Using maps and spatial analysis to inform global health decision making
Download as PPTX, PDF4 likes2,442 views
The document discusses how the Institute for Health Metrics and Evaluation at the University of Washington uses maps and spatial analysis to inform global health decision making. It summarizes their work on the Global Burden of Disease study which quantifies health loss from diseases, injuries and risk factors in 187 countries. It describes how they measure and analyze data inputs, manage missing data, use covariates and risk factors, conduct spatial-temporal regression, and perform small area estimation to analyze health patterns at subnational levels. Remaining challenges include adding more spatial covariates and conducting disease burden studies at local levels.
Using maps and spatial analysis to inform global health decision making
- 1. UNIVERSITY OF WASHINGTON Using maps and spatial analysis to inform global health decision-making Peter Speyer Director of Data Development @peterspeyer / speyer@uw.edu
- 2. Institute for Health Metrics and Evaluation • Independent research center at the University of Washington • Core funding by Bill & Melinda Gates Foundation and state of Washington • 160 faculty, researchers, and staff • Providing independent, rigorous, and scientific measurement and evaluations • “Our goal is to improve the health of the world’s populations by providing the best information on population health”
- 3. The Global Burden of Disease Study • A systematic, scientific effort to quantify the comparative magnitude of health loss due to diseases, injuries, risk factors • Created 1993, commissioned by the World Bank • GBD 2010 covers 291 causes, 67 risk factors in 187 countries for 1990, 2005, and 2010 by age and sex • GBD country hierarchy: 7 super-regions and 21 regions, based on geographic proximity and epidemiological profiles • Almost 600 country, disease, and risk factor experts from 80+ countries 3
- 5. Measuring burden of diseases and injuries 5 DALYs (Disability-Adjusted Life Years) Health Age Death Deaths Average life expectancy YLLs YLLs (Years of Life Lost) YLDs YLDs YLDs (Years Lived with Disability) Disability weight
- 6. GBD process & spatial challenges • Standards • Coverage • Representa- tiveness • Geographies over time 6 • Missing data • Missing values • Interactive visualizations • Mapping • Making data actionable Find & manage data Analyze data Get data used
- 7. GBD process & spatial challenges • Standards • Coverage • Representa- tiveness • Geographies over time 7 • Missing data • Missing values • Interactive visualizations • Mapping • Making data actionable Find & manage data Analyze data Get data used
- 8. Data inputs 8 •Surveys •Censuses •Vital registration •Verbal autopsy •Disease registries •Surveillance systems Population-based Encounter-level Other •Hospital/ ambulatory/ primary care records •Claims data •Literature reviews •Sensor data •Mortuaries/ burial sites •Police records
- 9. Global Health Data Exchange (http://www.ghdx.org) 9
- 10. 10
- 11. 11
- 12. 12
- 13. 13
- 14. GBD process & spatial challenges • Standards • Coverage • Representa- tiveness • Geographies over time 14 • Missing data • Missing values • Interactive visualizations • Mapping • Making data actionable Find & manage data Analyze data Get data used
- 15. 15
- 16. GBD covariates and risk factors • 300+ covariates, e.g., GDP per capita, access to water and sanitation, education • Gridded population used for several covariates (including AfriPop, AsiaPop, AmeriPop) – Population in coastal areas – Population-weighted average elevation, rainfall, temperature – Population density – Population at risk for causes like malaria • Ambient air pollution, ambient ozone pollution (satellite, surface monitor, TM5 global atmospheric chemistry transport model) 16
- 17. 17
- 18. • Show GBD Compare map for risk factors – Ambient air pollution 18
- 19. GBD – spatial-temporal regression • Capture more information than simple covariate models • Use weighted average of residuals, based on distance in time, age, and space • Geographic weights based on GBD regional hierarchy (country/region/super-region) • Vary weights based on data availability to increase/decrease smoothing 19
- 20. Add graph from COD Viz 20
- 21. GBD process & spatial challenges • Standards • Coverage • Representa- tiveness • Geographies over time 21 • Missing data • Missing values • Interactive visualizations • Mapping • Making data actionable Find & manage data Analyze data Get data used
- 22. 22
- 23. 23
- 24. 24
- 25. 25
- 26. 26
- 27. Small area estimation • Analyze health patterns, outcomes, and intervention coverage for 72 districts in Zambia • Most data only representative at country/province level • Modeling approaches – Pooling data over several years – Borrowing strength by exploiting spatial correlations – Using covariates • Add validation environment – Identify most appropriate measurement strategy – Establish minimum sample size for future data collection 27
- 28. 28
- 29. 29
- 30. 30
- 31. 31
- 32. 32
- 33. 33
- 34. Remaining tasks and challenges • Add more spatial covariates • Conduct burden of disease study at subnational level • Identify best practices for managing geographies (national, subnational) globally over time • Is there a portal for gridded data? 34