R. Moreno* P.Morais* Outcome prediction in intensive care: results of a prospective, multicentre, Portuguese study Received: 2 April 1996 Accepted: 24 October 1996 * On behalf of the Portuguese Severity Scoring Systems Study Groups of the Por- tuguese Intensive Care Society and of the Portuguese Society of Internal Medicine R. Moreno ( ) ⋅ P.Morais Intensive Care Unit, Hospital de Santo Anto ´ nio dos Capuchos, Alameda de Santo Anto ´ nio dos Capuchos, P-1150 Lisboa, Portugal FAX: +351 (1) 48 44 635 e-mail: r.moreno@mail.telepac.pt Abstract Objective: To compare the performance of the New Simpli- fied Acute Physiology Score (SAP- S II) and Acute Physiology and Chronic Health Evaluation (APACHE) II in an independent database, using formal statistical as- sessment. Design: Analysis of the database of a multicentre, prospective study. Setting: 19 intensive care units (ICUs) in Portugal. Patients: Data for 1094 patients con- secutively admitted to the ICUs were collected over a period of 4 months. Following the original SAPS II and APACHE II criteria, the analysis excluded patients younger than 18 years of age, read- missions, acute myocardial infarc- tion, burns, patients in the post-op- erative period after coronary artery bypass surgery, and patients with a length of stay in the ICU of less than 24 h. The group analysed comprised 982 patients. Interventions: Collection of the first 24 h admission data necessary for the calculation of SAPS II, APA- CHE II, Therapeutic Intervention Scoring System (TISS), Simplified TISS, organ system failure and basic demographic statistics. Vital status at discharge from the hospital was registered. Measurements and results: In this cohort, discrimination was better for SAPS II than for APACHE II (SAPS II: area under the receiver operating characteristic curve 0.817, standard error 0.015; APACHE II: 0.787, 0.015; p < 0.001); however, both models presented a poor cali- bration, with significant differences between observed and predicted mortality (Hosmer-Lemeshow goodness-of-fit tests H and C, p < 0.001). In a stratified analysis, this study was unable to demon- strate any definite pattern of associ- ation between the poor performance of the models and specific subgroups of patients except for the most se- verely ill patients, where both mod- els overestimated mortality. Conclusions: SAPS II performed better than APACHE II in this in- dependent database, but the results do not allow its use, at least without being customised, to analyse quality of care or performance among ICUs in the target population. Key words Severity of illness index ⋅ Intensive care ⋅ Critical care ⋅ Mortality prediction ⋅ Simplified Acute Physiology Score (SAPS II) ⋅ Acute Physiology and Chronic Health Evaluation (APACHE II) Intensive Care Med (1997) 23: 177–186  Springer-Verlag 1997 ORIGINAL