A 60 year old woman presented to the emergency department with 25 minutes of anterior, non-radiating, burning chest pain. She states that she might have felt short of breath, but she thinks it was just anxiety. There were no other symptoms, and she feels fine now. She has a normal ECG, and undetectable troponins on arrival and 6 hours after the onset of her pain. You think she is ready to be discharged, but you wonder if she needs to have a stress test.
This is the beginning of a series of posts looking at the evidence for stress testing. You would think that this would be a simple question: does stress testing help? But once you start digging, it becomes clear that the practice of stress testing is based on a series of assumptions, and each assumption needs to be considered.
When ordering a stress test after emergency department discharge, you are assuming:
- That the risk of ACS (pretest probability) is high enough to warrant further testing.
- That the stress test is accurate enough to detect patients with disease.
- That the disease detected by stress testing matters to the patient and is related to their chest pain.
- That finding the disease early, rather than waiting for the disease to become clinically obvious, is important.
- And, that once identified, we have a treatment that will improve outcomes for those patients with disease.
After sorting through a mass of literature, it isn’t clear that any of those assumptions are well founded. My conclusion is that stress testing has no role for low risk chest pain patients. To understand how I got there, the following 5 posts will summarize different areas of the medical literature.
Part 1 looks at the pretest probability of the patients we send for stress test, or our “miss rate”. With a standard chest pain workup, we miss very few patients with ACS. Therefore, the pretest probability for our stress test patients is incredibly low.
In part 2 I take some time to discuss the outcomes used in cardiovascular studies. Although an entire post dedicated to definitions might seem like overkill, these outcomes can be deceiving, and a firm grasp is required to make sense of the stress test literature.
Finally, in part 3 I will actually discuss the accuracy of the stress test. The bottom line: depending on what you are trying to detect, the stress test is either abysmal or just inaccurate.
Part 4 asks what we can do with the results of a stress test. Specifically, does identifying patients for invasive management lead to better outcomes? (Spoiler: invasive management does not help the type of patient we send for stress testing, which raises the question of how stress testing could change the management of these patients, even if it were an accurate test.)
These posts provide the background information for a podcast on the same topic done as part of the Emergency Medicine Cases Journal Jam series.
There is also a short (free) EM:RAP rant based on these posts.
If you enjoy these in-depth evidence reviews, you can find more here.