Currently about 15 per cent of hospital costs are attributed to intensive care. Research using statistical models has not adequately demonstrated that therapy in intensive care units (ICUs) is associated with reductions in the probability of death. In a study of 613 consecutive admissions to a multidisciplinary ICU, we reevaluate the relationship between ICU care and survival using a new acute physiology scoring system to control for the severity of illness of the patient population.When our severity of illness index was employed, we found a statistically significant and nonlinear relationship between the use of intensive medical care and the probability of survival. This statistical relationship Hospital-based intensive care accounts for about 15 per cent of hospital costs and is widely believed by physicians to be efficacious in treating individual patients.* However, studies of intensive care unit (ICU) populations have not been able to statistically demonstrate that the increased utilization of therapy in ICUs lowers the probability of death.2-4 Likewise, analyses of area-wide data have in many instances failed to find a measurable health benefit related to the availability of health care resources.5-9 In this paper we present results indicating that the impact of intense medical care on the probability of death is nonlinear and statistically measurable when the analysis controls for severity of illness.
Materials and Methods
Data BaseTo perform the analysis, we collected detailed data on 613 consecutive admissions to the 16-bed medical and surgi- produced a U-shaped curve with three distinct segments. The first segment exhibited an overall decrease in the probability of death with increasing therapy (275 admissions); the second segment, a fairly stable survival rate (281 admissions). Only in the third segment, where there were 57 admissions, did we find an overall increase in the probability of death as utilization of therapy increased.These findings suggest that quantitative measurement of severity of illness, when used in clinical studies, could produce improved insights into the relationship between therapy and health outcomes.(Am J Public Health 1982; 72:449-454.) cal ICU of the George Washington University Medical Center (GWUMC). The GWUMC ICU treats a wide variety of patients with acute medical and surgical problems with the exception of myocardial infarctions and burns.During the eight-month study period, we collected information on the patient's past medical history; indications for ICU admission; the amount and type of therapy each patient received; the severity of acute illness; and outcome, defined as dead or alive when discharged from the hospital. Information regarding past medical history included age, sex, operative status, smoking history measured in pack years,** and the use of alcohol if it contributed to past or present medical problems.To measure therapy we collected Therapeutic Intervention Scoring System (TISS) points during each shift the patient was in the ICU.'0 TISS us...