MPM-III Score Calculator
ICU Admission Mortality Estimate
Enter the patient's status at ICU admission or within the first hour, before ICU interventions materially alter the admission profile.
The “zero factors” term is derived automatically when all 14 MPM0-II risk-factor variables (excluding age) are absent. It is not a separate clinical diagnosis.
About MPM0-III
MPM0-III (Mortality Probability Model at ICU admission, version III) is a logistic-regression model developed to estimate the probability of hospital mortality using patient characteristics recorded at or within 1 hour of ICU admission. It is not a points-based bedside score.
The published model was developed using 124,885 eligible admissions from 135 ICUs at 98 hospitals participating in Project IMPACT (2001–2004). In its validation sample, discrimination was AUC 0.823; the reported standardized mortality ratio was 1.018 (95% CI 0.996–1.040).
How the probability is calculated
The calculator applies the published regression coefficients, including seven age-by-risk-factor interaction terms and the “zero factors” and “full code” terms.
Logit: β₀ + Σ(βᵢ × variableᵢ) + Σ(βage×i × age × variableᵢ).
Probability: 1 / (1 + e−logit) × 100.
Important operational note
The source table labels coma/deep stupor as GCS 3 or 4, while its Appendix A narrative describes coma as GCS 3 and deep stupor as GCS 4 or 5. This calculator follows the Appendix A clinical description (coma/deep stupor) rather than imposing a single GCS cutoff; users should apply the original clinical definition carefully.
Applicability and limitations
- Intended for adult ICU admissions; the development study excluded patients younger than 18 years, cardiac surgery, acute myocardial infarction, burns, and subsequent ICU readmissions during the same hospitalization.
- The model was developed predominantly in North American ICUs. External performance in other healthcare systems, including Indonesia, requires local validation or recalibration.
- MPM0-III estimates mortality at hospital discharge; it is not a 30-day mortality estimate and is not MPM24.
- The authors intended the model for groups and risk-adjusted ICU evaluation, not precise prediction for an individual patient or as a basis for treatment limitation or ICU admission decisions.
- Rapidly changing admission condition and specialty case-mix may reduce applicability. Cardiac surgery and acute myocardial infarction populations were excluded.
Model variables
| Variable | Regression coefficient |
|---|---|
| Constant | −5.362830 |
| Age (per year) | 0.0385582 |
| Coma / deep stupor | 2.050514 |
| Heart rate ≥150/min | 0.433188 |
| SBP ≤90 mmHg | 1.451005 |
| Chronic renal insufficiency | 0.5395209 |
| Cirrhosis | 2.070695 |
| Metastatic neoplasm | 3.204902 |
| Acute renal failure | 0.8412274 |
| Cardiac dysrhythmia | 0.8219612 |
| Cerebrovascular incident | 0.4107686 |
| Gastrointestinal bleed | −0.165253 |
| Intracranial mass effect | 1.855276 |
| CPR before admission | 1.497258 |
| Mechanical ventilation | 0.821648 |
| Medical/unscheduled surgical admission | 0.9097936 |
| Zero factors | −0.4243604 |
| Full code | −0.7969783 |
The seven interaction coefficients are applied in the calculation: age×coma −0.0075284; age×SBP<90 −0.0085197; age×cirrhosis −0.0224333; age×metastatic neoplasm −0.0330237; age×dysrhythmia −0.0101286; age×intracranial mass effect −0.0169215; age×CPR −0.011214.
References
Clinical notes
Document key findings and keep a structured record for your own reference.
My notes
Review, edit, or remove notes saved in this browser.
Report an issue
Your feedback helps us improve the reliability and usability of this clinical tool.