Objective:To determine whether the therapeutic intervention scoring system (TISS) reliably reflects the cost of the overall intensive care unit (ICU) population, subgroups of that population and individual ICU patients. Design: Prospective analysis of individual patient costs and comparison with TISS. Setting: Adult, 12 bedded general medical and surgical ICU in a university teaching hospital. Subjects: Two hundred fifty-seven consecutive patients including 52 coronary care (CCU), 99 cardiac surgery (CS) and 106 general ICU (GIC) cases admitted to the ICU during a 12-week period in 1994. A total of 916 TISS-scored patient days were analysed. Main outcome measures: A variable cost (VC) that included consumables and service usage (nursing, physiotherapy, radiology and pathology staff costs) for individual patients was measured daily. Nursing costs were calculated in proportion to a daily nursing dependency score. A fixed cost (FC) was calculated for each patient to include medical, technical and clerical salary costs, capital equipment depreciation, equipment and central hospital costs. The correlation between cost and TISS was analysed using regression analysis. Results: For the whole group (n = 257) the average daily FC was pound 255 and daily VC was pound 541 (SEM 10); range pound 23-pound 2,806. In the patient subgroups average daily cost (FC + VC) for CCU was pound 476 (SEM 17.5), for CS pound 766 (SEM 13.8) and for GIC pound 873 (SEM 13.6). In the group as a whole, a strong correlation was demonstrated between VC and the TISS for each patient day (r = 0.87, p < 0.001) and this improved further when the total TISS score was compared with the total VC of the entire patient episode (r = 0.93,p < 0.001). This correlation was maintained in CCU, CS and GIC patient cohorts with only a small median difference between actual and predicted cost (2.2 % for GIC patients). However, in the individual patient, the range of error was up to +/- 65 % of the true variable cost. For the whole group the variable cost per TISS point was pound 25. Conclusion: These results demonstrate that TISS reliably measures overall ICU population costs as well as those of the subgroups CCU, CS and GIG. However, the relationship between TISS and cost is less reliable for the individual patient.
