Occam: A Simulation Testbed for Studying Medical Organizations

1
Occam A Simulation Testbed for Studying Medical
Organizations

• Douglas B. Fridsma, M.D.
• Palo Alto VA Medical Informatics Fellow
• Stanford Medical Informatics
• Stanford University

2
Why study organizations?
3
(No Transcript)
4
(No Transcript)
5
Medical Care is Similar to Other High-risk
Activities

• Significant effort in medicine to understand,
  study, and evaluate work processes
• clinical trial protocols, guidelines, pathways
• Similarity to other high-risk maintenance tasks
• complex, highly interdependent system that
  requires on-going repair and monitoring
• cost of errors are high
• demand for high quality, cost-effective care
• coordination and information critical to success

6
Project Check-Up Subsea Oil Production Modules

• Organizational modeling and analysis helps to
  shrink design time back to 15 months, while
  maintaining required quality.
• Analysis
• Maintain functional organization structure to
  ensure technical quality
• Increase skill levels for five key team members
• Decentralize decision making for the 25 person
  structural design team
• Invest in CAD file sharing capability to speed up
  information exchange

7
Interaction between Product, Process
Organization
8
The Problem of Predicting Organization Behavior

• Organizations are complex
• Complex interactions among agents in organization
  
• Difficult to predict the effect of
• New technology
• Interacting agents
• Interacting activities
• Clinical work processes
• One time solutions are difficult to generalize
• Experimenting with a real organization can be
  expensive

9
The Goal of Occam

• To develop
• A theoretically sound framework
• Modeling language
• Simulation tools
• In which organizational questions can be
• Queried
• Tested with simulation tools
• Used to
• guide systematic organization and protocol design
• customize existing clinical protocols
• identify potential process errors
• improve cost-effectiveness

10
An Example The General Internal Medicine Clinic

• Clinical Problems
• Inconsistent use of protocols and guidelines
• Long wait times for an appointment
• Nurse Practitioners
• High utilization of physician consultation time
• Low productivity compared to peer institutions
• Physicians
• Difficulty retaining faculty because clinical
  responsibilities interfered with research time
• Low productivity compared to peer institutions

11
Organizational Questions to Consider

• How can we best integrate protocol-based care and
  best practice into our organization?
• What is the best use of our local expertise?
• Use of Nurse practitioners? Physicians?
• What policies would improve coordination and
  patient-care quality in our institution?
• Patient scheduling Continuity versus
  first-available?
• Nurse practitioner versus physician?
• How would new technology affect organizational
  efficiency?
• Electronic medical records
• Decision-support systems

12
(No Transcript)
13
(No Transcript)
14
Development of Occam
15
Organization Theory
16
What Organizational Theory Applies?

• Taylor and Scientific Management
• Mechanistic approach to interaction
• Best-practice and CQI (Purves, Pitty, 1995)
• Mayo and Human Relations
• Emphasis on cognitive processes and motivations
• Professional mistrust of best practices
  (Mintzberg, 1983)
• Galbraith and Information-processing Theory
• Organization structure contingent on external
  pressures
• Flow of information within and between
  organizations
• Organizational Design An Information-processing
  view, 1974.
• Simon and Bounded-rationality (1976)
• administrative man rather than economic man
• satisfice rather than optimize

17
What Organizational Theory Applies?

• Taylor and Scientific Management
• Mechanistic approach to interaction
• Best-practice and CQI (Purves, Pitty, 1995)
• Mayo and Human Relations
• Emphasis on cognitive processes and motivations
• Professional mistrust of best practices
  (Mintzberg, 1983)
• Galbraith and Information-processing Theory
• Organization structure contingent on external
  pressures
• Flow of information within and between
  organizations
• Organizational Design An Information-processing
  view, 1974.
• Simon and Bounded-rationality (1976)
• administrative man rather than economic man
• satisfice rather than optimize

18
Galbraiths Information-Processing Theory of
Organizations

• All work takes time
• Direct-work tasks
• Communication tasks
• Problem-solving tasks
• Decision-making tasks
• Organization hierarchy is a coordination and
  problem-resolution machine
• Actors in the organization are cognitively
  limited and linked by shared work

19
Elements of Theory to Model

• Agents
• Activities
• Communications
• Connections

20
Modeling and Representation
21
Information-Processing Model

• Agents
• Fill roles in an organization hierarchy
• Finite processing speed defined by agent skills
  and experience
• Activities
• Agents responsible-for activities
• Interdependencies between activities create
  communication needs
• Communication
• Generated by
• Information exchange
• Exceptions
• Exceptions resolved by organizational hierarchy

22
Components of Occam
Output

• Activity Delays
• Agent Backlogs
• Missed Communications
• Process Quality

Organization Hierarchy
Work Process
OPDL organization model
Monte Carlo Discrete Event Simulator
23
Traditional Representation of Information-Processi
ng Exceptions
MD
RN
Clinic Visit
Chemotherapy Treatment
Hospitalization
24
Inadequate Representation of Exceptions for
Medical Care Processes

• Traditional information-processing exceptions
• Static work processes
• Static agent-activity relationships
• Uniform exceptions
• Medical care work processes
• Dynamic work process
• New activity (hospitalization) added when an
  unexpected event (fever) occurs
• Dynamic relationships between agents and work
  processes
• Agent who generates exception ? agent who
  corrects exception
• Exceptions have different effects on the work
  process and organizations

25
A New Definition of Exception

• That part of the world which you chose to model,
  but not explicitly

26
Modeling is a Choice
ExplicitModel
Exception
Explicit Model
Exception
27
When is the Model Correct?
ExplicitModel
Exception

• Where you draw the line is determined by
• The definition of an exception
• The modeling stopping function
• Other factors?

28
Review of Clinical Work Processes

• Bone marrow transplantation clinical trial
  treatment protocol
• Protocol EST 2190 for breast cancer
• Dukes modified protocol for lung cancer
  treatment
• Breast cancer diagnosis algorithms
• Max Borten, "Gynecological Decision Making,
  1988.
• Urinary incontinence
• AHCPR guidelines
• Review of patient charts

29
BMT Clinical Trial Protocol
30
Exceptions in BMT protocols

• High-level overview of protocol describe routine
  activities of medical care
• Pages of exception-handling activities for
  unexpected events
• Exception-driven work processes
• Occam extends Galbraiths information-processing
  theory of organizations
• Focus on process of medical care
• Differential effect of exceptions on medical care
  process

31
Modeling Exceptions
32
New Process Exceptions in Occam

• Additions
• Treatment for nausea/vomiting
• Hospitalization for neutropenia
• Diagnosis of gastrointestinal pain
• Re-reading outside tests/xrays
• Coordination of records with
• Referring MD
• Radiation therapy
• Redo of venous access line
• Substitutions
• Follow-up visit in the hospital rather than in
  the BMT clinic

• Deletions
• Remove tamoxifen because of receptor status
• No use of acyclovir mentioned
• Re-Assignment
• Home care nurse rather than in-patient nurse
• Temporal Change
• Chemotherapy delay because of low WBC counts
• Bone marrow biopsy delayed

33
Addition
MD
RN
Clinic Visit
Chemotherapy Treatment
Hospitalization
34
Outline
35
Occam
Organization Hierarchy
Work Process
36
Occam
Organization Hierarchy
Work Process
37
Occam
Organization Hierarchy
Work Process
38
Occam
39
Outline
40
Directed Platelet Transfusion

• The Problem
• Directed platelet transfusion requires
  coordination across different clinic
• Delays postpone treatment
• Where are the places the process could be
  improved?
• What happens when the form is lost?
• What happens when the primary MD is gone?

41
(No Transcript)
42
(No Transcript)
43
(No Transcript)
44
(No Transcript)
45
(No Transcript)
46
(No Transcript)
47
Opportunities for Research

• Knowledge networks and virtual organizations
• Social Good Theory
• Networking Theory
• Transaction costs

• Quality Management
• error detection
• medical process re-engineering
• Testbed for novel healthcare delivery systems

• Systematic guideline development
• Process-oriented guideline modeling
• Exception modeling

• Docking with other cognitive and organizational
  models

48
Opportunities for Research

• Knowledge networks and virtual organizations
• Social Good Theory
• Networking Theory
• Transaction costs

• Quality Management
• error detection
• medical process re-engineering
• Testbed for novel healthcare delivery systems

• Systematic guideline development
• Process-oriented guideline modeling
• Exception modeling

• Docking with other cognitive and organizational
  models

49
1. Knowledge Networks

• Initial work expands the flexibility of the work
  process
• What of organizational flexibility?
• How does the internet and knowledge networks
  affect organizational performance?
• Relax the constraints on actor representation
• hierarchical to fully-connected
• Two NSF initiatives through IIS

50
2. Systematic Protocol Development

• Guidelines
• Facilitate decision-making
• Health-care provider focus

• Protocols
• Derived from guidelines
• Include both decision and process information
• Link guidelines and work processes

• Clinical work processes
• Trace the process of health care
• Health-care organization focus

51
Systematic Protocol Development

• Occam!

52
3. Bridging Micro-Theory and Macro-Theory
Sociology Organization Theory
53
4. A Tool to Identify Error-prone Work Processes

• "While we must still hold individuals responsible
  for high standards of performance, we now
  recognize that most errors result from faulty
  systems, not faulty people. To identify systems
  failures, we need to know about the errors they
  cause."
• -Lucian L. Leape, Boston Globe, 1999 Jan 12

54
Moving Quality Management Upstream in Health Care

• 80s Measure Product Quality
• Quality out-of-control by the time problems are
  detected
• Intervention JCAHO accreditation
• 90s Control Variance in Processes
• Predict quality problems as they arise
• Intervention Protocols and guidelines
• 2000 Design Work Process Organization
• Predict and manage quality problems before they
  arise
• Intervention Reconfigure work process and/or
  participants

55
A Testbed for Studying Medical Organizations

• Quality Management
• Error detection
• Medical process re-engineering
• Organization-specific studies of guideline
  implementation
• Healthcare policy research for populations
• Simulation for specific organizations
• Testbed for novel healthcare delivery systems
• Telemedicine
• New communication tools, electronic medical
  records

56
Occam A Simulation Testbed for Studying Medical
Organizations
57
Summary and Conclusions

• Linking guidelines to outcome measures requires
  understanding the process of medical care
• Organization theory provides tools to understand
  and model work processes, but requires practical
  and theoretic extensions to model medical
  problems
• Occam extends the information processing theory
  of organizations to allow modeling and simulation
  of medical organizations
• Occam is a theoretically-sound method in which to
  understand medical work and proactively design
  and implement new protocols or novel medical
  organizations

58
Types of Exceptions

• Knowledge Exceptions
• Changes the knowledge encoded within the model
• Data/knowledge base used for decision making
• Add criteria, add data about decision
• Process Exceptions
• Changes relationships between activities
• Successor/predecessor relationships
• New activities, new responsible agent, etc
• Activity Exceptions
• Changes activity characteristics
• Modify drug dose, without changing activity
  duration or responsible agent