Presentation

IARS 2017 Annual Meeting and International Science Symposium
Improving Health Through Discovery and Education
May 6-9, 2017 • Washington, DC
Clinical Decision Support Systems
in the Perioperative Setting
Bala G. Nair, PhD
Associate Professor & Director, PPiQSO
Anesthesiology and Pain Medicine
University of Washington, Seattle, WA
IARS 2017 Annual Meeting and International Science Symposium
Improving Health Through Discovery and Education
May 6-9, 2017 • Washington, DC
DISCLOSURES
Coulter Foundation Grant
Dexcom, Inc
Clinical Decision Support (CDS)
• Computer System
• Assimilates & processes EMR information
• Assists clinician in decision making towards:

Safer and optimal clinical care

Accurate documentation  improved billing & compliance

Better efficiency – cost & waste reduction
(EMR – Electronic Medical Record)
Perioperative setting
40-50% hospital revenue
Expensive care (~$50/min)
High risk
Major complication rate
= 3-17% (Half preventable)
> 5000 hospitals
30 million surgeries
Litigations
Major lawsuits: 150/y
Payment: ~$350,000
Waste
20-33 % of all hospital
wastes
• Complex, critical & dynamic environment
• Multi-sourced information flow
• Rapid shifts in patient conditions and information
Types of CDS
Passive
Active
User initiated
Automatic
• “Hard stops”
• Guideline documents
Non real-time
Real-time
• Post-hoc reports
• Preoperative summary
• Real-time alerts
• Real-time guidance
CDS Architecture
Rules Builder
Decision Rules
Data Acquisition
Module
Data processing
Module
PIMS / EMR
Database
Rules Processing
Module
Notification
Module
Message &
Response Logs
PIMS Workstation
Voice
Report
Phone
Pager
PIMS - Perioperative Information Management System
EMR – Electronic Medical Record
Visual
display of alerts
Integrated and add on
Advantages
Integrated  Self-contained
 Immediate access to data
 Seamless
Standalone
Disadvantages
 Lack of specialization and focus
 Simplistic, inflexible solutions
 Slower pace of enhancements
 Generally superior features  Indirect access to PIMS data &
 More customized to meet
data latency
perioperative needs
 Updates to the EMR could
 Generally faster pace of
cause data interface failures
enhancements
 Requires more maintenance
Examples
Integrated CDS – Advanced Clinical Guidance – Talis Clinical LLC
Examples
Add On CDS – Smart Anesthesia Manager with Cerner EMR
Examples
Add On CDS – AlertWatch
Benefits
Target item
Passive
Active:
Post-hoc
CDS description
Billing/compliance
 Required fields or hard stops
Clinical protocols
 Links to clinical protocols
Quality measure
 Email report on failed antibiotic
administration
Billing/compliance
 Individualized emails on
noncompliant documentation
Cost savings
 Individualized feedback on gas
flows
Benefits
Active:
Near
real-time
Target item
Quality
measures
CDS description
 Timely antibiotic initial dose and redoses
 Perioperative β-blocker administration
Billing/
Compliance
 Documentation of invasive lines
 Data elements required for billing
Cost savings
 Feedback on reducing fresh gas flow
Patient
Monitoring /
Anesthesia
management
 Gaps in non-invasive blood pressure
monitoring
 Hypotension or hypertension
 Activate patient monitor alarms after CPB
 Tidal volume for patients at risk of acute
lung injury
Clinical
protocols
 Glycemic management guidelines
 PONV risk and antiemetic drug therapy
 Traumatic Brain Injury
CPB - Cardio Pulmonary Bypass
PONV – Post-Operative Nausea & Vomiting
Hurdles
Access to data
•
Custom queries, FHIR, HL7
Data latency
•
Delayed alerts, False alerts
Interoperability
•
Domain restrictions
Operational impact
•
Production operation, User interaction
Considerations
Provider buy in, education
•
•
Educate and obtain provider buy in
Solicit feedback, refine rules
Workflow compatibility
•
•
Impact on workflow
Ability to perform CDS recommendations
Timing & Content
•
•
When & how to alert?
Feedback message content
Alert fatigue
•
•
Repeat alerts?
How frequently?
Future directions
Intraoperative
Current
Perioperative
Next
Future
Reactive
Predictive
Prescriptive
Suboptimal
(Machine learning,
predictive analytics)
(Precision medicine)
Manual
Optimal
Automatic
(Optimization techniques)
(Closed-loop systems)
Questions ?