Institute of Medicine Committee on Patient Safety Data Standards Meeting

1
Institute of Medicine Committee on Patient Safety
Data Standards Meeting

• James P. Bagian, MD, PE
• John Gosbee, MD, MS
• VA National Center for Patient Safety
• 23SEP02

2
Definitions

• Adverse Medical Event (IOM)
• A deviation from best practice in healthcare
  delivery that leads to unwanted harm to a patient
  or the mission of the organization
• Adverse Event (VA NCPS)
• Untoward incidents, therapeutic misadventures,
  iatrogenic injuries, or other adverse occurrences
  directly associated with care or services
  provided within the jurisdiction of a medical
  center, outpatient clinic or other VHA facility
  that result in harm to the patient.

3
Definitions

• Near Miss (IOM)
• A deviation from best practice in healthcare
  delivery that would have lead to unwanted harm to
  patient or to the mission of the organization,
  but was prevented through planned or unplanned
  actions.
• Close Call (VA NCPS)
• A Close Call is an event or situation that could
  have resulted in an Adverse Event but did not,
  either by chance or through timely intervention

4
Definitions - Issues

• Best Practices
• Says who and which ones?
• Condition Specific
• One Trick Ponies
• Wash Your Hands
• Not Generalizable or Systems Oriented

5
What Does (Self)Reporting Yield?

• Trends
• Of What?
• Only Tells You
• What is Recognized
• What Is Reported
• Not Necessarily What Is Important!
• Example MR Only 1 Case of gt3000
• Actually Omnipresent Gun Discharge, Floor
  Buffer, Nurses Pen, etc.
• MR Safe vs. Compatible

6
Process Overview Field NCPS

• Data Classification and Analysis
• Goal Is To Prevent Harm To The Patient
• Change Happens Locally
• Validate and Investigate For Widespread Use
• Pseudo Trends Can Point To Need For RCA

• Reports of Adverse Events Close Calls
• Prioritize SAC Score
• Safety Reports
• RCAs
• Use of Triage Cards

7
Data Collection
8
Categorize this Location, process involved and
Type of Event

• Patient is inadvertently given 5X overdose of
  Benzodiazepenes in 23 hr stay unit for
  parasuicidal gesture
• Is confused and decides to elope
• Falls on way to eloping in vestibule between room
  and hallway

9
Major influences

• 1998 VA Patient Safety Advisory Committee
• Narrative, narrative, narrative
• Avoid boxing people in
• James Farrier (aviation safety database expert)
• Narrative is key
• Premature categorization cheapens, hurts reports
• Even experts can not agree on agreed upon terms
• Chris Johnson (Univ. of Glasgow Accident Analysis
  Group)
• Most databases serve researchers and policy
  people
• Not designers, builders, operations poeple

10
Other Considerations

• Many categories sound logical, easy, fast,
• In real-life application, they are not
• NCPS cant use taxonomies that contradict major
  policies and philosophies
• Violation of policy is not a root cause
• Title of person involved with the event is not
  generally useful and potentially harmful
• If category does not inform us on a solution, it
  it is not useful

11
Root Cause/Contributing Factor is Categorized by
RCA Teams

• See Triage Cards
• Six major headings ( of keywords)
• Communication (14)
• Training (8)
• Fatigue/Scheduling (8)
• Environment/Equipment (23)
• Policies/Procedures (13)
• Barriers (12)

12
Five Categories Done at NCPS

• Location (49)
• Some nested
• Major and minor
• Event Outcome (8) (e.g., fall, suicide, other)
• Activity or Process (24)
• Actions (32)
• Outcome Measures (11)

13
Special Analysis and Classifications

• Completed and online (see www.patientsafety.gov)
• MRI hazards
• Oxygen Cylinders (see web site)
• Used to Develop Policy
• Patient Misidentification
• Wrong Site Surgery
• In Progress
• Suicide
• Elopement/wandering
• Wrong Tube, Wrong Hole, Wrong Connector
• Retained Sponges

14
Comparisons with other systems (examples)

• MedMARX
• We do not use word error
• Caregiver involved category is a slippery slope
  in wrong direction (blame cannot be extricated)
• Some perpetuating factors they list are not
  causal
• Eindhoven Classification
• Borrows from shifting cognitive theories (e.g.,
  Klein)
• Very hard to train non-human factors people
• Violations of rules and regulations is a second
  level category deviates from rule of causation

15
Natural Language Processing

• Early stages of scoping this work
• Synonyms for our keywords are many, and some hard
  to see in a sea of text
• As conceptual understanding changes, manual
  re-categorization unlikely
• It may lead to learning system that finds
  trends we could not across thousands of RCAs