Implementing Medical Algorithms to Reduce Medical Errors

1
Implementing Medical Algorithms to Reduce Medical
Errors

• Kathy A. Johnson, Ph.D.a, Gareth Kantor, M.D.b,
  John R Svirbely, M.D.c, MG Sriram, Ph.D.d, Jack
  W. Smith, M.D., Ph.D.a, Jorge Raul Rodriguez,
  M.D.e aUniversity of Texas - Houston, Houston,
  TXbCase Western Reserve University, Cleveland,
  OHcMcCullough-Hyde Memorial Hospital, Oxford,
  OHd Quanta Healthcare Solutions, San Francisco,
  CAeHospital Dr. Cosme Argerich, Buenos Aires,
  Argentina

2
Medical Error

• IOM definition of error
• the failure of a planned action to be completed
  as intended (i.e., error of execution) or the use
  of a wrong plan to achieve an aim (e.g., error of
  planning).
• IOM found 23 gaps including
• a need for more research in the area of
  communication and information sharing

3
Medical Algorithm

• Definition any computation, formula, survey, or
  look-up table that is useful in healthcare
• Purpose Improve the delivery of healthcare

4
Types of Medical Algorithms

• Coding look-up tables
• Comparison with normal population standards
• Data conversion
• Decision rules triaging
• Decision trees flow diagrams
• Diagnostic criteria
• Diaries symptom tracking
• Functional state description

• Grading and scaling
• Probability statistical analysis
• Prognostic scores
• Questionnaires
• Risk determination
• Simple classification
• Simple formulas
• Therapeutic indications contraindications

5
Algorithm Use Benefits

• Improved Performance and Efficiency
• Expanded Scope of Practice
• Quality Assurance
• Patient Self-Determination
• HOWEVER
• Often underutilized

6
Algorithm Representation Problems

• Lack of proper documentation
• Errors or oversights in representation
• Confusion about data units
• Confusion in expression of formulas
• Use of irrelevant or illogical data
• Uncertain transportability of data elements
• Difficulty in performing the requisite
  calculations without a calculator or reference
  table
• Inability to transport interpretations
• Unresponsiveness to change
• Uncertain purpose

7
Algorithm Use Problems

• Selecting the wrong algorithm for the patient,
  population or situation (I.e. using an algorithm
  irrelevant or unresponse to condition)
• Making an error in remembering the algorithm
• Using a version out of date or with an error
• Using an algorithm that is too simple
• Using an algorithm that is too complex
• Making an error during calculation or execution
  of the algorithm
• Using incorrect data
• Failure to question the output of the algorithm
• Inappropriate use of output

8
Automated Algorithm Benefits

• Reduced data entry errors
• Elimination of calculation errors
• Perfect recall of algorithm details
• Selection of algorithm with appropriate
  complexity
• Selection of the best algorithm for a population
  or situation
• Provide references to interpretation of the
  output
• Use of most up-to-date version
• Large Scale
• Integration with information systems
• Facilitation of evidence-based medicine
• Global resource (translation to other languages)
• Education and Research

9
Medical Software Development Constraints

• Legal Constraints
• Regulatory Constraints
• Security and Privacy Concerns
• Development Costs
• Validation of Algorithms

10
MEDAL

• MEDical Algorithms project
• http//www.medal.org
• Goal Provide a collection of algorithms in a
  format that supports clinicians, programmers, and
  validators
• Status Over 3900 algorithms in 45 chapters

11
MEDAL Algorithm Process

• Selection of algorithms
• Biomedical literature including journals and
  textbooks
• Documentation
• Abstracted from the medical literature
• Implementation as a spreadsheet
• Implementation in a software application

12
Software Application Representation
13
(No Transcript)
14
(No Transcript)
15
Excel Spreadsheet Implementation
16
(No Transcript)
17
Unit Conversions
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
Conclusions

• Medical algorithms are one important way to share
  medical information
• Automating medical algorithms can reduce errors
  in their application
• Automating medical algorithms can also facilitate
  validation
• More work to be done on automation
• More work to be done on expanding the set