Evaluating outcome from intensive care: a preliminary multihospital comparison

Area of Science:

• Critical Care Medicine
• Health Services Research
• Biostatistics

Background:

• Interhospital comparisons of patient outcomes are crucial for quality assessment.
• Variations in patient severity of illness can confound outcome comparisons.
• Standardized methods are needed to adjust for differences in patient populations.

Purpose of the Study:

• To evaluate the effectiveness of a general severity of illness classification system in comparing mortality rates across different intensive care units (ICUs).
• To assess the accuracy of projected death rates against observed mortality in multiple ICUs.

Main Methods:

• Surveyed 795 consecutive ICU admissions across 5 ICUs.
• Collected patient data including age, sex, admission indication, and severity of illness.
• Utilized a logistic multiple regression equation, calibrated with data from a sixth reference hospital, to project death rates.

Main Results:

• Significant differences in patient severity of acute illness were identified among the hospitals.
• These severity differences accounted for the majority of the variation in observed death rates.
• Projected death rates closely matched observed deaths in all evaluated ICUs.

Conclusions:

• A general severity of illness index, when combined with multivariate statistical techniques, shows promise for improving interhospital outcome comparisons.
• Further refinement and validation are recommended for this approach.
• This methodology can aid in more equitable assessments of ICU performance.