Send to

Choose Destination
See comment in PubMed Commons below
Crit Care Med. 1992 Jul;20(7):967-76.

Outcome in critical care patients: a multivariate study.

Author information

  • 1Department of Surgery, Albany Medical Center, NY 12208.



To identify patient variables that were significantly associated with outcome in the ICU.


Retrospective chart study of ICU patients.


Letterman Army Medical Center, Presidio of San Francisco.


A total of 110 consecutive patients who remained in the ICU for greater than 72 hrs.


Univariate analysis of 26 input variables, representing physiologic status and chronic health assessment at the time of ICU admission of the patients, and adverse events that occurred during their ICU stay. The dependent variable assessed was the outcome of patients after ICU intervention. A bad outcome included patient mortality or discharge from the ICU in worse condition or in a chronic vegetative condition. Five selected input variables were used in a multivariate logistic regression analysis to identify independent predictors of a successful outcome.


Univariate statistical techniques comparing the values of 26 input variables of patients who improved during their ICU stay (46 patients) with those values of patients who did not improve (52 deaths and 12 unimproved patients) indicated that the following nine variables were significantly different in these two groups: a) age (p = .04), b) acute physiologic score (p = .047), c) Acute Physiology and Chronic Health Evaluation (APACHE II) score (p = .040), d) the number of repeat ICU admissions during a given hospitalization (p = .019), e) development of respiratory complications while in the ICU (p = .033), f) sepsis developing while in the ICU (p = .0001), g) renal failure developing while in the ICU (p less than .0001), h) disseminated intravascular coagulation while in the ICU (p = .0028), and i) the occurrence of any iatrogenic complication while in the ICU (p less than .0001). Five of these nine variables were used in a stepwise logistic regression analysis to develop a model to predict ICU outcome. This model showed the following three variables to be significantly associated with unfavorable ICU outcome: a) occurrence of any iatrogenic event while in the ICU (p less than .001), b) development of renal failure in the ICU (p = .001), and c) occurrence of sepsis while in the ICU (p = .042). A common iatrogenic complication associated with unfavorable ICU outcome was related to drug therapy (frequently aminoglycoside toxicity). Twenty-one iatrogenic drug complications occurred in 64 patients who did not improve in the ICU, whereas only three iatrogenic drug complications occurred in 46 patients who improved (p less than .001).


For patients who remain in the ICU for greater than 72 hrs, events occurring after ICU admission are negatively associated with ICU outcome, more so than ICU admission status as reflected by such indices as APACHE II scores. Iatrogenic complications, often due to inappropriate drug therapy, have a significant association with adverse outcome by multivariate analysis. We suggest that iatrogenic complications influence ICU outcome, and that they are potentially preventable. By lessening their frequency, ICU outcome may be improved.

[PubMed - indexed for MEDLINE]
PubMed Commons home

PubMed Commons

How to join PubMed Commons

    Supplemental Content

    Loading ...
    Write to the Help Desk