Presentation is loading. Please wait.

Presentation is loading. Please wait.

Developing Analytic Forecasting Methodologies for Health Impact Assessment Rajiv Bhatia, MD, MPH San Francisco Department of Public Health.

Similar presentations


Presentation on theme: "Developing Analytic Forecasting Methodologies for Health Impact Assessment Rajiv Bhatia, MD, MPH San Francisco Department of Public Health."— Presentation transcript:

1 Developing Analytic Forecasting Methodologies for Health Impact Assessment Rajiv Bhatia, MD, MPH San Francisco Department of Public Health

2 Forecasting Page 1 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Presentation Overview The Distinction Among Assessment of Existing Conditions vs. Monitoring vs. Forecasting Three Examples: Existing Forecasting Method New Method Based on Existing Research New Method Based on New Research Implications for Alaska HIA

3 Forecasting Page 2 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Example 1: Assessment and Mitigation of Roadway Air Pollution Impacts on Sensitive Uses State and National air quality standards concern limited pollutants Regional monitoring does not capture intra-urban variation in exposure Regulations limit tailpipe emissions per mile but not vehicle intensity Local agencies do not regulate air quality land use conflicts related to high volume roadways

4 Forecasting Page 3 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Available Health Effects Assessment Methods For Air Quality Assessment Dose response functions can associate area- level air quality exposures with health effects Air quality dispersion models and other techniques can assess roadway related air quality exposure based on: Vehicle Flow, Speed Emissions Meteorology Relationship between Facilities and Sensitive Receptors

5 Forecasting Page 4 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Estimating Mortality Impacts From Exposure to PM 2.5 based on CARB CR Functions Mortality = R 0 [exp (-  *∆PM2.5 -1) ] P R 0 = Baseline Mortality Rate  = Coefficient Derived from Relative Risk P = Affected Population

6 Forecasting Page 5 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Spatial Extent of Vehicle PM 2.5 All Vehicle Sources using CAL3QHCR—West Oakland, CA

7 Forecasting Page 6 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Applications Location of Sensitive Uses Transportation System Planning Indoor Air Quality Ventilation Standards

8 Forecasting Page 7 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Example 2: Quantifying the Health Benefits of a Living Wage Few analyses health benefits of labor policies Plausible relationship meditated through material needs Consistent association among high quality epidemiologic studies on income and health looking at multiple health outcomes

9 Forecasting Page 8 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Data Required For Impact Analysis The baseline income of the population targeted by the living wage The estimated income gains of workers benefiting from the new wage A dose response function between income and health outcomes

10 Forecasting Page 9 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Inclusion Criteria For Studies Providing Dose-Response Relationships English language peer reviewed literature 1990-1998 Studies of income and mortality, hospitalizations, or health status indicators Subjects representative of the U.S. general population Income measured at the household, family or individual level Longitudinal design Statistical adjustment for age and gender year of income ascertainment provided Income applied as a continuous variable

11 Forecasting Page 10 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Estimated Health Effects Due To Living Wage Income Gains For Workers With A Current Family Income of $20,000

12 Forecasting Page 11 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Example 3: Area-level Model of Pedestrian Vehicle Collisions Transportation Analyses in EIA provides little analysis of Pedestrian Safety Impacts: Vehicle-pedestrian injuries and fatalities are preventable. Key area-level environmental determinants of collisions include: Traffic volumes Traffic speed Pedestrian activity

13 Forecasting Page 12 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 An Environmental Approach: Evident area-level patterns – correlate with the freeway network, concentrations of streets with heavy arterial traffic, pedestrian activity centers (e.g., downtown, Golden Gate Park). Vehicle-pedestrian injury collisions: San Francisco, California census tracts (2001–2005)

14 Forecasting Page 13 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Model development framework How do transportation, land use, and population factors predict change in pedestrian injury collisions in San Francisco census tracts? Travel Behaviors: walking,public transit, private vehicle use Population Characteristics: number of residents and workers, socio-demographic characteristics Built Environment: street and land use characteristics Vehicle-Pedestrian Collisions (Number): pedestrian injury and death

15 Forecasting Page 14 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Vehicle-Pedestrian Injury Collision Model oPublicly available data (SWITRS, U.S. Census, SF Planning, SF MTA) oTraffic Volume, Street Characteristics – SF DPH/UC Berkeley oContinuous, census-tract level variables oMultivariate, linear regression model – predicts the natural log of vehicle-pedestrian injury collisions: ln(PIC) = b0 + ∑biXi

16 Forecasting Page 15 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Vehicle-Pedestrian Injury Collision Final Model Variables Traffic volume (+) Arterial streets (+) Neighborhood commercial zoning (+) Employees (+) Residents (+) Land area (-) Below poverty level (+) Age 65 and over (-)

17 Forecasting Page 16 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Final Ordinary Least Squares Regression Model Vehicle-Pedestrian Injury Collisions: San Francisco, California, 2001-2005 (n=175 census tracts) a Excludes grade-separated street segments inaccessible to pedestrians.

18 Forecasting Page 17 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Comparing the Simple Bivariate Model to Our Multivariate Model Approach Holding all covariates constant (as above), the model is equivalent to a power function with β=0.753.

19 Forecasting Page 18 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Vehicle-Pedestrian Injury Collision Model: Eastern Neighborhoods Plans EIR Analysis a Areas defined based on SF Planning boundaries, and census tracts used for the Eastern Neighborhoods Rezoning Socioeconomic Impacts analysis. b Census Tract, Aggregate Traffic Volumes. c Based on the Air Quality Chapter, Eastern Neighborhoods Pre-draft Environmental Impact Report, 2007. d Population increases based on increased population and housing units projected in Rezoning Option B, detailed in the draft Eastern Neighborhoods Rezoning and Community Plans, Environmental Setting and Impacts, April 2007.

20 Forecasting Page 19 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Vehicle-Pedestrian Injury Collision Model: Eastern Neighborhoods Plans EIR Analysis 20% 21% 15% 24% Predicted % change in pedestrian injury collisions based on estimated changes in resident population and traffic volume.

21 Forecasting Page 20 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Vehicle-Pedestrian Injury Collision Model: Application Land Use Development Transportation Facilities Planning and Funding Congestion Pricing and other Transportation Policy

22 Forecasting Page 21 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 A General Approach to Predicting Health Effects Using Epidemiological Research 1.Develop Clear Analytic Objectives 2.Literature Review A.Develop study criteria and data needs B.Identify Sources C.Establish Adequacy of Existing Reviews D.Evaluate studies E.Formal summary and documentation of review 3.Make qualitative inferences on health effects 4.If appropriate and feasible, quantify effects A.Select or generate a summary effect measure B.Estimate Baseline and Changes to “Exposure” C.Predict Health Impacts (PAR, Forecasting) 5.Qualify certainty of assessment & predictions

23 Forecasting Page 22 Rajiv Bhatia Alaska Health Impact Assessment Training 2008 Developing Forecasting Methods for Alaskan HIA Some environmental - health relationships may be generalizable from general population studies HIA forecasting methods in Alaska probably requires new research environmental-health relationships Data collection and monitoring will support long term research efforts


Download ppt "Developing Analytic Forecasting Methodologies for Health Impact Assessment Rajiv Bhatia, MD, MPH San Francisco Department of Public Health."

Similar presentations


Ads by Google