thinking and error in emergency departments
DESCRIPTION
Errors in clinical decision making in the emergency department can be fatal! Through case studies, this presentation explores the factors contributing to error and strategies to overcome them.TRANSCRIPT
Thinking and error
Anne-Maree Kelly
December 5, 2012
Permissions This presentation may be reproduced in full or
in part on the condition that each slide used carries the following:
‘Reproduced with the permission of Professor Anne-Maree Kelly, Joseph Epstein Centre for Emergency Medicine Research @Western Health, Melbourne, Australia’.
@kellyam_jec
Diagnostic error Common
A top cause of medicolegal actions Up to 20% of autopsies
Emergency Medicine is a high risk environment Why?
Diagnostic error “No fault’ error
Silent or atypical disease Mimics something more common Lack of patient co-operation or presentation of symptoms Limitation of medical knowledge
Systems error Technical: test error, lack of test/resource Organisational: supervision, unavailability of expertise,
inefficient processes, cultural issues Cognitive error
Faulty data collection or interpretation Flawed reasoning Incomplete knowledge
Cognitive error: My questions
Is it predictable?
Is it unavoidable?
Your experience
Work with the two or three people around you.
Can you identify a case that you were involved in or heard about where thinking processes contributed to a diagnostic error?
Cases from the medicolegal world Ms X aged 42 Single mother of 4 Abdominal pain and vomiting Exam difficult due to obesity Three presentations to ED of a tertiary referral
hospital over 5 days Diagnosis #1: gastro (xray and bloods performed) Diagnosis #2: gastro; no further tests Diagnosis #3: bowel obstruction, ARF, gangrenous
gut Outcome: death
The issues #1:
Assessment was reasonable Xray was performed and clearly showed small
bowel obstruction #2:
Assessment was brief Assumed that previous diagnosis was correct Did not check results/ xray
The cost 4 children under 15 without a mother (or
interested father)
> 1 million dollars settlement
Case 2 HG aged 3 Rural setting, experienced mum 24 hours of D&V Seen by GP 1: gastro, home for fluids Seen by GP 2 next day: gastro; home for fluids Presented in ED in next town: gastro; home for fluids Day 3 admitted to small rural hospital by GP registrar for
oral fluids. Mother concerned re lack of urine. Day 4 evident that there had been no urine output for ~24
hours IV therapy Transferred to large hospital on Day 5 Cardiac arrest, died
The issues First GP assessment fine Second GP assessment
Failure of data collection: weight Assumption that all gastro settles with oral fluids Not listening to mum re intake / output
ED assessment Failure of data collection: weight Assumption that all gastro settles with oral fluids Not listening to mum re intake / output
In hospital management Failure of data collection: weight, fluid balance chart, frequent obs Assumption that all gastro settles with oral fluids Not listening to mum re intake / output ‘It will all be alright’ mentality blocking escalation of care to
specialist centre
Something closer to home 30-something woman Sore throat 24-48 hours Unable to swallow saliva Epiglottis suspected by ED team Difficulty engaging ENT team Eventually came and attempted endoscopic exam Acute hypoxia Surgical airway Alive...but close run thing
Issues Epiglottis now very uncommon
Dismissed the likelihood despite reasonable evidence
Failure to respect assessment of clinician actually seeing the patient
Types of cognitive predispositions to respond
Type of CDR Explanation
Aggregate bias Failure to believe aggregate data, eg guidelines‘My patient is different’
Anchoring Locking on to features in presentation too early and failing to adjust with further data
Ascertainment bias Thinking shaped by prior expectation eg gender bias, stereotyping
Availability Diagnosis is more likely if it readily comes to mind
Base-rate neglect Tendency to ignore the true prevalence of a disease, impacts Bayesian thinking
Commission bias Belief that harm can only be prevented by action; tendency to action rather than inaction
Confirmation bias The tendency to look for confirming evidence rather than evidence to refute
Diagnosis momentum Diagnoses are like sticky labels; once attached hard to remove
Types of cognitive predispositions to respond
Type of CDR Explanation
Feedback sanction Error not temporarily associated with immediate consequences
Framing effect The way we see things is influenced by how they are presented to us (gastro and positive stool for blood story)
Fundamental attribution error Blame patients for illness rather than look at situational factors
Gambler’s fallacy Pretest probability of a particular diagnosis influenced by previous but independent events (eg coin toss example)
Gender bias False belief of difference in probability of a diagnosis between genders
Hindsight bias Knowing what happened influences the perception of past events and inhibits realistic appraisal of why error occurred
Types of cognitive predispositions to respond
Type of CDR Explanation
Multiple alternatives bias Multiple options cause uncertainty; tendency to limit options to those we know and potentially ignore rarer alternatives
Omission bias Tendency towards inaction, usually for fear of doing harm
Order effects Information transfer is U shaped; we ‘hear’ better at the beginning and end. May miss important stuff in the middle
Outcome bias The tendency to opt for diagnoses with good outcomes
Over-confidence bias Belief that we know more than we do!
Playing the odds In ambiguous situations, a tendency to opt for the less serious diagnosis
Posterior probability error The tendency to be unduly influenced by what has gone on before (see case 1)
Types of cognitive predispositions to respond
Type of CDR Explanation
Premature closure Very powerful: Tendency to accept a diagnosis before it is fully verified
Psych-out error Tendency for error in psych patients especially missing of serious medical issues
Representativeness restraint Looks like a duck, quacks like a duck, is a duckMissing atypical presentations
Search satisfying Inappropriately calling off the search once something has been found
Sunk costs The more we ‘invest’ in a diagnosis, the less likely we are to relese it
Sutton’s slip Going for the obvious
Triage cueing Triage assignment falsely prompts bias towards serious/ non-serious illness
Unpacking principle Failure to elicit all relevant information
Types of cognitive predispositions to respond
Type of CDR Explanation
Vertical line failure Thinking in silos; inflexible thinking; failure to consider what else might this be?
Visceral bias Visceral arousal is associated with poor decisionsFeelings towards patients (positive and negative) may result in diagnoses being missed
Our pre-disposition to CDR depends on:•Personality•Experience•Self-awareness•Environment/ situation
Avoiding cognitive error: Exercise Group 1:
In pairs or threes, describe strategies that might help clinicians avoid CDR in patients they manage
Group 2:
In pairs or threes, describe strategies that might help supervising clinicians/ consultants avoid CDR in cases they are consulted about
Cognitive de-biasing strategies
Strategy Mechanism/ Action
Develop insight/ awareness Talking about and analysing diagnostic errorsSharing experience
Consider alternatives Establish processes that ‘force’ consideration of other diagnosesRoutinely asking What else might this be?Documenting why you consider something unlikely and why
Develop reflective approach to problem solving
Regularly ask yourself how you are thinking about diagnostic problems and how you might do it better
Decrease reliance on memory System level: cognitive aids, guidelines, etcPersonal level: Don’t rely on memory. Look things up!
Specific training In CDRIn probability theory and Bayesian thinking
Simulation Both as case discussion and in simulator training
Cognitive de-biasing strategies
Strategy Mechanism/ Action
Cognitive forcing strategies Develop specific strategies for particular high risk situations eg medical clearance of psychiatric patients
Make it easier More information readily available
Minimise time pressures More time to think usually means better decisions
Accountability Clear accountability and followup of decisions made
Feedback Rapid and reliable feedback esp. re diagnostic error or ‘good picks’ assists diagnostic ‘calibration’
Teamwork Two heads are better than one. Information sharing/consultation with other team members eg nurses, other doctors, allied health etc.
Summary Diagnostic error and how we think are
intimated associated
Cognitive errors can be reduced by: System measures to promote information
availability and ‘force’ consideration of high risk groups/ diagnoses
Personal measures such as self-awareness, de-biasing strategies
Training Teamwork