UW MED

1
UW MED PHINEXUniversity of Wisconsin
Medical Record Public Health Information
Exchange

• Wisconsins Clinical EMR Public Health Data
  Exchange Pilot
• Theresa Guilbert, MD, MS
• Project PI
• University of Wisconsin-Madison
• Department of Pediatrics
• tguilbert_at_wisc.edu

2
Why study chronic disease risk factors present in
the environment community?
3
(No Transcript)
4
Multi-Level Approach

• A multilevel approach that includes an ecological
  viewpoint may help to explain heterogeneities in
  chronic disease expression across socioeconomic
  behavioral, and geographic boundaries that remain
  largely unexplained
• Improved knowledge regarding disease disparity is
  important in order to develop intervention
  strategies

5
Overall Hypothesis

• Data exchange between UW Dept Family Medicine
  (DFM) clinics and the Wisconsin State Division
  of Public Health (DPH) and subsequent linking of
  these data to public databases on geographical,
  environmental, socioeconomic, and demographic
  profiles will highlight areas of disparity and
  discover novel chronic and communicable disease
  risk factors

6
Rationale

• By having such a large clinical data set and
  using sophisticated spatial and multivariate
  modeling and data mining tools, areas of
  healthcare disparities will be highlighted
• New information about risk factors will be
  discovered to guide
• Clinical care
• Inform clinical quality improvement
• Design public health interventions
• Facilitate further research

7
Specific Aims

• Establish a health information exchange between
  DFM clinics the DPH using a HIPAA privacy rule
  compliant limited data set
• All Personal Health Identifiers are removed
  except for gender, ethnicity/race, birth
  year/month, dates of service, zip code, and
  census block group
• This approach has been proved by the UW IRB
• Determine areas and populations of chronic and
  communicable disparity through collaboration with
  the UW Applied Population Laboratory (APL)

8
Specific Aims

• GIS and spatial analyses of population trends to
  chart areas of disparity and geographic
  characteristics of those communities that can
  lead to hypotheses regarding etiology
• Assess novel environmental and community risk
  factors by matching CBG coded EHR to its
  community level demographic and socioeconomic
  characteristics using data bases available
  through the APL and DPH.

9
Specific Aims

• Use multivariate (logistic and Poisson
  regression, fixed and random effects regression
  modeling) and data mining techniques at DPH to
  create predication models that specify risk
  factors associated with asthma among many
  environmental and community based factors from
  the census and commercial databases
• Using statistical clustering techniques analyze
  and determine prominent within patient disease
  co-morbidity groupings and determine the
  individual and community risk predictors of these
  clusters

10
Specific Aims

• DPH operates the Public Health Information
  Network (PHIN), a secure, web based system
• Advanced statistical and GIS modeling services
• SAS Business Intelligence Server/Enterprise Miner
• ESRI ArcGIS server
• Available community level databases include
• Census Demographic
• Tapestry Segmentation
• Consumer Spending
• Business Summary and Location
• Retail Market Place

11
Multi-Level Modeling and Data Mining of Disease
Risk, Disparity, and Health Outcome Quality
Outcomes Patient Clinician Clinic Community
Outcomes Factors Factors Factors Factors
Asthma Age Age Location Census Block Group
Diabetes Gender Gender Capabilities Poverty
CVD / CHF Race/ethnicity Certifications Processes Education level
Immunizations Co-morbidities Graduation   Built environment
Obesity Medications date   Traffic
Hypertension Education Years of practice   Recreation / parks
Smoking Literacy   Safety / crime
Alcohol Language     Psycho-demographics
A1c level Insurance     Restaurant mix
LDL Urban / Rural     Fast food sales
HDL Census Block Group     Fresh fruit vegetable sales / consumption
BP        
Hospitalizations       Public Health Program Information
Health Care -Process factors        
(e.g, time to repeat follow-up)        
Electronic Health Record Hospitalization Data Electronic Health Record Hospitalization Data Electronic Health Record Hospitalization Data Electronic Health Record Hospitalization Data Census / ESRI BA Data
12
Data Sets

• Public Health Behavioral Risk Factor
  Surveillance System 2004-2009
• Clinical UW Family Medicine UW Hospitals and
  Clinics (demographics, diagnoses, problem lists,
  laboratory test results, vital signs, procedures,
  medication lists)
• Community Data ESRI geo-coded data (CBG)

13
ESRI Data Bases Fresh Fruit Vegetable
Consumption IndexMilwaukee Suburbs Census
Tracts
Color Ramp Grey Lowest White-Low Cream-Medium
Yellow-High Red-Very High Source ESRI / BLS
Consumer Expenditure Survey
14
Fresh Fruit Vegetable Consumption IndexWith
Individual Store Location / Sales
VolumeMilwaukee Suburbs Census Tracts
Color Ramp Grey Lowest White-Low Cream-Medium
Yellow-High Red-Very High Circle size store
sales volume Source ESRI / BLS Consumer
Expenditure Survey
15
Disparity in Dane County?

• 50 of K-12 students in Madison schools are
  economically disadvantaged (gt 70 in some)
• 50 of the kids in the Madison Metropolitan
  School District are of racial/ethnic minority
  groups (poor access to care)
• Disparity does not always correlate with poverty
• Falk Elementary School (West Madison) has 9
  children with asthma and a 65 poverty rate with
  70 minorities
• Mendota Elementary School (North Madison) has 22
  children with asthma and a 70 poverty rate with
  74 minorities

16
Collaborative Effort

• Brian Arndt-UW DFM
• Bill Buckingham-UW APL
• Tim Chang-UW Biostats
• Dan Davenport-UW Health
• Kristin Gallager-UW Pop Health
• Theresa Guilbert (PI)-UW Peds
• Larry Hanrahan-DPH

• David Page-UW Biostats
• Mary Beth Plane-UW DFM
• David Simmons-UW DFM
• Aman Tandias-DPH
• Jon Temte-UW DFM
• Kevin Thao-UW DFM
• Carrie Tomasallo-DPH

17
What have we learned so far?
18
Presenters

• Brian Arndt MD-UW Family Medicine
• UW MED- PHINEX Diabetes Obesity Use Case
  Clinician Lead
• Estimating the Prevalence of Diabetes in
  Wisconsin
• Kevin Thao-UW Family Medicine
• The Prevalence of Type 2 Diabetes Mellitus in a
  Wisconsin Hmong Patient Population

19
Presenters

• Carrie Tomasallo, PhD, MPH-Wisconsin Division of
  Public Health
• Wisconsin Asthma Program
• Estimating Wisconsin Asthma Prevalence Using
• Clinical Electronic Health Records and Public
  Health
• Data