Cost-utility in medical intensive care patients. Rationalizing ongoing care and timing of discharge from intensive care.

ABSTRACT

RATIONALE Intensive care unit (ICU) treatment costs pose special challenges in developing countries. OBJECTIVES: To determine the prognostic value of the "utility" score and evaluate the relationship of willingness to pay assessment to utility score during ICU admission. METHODS: We performed a prospective study spanning 12 months in a 24-bed medical ICU in India. Treatment cost was estimated by direct measurement. Global utility score was assessed daily by healthcare providers on a Likert scale (0-1 in increments of 0.1, with 0 indicating death/severe disability and 1 indicating cure/perfect health). The sensitivity, specificity, and likelihood ratios of utility in predicting ICU mortality was calculated. Receiver operating characteristic curves were generated to compare Day 2 utility with APACHE II. The caregiver's willingness to pay for treatment was assessed on alternate days using the bidding method by presenting a cost bid. Based on the response ("yes" or "no"), bids were increased or decreased in a prespecified manner until a final bid value was reached. Simultaneously, treating doctors were asked how much institutional funds they would be willing to spend for treatment. MEASUREMENTS AND MAIN RESULTS: Primary diagnosis in 499 patients included infection (26%) and poisoning (21%). The mean (SD) APACHE II score was 13.9 (5.8); 86% were ventilated. ICU stay was 7.8 (5.5) days. ICU mortality was 23.9% (95% confidence interval, 20.3-27.8). Survival without disability was 8.3% (2/24) for Day 2 utility score ≤0.3 and 95.8% (53/56) for Day 5 score >0.8 (P < 0.001). The likelihood ratio to predict mortality increased as utility values decreased and was highest (5.85) for utility 0.2. Area under the receiver operating characteristic curves for utility and APACHE II were similar. Willingness to pay by the caregiver was 53% of treatment cost and was not influenced by utility. Willingness to pay by ICU doctors showed an inverted U-shaped relationship with utility. CONCLUSIONS: Utility scores help prognosticate, with Day 2 score ≤0.3 associated with poor outcome and ≥0.8 Day 5 score with survival. The caregiver's willingness to pay was inadequate to meet treatment cost. ICU doctors were willing to spend more for moderate utility scores than for very high or low utility values. Further prospective studies are needed to optimize the utilization of scarce ICU resources by identifying patients for appropriate step-down care using utility and willingness to pay.