitalizations. These time-series were highly correlated after removing the day-of-week, holiday and seasonal trends. The estimated percent excess risks in the cold season (October to March) were 1.9% (95% confidence interval (CI): 0.6, 3.2), 2.1% (95% CI: 0.9, 3.3), and 1.8% (95%CI: 0.5, 3.0) per same-day 10 µg/m³ increase in PM_2.5 for cardiac-only syndromic data, cardiovascular syndromic data, and hospitalizations, respectively. Conclusions/Significance: Near real-time emergency department chief complaint data may be useful for timely surveillance of cardiovascular morbidity related to ambient air pollution and other environmental events. The result from this analysis has been recently published (Mathes et al., 2011).

The association of peaks in daily tree pollen concentration and allergy medication sales in New York City: This analysis has been conducted in collaboration with the researchers from Columbia University Mailman School of Public Health (Patrick Kinney and Kate Weinberger), Mount Sinai School of Medicine (Perry Sheffied), and Louis Calder Center,  Fordham University (Guy Robinson). As part of the syndromic surveillance data, NYCDOH keeps track of over-the-counter (OTC) allergy medication sales in NYC. Dates of select peak tree pollen concentrations were obtained from a primary New York City monitoring station for the years 2003-2008. Time series models were fit with the logarithm of daily allergy medication sales reported to the city health department as a function of the same-day and lagged binary tree pollen peak indicators, adjusting for season, year-to-year variation, temperature, and day-of-week. Multi-day effects also were estimated by including the seven lag days. Significant associations were found between the maple, oak and birch tree pollen peaks and allergy medication sales, with the strongest association at 2-day lag (excess sales of 28.7%  [95%CI: 17.4-41.2] over the average sales during the study period). Improved public health advisories and more specific pollen season charts will be possible with further refinement of exposure metrics and additional health outcomes. A manuscript of this analysis was published in 2011 (Sheffied et al., 2011).

Particulate matter source types associated with cardiovascular hospitalizations and mortality: Recent time-series studies have indicated that both cardiovascular disease (CVD) mortality and hospitalizations are associated with particulate matter (PM). However, seasonal patterns of PM associations with these outcomes are not consistent, and PM components responsible for these associations have not been determined. We investigated this issue in NYC, where PM originates from regional and local combustion sources. In this study, we examined the role of PM_2.5 chemical components on both CVD hospitalizations and on mortality in NYC. We analyzed daily deaths and emergency hospitalizations for cardiovascular diseases (CVDs) among those persons ≥ aged 40 years of age for their associations with PM_2.5, its chemical components, nitrogen dioxide (NO₂), carbon monoxide, and sulfur dioxide for the years 2000–2006 using a Poisson model adjusting for temporal and seasonal trends, temperature effects, and day of the week. We estimated excess risks per inter-quartile-range increases at lags 0 through 3 days for warm (April–September) and cold (October–March) seasons. The CVD mortality series exhibit strong seasonal trends, whereas the CVD hospitalization series show a strong day-of-week pattern. These CVD outcome series were not correlated with each other but were individually associated with a number of PM_2.5 chemical components from regional and local sources, each with different seasonal patterns and lags. Coal combustion-related Coal-combustion–related components (e.g., selenium) were associated with CVD mortality in summer and CVD hospitalizations in winter, whereas elemental carbon and NO₂ showed associations with these outcomes in both seasons. Local combustion sources, including traffic and residual oil burning, may play a year-round role in the associations between air pollution and CVD outcomes, but transported aerosols may explain the seasonal variation in associations shown by PM_2.5 mass. The result of this analysis has been recently published (Ito et al., 2011).

URLs/Downloads:

Final Progress Report

2010 Progress Report

2009 Progress Report

2008 Progress Report

Record Details:

Record Type:PROJECT( ABSTRACT )

Start Date:12/01/2007

Completion Date:11/30/2010

Record ID: 187267

Show Additional Record Data

Related Organizations:

Role :OWNER

Organization Name :NEW YORK CITY DEPARTMENT OF HEALTH AND MENTAL HYGIENE

Mailing Address :125 Worth Street, CN 33

Citation :New York

State :NY

Zip Code :10013



Role :OWNER

Organization Name :NYU SCHOOL OF MEDICINE

State :NY

Project Information:

Approach :

Our initial model development involves analyses of past weather, air pollution data, and respiratory and asthma ED visits. The main air pollutants of interest are particulate matter less than 2.5 μm (PM_2.5) and ozone, but other gaseous air pollutants (carbon monoxide, sulfur dioxide, and nitrogen dioxide) will also be considered. Continuous hourly-average PM_2.5 and gaseous pollutants data are available for multiple sampling locations in NYC, allowing modeling of spatial variation in air pollution and morbidity. While past models in air pollution epidemiology focused on conservatively estimating risks for air pollutants while adjusting for weather, day-of-week, and temporal trends treated as independent variables, we will instead explore alternative models that consider weather-air pollution interactions and characterize the sequence of events that predict adverse respiratory system effects. We will also test the incorporation of the existing air pollution episode prediction (forecasting) models into our health effects models. Our aim is to develop health effects models that are useful for predictions of air pollution health impact days based on real-time environmental and health data, and, ultimately, for use in public health impact prevention programs. We will also examine the influence of special events (e.g., Asian dust storms, Canadian forest fires) to evaluate how they impact health and the model predictions. Back-trajectory analysis will be conducted to identify/confirm specific source impacts. Temporal relationships across health outcomes (e.g., ED visits and mortality) in the past data will also be examined to test models’ applicability for avertive/preventive measures. We will develop candidate alternative models in the first half of the project based on past data, and then independently test our models on a prospective real-time basis in the 2nd half of the project. Thus, we will be able to conduct real-world tests of al

Cost :$494,552.00

Research Component :Health Effects

Project IDs:

ID Code :R833623

Project type :EPA Grant