Diagnostic Reasoning I and II
-
Upload
openmichigan -
Category
Health & Medicine
-
view
1.143 -
download
2
Transcript of Diagnostic Reasoning I and II
![Page 1: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/1.jpg)
Author(s): Rajesh Mangrulkar, M.D., 2011
License: Unless otherwise noted, this material is made available under the terms of the Creative Commons Attribution–Non-commercial–Share Alike 3.0 License: http://creativecommons.org/licenses/by-nc-sa/3.0/
We have reviewed this material in accordance with U.S. Copyright Law and have tried to maximize your ability to use, share, and adapt it. The citation key on the following slide provides information about how you may share and adapt this material.
Copyright holders of content included in this material should contact [email protected] with any questions, corrections, or clarification regarding the use of content.
For more information about how to cite these materials visit http://open.umich.edu/education/about/terms-of-use.
Any medical information in this material is intended to inform and educate and is not a tool for self-diagnosis or a replacement for medical evaluation, advice, diagnosis or treatment by a healthcare professional. Please speak to your physician if you have questions about your medical condition.
Viewer discretion is advised: Some medical content is graphic and may not be suitable for all viewers.
![Page 2: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/2.jpg)
Attribution Keyfor more information see: http://open.umich.edu/wiki/AttributionPolicy
Use + Share + Adapt
Make Your Own Assessment
Creative Commons – Attribution License
Creative Commons – Attribution Share Alike License
Creative Commons – Attribution Noncommercial License
Creative Commons – Attribution Noncommercial Share Alike License
GNU – Free Documentation License
Creative Commons – Zero Waiver
Public Domain – Ineligible: Works that are ineligible for copyright protection in the U.S. (17 USC § 102(b)) *laws in your jurisdiction may differ
Public Domain – Expired: Works that are no longer protected due to an expired copyright term.
Public Domain – Government: Works that are produced by the U.S. Government. (17 USC § 105)
Public Domain – Self Dedicated: Works that a copyright holder has dedicated to the public domain.
Fair Use: Use of works that is determined to be Fair consistent with the U.S. Copyright Act. (17 USC § 107) *laws in your jurisdiction may differ
Our determination DOES NOT mean that all uses of this 3rd-party content are Fair Uses and we DO NOT guarantee that your use of the content is Fair.
To use this content you should do your own independent analysis to determine whether or not your use will be Fair.
{ Content the copyright holder, author, or law permits you to use, share and adapt. }
{ Content Open.Michigan believes can be used, shared, and adapted because it is ineligible for copyright. }
{ Content Open.Michigan has used under a Fair Use determination. }
![Page 3: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/3.jpg)
Patients and PopulationsMedical Decision-Making: Diagnostic
Reasoning I and II
Rajesh S. Mangrulkar, M.D.University of Michigan
Department of Internal MedicineDivision of General Medicine
Fall 2011
![Page 4: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/4.jpg)
Industry Relationship Disclosures
Industry Supported Research and Outside Relationships
• None
![Page 5: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/5.jpg)
Ask
Acquire
Appraise
Apply
Ask
Acquire
Thread 1: Information Retrieval
![Page 6: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/6.jpg)
PICO: A Tool to Structure the Foreground Question
TherapyTherapy DiagnosisDiagnosis
P Patient Pop Disease
I Intervention Test
C Comparison Gold Standard
O Outcome Accuracy
![Page 7: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/7.jpg)
Foreground Questions - Case
Using the PICO model, jot down 1 foreground question with your partner that will help you care for this patient:
A 42 year old woman comes to her primary care practitioner’s office for follow up of her diabetes. She is currently on glyburide 10 mg twice daily. However, her blood sugars still stay elevated. After you see this patient, your attending asks whether you think she should add metformin to her regimen.
Patient - Intervention - Comparison - Outcome
![Page 8: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/8.jpg)
Foreground Questions - Therapy
• In type II diabetics, is metformin and glyburide better than glyburide alone at lowering blood sugar?
• Among women with type II diabetes, are there more instances of low blood sugar events in patients on both metformin and glyburide, compared to glyburide alone?
![Page 9: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/9.jpg)
Sources for Foreground Questions
• MEDLINE
• Practice Guidelines
• Evidence Based-Databases– Cochrane Library– ACP Journal Club
![Page 10: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/10.jpg)
Ask
Acquire
Appraise
Apply
Appraise
ApplyThread 3: Diagnostic Reasoning
![Page 11: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/11.jpg)
Initial Diagnostic ReasoningThe Odyssey Reloaded
• Failure to entertain all possibilities
• Failure to pay attention to all symptoms
• Failure to inform customer
• Failure to perform diagnostic tests
• Entertain all important possibilities
• Elicit and pay attention to description of all symptoms
• Inform and involve patients
• Perform effective diagnostic tests
The Mechanic The Clinician
![Page 12: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/12.jpg)
The Odyssey: Conclusion
Initial Possibilities#1: Trunk latch defect (recall pending)#2: Ajar sensing defect on side door#3: Side door not closing properly
50 Prime, flickr
![Page 13: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/13.jpg)
The Odyssey: Conclusion
Initial Possibilities#1: Trunk latch defect (recall pending)#2: Ajar sensing defect on side door#3: Side door not closing properly
The Answer
#2: Ajar sensing defect on side door
50 Prime, flickr
![Page 14: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/14.jpg)
Learning Objectives
By the end of this lecture, you will…•demonstrate diagnostic question formulation•define and calculate sensitivity, specificity, and predictive values for diagnostic tests•explain how risk factors drive prior probabilities, and how this concept relates to prevalence•modify probabilities from test results through 2x2 table calculations, Bayesian reasoning, and Likelihood Ratios
![Page 15: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/15.jpg)
Case: Diagnostic Reasoning
• The case: A 60 year old man without heart disease presents with sudden onset of shortness of breath.
• Description of the problem: Yesterday, after flying in from California the day before, the patient awoke at 3AM with sudden shortness of breath. His breathing is not worsened while lying down.
![Page 16: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/16.jpg)
Diagnostic Reasoning: Your Intake
• Q: “What other symptoms were you feeling at the time?”
• A: He has had no chest pain, no leg pain, no swelling. He just returned yesterday from a long plane ride. He has no history of this problem before. He takes an aspirin every day. He smokes a pack of cigarettes a day.
![Page 17: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/17.jpg)
Diagnostic Reasoning: First Steps
The differential diagnosis
Basic Tasks:• Assign likelihoods to each possibility
– E.g. P(X) = probability that “X” is the cause of the patient’s symptoms
• Place the possibilities in descending order of likelihood
• State why (rationale)
![Page 18: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/18.jpg)
My list
My differential diagnosis– Pulmonary embolism – Congestive heart failure– Emphysema exacerbation– Asthma exacerbation
![Page 19: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/19.jpg)
Probabilities
(1) PE P(PE) = 40%
(2) CHF P(CHF) = 30%
(3) Emphysema P(emphysema) = 20%
(4) Asthma P(asthma) = 10%• What is the probability that the shortness of
breath is due to either PE or CHF? 70%*
*provided that both do not happen simultaneously (i.e., they are “mutually exclusive”). If there is a 10% chance that 2 events may happen, then this number is 60% (make sure you understand why).
![Page 20: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/20.jpg)
Prior Probabilities
• Based on many factors:– Clinician experience– Patient demographics– Characteristics of the patient presentations
(history and physical exam)– Previous testing– Basic science knowledge
• Quite variable but can be standardized– Clinical Prediction Rules– http://medcalc3000.com/PulmonaryEmbRiskPisa.htm
![Page 21: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/21.jpg)
More information
• Family history: he has had a DVT in the past (age 40)• Physical Exam:
– His blood oxygen saturation is normal on room air– His respiratory rate is 16, but his pulse rate is 105
beats per minute– Examination of his lungs reveals some crackles and
wheezes, but no pleural rub or evidence of consolidation.
– Swollen right leg, with firm vein below the knee• CXR: normal• EKG: sinus tachycardia
http://medcalc3000.com/PulmonaryEmbRiskPisa.htm
![Page 22: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/22.jpg)
Diagnostic Reasoning: Testing
• If a Test existed that could “rule in” PE as the diagnosis with 100% certainty:
then P(PE | Test+) = 100%
• Two questions:– What is this test called?– Does P(CHF | Test+) = 0%?
Gold StandardNo
![Page 23: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/23.jpg)
Diagnostic Testing
• Facilitates the modification of probabilities.
• Can include any/all of the following:– Further history taking– Physical Examination maneuver– Simple testing (laboratory analysis,
radiographs)– Complex technology (stress testing,
angiography, CT/MRI, nuclear scans)$$$
![Page 24: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/24.jpg)
P
I
C
O
PICO: The Anatomy of a Diagnostic Foreground Question
• Patient: define the clinical condition or disease clearly.
• Intervention: define the diagnostic test clearly
• Comparison group: define the accepted gold standard diagnostic test to compare the results against.
• Outcomes of interest: the outcomes of interest are the properties of the test itself (e.g., performance and others we’ll discuss).
D
T
G
P
![Page 25: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/25.jpg)
Practice PICO
Case: A 60 year old man without heart disease presents with sudden onset of shortness of breath. Considering PE.
Diagnostic Test to consider: Ventilation / Perfusion Scanning
Gold standard: Pulmonary angiography
Need: Diagnostic performance
![Page 26: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/26.jpg)
Practice PICO
Case: A 60 year old man without heart disease presents with sudden onset of shortness of breath. Considering PE.
Diagnostic Test to consider: Ventilation / Perfusion Scanning
Gold standard: Pulmonary angiography
Need: Diagnostic performance
PP
![Page 27: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/27.jpg)
Practice PICO
Case: A 60 year old man without heart disease presents with sudden onset of shortness of breath. Considering PE.
Diagnostic Test to consider: Ventilation / Perfusion Scanning
Gold standard: Pulmonary angiography
Need: Diagnostic performance
PP
II
![Page 28: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/28.jpg)
Practice PICO
Case: A 60 year old man without heart disease presents with sudden onset of shortness of breath. Considering PE.
Diagnostic Test to consider: Ventilation / Perfusion Scanning
Gold standard: Pulmonary angiography
Need: Diagnostic performance
PP
IICC
![Page 29: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/29.jpg)
Practice PICO
Case: A 60 year old man without heart disease presents with sudden onset of shortness of breath. Considering PE.
Diagnostic Test to consider: Ventilation / Perfusion Scanning
Gold standard: Pulmonary angiography
Need: Diagnostic performance
PP
IICC
OO
![Page 30: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/30.jpg)
Can the test be used?Step 1 - Accuracy and Precision
• Accuracy - The result of the test corresponds consistently with the true result.– The test yields the correct value
• Precision - The measure of the test’s reproducibility when repeated on the same sample.– The test yields the same value
![Page 31: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/31.jpg)
Accuracy vs. Precision
![Page 32: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/32.jpg)
Accuracy vs. Precision
Proceed to Step 2
![Page 33: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/33.jpg)
Accuracy vs. Precision
Proceed to Step 2 Calibrate Equipment
![Page 34: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/34.jpg)
Accuracy vs. Precision
Proceed to Step 2 Calibrate Equipment Start Over
![Page 35: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/35.jpg)
Can the test be used?Step 2 - Diagnostic Performance
1. A well-defined group of people being evaluated for a condition undergo:
- an experimental test, and - the gold standard test.
2. Comparison is made between the result of the new test and that of the gold standard.
![Page 36: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/36.jpg)
Diagnostic Performance: Statistical Significance
• Statistical significance: strength of the association between…– Diagnostic study results (for the diagnosis of a
particular disease)
– Gold standard results (for the diagnosis of the same disease, in the same population)
• Strength = degree of correlation
![Page 37: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/37.jpg)
Diagnostic Performance: Clinical Significance
• Clinical significance: how likely is the diagnostic test going to affect patient care?– Magnitude of the association between test results and
the accepted gold standard
– Other literature (including those of the gold standard)
– Cost of the test, reproducibility of test
– Disease characteristics (will the test result affect management of the disease?)
![Page 38: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/38.jpg)
Diagnostic performance is an association between test result and diagnosis of a condition (as assessed by the gold standard)
A B
C D
Disease + Disease -
Test +
Test -
What are the results - Diagnosis
BONUSWhat type of
variable is disease state?
TP
FN TN
FP
![Page 39: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/39.jpg)
Which test characteristics?
• There are prevalence-dependent and prevalence-independent measures in diagnostic tests.
• Prevalence-independent: sensitivity and specificity.
• Prevalence-dependent: positive and negative predictive values.
![Page 40: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/40.jpg)
Test Characteristics: SeNsitivity
Sensitivity: • The probability that the test will be positive
when the disease is present.
• Of all the people WITH the disease, the percentage that will test positive.
• A seNsitive test is one that will detect most of the patients who have the disease (low false-Negative rate).
P (Test + | Disease +)
![Page 41: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/41.jpg)
Test Characteristics: SPecificity
Specificity: • The probability that the test will be
negative when the disease is absent.
• Of all the people WITHOUT the disease, the percentage that will test negative.
• A sPecific test is one that will rarely be positive in patients who don’t have the disease (low false-Positive rate).
P (Test - | Disease -)
![Page 42: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/42.jpg)
Test Characteristics: Predictive Values
• Positive predictive value: the probability that a patient has a disease, given a positive result on a test.
P (Disease + | Test +)
• Negative predictive value: the probability that a patient does not have a disease, given a negative result on a test.
P (Disease - | Test -)
![Page 43: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/43.jpg)
A B
C D
Dx+ Dx-
T+
T-
A+C B+D
Diagnostic Test Characteristics
• Sens = A/(A+C)
• Spec = D/(B+D)
• PPV = A/(A+B)
• NPV = D/(C+D)
![Page 44: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/44.jpg)
To reflect upon...
Why?
Sensitivity and Specificity
Prevalence-Independent characteristics
Positive and Negative Predictive Values Prevalence-Dependent characteristics
![Page 45: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/45.jpg)
Let’s try it out
Case: To determine the diagnostic performance of V/Q scans for detecting pulmonary embolism, a study was conducted where 300 patients underwent both a V/Q and pulmonary angiogram. 150 patients were found to have a PE by PA gram. Of those, 75 patients had a high probability VQ scan. Of the 150 patients without a PE, 125 had a non-high probability VQ scan.
V/Q scan
Pulmonary Angiogram
![Page 46: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/46.jpg)
Let’s try it out
Case: To determine the diagnostic performance of V/Q scans for detecting pulmonary embolism, a study was conducted where 300 patients underwent both a V/Q and pulmonary angiogram. 150 patients were found to have a PE by PA gram. Of those, 75 patients had a high probability VQ scan. Of the 150 patients without a PE, 125 did not have a high probability VQ scan (VQ other).
75 25
75 125
PE+ PE-
VQ hi
VQ
other
150 150
![Page 47: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/47.jpg)
Let’s try it out
75 25
75 125
PE+ PE-
VQ hi
VQ
other
150 150
• Sens
• Spec
• PPV
• NPV
= 75/(75+75) = 50%
= 125/(125+25) = 83%
= 75/(75+25) = 75%
= 125/(125+75)= 63%
![Page 48: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/48.jpg)
Pretest Test resultProbability Test changes theP (Disease) Result probability of
disease P (Disease|Test Result)
Modification of Probability
![Page 49: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/49.jpg)
A B
C D
Dx+ Dx-
T+
T-
A+C B+DDisease
Prevalence
Test Characteristics and Prevalence
• Sens = A/(A+C)
• Spec = D/(B+D)
• PPV = A/(A+B)
• NPV = D/(C+D)
![Page 50: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/50.jpg)
Prevalence
75 25
75 125
PE+ PE-
VQ hi
VQ
other
150 150
• Sens = 50%• Spec = 83%• PPV = 75%• NPV = 63%• Prevalence = ???50%
![Page 51: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/51.jpg)
Populations and Patients
Population view• Prevalence reflects
the number of people with the disease at a given moment
Patient view• Same concept
implies how likely an individual patient has the disease
• P (Disease)
![Page 52: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/52.jpg)
Pretest Test resultProbability Test changes theP (Disease) Result probability of
disease P (Disease|Test Result)
Modification of Probability
Disease
Prevalence
![Page 53: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/53.jpg)
An Important Question and Assumption
Question: Are certain test characteristics fixed?
Answer: Generally, yes.
Sensitivity and specificity are constants, regardless of the prevalence of the disease in the studied population (prevalence-INdependent)*
*Exceptions and caveats to this assumption are real, but are beyond the scope of this course
![Page 54: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/54.jpg)
Pretest Test resultProbability Test changes theP (Disease) Result probability of
disease P (Disease|Test Result)
Modification of Probability
Disease
Prevalencesensitivity
specificity
![Page 55: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/55.jpg)
Importance of Pre-Test Probability
• Hi-prob V/Q: Sens = 50%, Spec = 83%
Pre-TP/Prev PPV NPV
50%
PV
75
75
25
125
D+
T-
D-
T+
75% 63%
How do our predictive values relate to our probability after the test result is obtained (our
post-test probabilities)?
Post-TP
![Page 56: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/56.jpg)
Importance of Pre-Test Probability
• Hi-prob V/Q: Sens = 50%, Spec = 83%
Pre-TP/Prev PPV NPV
50%
PV
75
75
25
125
D+
T-
D-
T+
75% 63%
• If our Pre-test Probability was 50%, and we obtain a hi-prob V/Q scan on this patient, what is our Post-test probability?
Post-TP
75%
![Page 57: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/57.jpg)
Importance of Pre-Test Probability
• Hi-prob V/Q: Sens = 50%, Spec = 83%
Pre-TP/Prev PPV NPV
50%
PV
75
75
25
125
D+
T-
D-
T+
75% 63%
• If our Pre-test Probability was 50%, and we obtain a V/Q-other scan on this patient, what is our Post-test probability?
Post-TP
37% (tricky: 1-63%)
![Page 58: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/58.jpg)
What did we just do?
50%
0
100
75% = P(PE|T+)
37% = P(PE|T-)
P (PE)
VQ hi
VQ other
P (PE | Test)
![Page 59: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/59.jpg)
Pretest Test resultProbability Test changes theP (Disease) Result probability of
disease P (Disease|Test Result)
Modification of Probability
Disease
Prevalencesensitivity
specificity Predictive Values(Positive and Negative)
![Page 60: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/60.jpg)
Fundamental Assumptions
Sensitivity and specificity are constants, regardless of the prevalence of the disease in the studied population (prevalence-INdependent)*
Positive and Negative Predictive Values are dependent on the prevalence of the disease in the studied population (prevalence-DEpendent)
*with exceptions
![Page 61: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/61.jpg)
Now, what do we do?*clickers
75% = P(PE|T+)
Q1: Choices:a)Treat as if patient has PEb)Decide to get another testc)Decide that patient does not have a PE
What factors do you consider when making the next decision?
![Page 62: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/62.jpg)
Now, what do we do?*clickers
37% = P(PE|T-)
Q2: Choices:a)Treat as if patient has PEb)Decide to get another testc)Decide that patient does not have a PE
What factors do you consider when making the next decision?
![Page 63: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/63.jpg)
Now, what do we do?
75% = P(PE|T+)
37% = P(PE|T-)
Choices:•Treat as if patient has PE•Decide to get another test•Decide that patient does not have a PE
Choices:•Treat as if patient has PE•Decide to get another test•Decide that patient does not have a PE•Decide to get another test
•Treat as if patient has PE
What factors do you consider when making the next decision?
![Page 64: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/64.jpg)
What if we change our pretest probability?
• In essence, we are simultaneously changing the prevalence:– Original pre-TP = P(PE) = 50%– New pre-TP = P(PE) = 25%
HIGH RISK
MED RISK
• Assuming that sensitivity and specificity are fixed…then we must recalculate our predictive values to determine our new post-test probabilities.
![Page 65: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/65.jpg)
Importance of Pre-Test Probability
• Hi-prob V/Q: Sens = 50%, Spec = 83%
Pre-TP/Prev PPV NPV
50%
25%
Post-TP75
75
25
125
38
37
38
187
D+
T-
D-
T+
D+ D-
T+
T-
75% 63%
50% 83%38/(38+38) 187/(187+37)
hi riskmed risk
Our Pre-test Probability was 25%, we obtain a V/Q-other scan on this patient, our Post-test probability is now…17%
![Page 66: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/66.jpg)
Decision time*clickers
50% = P(PE|T+)
Q3: Choices:a)Treat as if patient has PEb)Decide to get another testc)Decide that patient does not have a PE
![Page 67: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/67.jpg)
Decision time*clickers
17% = P(PE|T-)
Q4: Choices:a)Treat as if patient has PEb)Decide to get another testc)Decide that patient does not have a PE
![Page 68: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/68.jpg)
Decision time
50% = P(PE|T+)
17% = P(PE|T-)
Choices:•Treat as if patient has PE•Decide to get another test•Decide that patient does not have a PE
Choices:•Treat as if patient has PE•Decide to get another test•Decide that patient does not have a PE•Decide to get another test
•Decide to get another test
![Page 69: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/69.jpg)
Let’s change it again…
• Again, we are changing the prevalence:– Young woman, no risk factors, some
dyspnea, no history, normal exam– If we consult our clinical prediction rule:
• New pre-TP = P(PE) = 5%: LOW RISK
![Page 70: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/70.jpg)
Importance of Pre-Test Probability
• Hi-prob V/Q: Sens = 50%, Spec = 83%
Pre-TP/Prev PPV NPV
50%
5%
Pred Val75
75
25
125
8
7
47
238
D+
T-
D-
T+
D+ D-
T+
T-
75% 63%
15% 97% 8/(8+47) 238/(238+7)
hi risklo risk
![Page 71: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/71.jpg)
What did we just do?
50%
0
75% = P(PE|T+)
37% = P(PE|T-)
P (PE)
VQ hi
VQ other
P (PE | Test)
5%15% = P(PE|T+)
3% = P(PE|T-)
VQ hi
VQ other
ObservationAs prevalence (pre-test probability) decreases,
positive tests are more likely to be false-positives
![Page 72: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/72.jpg)
Fundamentally...
Question: If you get a high probability V/Q scan for the diagnosis of pulmonary embolism, is it more likely to represent a false positive test if the patient presented with…
(a) many clinical features of PE (shortness of breath, chest pain, long plane ride), or
(b) no clinical features of PE (no shortness of breath, no chest pain, no leg swelling, no long plane ride)?
![Page 73: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/73.jpg)
Alternative Vocabulary - Rates
• True Positive Rate = sensitivity
• False Positive Rate = 1-specificity
• False Negative Rate = 1-sensitivity
• True Negative Rate = specificity
![Page 74: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/74.jpg)
Combining Rates - Methods
• Likelihood Ratios
• ROC Curves
![Page 75: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/75.jpg)
Combining Rates - Method 1Likelihood Ratios (LR)
• Concept - LRs depict the relationship between true and false rates– TPR/FPR = LR for a positive test result– FNR/TNR = LR for a negative test result
TPR sens LR = --------- = -------------
FPR 1-spec
Typically >1, excellent >10
FNR 1-sens LR = --------- = -------------
TNR spec
Typically <1, excellent <0.1
![Page 76: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/76.jpg)
ApplicationLikelihood Ratios (LR)
Key Concept: LRs can be combined with pre-test odds to get post-test odds
Remember our scenario:
High risk pt - 50% (PreTP)
0.50
LR (VQ hi) = --------- = 2.94
1-0.83
50% 1.0 2.94 2.94 75%
*converting odds to probability and vice and versa - many references online
PreTP
LRPreTO
PostTO
PostTP
x =* *
![Page 77: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/77.jpg)
Combining Rates - Method 2ROC Curves
Visual depiction of LR• Tests with continuous
values only• Sensitivity-specificity
tradeoff at different cutoffs
• TPR plotted against FPR
![Page 78: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/78.jpg)
ApplicationROC Curves
ROC Curves• Area under the curve
determines overall utility of the test
• Inflection point reflects optimal threshold
• More in Small Group Exercise– Assignment 3
![Page 79: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/79.jpg)
Take Home Points
• Research studies of diagnostic tests give you test characteristics, not predictive values.
• Relationships between sensitivity and specificity can be captured in ROC curves (for tests with thresholds) and Likelihood Ratios (LRs)
• Appropriate use of tests stem from large differences between pre-test and post-test probabilities, resulting from LRs that strongly deviate from 1.
• If your pre-test probability is very low (<10%) or very high (>90%), it is rare that a single test can help.
![Page 80: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/80.jpg)
Diagnostic ReasoningThe Odyssey Returns
• Failure to entertain all possibilities
• Failure to pay attention to all symptoms
• Failure to inform customer
• Failure to perform diagnostic tests
• Entertain all important possibilities
• Elicit and pay attention to description of all symptoms
• Inform and involve patients
• Perform effective diagnostic tests
The Mechanic The Clinician
![Page 81: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/81.jpg)
ApplyAcquire
Ask
Appraise
50 Prime, flickr
![Page 82: Diagnostic Reasoning I and II](https://reader033.fdocuments.net/reader033/viewer/2022052910/559ab2171a28abc5378b4637/html5/thumbnails/82.jpg)
Additional Source Informationfor more information see: http://open.umich.edu/wiki/AttributionPolicy
Slide 09: http://www.ncbi.nlm.nih.gov/pubmed/ ; http://www.guideline.gov/ Slide 12: 50 Prime, flickr, http://www.flickr.com/photos/pernett/1544045987/, CC: BY http://creativecommons.org/licenses/by/2.0/deed.enSlide 73: 50 Prime, flickr, http://www.flickr.com/photos/pernett/1544045987/, CC: BY http://creativecommons.org/licenses/by/2.0/deed.en