Critical care utilization in Canada increased from 17 ICU patient days per thousand population in 1969 to 42 ICU patient days per thousand population in 1986) This represented approximately 8% of 1986 total inpatient costs. In the same year, intensive care utilization in the United States was 108 patient days per thousand population, accounting for 20% of all inpatient costs. This increase in utilization has taken place without knowledge of either the efficacy or cost-effectiveness of the care delivered. The need to control costs while demonstrating effective utilization of resources lends an urgency to the development of appropriate databases and, more importantly, processes through which outcome analysis can be translated into effective, high quality care at low cost. In their article in this issue, Byrick and Caskennette describe an automated system for critical care audit. The authors correctly identify two major elements of such utilization management and quality assurance systems. These are development and testing of an appropriate data base, and implementation of an audit process. The audit process must include definition of appropriate outcomes, interdisciplinary data analysis, development of an action plan to address important issues, and implementation of required changes. The final element in "closing the loop," namely evaluation of the impact of the audit process on quality of care and utilization is not addressed by the authors. In fact, there are few well-designed outcome studies which analyze the results of utilization review in the literature. 2 Much of the information on utilization management comes from the United States where rapidly inflating health care costs and complex interrelationships of governmental and private payers have resulted in pressures to control costs while assuring quality care. Utilization management has evolved from early quality assurance efforts by providers, to increased third party involvement which has moved the process from retrospective, to concurrent, and even prospective (preadmission) review. Despite these massive efforts, the scientific merit of these endeavours remains unclear. A recent survey on utilization management 3 revealed that ap-