Disease Mapping: A Renewed Opportunity for Epidemiologic Investigation Dan Wartenberg UMDNJ-RW Johnson Medical School, Piscataway, NJ and the Cancer Institute of New Jersey Research supported by: • R01 CA92693-02 from the National Cancer Institute • U61/ATU272387 from ATSDR (CDC) • New Jersey Legislature [email protected]
50
Embed
Disease Mapping: A Renewed Opportunity for Epidemiologic Investigation
Disease Mapping: A Renewed Opportunity for Epidemiologic Investigation. Dan Wartenberg UMDNJ-RW Johnson Medical School, Piscataway, NJ and the Cancer Institute of New Jersey Research supported by: • R01 CA92693-02 from the National Cancer Institute - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Disease Mapping: A Renewed Opportunity for Epidemiologic Investigation
Dan WartenbergUMDNJ-RW Johnson Medical School, Piscataway, NJand the Cancer Institute of New Jersey
Research supported by: • R01 CA92693-02 from the National Cancer Institute • U61/ATU272387 from ATSDR (CDC) • New Jersey Legislature
Disease and Exposure Mapping Descriptive studies: Issues of scale, accuracy, missing data, interpretation Cluster Detection and Disease Surveillance Assessment and Availability of Health Services: Screening, access to care Exposure Assessment: Traffic density and Pesticide Studies
Data Linkage Geographical Correlation Studies and Hypothesis Generation The Assessment of Risk from a Point or Local Source Lead poisoning: Linking risk factors, case identification and intervention
Data Integration Habitat Suitability for Disease Vectors—Lyme disease, West Nile,… Infection and Cancer: ‘Population Mixing’ and the role of hygiene/SES
Novel Epidemiologic Study Designs: Population identification and recruitment Rare Exposures AND Outcomes: Magnetic fields and Childhood leukemia
Why has disease mapping growing so rapidly?
Development of GIS Geographically indexed relational database Computer program to map and analyze
spatial data Increasing availability of georeferenced
data Ability to geocode, use GPS Demographics, disease outcomes,
environmental quality, health services
Disease Mapping
The history of disease mapping Several excellent reviews, e.g.,
Cliff and Haggett (1988) Howe (1989) Walter and Birnie (1991) Lawson et al. (1999) Walter (2000) Bithell (2000)
Evolution of Data Displays Univariate mapping
Cases (dot or spot maps) Area counts—aggregate to numerators Rates (choropleth maps)—added denominator Adjusted rates—accommodate confounding Smoothed rates—easier interpretation 3-Dimensional plots—alternative displays
Multivariate maps Bivariate displays Maps of multivariate analyses (e.g., PCA)
Cholera in London, 1849Dot Map+Marginal Histograms
From Cliff and Haggett 1988
Thematic and Rate Maps
From Cliff and Haggett 1988
Three-Dimensional View
From Cliff and Haggett 1988
Space-Time Maps
From Cliff and Haggett 1988
Cartograms
Effect of Age Adjustment
From Monmonier 1997
Displaying Multiple Variables
Aggregation Bias
Rate Instability Rate estimates for areas with small
populations often are unstable Previous approaches
Smoothing (e.g., empirical Bayes) Can miss important peaks
Aggregation of geographic units Need consistent rule to identify grouping criteria Need replicable algorithm for aggregation process
Clarify of misconceptions Allay concerns that are unfounded Initiate study when concerns are well founded
Encourage Remediation and Prevention Determine if situation is a sentinel of a larger
problem Identify unknown exposure situations
Facilitate Scientific Discovery Identify new exposure-disease link Identify new carcinogens
New Opportunities Rapid response to resident concerns (e.g., SAHSU)
Online, real-time display of cancer data, known point sources, other environmental data
Space-time pattern of cancer incidence at small scale
How many ‘clusters’ are there? Where are the lows as well as the highs, and why?
Prioritize potential investigations What warrants follow up?
Implementation of prospective surveillance Active vs. Passive approach
Exposure Assessment/Estimation
An Epidemiologist’s Perspective on Exposure Assessment
A tool rather than an end in and of itself
User rather than developer Specific uses
Conduct otherwise undoable epidemiology
Improve scope, focus or interpretability
Point and Nonpoint Source Pollution Point source pollution
Toxics, birth defects and low birth weight (Geschwind et al. 1992; Stallones et al. 1992; Berry
and Bove 1997; Croen et al. 1997) Incinerators and cancer (Gatrell and Dunn
1995) Nonpoint source pollution
Drinking Water (Nuckols et al. 1995) Pesticides (Nuckols et al. ) Traffic derived air pollution
(English et al. 1999; Ritz et al. (2000, 2002, 2003; Brauer et al. 2003)
Air Pollution from Traffic
From Wilhelm and Ritz 2003
Data Linkage
Map-Based Correlational Studies
Various historical efforts New impetus triggered by NCI Atlas (1970s)
Compared mortality maps to possible exposures Then validated with traditional epidemiology
Bladder cancer and chemical manufacturing Nasal adenocarcinoma and furniture manufacturing Lung cancer and shipyards Oral cancers among women and snuff use
Approach reinvigorated in past few years New tools and geocoded databases
Despite the ‘Bad Press’ these can be useful Must be careful of limitations of ‘ecologic analysis’
Preventing Childhood Lead Poisoning
Screening data readily available Using the GIS
screening assessing the impact of targeted screening evaluating predictive equations
Are SES, housing age, other factors predictive
surveillance combining maps of exposure and disease
target educational and remediation programs
Preliminary Lead Study Proposed strategy for targeted screening
use demographics and local hazards to predict lead poisoning rates
map cases validate by comparing cases to predictions and
then adjust prediction equation Developed hypothetical model and data Implement appropriate
screening/intervention (Wartenberg 1992)
Childhood Lead Exposure:Advantages of Using a GIS
Can expand to nationwide demographic evaluation
Can explore prediction models using individual data
Can look for other risk factors Can adjust for bias due to spatial
autocorrelation Can design/evaluate screening and
interventions
Population Mapping
Magnetic Fields and Childhood Leukemia
Do magnetic fields cause cancer? Blue Ribbon panels--possible Combined Analyses of Studies Show
Small but consistent elevations of risk A moderate exposure-response gradient Few subjects and “high” exposures
Exposure Metric Pooled Analysis Meta-Analysis Ahlbom et al. (2000) Greenland et al. (2000) Wartenberg (2001)
Used seroprevalence in dogs as surrogate for human risk
Combined to predict overall human risk
Lyme Disease RiskLyme Disease Risk based on co-Kriging
Habitat Suitability and Seroprevalence Data in Dogs
Lyme Disease in NE US
FIGURE 2: Habitiat Suitability Spatial Structure
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Average Distance Between Observations (KM)
Mo
ran
's I
Estimated Tick Establishment
Childhood Leukemia and Clusters Observed cluster of childhood cancer in
Seascale, UK near nuclear power station Few causes known of childhood leukemia Government concerned (much publicity)
Commissioned series of studies Results suggest only risk was paternal
exposure Seemed contrary to radiation risk
Others sought alternative explanation
Hypotheses
Kinlen—Population Mixing Children from isolated populations, with
decreased exposure to infections, when exposed to others from regions of greater population density are at increased risk of leukemia, likely due to viral etiology
Greaves—SES and Hygiene Delay of exposure to infections from infancy to
each childhood, such as due to improvements in socioeconomic status and hygiene, puts children at increased risk of leukemia and lymphomas.
Evaluations of Hypotheses Several studies in UK, generally supportive Few in other countries, mixed This study: Analysis of SEER Data
Use data from states with rural counties Test hypothesis for rates of ALL Compare rates by
Size of population change Size of income
Compare to pattern of CNS cancers as reference (to control for methodology)
Results for IA, NM, UTPercentage Change in Population Size