learning curves in medical education
DESCRIPTION
Dr. Martin Pusic (Columbia University) speaks about Learning Curves and Degradation of knowledge as it applies to medical education through his studies on medical education assessments and educational technology.TRANSCRIPT
Learning Curves for Medical Education
Martin V. Pusic MD PhD
Division of Pediatric Emergency MedicineChildren’s Hospital of NY-Presbyterian
Columbia University
Acknowledgements
• Kathy Boutis MD
• Martin Pecaric PhD
• Royal College Medical Education Grants
Acknowledgements
Outline
• Learning Curve General Principles
• Medical Example of Learning Curves
• Implications
Ericsson 1993
What is the Atomic Unit of
Emergency Medicine Practice?
The Atomic Unit of Emergency Medicine
DECISION
The Atomic Unit of Emergency Medicine
DO TEST
DON’T
The Atomic Unit of Emergency Medicine
TREAT
DON’TTREAT
The Atomic Unit of Emergency Medicine
ADMIT
DISCHARGE
The Atomic Unit of Emergency Medicine
ADMIT
ADMIT
ECG
Radiograph
Ankle Radiology Example
• PERC collaborative project
• Div of Pediatric Emergency Medicine– University of Toronto– Columbia University
• Image Bank of 234 Ankle X-ray Cases– Others in Development
Screen shot of application
Item Bank
Case 1
Item Bank
Case 1 Case2
Item Bank
. . . Case 1 Case2 Case3 Case4 Case 5 Case xx
Advantages of Image Bank
• Fidelity
• Repetitive Deliberate Practice
Dilemma #1
How do we measure the learning?
2 x 2 Table
Significant Pathology
+
No Pathology
--
User Says + a b a+b
User Says -- c d c+d
a+c b+d a+b+c+d
Ankle Study
• Recruited broad spectrum of subjects:
• All responses logged in a MySQL database
Medical Students
20
Residents 6
Fellows 12
PEM 5
Radiologists 3
Results
• Five institutions participated
• Took approximately 4 hours to complete all 234
• Overall Cohen’s Effect Size for the intervention was 0.55 (95% CI: -0.4, 1.5)
Five Levels of Expertise-.
50
.51
1.5
2d
Prim
e
MS RES FELLOW PEM RAD
Dilemma #2
How do we detect the learning that is happening along the way?
Five Levels of Expertise-.
50
.51
1.5
2d
Prim
e
MS RES FELLOW PEM RAD
Rater 99 LR+ and LR-
0.00
0.50
1.00
1.50
2.00
2.50
3.00
1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235
Case Number
LR
Est
imat
e
LR-- LR+
Rater 101 LR+ and LR-
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235
Case Number
LR
Est
imat
e
LR-- LR+
Rater 105 LR+ and LR-
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235
Case Number
LR
Est
imat
e
LR-- LR+
Rater 107 LR+ and LR-
0.00
0.50
1.00
1.50
2.00
2.50
3.00
1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235
Case Number
LR
Est
imat
e
LR-- LR+
LR+ and LR-
0.00
1.00
2.00
3.00
4.00
5.00
6.00
1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235
Case Number
LR
Est
imat
e
LR-- LR+
Rater 84
Rater 107 LR+ and LR-
0.00
0.50
1.00
1.50
2.00
2.50
3.00
1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235
Case Number
LR
Est
imat
e
LR-- LR+
Five Levels of Expertise-.
50
.51
1.5
2d
Prim
e
MS RES FELLOW PEM RAD
0.1
.2.3
.4.5
.6.7
.8.9
1(m
ean)
dP
rim
e
0 50 100 150 200 250Sequence
MS RES
FELLOW PEM
RAD
dPrime
Summary
• Roughly 2/3rd follow the Thurstone Pattern
• Not all curves same form:– Medical Students Shallow Curves– Residents and Fellows Steepest– Attendings Plateau
• Overall Education Management
• Individual Feedback
Ericsson 1993
Distributed Practice
Ability to interpret heart sounds over Ability to interpret heart sounds over time (Butterworth 1960)time (Butterworth 1960)
Outline
• Learning Curve General Principles• Useful Parameters
• Medical Example of Learning Curves• Examples abound• Group vs. Individual Considerations
• Implications• Trajectories more accurate representation