Does intensive care improve outcome?

Introduction

'Does intensive care improve outcome?' is a question of fundamental importance. It is unlikely to ever be answered conclusively, because of ethical constraints on performing a controlled study of two groups randomised to 1CU admission or no ICU admission. It is possible however, to assess ICU admission as an independent predictor of survival when considered alongside other possible predictors.

Method

Prospective cohort study of all adult patients referred for emergency admission to the only adult ICU of a university hospital during a 3-month period. Because of the limited number of alternative ICU beds patients refused admission are not transferred to an ICU in another hospital. Exclusion criteria were: direct ICU transfers from other hospitals, patients referred when the ICU was full, patients with acute burn injury and cardiac surgery patients. (The latter two groups were excluded from the original MPM II derivation and validation sets). MPM IIM₀ score was calculated for each patient and the following data were collected: sex, referring specialty, APACHE II diagnostic weighting, admission or refusal of admission to ICU and hospital survival. MPM II₂ consists of 14 physiological and diagnostic variables and is the only severity scoring system available at ICU admission. Logistic regression analysis using a forward stepwise conditional method was performed using SPSS for Windows. Variables included as possible predictors of survival were sex, MPM II₀, APACHE II diagnostic weighting, ICU admission, and interactions between admission and MPM II₀ and admission and APACHE II weighting.

Results

Three hundred and eighty-three patients were studied of whom 229 were admitted. Low MPM II₀ low APACHE II diagnostic weighting and admission to ICU were found to be independent predictors of hospital survival with a hospital survival odds ratio for ICU admission/ refusal of 2.41 (95% CI: 1.48-3.93).

Discussion

This result demonstrates that admission to ICU is an independent predictor of survival when compared with available likely predictors. It is unlikely that the association between admission and survival is due to selective admission of those patients more likely to survive as the association was independent of severity of illness, type of illness and sex. Our results therefore strongly support a positive answer to the question 'Does admission to ICU improve outcome?'.