Medicare intensive care unit use: Analysis of incidence, cost, and payment*

Abstract

To determine the incidence, cost, and payment for intensive care unit services among Medicare beneficiaries. Retrospective observational database cohort study. All nonfederal hospitals with intensive care unit beds (n = 5003) paid through the inpatient prospective payment system (IPPS). We used all fiscal year 2000 Medicare IPPS hospitalizations with consistent payment information (n = 10,657,587). None. We examined the distribution of cost and payments overall, by hospital type, and by diagnosis related group. Intensive care was used in 2,353,208 cases (21.1%). The overall incidence was 59.8 cases per thousand beneficiaries in the aged (65+) population, increasing with age from 36.2 (65–69) to 91.6 (85+). Intensive care unit patients cost nearly three times floor patients ($14,135 vs. $5,571), with two thirds of costs associated with the intensive care unit portion of the stay, $2,278 per intensive care unit day. However, intensive care unit cases were paid at a rate only twice floor cases ($11,704 vs. $5,835). Only 83% of costs were paid for intensive care unit patients, compared with 105% for floor patients, generating a $5.8 billion loss to hospitals when intensive care unit care is required. There was a linear association between the percent intensive care unit in a diagnosis related group and the percent paid, with payment >90% of cost only in diagnosis related groups with ≥60% intensive care unit cases. We found that teaching hospitals were better paid than nonteaching hospitals (87% vs. 78% of costs, respectively), but this was only due to indirect medical education payments. Intensive care is common, expensive, and poorly paid in the Medicare population. Few diagnosis related groups have a large enough intensive care unit population to ensure adequate payment. Additional diagnosis related groups for conditions common to the intensive care unit would improve payment and enable incentives for efficiency.