Transcript Keynote Address - The University of Texas at Dallas

Applying Data Warehousing to
Community Health Assessment
WITS’99 Keynote Address
Alan R. Hevner
University of South Florida
[email protected]
Preface - WITS Retrospective
As we approach 2000, a quick look back:
WITS’91 - Boston (Ram and Wang)
WITS’92 - Dallas (Storey and Whinston)
WITS’93 - Orlando (Hevner and Kamel)
WITS’94 - Vancouver (De and Woo)
WITS’95 - Amsterdam (Jarke and Ram)
WITS’96 - Cleveland (Ernst and Sen)
WITS’97 - Atlanta (Segev and Vaishnavi)
WITS’98 - Helsinki (Bubenko and March)
WITS’99 - Charlotte (Narasimhan and Sarkar)
December 11, 1999
WITS'99 Keynote Address
2
Outline
Research Motivation - Community Health
Measurement and Assessment
The CATCH Methodology
A Data Warehousing Solution
Data Dissemination Modes
Community Health Decision Making
A CATCH Demonstration
December 11, 1999
WITS'99 Keynote Address
3
Acknowledgements
Co-Principal Investigators
James Studnicki - College of Public Health, USF
Don Berndt - College of Business Admin., USF
Research Staff
Center for Health Outcomes Research Staff
Doctoral and Masters Students
Funding
U.S. Dept. of Commerce TIIAP Grant
Bear Stearns Research Laboratory
Florida Communities
December 11, 1999
WITS'99 Keynote Address
4
Research Motivation
 U.S. has the Highest Per Capita Health Expenditures in the
World
 Low Rank of U.S. as defined by Health Status Indicators
 Transition from a Disease to Health focus and from a
Treatment to a Prevention strategy
 Health Priorities defined by Political Agendas and the
Managerial Objectives of Health Organizations rather than
Objective Evaluation
 Pluralistic, Non-Integrated Health Care Systems
 No Single Organization is Responsible for the Health of the Community
 No Uniform Method to define the “Health of the Community”
which is Universally Accepted and Consistently Applied
December 11, 1999
WITS'99 Keynote Address
5
Community Health Planning
Institute of Medicine (IOM) 1988 Report on
the Future of Public Health
Recommends a regular and systematic collection,
assemblage, and analysis of information on the
health status and needs of communities.
IOM 1997 Report on Using Performance
Monitoring to Improve Community Health
Calls for a Community Health Profile which can be
used to support priority setting, resource allocation
decisions, and the evaluation of health program
impacts.
December 11, 1999
WITS'99 Keynote Address
6
Collaborative Health Decision Making
Multi-Sector Community Health Stakeholders
Health Organizations
Public Sector Agencies
Medical Care Providers
Businesses
Religious Community
Educational Institutions
Government Agencies
Decisions must be based on Unbiased, Timely
Information
December 11, 1999
WITS'99 Keynote Address
7
CATCH Methodology
Comprehensive Assessment for Tracking
Community Health (CATCH)
Project initiated in 1991
14 Florida County Applications
Marion County, Indiana (Indianapolis)
Potential Regional, National, and
International Applications
December 11, 1999
WITS'99 Keynote Address
8
Indicator 1
Indicator 2
.
.
.
Indicator i
.
State Averages
Indicator 1
Indicator 2
.
.
.
Indicator i
.
.
Community
Health Indicators
Indicator 1
Indicator 2
.
.
.
Indicator i
.
CATCH
Methodology
State
Favorable Unfavorable
F
Fav.
Peer
Fav/Fav
Indicators
Fav/Unfav
Indicators
I
L
T
Unfav. Unfav/Fav
Peer Community
Averages
Additional Health
Standard
Comparisons
Indicators
Health
Challenges
E
R
1. Indicator i
2. Indicator j
,
.
.
S
CATCH N-Dimensional
Comparison Matrix
Prioritized List of
Community
Health
Challenges
Data Collection and Analysis
Ten Indicator Groups
Demographics
Socioeconomic
Maternal and Child Health
Social and Mental Health
Physical Environmental Health
Health Status: Morbidity/Mortality
Sentinel Events
Infectious Diseases
Health Resource Availability
Behavioral Risk Factors
December 11, 1999
WITS'99 Keynote Address
10
Priority Filters
Number Affected
Economic Impact
Availability of Efficacious Intervention
Magnitude of Difference
Trend Analysis
December 11, 1999
WITS'99 Keynote Address
11
Peer Comparison
CRITERIA
Hillsborough
Peer
Group
Duval
25.41%
26.58%
24.84% 24.46%
Orange
Polk
% Population
< Age 18
24.86%
% Population
> Age 64
12.71%
13.01% 11.27%
11.51% 18.37%
% Non-white
Population
15.32%
21.13% 27.20%
19.08% 14.76%
% Families Below
Poverty Level
9.5%
9.0%
Source: Florida County Comparisons 1995
9.8%
7.8%
9.4%
Comparison Matrix
CATEGORY
INDICATOR CO PEER ST
Socioeconomic % Labor force 5.2% 5.8% 6.6%
unemployed
Maternal &
Infant mortality: 12.6
14.4 11.9
Child health
non-white
STATE
FAVORABLE UNFAVORABLE
Infectious
Disease
Tuberculosis
cases
0.31
Health Status
Colorectal
cancer
11.3
10.8 12.3
FAV
51.3
41.7 45.6
PEER
Sentinel Events Cervical
cancer
late stage
Resource
Licensed
Availability
hosp. beds
Physical/
Environmental
Drowning
fatalities
Domestic
viol. cases
Current
Behavioral Risk smokers
Social & Mental
0.25
0.57
% Labor force
unemployed
Infant mortality:
non-white
Challenges:
UNFAV
5.9
4.7
4.5
2.4
2.0
2.7
1041.0 1041.8 864.1
24.8
26.9
23.1
Drowning
fatalities
Late stage
cervical cancer
Further Screening
Priority Filters
PRIORITIZATION
SCREENS
Availability
of
Efficacious
Intervention
Economic
Impact
Number of
People
Affected
Magnitude
of
Difference
SAMPLE HIGH
PRIORITY AREAS
Trend
Direction
and
Magnitude
Avoidable Hosp.:
Asthma
Low birthweight
Gonorrhea cases
Stroke
Cervical cancer:
%late stage
Pneumonia/
Influenza
Social and Mental Health
INDICATORS COMPARED TO STATE & PEER VALUES
STATE
FAVORABLE
FAVORABLE
P
E
E
R
UNFAVORABLE
AA = Age Adjusted
UNFAVORABLE
Child maltreatment
Elderly abuse
Homicide AA mortality
Crude homicide rate:non-white
Burglary offenses
Forcible sex assaults
Crude homicide rate: total
Illegal drug sales
Crude suicide rate: white
Domestic violence cases
Simple assaults
Aggravated assaults
Illegal drug possession
Crude homicide rate: white
Suicide AA mortality
Crude suicide rate: total,
non-white
Intentional injury AA mortality
Alcohol related motor vehicle
accidents
Alcohol related motor vehicle
mortality
Psychiatric admissions
% w/ good mental health
Indicator Fact Sheet
INDICATOR: AIDS CASES
1994 AIDS CASES,
Incidence rate per 100,000 population
FIVE YEAR TREND ANALYSIS
80
80
60
48
40
20
16
90
0
County
Peer
91
92
93
94
Florida
KEY: Thick line = County value, Thin line = Florida value
1990
1991
1992
1993
1994
________________________________________________________________
County:
19.5
24.6
26.2
55.3
27.6
Florida:
29.6
41.5
41.7
77.2
61.5
Source: PHIDS
CATCH Data Warehouse
Manual CATCH Limitations
Labor-Intensive and Slow
Four months per report
Longitudinal Trend Analyses are Cost Prohibitive
Extension of County Reports to State, National,
and International Reports
Knowledge Discovery Potential not Realized
CATCH Data Warehouse Solution
December 11, 1999
WITS'99 Keynote Address
17
Data Warehouse Challenges Construction
Data Collection
Data Sources
Data Quality
Extraction, Transformation, and Transportation
Data Warehouse Design
Star Schemas
Data Staging
Sizing and Cleansing
Quality Assurance
December 11, 1999
WITS'99 Keynote Address
18
Hospital Discharge Star Schema
INDICATOR
LOAD EVENT
#
*
*
*
o
o
o
o
o
o
o
o
ID
USERNAME
STATUS
START
END
PROCESS
VERSION
TYPE
ROWS_PROCESSED
ROWS_REJECTED
DESCRIPTION
NOTE
GENDER
# ID
* DESCRIPTION
* ABBREVIATION
discharge fact
discharge fact
discharge fact
#
*
*
o
o
o
o
o
o
o
o
o
o
o
o
ID
NAME
DESCRIPTION
ABBREVIATION
NOTE
TYPE
FREQUENCY
GEO_GRAIN
TIME_GRAIN
ELECTRONIC
MULTIPLIER
ECO_IMPACT
EFFICACY
LAST_LOAD
NEXT_LOAD
gender dimension
indicator dimension
load dimension
CT DISCHARGE
discharge fact
* VALUE
YEAR
#
o
o
o
YEAR
HALF DECADE
DECADE
NOTE
RACE
# ID
* CATEGORY
o ABBREVIATION
discharge fact
year dimension
race dimension
age dimension
county dimension
discharge fact
AGE
#
*
*
o
o
o
o
December 11, 1999
ID
AGE
UNIT
CATEGORY
Y5 BAND
Y10 BAND
CUST BAND 1
discharge fact
COUNTY
#
*
o
o
o
o
o
*
ID
NAME
MIL BASE
MIL BASE CNT
COASTAL
REGION
HEALTH DISTRICT
VITAL STATS ID
WITS'99 Keynote Address
19
ICD-9 Code Dimension Hierarchy
ICD9 PROC CHAPTER
# ID
* DESCRIPTION
includes
included in
ICD9 PROC SECTION
includes
# ID
* DESCRIPTION
CCHPR PROCEDURE
includes
# ID
* DESCRIPTION
ICD9 DX CHAPTER
# ID
* DESCRIPTION
CCHPR DX
# ID
* DESCRIPTION
includes
included in
includes
ICD9 DX SECTION
# ID
* DESCRIPTION
included in
includes
included in
ICD9 CODE
included in
December 11, 1999
# CO DE
* DISEASE
o CATEGO RY
WITS'99 Keynote Address
included in
20
Data Warehouse Challenges Operations
User Interfaces
Performance
Security
Backup and Recovery
Knowledge Discovery
Data Mining
December 11, 1999
WITS'99 Keynote Address
21
Data Dissemination Modes
Effective Presentation of CATCH Information
to Community Decision Makers
Data Dissemination Modes
Pre-defined Reports
Data Browsing
Ad-hoc Queries
Internet Access
Hypertext Information Screens
Dynamic Access to Data Warehouse
December 11, 1999
WITS'99 Keynote Address
22
Community Group
Decision Making
Research Field: IT Support for Group
Decision Making
Research Question: How will communities
make most effective use of the CATCH data
for health care decision making?
Research Testbed: During 2000 we will
provide CATCH reports to all 67 Florida
counties.
December 11, 1999
WITS'99 Keynote Address
23
Group Decision Making Issues
 Motivation of community to use data
 Presence of a champion for specific actions
 Size and make-up of the decision making group
 Speed of the decision making process
 Stakeholders around the table and their influence
 Resource constraints
 Political nature of the process
 Differential accesses to data among communities
 Ease of access and usefulness of the data
 Requests for customized analyses
 Information exchange patterns and practices
December 11, 1999
WITS'99 Keynote Address
24
CATCH Data Warehouse
Demonstration
Policy Question on Racial Disparity in Infant
Mortality in Florida:
“What is the pattern of variation in infant mortality
between whites and non-whites throughout Florida?
What factors best explain this variation?”
December 11, 1999
WITS'99 Keynote Address
25
Data Browsing Strategy
Produce a Table of Florida Counties and
Infant Mortality Data
Sort and Graph the Information
Cluster the Counties into Four Groupings
Select Factors for Analysis and Correlation
Perform Further In-Depth Analyses
Data Mining
Multivariate Statistics
December 11, 1999
Neural Networks
WITS'99 Keynote Address
26
December 11, 1999
WITS'99 Keynote Address
27
December 11, 1999
WITS'99 Keynote Address
28
December 11, 1999
WITS'99 Keynote Address
29
December 11, 1999
WITS'99 Keynote Address
30
December 11, 1999
WITS'99 Keynote Address
31
December 11, 1999
WITS'99 Keynote Address
32
December 11, 1999
WITS'99 Keynote Address
33
December 11, 1999
WITS'99 Keynote Address
34
December 11, 1999
WITS'99 Keynote Address
35
December 11, 1999
WITS'99 Keynote Address
36
December 11, 1999
WITS'99 Keynote Address
37
December 11, 1999
WITS'99 Keynote Address
38
Conclusions
The Application of Data Warehousing
Technology to Community Health Care can
make a Social Contribution
Technical Research Challenges
Collaborative Group Decision Making: What
factors are associated with effective
community use of CATCH data?
Leadership
Infrastructure
Decision-Making Process
Public/Private Sector Cooperation
December 11, 1999
WITS'99 Keynote Address
39
Appendix:
CATCH Data Indicators
December 11, 1999
WITS'99 Keynote Address
40
Data Indicators
DEMOGRAPHIC CHARACTERISTICS
% Total population by gender
% Total population by age
% Total population by race
% Population rural
% Labor force by gender
Median Age
Net migration
Live births per 1,000 population
Deaths per 1,000 population
December 11, 1999
WITS'99 Keynote Address
41
Data Indicators
SOCIOECONOMIC CHARACTERISTICS
 Non-graduates of high school
High school dropouts
 Per capita income
Labor force unemployed
 Persons below poverty level
WIC eligibles
 Medicaid eligibles
% Medicaid births
 HMO enrollment
% enrolled in a health plan
 Families with children < age 18 below poverty level
 Population receiving food stamps
 Students eligible for free/reduced lunch program
 %Low income persons with access to dental care
December 11, 1999
WITS'99 Keynote Address
42
Data Indicators
MATERNAL AND CHILD HEALTH
 Infant Mortality
Child mortality
 Neonatal mortality
Post neonatal mortality
 Low birthweight
Very low birthweight
 Perinatal condition mortality
 Birth Defects Mortality
 % Live births w/1st trimester prenatal care
 % Live births w/3rd trimester prenatal care
 % Live births w/ no prenatal care
 Live births to mothers < age 15
 Live births to mothers age 15 - 17
 Live births to mothers age 18 - 19
 Repeat births to teens
December 11, 1999
WITS'99 Keynote Address
43
Data Indicators
PHYSICAL ENVIRONMENTAL HEALTH
 Salmonella cases
Campylobacter cases
 Shigella cases
Rabies in animals
 Lead poisoning
Fluoridated water
 Firearm fatalities
Drowning fatalities
 Poisoning fatalities
Bicycle fatalities
 Contaminated wells
Septic tank repair permits
 Enteric disease cases: total and in children < age 6
 Foodborne and waterborne outbreaks
 Motor vehicle mortality - age adjusted
 Unintentional injury mortality - age adjusted
December 11, 1999
WITS'99 Keynote Address
44
Data Indicators
INFECTIOUS DISEASE
 AIDS incidence, cumulative cases, & mortality
 HIV seropositivity
 Infectious Syphilis cases
 Congenital Syphilis cases
 Gonorrhea cases
 Chlamydia cases
 Hepatitis A and B cases
 Meningitis cases
 Tuberculosis cases
 Tuberculosis mortality - age adjusted
 % Vaccinated by kindergarten
December 11, 1999
WITS'99 Keynote Address
45
Data Indicators
SOCIAL AND MENTAL HEALTH
 Alcohol Related motor vehicle accidents & mortality
 Assaults: Forcible sex, Burglary, Simple and Aggravated
 Juvenile delinquency rates
 Suicide - crude & age adjusted
 Intentional injury - age adjusted
 Homicide - crude & age adjusted
 Child Abuse, Elderly Abuse - reported and confirmed cases
 Domestic Violence - Reported cases
 Mental health of adults: days/month w/o good mental health
 Hospitalization rates for:
 Baker Act, Psychoses, Depression, Alzheimer's Disease, Alcohol
abuse & Drug abuse
December 11, 1999
WITS'99 Keynote Address
46
Data Indicators
HEALTH STATUS INDICATORS
Morbidity Cases
Melanoma
Breast cancer
Colorectal cancer
Smoking related cancers
Prostate cancer
Cervical cancer
Lung & bronchus cancer
Age Adjusted Mortality Rates (Crude)
Chronic liver disease & cirrhosis (crude)
Pneumonia/Influenza (crude)
Diabetes Mellitus (crude)
Cardiovascular disease
Heart disease (crude)
Stroke (crude)
C.O.L.D.
YPLL
December 11, 1999
WITS'99 Keynote Address
Melanoma
Breast cancer
Cervical cancer
Colorectal cancer
Lung/smoking rel. cancer
Preventable cancer
Prostate cancer
All cancers (crude)
47
Data Indicators
SENTINEL EVENTS
Vaccine Preventable Diseases
Measles
Mumps
Rubella
Pertussis
Late Stage Cancers
Breast cancer cases - % late stage
Cervical cancer cases - % late stage
Avoidable Hospitalizations
Asthma
Cellulitis
Congestive heart failure
Diabetes
Gangrene
Hypokalemia
December 11, 1999
Immunizable conditions
Malignant hypertension
Perforated/bleeding ulcer
Pneumonia
Pyelonephritis
Ruptured appendix
WITS'99 Keynote Address
48
Data Indicators
HEALTH RESOURCE AVAILABILITY
Licensed Beds
Hospitals
Nursing homes
Licensed Professionals
Doctors
Dentists
RNs
Pharmacists
Nurse Midwives
Opticians/optometrists
LPNs
Dieticians
Psychologists
Ratio of Medicaid Eligibles to Participating
Physicians
December 11, 1999
WITS'99 Keynote Address
49
Data Indicators
BEHAVIORAL RISK FACTORS
Mammograms
Pap smears
Blood pressure screening
Cholesterol screening
Smoking
Obesity
Seat Belt Use & Child Seat Use
Bicycle Helmet Use
Check-up in last year
Health Care Foregone due to cost
December 11, 1999
WITS'99 Keynote Address
50