Simplified Acute Physiology Score (SAPS) 3

Why use it

SAPS 3 provides objective, contemporarily calibrated outcome prediction enabling: (1) fair comparison of case-mix-adjusted mortality outcomes between ICUs (standardized mortality ratio benchmarking); (2) transparent prognosis communication to multidisciplinary teams and families; (3) stratification of patients in clinical trials of critically ill adults; (4) identification of unusually high or low performing ICUs for quality improvement initiatives; (5) justification of resource allocation in high-acuity settings. Unlike APACHE II (derived from 1970s–1980s US data), SAPS 3 reflects contemporary global ICU populations, treatments, and admission patterns, including post-cardiac surgery, medical emergencies, trauma, and post-operative monitoring.

About

The Simplified Acute Physiology Score 3 (SAPS 3) was developed by Metnitz, Moreno, and colleagues and published in two companion papers in Intensive Care Medicine in 2005, derived from the first European Prevalence of Infection in Intensive Care (EPIC II) study and other international ICU registries. The derivation cohort comprised 21,571 patients from 303 ICUs in 35 countries, making it the largest multinational ICU outcomes dataset at the time of publication. SAPS 3 was specifically designed to reflect 21st-century ICU populations, incorporating contemporary therapeutic interventions and a far broader geographic case-mix than its predecessors (SAPS and SAPS II). The scoring system evaluates 20 clinical variables organized across three domains: Box I assesses patient characteristics existing prior to ICU admission (age, comorbidities, intrahospital location before ICU, prior vasoactive therapy, and length of hospital stay before ICU); Box II captures circumstances of ICU admission (surgical status, reason for ICU admission, and presence of infection at admission); and Box III records acute physiological derangements measured within ±1 hour of ICU admission (Glasgow Coma Scale, heart rate, systolic blood pressure, bilirubin, body temperature, creatinine, urine output, oxygenation on mechanical ventilation, white blood cell count, platelet count, and arterial pH). The sum of variable scores yields the SAPS 3 total. Predicted hospital mortality is derived using logistic regression equations calibrated to nine global geographic regions (worldwide, Western Europe, Eastern Europe, Australasia, North America, Latin America, Sub-Saharan Africa, Southeast Asia/Pacific Islands). SAPS 3 demonstrated superior discrimination and calibration compared to SAPS II and APACHE II in the international validation cohort, achieving an AUROC of 0.848 for hospital mortality prediction in the worldwide model.

When to use

• Any adult patient (age ≥ 18) admitted to a medical, surgical, cardiac, neurological, or mixed ICU
• All variables should reflect the patient status at ICU admission or the worst value within ±1 hour of ICU arrival
• Use the WORST values within the ±1 hour window for all Box III physiological parameters
• Benchmarking: calculate SAPS 3 at ICU admission and compare observed-to-expected (O/E) mortality ratios across units and over time
• Clinical trial stratification — widely required as a stratification variable in international critical care trials
• NOT validated for: paediatric ICUs (use PRISM III or PIM3 instead); burn unit patients; post-cardiac surgery patients in some regional calibrations; admission assessments performed more than 1 hour after ICU arrival
• NOT a substitute for clinical judgment — SAPS 3 provides probabilistic population-level estimates, not individual outcome certainty

Formula

Pearls & pitfalls

• ALL Box III physiological variables must reflect the WORST values recorded within the 1-hour window before or after ICU admission — not the admission-time value or the best value. Missing physiological data should prompt reassessment, not assumption of normality.
• SAPS 3 predicts HOSPITAL mortality (not ICU mortality) — the predicted probability includes deaths occurring after ICU discharge but during the same hospitalization, consistent with the study endpoint used in the derivation cohort.
• Nine region-specific logistic regression equations are published (Moreno et al. ICM 2005, Table 6): worldwide, Western/Central Europe, Eastern Europe, Australasia, North America, Latin America, and Sub-Saharan Africa/Southeast Asia/Pacific Islands. The worldwide equation is appropriate when regional calibration is unknown, but regional equations provide better calibration in their respective populations.
• Metastatic cancer (11 points) is the single heaviest comorbidity item in SAPS 3, substantially outweighing all other Box I comorbidities. Any patient with known distant metastases will carry a significant baseline score increment irrespective of acute physiology.
• The surgical status component (Box II) must distinguish carefully between elective post-operative (lowest acuity), scheduled post-operative (planned for this hospitalization), unscheduled/emergency post-operative, and non-surgical medical admission — both unscheduled surgery and medical admission score identically (+6), reflecting their similar outcome profiles in the derivation cohort.
• GCS in Box III should reflect the patient's NEUROLOGICAL state, not sedation-induced depression. If a patient is pharmacologically sedated, use the pre-sedation GCS or the neurological exam performed before sedation was initiated.
• Urine output requires a MEASURED 24-hour collection via urinary catheter or a projected rate from an early measurement. Estimated urine output without catheterization should not be used — this variable is important for detection of renal impairment at admission.
• SAPS 3 outperforms APACHE II particularly in medical ICU admissions, non-operative patients, and non-North American populations. APACHE IV may retain advantages in specific North American surgical populations.
• The Standardized Mortality Ratio (SMR = observed deaths / SAPS 3 predicted deaths) is the primary benchmarking metric. SMR > 1.0 indicates excess mortality relative to prediction; SMR < 1.0 indicates better-than-expected performance. SMR confidence intervals (exact Poisson) are required for valid comparison.
• Unlike the original SAPS and SAPS II (which scored variables over the first 24 hours of ICU stay), SAPS 3 uses the 1-hour admission window exclusively — enabling immediate outcome prediction at ICU arrival without waiting for 24-hour data.

Next steps

Score < 40 (Very Low Risk): Hospital mortality typically < 10%. Standard ICU monitoring and care. Reassess trajectory daily. Score 40–59 (Low–Moderate Risk): Hospital mortality approximately 10–25%. Standard ICU management. Daily goals-of-care assessment. Score 60–79 (Moderate–High Risk): Hospital mortality approximately 25–50%. High-dependency multidisciplinary care. Initiate or advance goals-of-care conversation. Score 80–99 (High Risk): Hospital mortality approximately 50–70%. Intensive organ support. Goals-of-care discussion mandatory. Integrate palliative care planning. Score ≥ 100 (Very High Risk): Hospital mortality typically > 70%. Balance aggressive resuscitation with realistic prognosis. Early specialist palliative care involvement. Advance directive review with patient and family. Avoid futile interventions. Note: For precise predicted mortality probability, use the official SAPS 3 calculator at saps3.org with region-specific logistic regression equations (Moreno et al. ICM 2005, Table 6).

Evidence

Metnitz PGH et al. SAPS 3 — From evaluation of the patient to evaluation of the intensive care unit. Part 1: Objectives, methods and cohort description. Intensive Care Med. 2005;31(10):1336-1344. PMID:16132893. Moreno RP et al. SAPS 3 — From evaluation of the patient to evaluation of the intensive care unit. Part 2: Development of a prognostic model for hospital mortality at ICU admission. Intensive Care Med. 2005;31(10):1345-1355. PMID:16132892. Derivation cohort: n=21,571 patients, 303 ICUs, 35 countries. AUROC = 0.848 (worldwide). Keegan MT et al. Comparison of APACHE III, APACHE IV, SAPS 3, and MPM0 III and influence of resuscitation status on model performance. Chest. 2012;142(4):851-858. PMID:22459930. External validation in multiple European, Australian, and Asian ICU registries confirming discrimination and calibration superiority over SAPS II.

References

1. Metnitz PGH et al. SAPS 3 — Part 1: Objectives, methods and cohort description. Intensive Care Med. 2005;31(10):1336-1344. — link
2. Moreno RP et al. SAPS 3 — Part 2: Development of a prognostic model for hospital mortality at ICU admission. Intensive Care Med. 2005;31(10):1345-1355. — link
3. Keegan MT et al. Comparison of APACHE III, APACHE IV, SAPS 3, and MPM0 III and influence of resuscitation status on model performance. Chest. 2012;142(4):851-858. — link
4. Sakr Y et al. Use and effectiveness of different prediction models for mechanical ventilation in patients with sepsis. ICM. 2012. — link
5. Vincent JL et al. EPIC II Group. International study of the prevalence and outcomes of infection in intensive care units. JAMA. 2009;302(21):2323-2329. — link