Using Clinical Data for Scholarly Projects

Report
DATA
Role of data in QI and Scholarship
Sources/categories of data
Characteristics of “good” data
Administrative databases – pros &cons
New Informatics support for
Scholarship, QI, and Translational
Research
Data are the tools for quality improvement
“Learning Healthcare System”
Data Sources
Clinical Data
Review medical records
Registries
Clinical Trials
Administrative
Data Bases
Proprietary
UHC, Premier,
HMO’s
Government
VAH, CMS
Specialty
organizations
NIH funded
Industry/FDA
Industry
registries
CDC, States
wwww.ClinicalTrials.gov
Difference between Clinical Data and
Administrative Data Bases
•
Clinical data (National Surgical Quality Improvement Program)
–
–
–
•
–
Prospective data collection, chart abstraction
Expensive, labor-intensive
Face validity among physicians
Administrative data base (UHC’s CDB, Premier, ThomsonReuters)
– Always retrospective, Claims data (medical record coding)
– Can study resource use and cost of care
– Very efficient way to collect data
I2B2 – Integrating Informatics for Biology and Bedside
– HERON (Healthcare Enterprise Repository Ontological
Narration) at KUMC
– Software program - integration of EPIC, Clinical information,
IDX of retrospective data
Where do the data elements come from?
Physician: Documentation of patient care
Coders: Assignment of codes to diagnoses and procedures
Creation of a ‘CLAIM’ with patient demographics; DRG;
diagnoses and procedures; LOS; charges;
admission/discharge dates, status; physician; etc.
Payers (e.g.
CMS, BCBS)
State
UHC Clinical
Data Base
(CDB)
Died
Survived
Risk Model
Low Risk
High Risk
A robust model should assign higher probability of death to patients who died
than to those who survived, at least 70% of the time (i.e. c-index >= 0.70)
UHC Risk Adjustment Overview 2008
Potentially avoidable
complications (not
input into the model)
Age
Gender
Race
Socioeconomic status
(Medicaid, self pay, charity, no
charge)
Admission status (emergency)
Transfer status, acute hospital,
nursing home
Palliative care
DRG-specific conditions
Ventilator on Day 1
Severity-of-illness class for
DRG based models
risk of mortality
Inputs
Up to 30 comorbid or
chronic conditions (e.g.
diabetes, liver disease,
obesity)
Separate regression
models for Cost, LOS,
mortality for each DRG
Expected mortality
Expected cost
Expected LOS
What Variables Are Studied
Almost anything having to do with an inpatient stay
(ambulatory variables currently in development)
 Risk Adjusted Outcomes –
Observed and Expected (O/E) for LOS, Mortality and Cost
 Complications, Readmissions, AHRQ Patient Safety Indicators
Performance based on:
Resource Utilization*:













Hospitals
Product Lines
DRGs & MS-DRGs
Diagnoses / Procedures
Physicians
Discharge Date/Month/Year
Patient Demographics
Blood Products
Drugs
Imaging Tests
ICU
Med/Surg Supplies
Pharmacy
* Resource Manager
If you want to use UHC database?
• Develop your proposal
• Contact : Chris Wittkopp – Organizational improvement
• Discuss your proposal and her assessment of data retrieval strengths
• Write short proposal with background, purpose, methods
• Submit proposal to Human Subject review
• If QI project can get exemption
Frontiers (CTSA)Biomedical Informatics
Goals
• Portal for investigators to access clinical and transitional
research resources, track usage, and provide
informatics consultative services
•
Create a platform, HERON, to integrate clinical and
biological data for translational research
•
Link biological tissues to data generated by research
cores
•
Leverage statewide telemedicine and Health
Information Exchange (HIE) to support community
based translational research
What is HERON?
HERON (Healthcare Enterprise Repository for Ontological
Narration) is a search discovery tool that allows you to search deidentified data from various hospital and medical center sources that
include but are not limited to Epic/O2 (the hospital electronic medical
record), IDX (the clinical billing system), KU Hospital Cancer Registry,
KU Biospecimen Repository, REDCap (selected projects), Social
Security Death Index, and University HealthSystem Consortium
(Quality Measure Data). By combining the various data sources,
researchers can look at the data in new ways not available when
viewing data in one source at a time.
Why should I use HERON?
HERON is a powerful tool that can save time during your research
process. Searching across multiple data resources allows you to
view data trends, key in on your research criteria, modify your search
requirements and see how the data changes. This is a good tool to
employ at the start a research project as it saves time by helping you
focus and define your research. The HERON tool also provides
analysis tools, such as the Timeline and the Cancer Survival Analysis
tools.
Larger projects where biostats sets up data sets, does
monitoring and auditing, ie funded RCT
Data management tool, each investigator
enters and monitors own data
Frontiers (CTSA)Biomedical Informatics
Goals
• 1) Portal for investigators to access clinical and
transitional research resources, track usage, and
provide informatics consultative services
•
2) Create a platform, HERON, to integrate clinical and
biological data for translational research
•
3) Link biological tissues to data generated by research
cores
•
4) Leverage statewide telemedicine and Health
Information Exchange (HIE) to support community
based translational research
Summary
• Data is essential for Scholarship, Quality
Improvement and Education
• Sources of data are multiple
•
•
•
•
Clinical
Administrative
Registry
Informatics for Integrating Biology with Bedside
– HERON
– Data Management Systems
– CRIS
– RedCap

similar documents