Advanced Topics in Medicine MSSU Lecture Series

1
Advanced Topics in MedicineMSSU Lecture Series

• Diagnostic Reasoning

2
50 year-old man with 50 probability of
disease.What is the probability after the
following test?
3
To calculate the probability of disease

• pretest odds x LR post-test odds
• Where,
• pretest odds odds of disease before test
• LR likelihood ratio for test result
• post-test odds odds of disease after test

4

• without disease

with disease
5
Probability of Disease

• with disease
• with and without disease

6
Odds of Disease

• with disease

without disease
7
Probability
Odds

• with disease
• with and without disease

with disease
without disease
8
(No Transcript)
9
Likelihood Ratios Simplified2-5-10, 15-30-45
10
Sensitivity and Specificity
Within 6 hours of symptoms, cTnI
has Sensitivity 58 Specificity 94 Diagnostic
Marker Cooperative Study. Circ 1999991671

• Problem Describe in terms that an
  anxious patient with chest pain would
  understand

11
Likelihood Ratio

• probability of test result in people with
  disease
• probability of test result in people without
  disease

12
Positive Likelihood Ratio

• probability of test result in people with
  disease
• probability of test result in people without
  disease

LR
sensitivity (1 - specificity)
LR
13
Negative Likelihood Ratio

• probability of - test result in people with
  disease
• probability of - test result in people without
  disease

LR -
(1 sensitivity) specificity
LR -
14
Case Iron deficiency anemia?

• An 18 year-old woman has a pretest probability of
  iron deficiency of 20 based on a history (JRA)
  and examination of the peripheral smear.
• At U of C laboratory, a ferritin of lt 12 ng/ml is
  considered a positive test for iron deficiency
  with
• sensitivity 65
• specificity 99
• The patients ferritin of 30 is a negative
  test.
• LR -

15
Case Iron deficiency anemia?
probability of - test result in people with
disease probability of - test result in people
without disease
LR -
(1 sensitivity) specificity
0.35 0.99
LR -

post-test odds pretest odds x LR -
Conclusion Does probability of iron
deficiency after testing (ferritin 30)
INCREASE or DECREASE?
16
Disease
100
Test
80
50
Test -
20
40
50
Test
60
No Disease
Test -
LR
LR -
100
17
Disease
100
High
50
60
Low
Medium
20
20
20
20
High
Medium
50
60
LR (high)
No Disease
Low
LR (medium)
LR (low)
100
18
Disease
100
50
Test Level
No Disease
50
LR (0-1)
LR (1)
LR (lt2)
100
19
Meningitis or Bacteremia in Febrile Infants?
Generate a 2 x 5 Contingency Table
20
Meningitis or Bacteremia in Febrile Infants?
Generate a 2 x 5 Contingency Table
21
Case Iron deficiency anemia?

• Since ferritin is not a dichotomous test,
  recalculate this patients probability of iron
  deficiency using the following data from Guyatt
  (J Gen Intern Med 19927145-53).

22
(No Transcript)
23

• LR

24
D-Dimer for Acute DVT and PE
Ann Intern Med 2004140589
25

• LR (low prob)

26
(No Transcript)
27
(No Transcript)
28
V/Q Study for Diagnosing PE
PIOPED. JAMA 19902632753.
29
V/Q Study for Diagnosing PE
30

• LR (low prob)
• Post-test odds LR (low prob) x pretest odds
• Conclusion

31

• LR -

32
ACC/AHA 2002 Guideline Update for the Management
of Patients With Chronic Stable Angina
33
Work-Up Bias Example

• Investigators randomly sampled 550 children from
  a wide variety of primary care practices to
  evaluate a screening test to detect delay in
  language skills. The gold standard was an
  extensive structured interview. This was given to
  all 50 children who were positive on the
  screening test and to a random sample of 50
  children who were negative on the screening test.
  Interviewers (applying the gold standard) were
  not aware of the screening test results.

34
Work-Up Bias Example
Results Sensitivity Specificity Study
Conclusion Is the sensitivity high enough to
make this an acceptable screening test?
35
Unbiased All patients worked up with a gold
standard
Biased Patients with negative test less likely
to have gold standard
Sensitivity A/(A C) Specificity D/(B D)
Sensitivity Specificity
36
Uncorrected
Corrected
Sensitivity Specificity
Sensitivity 90 Specificity 75
Study Conclusion
37
Uncorrected
Corrected
Sensitivity 47 Specificity 97
Sensitivity 90 Specificity 75
Study Conclusion
38
ACC/AHA 2002 Guideline Update for the Management
of Patients With Chronic Stable Angina
39
Prognosis from Dobutamine EchoIn 300 Patients
Followed for 7.3 2.8 Years
1 Indeterminate tests (stopped early) were
considered normal. 2 Ischemia during stress or
fixed wall motion abnormalities. Schinkel AF et
al. Am J Med 20041171-9.
40
50 year-old man with 50 probability of
disease.What is the probability after the
following test?