What is the relevance of nursing informatics to health disparities

1
What is the relevance of nursing informatics to
health disparities?

• Patricia Flatley Brennan, RN, PhD
• With the assistance of
• Teresa Zayas-Caban
• Kanittha Volrathongchia

School of Nursing University of Wisconsin-Madison
2
Acknowledgements

• The work of colleagues over the past 50 years
• Conversations and debates with many including
• AMIA colleagues Zak Kohane, Russ Altman, Mark
  Musen, Gil Kuperman, Carol Friedman, Bonnie
  Kaplan, Charley Safran
• UW-Madison colleagues Rima Apple, Richard
  Staley, Barbara Bowers, Josette Jones the
  Brennan research team, in particular Teresa
  Zayas-Caban and Kanittha Volrothongchia
• The Moehlman Bascom Professorship
• NIH

3
Our team
4
Goals for today

• Describe accomplishments in health informatics
• Illustrate how techniques and approaches in
  health informatics aid in detecting,
  understanding, and mitigating health disparities
• Identify points of intersection between health
  informatics and health disparities

5
Health Informatics
Health Biomedicine Health Care Health
Policy Public Health Genomics
Formalization, Transformation Manipulation,
Application of Information
Semantics
Syntax
6
Nursing Informatics
Knowledge RepresentationKnowledge
ManagementKnowledge Discovery
Diagnosis Treatment of Human Response
Formalization, Transformation Manipulation,
Application of Information
7
Health Policy Clinical Practice
Health Care
System Definition
System Evaluation
System Construction
System Validation
Informatics
System Description
8
Progress In Health Informatics Research
Health Policy Clinical Practice
System Definition
System Evaluation
System Validation
System Construction
System Description
9
NAMING AND LABELING
ReferenceModels
Ontologies
Formal Languages
Vocabularies
Controlled Terms
Clinical Practice
System Definition
System Evaluation
Terms
System Validation
System Construction
System Description
10
Genomic Data Mining
Bayesian Belief Nets
Explanation
Clinical Practice
AI, Probability Models and certainty factors
System Definition
System Evaluation
System Construction
System Validation
System Description
Decision Support for Diagnostics Therapeutics
11
Alerts Reminders
Clinical Information systems
Bringing knowledge to the point of care
12
Health Policy Clinical Practice
Systems Implementations
Knowledge Acquisition Strategies
System Definition
System Evaluation
System Construction
System Validation
Terminology and Data Models
Basic Medical Information Science
System Description
13
So what does all of this mean for people
interested in health disparities?
14
What constitutes health disparities?
Underserved
minority health
Burden of Illness
15
Health Disparities

• Variability in health states or access to health
  services based on characteristics of the person
  or group that should not, independently, cause
  such variability

16
An informaticists view of health disparities

• Differential access to known health services
• Differential experience of known health care
  problems
• Differential manifestation of health phenomena
• Differential response to health, illness and
  developmental challenges
• Differential approaches to management of health,
  illness, and development challenges

17
Health Informatics brings to the study of
health disparities

• Representational tools
• Strategies to extract knowledge from existing
  data sets
• Innovations for reaching individuals unreachable
  through the existing health care delivery system

18
At the beginning-- Its all a name game
Abstraction
Minimum Data Set
Clinical Record
Document
Label or Name
Real world state or experience
19
R
Formalismthat is comprehensible, computable,
and translatable
RReal World Experience
NLP
Artifact (e.g. laboratory report, discharge
summary)
20
Representation schemes

• Health disparities investigators offer
• Language
• Stakeholders
• Range of diversity
• Vernaculars

• Informatics offers
• Representation schemes
• Terminology models
• Information Models
• Integration mechanisms

21
Using existing data sets to understand
experiences of diverse groups
22
Select DataSet
Evaluate Congruence Verdicality
Integrity AAA PreProcess
Conduct analysis Interpret findings
Choose Model
23
An Illustration Using the MDS to predict Falls
Risk(Kanittha Volrothongchai)

• Selecting the data set
• Congruence with the research aims
• MDS created as a finance and authorization tool
• Veridicality
• Data Definitions
• Acceptable but changed mid-period
• Data Source (Compliance reports)
• Integrity
• How were the data originally collected?
  (clinician-reported)
• How were the data stored? (paper --gt computer --gt
  database)
• AAA Availability, Authorization, Access

24
PreProcessing of Large Data Sets

• Descriptions
• By Case
• Number (35,000)
• Non-redundant and unique (25,000)
• Complete (6,000)
• By Variables
• Stability of data definition (changed over time
  dual purpose)
• Controlled terminologies (local modification)
• Complete (no)
• Patterns of missingness
• Handling missing data (imputation, hot deck, mean
  value)

25
Selecting an analysis strategy

• Revisit the purpose of the work
• Explore data mining and knowledge discovery in
  data (KDD) methods
• General rule select the most robust model that
  the data can support
• Model Validation

26
Using sophisticated models to understand
experiences of diverse groups
27
Simulation tools to model systems dynamicsTeresa
Zayas-Caban

• Question Does race and ethnicity affect access
  to care?
• Financial barriers impede access
• African-American and Latinos receive fewer
  services when compared with whites
• Underserved Minority states co-occur
• What governs access to care?
• Individuals ability to pay
• Clinicians willingness to accept payment
  mechanism

28
Simulation model
29
What is needed to make a simulation model?

• Estimates of service time by insurance type
• Distribution of insurance type
• Distribution of insurance type by ethnic group
• A formula relating all of them
• Arrival Rate
• Assigned probability of insurance and ethnicity
• ServiceTimeINT(ArrivalTime) (7 minutes)
• Adjustment of service time by ethnicity

30
Data Obtained from Public Sources
County Health Assessment Status Report
31
Data obtained from Published Papers

• Service times

32
Estimated Service Demand using Local Information

33
Critical Assumptions in the Simulation

• Uninsured will not get served
• Medicaid-insured will get served half of the time
• Insurance state and ethnicity independent
• Standard simulation distributional and behavioral
  assumptions

34
Does ethnicity and insurance type influence
access to care?
35
Using information technology to reach diverse
groups with relevant interventions
36
HeartCare

• Web-based, tailored coaching and communication
  service established to ease hospital discharge

• Patients recovering from Cardiac Surgery
• Average age 57
• Accessed HeartCare 2-3 times a week
• Is it really better than standard care?
• 18-month community experiment
• Compared HeartCare to Usual Care and to CHIP, an
  audio-tape

37
(No Transcript)
38
(No Transcript)
39
Does access to HeartCare improve recovery from
CABG?

• Yes!

40
Access to HeartCare reduces Symptom Impact
SIP
CHIP
HeartCare
Months Since Surgery
41
Access to HeartCare reducesPost-surgery
Depression
CHIP
Depression
HeartCare
Months Since Surgery
42
Some groups find difference in use based on race,
ethnicity, and age
43
Its hard for informaticists to think about
health disparities!
44
Why does the concept of health disparities
challenge informatics?

• Formalisms, a foundational concept, require
  stability in the underlying phenomena represented
  by words (A rose is a rose)
• Socially-constructed concepts like race or
  ethnicity lack stability and are used
  idiosyncratically
• Technology is inherently value-laden, both in its
  development and in its deployment
• We lack incentives strategies to make values
  explicit
• Homomorphisms rule, Occams Razor guides

45
Guidance for the Future

• Capitalizing and responding to points of
  intersection between health informatics and
  health disparities

46
A rose by any other name
47
Point of Intersection Labeling Naming

• Current State
• Informatics developments arise from within the
  formal health care system
• The formal health care system is based on
  majority experiences, values, and investments
• Opportunity
• Enrich clinical vocabularies with terms more
  reflective of human response
• Critically appraise modify existing
  vocabularies for representativeness

48
Genes are more than bits of people, andPeople
are more than bits of genes
49
Point of Intersection Genetic Health

• Current State
• One race, ONE RACE!
• Health consequences appears to result from
  cultural, environmental and behavioral factors
  more so than genetic factors related to race
• None the less, some genetic structures cluster
  within recognized groups
• Opportunity
• Expand clinical and personal health records to
  permit careful exploration of phenotype
  consequences

50
Decision support works behind the scene
51
Point of Intersection Decision Rules, Alerts
Guidelines

• Current State
• Financial and policy pressures towards stable,
  reproducible guidance for care investment,
  distribution, and evaluation decisions
• Increasing awareness that the success of health
  care rests on articulation with every-day life
• Opportunity
• Policy Participation (NHII)
• Expand research to show consequences of current
  thinking

52
Research into the experience of health
disparities leads to richer, more appropriate
data elements, data definitions, and reference
data models
53
ask yourself if the step you contemplate is
going to be of any use to the poorest and
weakest man whom you have seenWill he gain
anything by it?Will it restore him to control
over his life and destiny?then you will find
your doubts and yourself melting away

• Gandhi,1947

54
Thank you for your interest and
attention!healthsystems.engr.wisc.edu

• pbrennan_at_engr.wisc.edu