Presentation is loading. Please wait.

Presentation is loading. Please wait.

Modeling as an exposure estimation approach for use in epidemiologic studies Part 2: Example applications KL Dionisio 1, LK Baxter 1, V Isakov 1, SE Sarnat.

Similar presentations


Presentation on theme: "Modeling as an exposure estimation approach for use in epidemiologic studies Part 2: Example applications KL Dionisio 1, LK Baxter 1, V Isakov 1, SE Sarnat."— Presentation transcript:

1 Modeling as an exposure estimation approach for use in epidemiologic studies Part 2: Example applications KL Dionisio 1, LK Baxter 1, V Isakov 1, SE Sarnat 2, JA Sarnat 2, J Burke 1, H Özkaynak 1 1 U.S. EPA, RTP, NC 2 Emory University, Atlanta, GA

2 Emory-GA Tech cooperative agreement Objective: evaluation of alternative exposure metrics of ambient air pollution exposure for use in epidemiologic studies Study period: 1999-2002 225 ZIP codes in the Atlanta, GA metropolitan area 3 tiers of exposure metrics 6 pollutants –PM 2.5, EC, SO 4, NO x, CO, O 3

3 Exposure tiers Tiers of Exposure Metrics Personal Behavior/Time Activity Microenvironmental Characteristics Central Site Monitoring Land-Use Regression Modeling Air Quality Modeling (CMAQ, AERMOD, hybrid) Exposure Modeling (SHEDS, APEX) Data Blending Monitoring Data Emissions Data Meteorological Data Land-Use/Topography Monitoring Data Emissions Data Meteorological Data Land-Use/Topography Input data Satellite Data Central site monitor measurements (“CS”) Modeled ambient estimates (“Hybrid”) Modeled exposure estimates (“Personal exposure”)

4 Central site monitor measurements (“CS”) Modeled ambient estimates (“Hybrid”) –Regional background (statistical model) –Local air quality emissions model (AERMOD) Modeled exposure estimates (“Personal exposure”) –APEX (CO, NO x ) –SHEDS (PM 2.5, EC, SO 4, O 3 ) Exposure tiers

5 Central site monitor measurements (“CS”) Modeled ambient estimates (“Hybrid”) –Regional background (statistical model) –Local air quality emissions model (AERMOD) Modeled exposure estimates (“Personal exposure”) –APEX (CO, NO x ) –SHEDS (PM 2.5, EC, SO 4, O 3 ) No spatial resolution ZIP code-level estimates All tiers: 24-hr average concentrations (8-hr max for O 3 ) Exposure tiers

6 Hybrid Personal exposure CS Dionisio et al., 2013, JESEE Observed spatial variation

7 CS Regional background AERMOD Hybrid Personal exposure CS Regional background AERMOD Hybrid Personal exposure PM 2.5 NO x Exposure metrics

8 CS Regional background AERMOD Hybrid Personal exposure CS Regional background AERMOD Hybrid Personal exposure PM 2.5 NO x Exposure metrics

9 RR=1 RR>1 RR<1 Relative risk (RR): the risk of developing the health outcome related to exposure (exposed vs. non-exposed, or per incremental increase in exposure) RR Exposure is associated with outcome Exposure prevents outcome No relationship between exposure and outcome Health effect estimates: RR

10 Relative risk of mortality for smokers compared to those who had never smoked Source: Thun et al., 2013, NEJM *Excerpt from Table 2 Lepeule et al., 2012, EHP Relative Risk: examples

11 Relative risk of mortality for smokers compared to those who had never smoked RR ~ 25  Men who smoke 10-19 cigarettes per day have 25 times the chance of dying of lung cancer compared to men who have never smoked. Source: Thun et al., 2013, NEJM *Excerpt from Table 2 Lepeule et al., 2012, EHP RR=1.14  There is a 14% increased risk of all-cause mortality per 10 µg/m 3 increase in PM 2.5 Relative Risk: examples

12 PM 2.5 NO x PM 2.5 NO x RR by exposure tier: Atlanta *Excerpt from Figure 5 Sarnat et al., 2013, JESEE

13 *Excerpt from Figure 5 Sarnat et al., 2013, JESEE PM 2.5 NO x PM 2.5 NO x CSHybrid Personal exposure CS Hybrid Personal exposure RR by exposure tier: Atlanta

14 Hybrid and exposure model estimates exhibit spatial variability for locally-generated pollutants Modestly stronger associations for ASW (larger RR and narrower CI) for NO x using hybrid or personal exposure model estimates compared to CS Modestly stronger association for ASW for PM 2.5 using hybrid model estimates, slightly weaker association with personal exposure model estimates, compared to CS Conclusions


Download ppt "Modeling as an exposure estimation approach for use in epidemiologic studies Part 2: Example applications KL Dionisio 1, LK Baxter 1, V Isakov 1, SE Sarnat."

Similar presentations


Ads by Google