Grantee Research Project Results
Final Report: The Detroit Asthma Morbidity, Air Quality and Traffic (DAMAT) Study
EPA Grant Number: R833628Title: The Detroit Asthma Morbidity, Air Quality and Traffic (DAMAT) Study
Investigators: Wahl, Robert L , Batterman, Stuart A. , Wasilevich, Elizabeth , Hultin, Mary Lee , Dombkowski, Kevin , Mukherjee, Bhramar , Michalak, Anna
Institution: Michigan Department of Community Health , University of Michigan
EPA Project Officer: Hahn, Intaek
Project Period: September 30, 2007 through September 29, 2010 (Extended to September 29, 2011)
Project Amount: $499,777
RFA: Development of Environmental Health Outcome Indicators (2006) RFA Text | Recipients Lists
Research Category:
Objective:
There were several specific aims for each hypothesis (see below).
- Hypothesis 1 examined whether daily changes in asthma morbidity among the pediatric Medicaid population in Detroit from 2004 to 2006 were attributable to fluctuations in concentrations of ambient air pollutants including carbon monoxide (CO), oxides of nitrogen (NOx), particulate matter less than 10 micrometers in diameter (PM10), particulate matter equal or less than 2.5 micrometers in diameter (PM2.5), ozone (O3,), and sulfur dioxide (SO2).
- Hypothesis 2 examined whether daily changes in asthma morbidity among the same pediatric Medicaid population and study period (2004-2006) could be separated into effects caused by regional (“background”) and local (“urban increment”) components associated with air pollutants named above.
- Hypothesis 3 examined whether the associations between air pollutants and asthma morbidity were strengthened when we accounted for residential location and proximity to pollutant emissions from traffic and industry.
- Hypothesis 4 examined whether the spatial pattern of pediatric asthma morbidity, specifically the odds ratio (relative likelihood) of asthma-related urgent care use (cases) relative to all other pediatric claims (controls), is increased with exposure to industrial and traffic-related pollutants as determined using residence locations of children and spatial information on exposures resulting from local emission sources.
- Hypothesis 5 examined whether indicators derived using epidemiological-based methods, as demonstrated using DAMAT study results, could serve as direct and meaningful measures of health-based impacts due to air pollutants that could inform decisions related to assessment, management and policy.
Summary/Accomplishments (Outputs/Outcomes):
The project was completed successfully. In Project Year 1, the health outcome and exposure data sets were collected, cleaned and processed, including specifically:
- A set of daily exposure measures, based on measurements of the selected pollutants measured at Detroit air quality monitors for the primary study period (2004-2006), was developed.
- Daily Medicaid data on all children (1-17 years of age) residing in Detroit using urgent care facilities for asthma related claims, including hospital visits and urgent care visits, for the study period were obtained (Table 1).
Figure 1. Trend of daily counts of asthma events for the pediatric Medicaid population (children 2–18 years of age) in Detroit, Michigan, 2004–2006. Daily observations are shown as points with locally estimated scatter-plot smoothing trend shown as overlaying fitted curve. The endpoints of asthma events include emergency department visits without hospitalization, direct admission for hospitalization, and hospitalizations admitted through the emergency department.
During Project Year 2, Specific Aims 1 through 4 were all addressed as follows:
- The Medicaid and air quality data sets were linked (Hypothesis 1);
- Indicators for the urban increment of air pollution were developed by estimating background pollutant levels derived from measurements at outlying and upwind monitoring sites and subtracting these values from measurements monitored in Detroit. Associations between the regional and urban increments of specific criteria air pollutants (CAPs – PM10, PM2.5, O3, and SO2) and daily urgent care use for asthma were determined (Hypothesis 2);
- Maps showing the spatial patterns of traffic-related pollutants using a geographical information system (GIS) analysis and simplified dispersion models were constructed. Traffic data included total and commercial vehicle flows; the latter was a surrogate for heavy-diesel vehicles. We began to classify the likely impact of traffic and other local pollutant sources on each Medicaid claim during the study period using the children’s residence location (each of which has been previously geo-coded (Hypothesis 3); and
- Residence location of each child making an asthma-related Medicaid claim was classified with regard to potential exposure from local emission sources; controls were selected randomly from claims for non-respiratory related events and appropriately matched to the asthma claims; and analysis was initiated to determine the relative probability of asthma-related morbidity in areas affected and not affected by local pollution sources using case/control methods, including multi-nominal multiple logistic regression models to analyze subtypes of asthma claims (Hypothesis 4).
In Project Year 3, all five Specific Aims were addressed as follows:
- The relationship between daily fluctuations in pollutant concentrations and daily urgent care use was investigated using a longitudinal analysis employing case/cross-over Poisson (time series) regression models (Hypothesis 1);
- The analyses to examine relationships between regional and urban increments of specific CAPs and asthma outcomes were completed (started in year 2; Hypothesis 2);
- The classification of the likely impact of traffic and other local pollutant sources on each Medicaid claim was completed and efforts to determine the effect of traffic and other local pollutant sources on the strength of the association between daily asthma-related morbidity and daily air pollution exposures, using the longitudinal models developed in this project, was started and almost completed (Hypothesis 3);
- The case control study was completed to determine the relative probability of asthma-related morbidity in areas affected and not affected by local pollution sources, including multi-nomial multiple logistic regression models to analyze subtypes of asthma claims (Hypothesis 4); and
- Three scenarios portraying reasonable but contrasting future emission patterns were derived, including (1) current conditions; (2) reduced emissions from local mobile sources; and (3) reduced regional transport (Hypothesis 5).
In Project Year 4 (no-cost extension), Specific Aims 3 and 5 were addressed as follows:
- Assessment of the effects of traffic and other local pollutant sources on the strength of the association between daily asthma-related morbidity and daily air pollution exposures, using the longitudinal models developed in this project, was completed (started in year 3; Hypothesis 3);
- Air pollution impacts were estimated for each of three scenarios portraying future emission patterns (see Project Year 3 above) using the results of the longitudinal models (Hypotheses 1, 2, and 3) and results of the spatial analysis (Hypothesis 4; Hypothesis 5).
- Evaluated the uncertainty in the indicators, using both the estimated error as well as cross-validations that compared results across each study year.
- Analyzed and compared indicators for the spatial and temporal modes, including effects of prediction uncertainty on both population and individual means.
- We developed four manuscripts based on our work (see publications). We also have developed poster and oral presentations for team members to present at scientific meetings and conferences. Writing manuscripts and presenting our results will continue beyond the project period.
Conclusions:
Hypothesis 1
Evidence of significant increases in daily acute asthma events was found for exposure to SO2 and PM2.5, and a significant threshold effect was estimated for exposure to PM2.5 at 13 and 11 µg m-3, using generalized additive models and conditional logistic regression models, respectively (Figure 2). Stronger effect sizes above the threshold were typically noted, as compared with a standard linear relationship, e.g., in the time-series analysis, an interquartile range increase (9.2 µg m-3) in PM2.5 (5-day-moving average) had a risk ratio of 1.030 (95% CI: 1.001, 1.061) in the generalized additive models, and 1.066 (95% CI: 1.031, 1.102) in the threshold generalized additive models. The corresponding estimates for the case-crossover design were 1.039 (95% CI: 1.013, 1.066) in the conditional logistic regression, and 1.054 (95% CI: 1.023, 1.086) in the threshold conditional logistic regression (Table 1). See Li S, et al. (2011) for more detailed discussion.
This study indicates that the associations of SO2 and PM2.5 concentrations with asthma emergency department visits and hospitalizations, as well as the estimated PM2.5 threshold were fairly consistent across time-series and case-cross over analyses, and suggests that effect estimates based on linear models (without thresholds) may underestimate the true risk.
Hypotheses 3 and 4
Acute asthma outcomes, including inpatient and emergency department visits, were associated with proximity to primary roads with an odds ratio of 0.97 (95% CI: 0.94, 0.99) for a 1 kilometer (km) increase in distance using conditional logistic regression, implying that asthma events are less likely as the distance between the residence and a primary road increases. Similar relationships and effect sizes were found using polychotomous conditional logistic regression. Another plausible exposure metric, a reduced form response surface model that represents atmospheric dispersion of pollutants from roads, was not associated with acute asthma outcomes under that exposure model.
There is moderately strong evidence of elevated risk of acute asthma outcomes close to major roads based on the results obtained in this population-based matched case-control study (Figure 3). See Li S, et al. (2011) for detailed discussion.
Figure 2. Estimated spline terms of concentration with 95% confidence bands showing non-linear relationships between pollutant concentrations and daily asthma events, using 5-day moving average of pollutant concentrations. Vertical reference lines show estimated threshold parameters using profile likelihood method corresponding to each pollutant and model (*CLR: conditional logistic regression; GAM: generalized additive model.).
Hypotheses 2 and 5
Predicted CO concentrations showed reasonable agreement with annual average and 24-hour measurements, e.g., 59% of the 24-hr predictions were within a factor of two of observations in the warmer months when CO emissions are more consistent. The highest concentrations of both CO and PM2.5 were predicted to occur near intersections and downwind of major roads during periods of unfavorable meteorology (e.g., low wind speeds) and high emissions (e.g., weekday rush hour). The spatial and temporal variation among predicted concentrations was significant, and resulted in unusual distributional and correlation characteristics, including strong negative correlation for receptors on opposite sides of a road and the highest short-term concentrations on the “upwind” side of the road (Figure 4). See Batterman SA, et al. (2010) for detailed discussion.
Table 1. Comparison of GAMsa and time-stratified case-crossover CLRsa with thresholds; estimated risk and odds ratios above the threshold for daily asthma events and a one interquartile range increase of a pollutant exposure for the pediatric (children 2–18 years of age) Medicaid population in Detroit, MI, 2004–2006.
| GAM with threshold |
| Case-crossover CLR | ||||||
with threshold | |||||||||
| RR* | 95% CI* | QAIC* | OR* | 95% CI | AIC | |||
NO2 (ppb) ξ=23 | |||||||||
3-day lag | 0.957 | 0.882 | 1.039 | 6222.9 | 0.960 | 0.894 | 1.030 | 100751.5ª | |
5-day lag | 1.015 | 0.939 | 1.097 | 6203.1ª | 1.003 | 0.936 | 1.075 | 101197.3 | |
3-day moving average* | 0.937 | 0.845 | 1.039 | 6317.9 | 0.945 | 0.867 | 1.031 | 103616.2 | |
5-day moving average | 0.984 | 0.876 | 1.106 | 6334.9 | 0.958 | 0.867 | 1.059 | 104038.4 | |
SO2 (ppb) ξ=8.25 | |||||||||
3-day lag | 1.036 | 0.983 | 1.093 | 6199.6 | 1.022 | 0.974 | 1.074 | 101064.3ª | |
5-day lag | 1.049 | 0.994 | 1.106 | 6192.4ª | 1.057 | 1.006 | 1.110 | 101668.5 | |
3-day moving average | 0.948 | 0.861 | 1.043 | 6317.4 | 0.917 | 0.842 | 0.998 | 103865.1 | |
5-day moving average | 1.026 | 0.900 | 1.170 | 6315.8 | 0.960 | 0.855 | 1.078 | 104019.5 | |
CO (ppm) ξ=0.45 | |||||||||
3-day lag | 0.998 | 0.972 | 1.025 | 6325.2 | 1.003 | 0.979 | 1.026 | 103930.8 | |
5-day lag | 1.008 | 0.982 | 1.034 | 6305.0ª | 1.009 | 0.986 | 1.033 | 103829.5ª | |
3-day moving average | 0.956 | 0.917 | 0.997 | 6328.3 | 0.966 | 0.932 | 1.002 | 104038.4 | |
5-day moving average | 0.968 | 0.920 | 1.018 | 6328.4 | 0.975 | 0.933 | 1.017 | 104032.4 | |
PM2.5 (μg m-3) ξ=12 | |||||||||
3-day lag† | 1.046 | 1.017 | 1.075 | 6320.4 | 1.030 | 1.005 | 1.056 | 103921.0 | |
5-day lag† | 1.055 | 1.026 | 1.084 | 6308.4ª | 1.041 | 1.015 | 1.066 | 103824.9ª | |
3-day moving average | 1.024 | 0.992 | 1.057 | 6334.3 | 1.007 | 0.980 | 1.036 | 104038.8 | |
5-day moving average† | 1.066 | 1.031 | 1.102 | 6317.9 |
| 1.054 | 1.023 | 1.086 | 104033.1 |
Bold: the RRs or ORs are statistically significant (P-value < 0.05).
Underline: the QAIC/AIC of the threshold model is smaller than the QAIC/AIC from the corresponding linear model.
a AIC: Akaike information criterion; QAIC: quasi-Akaike information criterion; CI: confidence interval; CLR: conditional logistic regression; GAM: generalized additive model; RR: risk ratio; OR: odds ratio; moving average: average for the specified number of days preceding the asthma events.
b The QAIC or AIC is the smallest among all lags for each pollutant.
The case study findings likely can be generalized to many other locations, and they have important implications for epidemiological and other studies. The reduced-form model is intended for exposure assessment, risk assessment, epidemiological, geographical information systems and other applications.
Maintenance of the quality assurance requirements of 40 C.F.R. 30.54 and the agreement
Use of Existing/Secondary Data - Air Quality, Meteorological Data and Analytical Methods
The US EPA has set forth specifications in the Code of Federal Regulations (parts 50, 53, 58) for instruments measuring the criteria pollutants: SO2, CO, NO2, PM10, O3, lead, and PM2.5. These specifications are used to classify instrumentation as meeting the Federal Reference Method (FRM), a Federal Equivalent Method (FEM) or other method. All of the criteria pollutant ambient monitoring that is performed by MDEQ falls into the FRM category, if one has been established. For continuous PM2.5 measurements, a FRM or FEM instrument currently is not available. MDEQ has opted to use a TEOM to collect PM2.5 continuous measurements. These measurements are subjected to intensive evaluation by MDEQ's participation in various internal and external audit and quality assurance (QA) programs.
Figure 3. Estimated odds ratios with 95% confidence intervals of asthma for distance thresholds from primary roads. Estimated odds ratios with 95% confidence intervals of asthma for different distance thresholds from primary road, as well as the number of subjects lying in the two level indicator of distance from primary road by the case-control status. Each estimate is based on conditional logistic regression results using distance as a dichotomous factor indicating residence location inside or outside the corresponding buffer, for the population-based matched case-control data set of asthma from the pediatric Medicaid population in Detroit, MI, 2004-2006.
The U.S. EPA also specifies siting criteria used to select the location for a monitoring station, which depend upon the pollutant being measured, the size of the targeted air mass, and the monitoring objective, e.g., determining the maximum pollutant concentration or estimating population exposure. Also specified are population distributions, geographical characteristics and distances from roadways, major sources, large trees, vents, buildings and the inlet probes of other monitoring equipment. EPA also specifies the height of the inlet probes as within the breathing zone. The inlets meet minimum probe separation criteria; in addition, the inlets are removed from the influence of large buildings, trees, vents etc. The MDEQ monitoring network has passed a systems audit conducted by U.S. EPA Region 5.
Calibration and performance evaluation of the sampling
Ambient air monitoring data undergoes various phases of quality control from the minute day-to-day operations conducted by a site operator to full scale monitoring network audits covering all phases of operation that are conducted by the U.S. EPA. These activities range from instrument maintenance, calibrations and zero checks, flow audits and performance audits, and systems wide audits. Additional quality control aspects include participation in round robin and performance evaluation samples by laboratories analyzing 24-hour samples for the non-continuous measurements. All monitoring activities are conducted with a quality assurance project plan (QAPP) that has been fully approved by EPA Region 5. Standard operating procedures (SOPs) and chains of custody also are used as part of the quality assurance plan. The environmental data used in the proposed study were collected in accordance with these checks, and were further checked for consistency, outliers, etc.
Figure 4. Predicted concentrations across the case study area. A. Annual average CO concentration; B. Annual average PM2.5 concentration; C. Worst-case CO day; D. Best-case CO day. M39 and GR (Grand River) are the modeled roads.
Contributions to the understanding of health impacts from air pollutant exposures
The work produced a number of results that were significant, relevant and practical. These are noted briefly below.
a. The work’s significance to the field included, first, the value of using Medicaid data to explore, quantify and understand the relationship between air pollutant exposures and health outcomes. Second, we tested a number of innovative techniques to explore and refine those relationships, including the use of point source and traffic information, and several innovative air pollution exposure indicators. We also explored the development of exposure indicators, e.g., use of traffic measurements versus model predictions.
b. The general goals of this grant were to use existing databases of environmental (ambient), biological and/or health-related data to develop indicators that reliably signaled the impact of changes in environmental conditions, management approaches or policies on human health. This project met these goals in that we predicted changes in health care use for asthma based on environmental exposure measures. The mission of EPA is to protect human health and the environment; this project increased the knowledge base necessary to protect human health in Michigan.
c. The work had practical applications in understanding causes of asthma aggravation, and, in terms of exposure indicators, developed epidemiological models and general indicators of air pollutant stress. In addition, an understanding of the effects of regional and local increments of air pollutant levels can be used to help plan monitoring networks and to help interpret monitoring data.
Journal Articles on this Report : 5 Displayed | Download in RIS Format
Other project views: | All 5 publications | 5 publications in selected types | All 5 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Batterman SA, Zhang K, Kononowech R. Prediction and analysis of near-road concentrations using a reduced-form emission/dispersion model. Environmental Health 2010;9:29. |
R833628 (2010) R833628 (Final) |
Exit Exit Exit |
|
Li S, Batterman S, Wasilevich E, Elasaad H, Wahl R, Mukherjee B. Asthma exacerbation and proximity of residence to major roads: a population-based matched case-control study among the pediatric Medicaid population in Detroit, Michigan. Environmental Health 2011;10:34. |
R833628 (2010) R833628 (Final) |
Exit Exit Exit |
|
Li S, Batterman S, Wasilevich E, Wahl R, Wirth J, Su F-C, Mukherjee B. Association of daily asthma emergency department visits and hospital admissions with ambient air pollutants among the pediatric Medicaid population in Detroit: time-series and time-stratified case-crossover analyses with threshold effects. Environmental Research 2011;111(8):1137-1147. |
R833628 (2009) R833628 (2010) R833628 (Final) |
Exit Exit |
|
Li S, Mukherjee B, Batterman S. Point source modeling of matched case-control data with multiple disease subtypes. Statistics in Medicine 2012;30(28):3617-3637. |
R833628 (2010) R833628 (Final) |
Exit |
|
Li S, Mukherjee B, Batterman S, Ghosh M. Bayesian analysis of time-series data under case-crossover designs: posterior equivalence and inference.. Biometrics 2013;69(4):925-936. |
R833628 (Final) |
Exit |
Supplemental Keywords:
RFA, Scientific Discipline, Health, Air, HUMAN HEALTH, particulate matter, Health Risk Assessment, air toxics, Exposure, Epidemiology, Susceptibility/Sensitive Population/Genetic Susceptibility, Risk Assessments, Health Effects, genetic susceptability, Biology, copollutant exposures, sensitive populations, atmospheric particulate matter, asthma, airway epithelial cells, cardiopulmonary responses, fine particles, PM 2.5, inhaled pollutants, acute cardiovascular effects, acute lung injury, stratospheric ozone, morbidity, air pollutants, motor vehicle emissions, automotive emissions, motor vehicle exhaust, air pollution, susceptible subpopulations, cardiac arrest, diesel exhaust, chronic health effects, lung inflammation, oxidant gas, particulate exposure, cardiopulmonary response, heart rate, human exposure, atmospheric aerosols, Acute health effects, inhaled, chronic obstructive pulmonary disease, human susceptibility, cardiotoxicity, cardiopulmonary, mortality, concentrated particulate matter, air contaminant exposure, air quality, environmental hazard exposures, toxics, airborne urban contaminants, cardiovascular disease, acute exposureProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.