Prognostication Following Cardiac Arrest

Introduction

In 2019 the AHA reviewed data for cardiac arrest finding that out-of-hospital cardiac arrest (OHCA) occcured in ~356,000 patients with ~209,000 patients suffering in-hospital cardiac arrest (IHCA). Acheivement of return of sponateous circulation (ROSC) although the first step in treatment, does not always result in a good outcome with only about 12% of OHCA patients surviving to discharge and only 8% survive with a good outcome. Accurate prediction of neurologic outcomes becomes hugely important for determining who to perform addition testing on, and who to give more time to. It also is very important to decide who has brain injury so severe that no further testing or treatment will be helpful.

This brings up another major point about outcomes following cardiac arrest and how we study them. We define them as good vs poor. Thats it, there is no middle ground. Yes, this is defined by the statistical analysis we are utilizing when studying tests and outcomes, but there is a dichotomy of outcomes in research (not just cardiac arrest) that is not perfect. It is very important to consider this fact, along with the family and patient wishes when having discussions regarding any consideration of prediction or WLST. 

Prognostic accuracy is different than diagnostic accuracy. What I mean by this is that when assessing diagnostic accuracy you are evaluating a test or finding against a gold standard to determine how well it performs. 

• For example - If you want to know how well an ultrasound is at confirming tube placement after intubation, you compare this to a chest x-ray, the gold standard or reference that is used clinically

When you evaluate a test, say pupillary light reactivity, and use good vs bad as an outcome, you are using the presense or absence of pupillary light reactivity as a surrogate for severe brain injury and the development of the outcome. 

Most tests that are performed or aimed to assist in prediction of outcomes after cardiac arrest are aimed at identifying poor outcome. The most commonly reported variables are

• False-positive rate (FPR) - The proportion of patients who ultimately have a good outcome when they were assigned a poor outcome because of the test result

• Specificity - When attempting to predict poor outcomes, this is the patients that have a positive test and a poor outcome vs those with a positive test with a good outcome (TP vs FP)

A key component to assessment of any exam or test and how it leads to poor outcome is avoidance of falsely pessimistic predictions. Below is a chart about the self-fulfilling prophecy in medicine but most notably neurocritical care. Many more patients have survived and some have great outcomes purely because we give them a chance. This begins when one (clinician) decides that based on a test or exam finding a patient will have a poor outcome and further care is pointless. This may be born from previous similar cases, witnessed poor outcomes or purely because as ICU providers we see patients only in the very beginigng of their illness when they appear the worst. This belief, that the outcome will be bad, can then influence other team members, tests that are completed (or not), and eventually expectations of the family and the team. This may lead to WLST and to he patients death that reinforces the prediction of the poor outcome only to repeat again. Notice that once we have been influenced to convince a family to WLST on a patient they do not have any chance for recovery, for without these treatments they will die.