Outcome measures

Clinically meaningful outcomes: something meaningful to the patient; eg.

- survival

- quality of life

Surrogate outcomes: something meaningful to the intensivist; eg.

- oxygenation

- blood pressure, urine output

So, surely surrogate outcomes are worth pursuing? There must be some benefit to maintaining normoxia, and that must have some sort of positive influence on survival... right?

No. In general, surrogate outcomes are not as useful as the clinically meaningful outcomes. Not unless the surrogate outcome has been validated as a good surrogate for a clinically meaningful outcome. Eg. if one is able to produce a study which directly links normoxia with increased survival, then oxygenation becomes a valid surrogate for mortality.

Outcome measures

LITFL has this well wrapped up.

ICU mortality

  • Advantages:
    • Simple, important outcome measure, and already recorded by the hospitals.
  • Disadvantages:
    • Mortality and quality of care are different issues.
    • ICU mortality can be manipulated by redefining what "ICU" means, and by discharging patients who are about to die.


Hospital mortality

  • Advantages:
    • All the benefits of "mortality" as a simple and important outcome measure
    • Avoids the confounding issue of ICU mortality (i.e. the definition of ICU, and the disposal of perimortem patients into palliative care wards)
  • Disadvantages:
    • Confounding factor: ICU care and ward care are mixed up together, impossible to separate the influences.
    • Survival has little merit as an outcome measure unless it is meaningful survival.


90 day mortality

  • Advantages:
    • All the benefits of "mortality" as a simple and important outcome measure
  • Disadvantages:
    • Totally arbitrary (why not 91 days?)
    • Some may be lost to follow up. Are they dead?

Functional status at 1 year

  • Advantages:
    • More meaningful than mortality, some might argue
  • Disadvantages:
    • Scoring is arbitrary
    • Difficult to collect the data
    • Many are lost to follow up.
    • Is the ongoing disability a consequence of the critical illness or the normal progression of disease? Impossible to say.

Surrogate outcomes

Things like inflamatory markers, noradrenaline dose, cardiac output, and so forth.

Things which are important to those ICU nerds, but perhaps less important in the real world.

  • Advantages:
    • Cheap and easy to measure
    • More responsive to interventions than hard clinical outcomes (eg. mortality)
  • Disadvantages:
    • Surrogate outcomes often fail to correlated with "real" outcomes
    • Not patient-centered, and distracting from patient-centered care
    • Performance indicators based on surrogate outcomes tend to spawn administrative abominations like the 4 hour rule.