Cardiac arrest: physicians could not predict in-hospital survivors

Clinical bottom line (level 4)

  1. Physicians could not accurately predict which patients will survive in-hospital cardiac arrest using information available in the first 24 hours.
Ebell et al: Journal of General Internal Medicine 1996; 11: 16-22
Expires October 2003

The study

Setting: 2 acute hospitals, USA

180 patients (aged ?, ?% male) who had a cardiac arrest

Excluded if
  • use of chest compressions, artificial ventilation or rescue breathing and cardiac medications were not documented
  • resuscitated in the operating room
  • resuscitation occurred in the emergency room



    • ?independent unblinded reference standard, applied in all patients from a non-consecutive appropriate spectrum.
    Reference standard:
    • outcome from CPR
    Diagnostic test: case summary based on information available in first 24 hours of admission
    • physicians were asked to predict the outcome of in-hospital cardiopulmonary resuscitation., producing 713 predictions

    The evidence

    pre-test probability of surviving cardiac arrest: 36.5%, (95% CI: 32.9% to 40.0%)

    diagnostic test survival death LR
    (95% CI)    		 post-test probability
    physician's prediction < 10% 61 94 1.13
    (0.85 to 1.5)     		39%
    11% to 20% 46 72 1.11
    (0.79 to 1.56)     		39%
    21% to 40% 56 91 1.07
    (0.80 to 1.4)     		38%
    41% to 60% 43 92 0.81
    (0.59 to 1.13)     		32%
    61% to 80% 34 70 0.85
    (0.58 to 1.24)     		33%
    > 80% 21 33 1.11
    (0.66 to 1.88)     		39%
    total 261 452

    • Physicians were worse at predicting survival to discharge (25% predicted compared with 3.8% actually discharged).

    Comments

    1. The ability to predict survival did not change based on a physician's experience or speciality (internal medicine or family medicine)
    2. The basis of the hypothesis is a little unusual. Physicians don't typically predict any individual patients response to treatment ie/ which leukaemia will respond to chemotherapy. Instead patients are given the general outcomes to help them make an informed decision. We can accurately inform patients that their overall chance of being resuscitated is poor outside of very limited circumstances.

    Citation

    1. Ebell MH, Bergus GR, Warbasse L, et al: The inability of physicians to predict the outcome of in-hospital resuscitation. Journal of General Internal Medicine 1996; 11: 16-22
    Contributor: Chris Ball and Clare Wotton, October 1999
    Reviewer: Michael Christian

    Clinical Question.
    Patient cardiac arrest
    Intervention or Exposure physician prediction
    Outcome mortality