Cost calculation and prediction in adult intensive care: a ground-up utilization study

The ability of various proxy cost measures, including therapeutic activity scores (TISS and Omega) and cumulative daily severity of illness scores, to predict individual ICU patient costs was assessed in a prospective "ground-up" utilization costing study over a six month period in 1991. Daily activity (TISS and Omega scores) and utilization in consecutive admissions to three adult university associated ICUs was recorded by dedicated data collectors. Cost prediction used linear regression with determination (80%) and validation (20%) data sets. The cohort, 1333 patients, had a mean (SD) age 57.5 (19.4) years, (41% female) and admission APACHE III score of 58 (27). ICU length of stay and mortality were 3.9 (6.1) days and 17.6% respectively. Mean total TISS and Omega scores were 117 (157) and 72 (113) respectively. Mean patient costs per ICU episode (1991 dollar AUS) were dollar 6801 (dollar 10311), with median costs of dollar 2534, range dollar 106 to dollar 95,602. Dominant cost fractions were nursing 43.3% and overheads 16.9%. Inflation adjusted year 2002 (mean) costs were dollar 9343 (dollar AUS). Total costs in survivors were predicted by Omega score, summed APACHE III score and ICU length of stay; determination R2, 0.91; validation 0.88. Omega was the preferred activity score. Without the Omega score, predictors were age, summed APACHE III score and ICU length of stay; determination R2, 0.73; validation 0.73. In non-survivors, predictors were age and ICU length of stay (plus interaction), and Omega score (determination R2, 0.97; validation 0.91). Patient costs may be predicted by a combination of ICU activity indices and severity scores.