Decision Analysis Case

<!-- InstanceBeginEditable name="Title" -->Title of page<!-- InstanceEndEditable -->


Medical Epidemiology and Evaluating Medical Evidence

Decision Analysis Case

An exercise in decision analysis: When is coronary angiography required in aortic valve replacement?

In many cases of aortic stenosis, it becomes clinically advisable to perform surgery for aortic valve replacement (AVR). In such cases, the presence of associated occult coronary artery disease (CAD) is known to both increase the perioperative mortality and to shorten the average life expectancy following the procedure. In patients with CAD, a coronary artery bypass graft (CABG) procedure lessens the perioperative mortality of the AVR enough to outweigh the perioperative mortality associated with CABG itself. Therefore, in principle it is desirable to know the CAD status of a candidate for AVR prior to surgery, in order to choose between two surgical interventions: AVR alone, or CABG and AVR.

Occult coronary artery disease may be diagnosed via cardiac catheterization and angiography, a "gold standard" procedure. Unfortunately, however, there is no free lunch: catheterization has an associated mortality of its own, irrespective of the presence of CAD. There are other tests that are useful in the diagnosis of occult CAD, but their sensitivity and specificity are less than the gold standard. Here we consider the possible utility of one such test for a particular patient.

Statement of the clinical situation and reasonable choices

A 63 year old woman has exertional syncope without history of chest pain, myocardial infarction, or congestive heart failure. She has physical, electrocardiographic (ECG) and echocardiograms typical of severe aortic stenosis. Aortic valve replacement is recommended. The only utility that the patient wishes you to consider in making treatment decisions is life expectancy. (More specifically, cost is not to be considered in this decision, nor does the patient wish to impose any extra penalty, beyond the number of years of expected life lost, for premature death due to surgery or catheterization.) Thus, mortality at catheterization or perioperative mortality is considered to have a utility of 0 years, and the utilities of other events are the associated life expectancies in years.

Of the various tests that may provide an indication of occult CAD, you consider using the dipyridamole-thallium with handgrip exercise (DTH) test. This test is considered abnormal if perfusion defects are seen. The test has no associated mortality or other complications of any consequence. It has sensitivity of 0.85 and specificity of 0.86.

In this context, consider three reasonable clinical strategies:

  1. Perform the AVR procedure without CABG, and without prior evaluation of CAD by DTH or by cardiac catheterization. (No Cath)
  2. Perform a cathetization. If this procedure is negative for CAD, perform AVR alone. If catheterization detects CAD, then perform a CABG followed by AVR. (Cath)
  3. Employ serial testing with DTH and catheterization. Perform the DTH test first. If positive, follow the procedure in choice (2) above. If negative, follow choice (1) above. (Serial test)

In this context, to choose the decision expected to be most beneficial to your patient on her terms, you will need the additional information below.

1. Prevalence of occult CAD

Based on a study compiling autopsy findings in patients who died of noncardiac causes, you use the prevalence of occult CAD in a 63 year old woman as 0.067.

2. Probabilities of Short-term events



Probability of dying at catheterization/angiography



Conditional probabilities of perioperative death


AVR alone, patient has no CAD


AVR with CAD; patient does not receive CABG


AVR with CAD; patient undergoes CABG


3. Probabilities of long-term survival

For this exercise, these are the life expectancy for a 63 year old woman for each of the clinical states

Clinical state

Life expectancy (years)

Aortic stenosis after AVR (no CAD)


Aortic stenosis after AVR (CAD, no CABG)


Aortic stenosis after AVR (CAD & CABG)

8.8 years


  1. Make a decision tree to help choose among these strategies, using the data given.
    (You may find it easier to separate the tree into several pages, with each major branch on a separate page, rather than squeeze everything on one page.)
  2. Fold back the tree and specify the strategy with the highest expected value.
  3. Be prepared to discuss and explain what factors are responsible for the difference in expected value between strategies.
  4. If occult CAD was more prevalent (say in an older patient, where CAD rate is known to be higher), what effect would this have on the expected utility of catheterization?
  5. At what value for the prevalence of occult CAD would you be indifferent between strategies (1) and (2)?

Note: All students are encouraged to complete this assignment individually or in groups before the class session on 11/10. A solution to this assignment will eventually be posted on the course website.

Emergency/Extreme Weather
If conditions require the cancelling of classes or closure, the college will broadcast an announcement on this website, Facebook, Twitter, and Google+.

Review the full Emergency Action Plan.
Faculty/Staff | Students
Twitter Twitter
Facebook Facebook
Tumblr Tumblr
Google+ Google+
Youtube Youtube
University of Illinois
College of Medicine at Urbana-Champaign

506 S. Mathews Ave.
Urbana, IL 61801

Email Us
Phone: (217) 333-5465
Fax: (217) 333-8868

College of Medicine Campuses: Chicago | Peoria | Rockford | Urbana :: © 2017 University of Illinois College of Medicine at Urbana-Champaign