APACHE II Score: A Clinical ICU Tool, Not a Public Risk Calculator

What APACHE II Measures

APACHE II stands for Acute Physiology and Chronic Health Evaluation, version II. It is a severity-of-illness scoring system used in adult intensive care units to estimate mortality risk during the first 24 hours of an ICU admission. The score is one of the most widely cited clinical decision-support tools in critical care medicine and has been used since 1985, when it was published by William Knaus and colleagues at George Washington University in Critical Care Medicine. It replaced the original APACHE system from 1981, simplifying the variable count while preserving predictive performance.

The APACHE II score combines twelve physiologic variables, the patient's age, and a chronic health adjustment to produce a single number between 0 and 71. A higher score indicates greater physiologic derangement and a higher predicted mortality. The score is then mapped, via published equations specific to the patient's primary diagnosis, to an estimated probability of in-hospital death. APACHE II is one component of the broader APACHE family, which now also includes APACHE III (1991) and APACHE IV (2006), each of which uses more variables and updated reference data sets. APACHE II remains in widespread use because it is the version most extensively published in the literature and most widely taught.

The intended uses of APACHE II are clinical decision support, research stratification, and quality benchmarking. It allows ICUs to compare their observed mortality with the mortality predicted by the score, adjusting for the severity of illness in their patient mix. It supports research by allowing severity to be controlled across study groups. It can inform discussions about prognosis with families. It does not replace clinical judgment, and it is not used in isolation to make treatment decisions for an individual patient.

Where the Inputs Come From

The twelve physiologic variables in APACHE II are drawn from the worst values recorded during the first 24 hours of ICU admission. They are not a single snapshot. The clinician or data analyst computing the score reviews the entire 24-hour observation period and identifies, for each variable, the value that contributes the most points (typically the value furthest from the normal range, in the direction of clinical concern).

The variables include: temperature (rectal), mean arterial pressure, heart rate, respiratory rate, oxygenation (computed from arterial blood gas as either A-a gradient or partial pressure of oxygen depending on the inspired oxygen fraction), arterial pH, serum sodium, serum potassium, serum creatinine (with a doubled point allocation if there is acute kidney failure), hematocrit, white blood cell count, and Glasgow Coma Scale (where the contribution to the score is 15 minus the GCS value). Most of these inputs are not vital signs the patient or family knows. Mean arterial pressure is calculated from invasive arterial monitoring or from non-invasive blood pressure values via a specific formula. Arterial pH and oxygenation come from a blood gas drawn from an arterial line. Sodium, potassium, creatinine, hematocrit, and white blood count are laboratory results. Glasgow Coma Scale is a structured neurological assessment performed by a trained clinician.

To these physiologic points are added age points (0 points under 44 years, scaling up to 6 points at 75 and older) and a chronic health points adjustment (2 points for elective post-operative admission with a severe organ system insufficiency or immunocompromised state, 5 points for emergency post-operative or non-operative admission with the same conditions). The chronic health adjustment is itself a clinical judgment call requiring documented evidence of severe insufficiency in the liver, cardiovascular system, respiratory system, kidney, or immune system as defined in the original publication.

In modern practice, APACHE II is rarely calculated by hand at the bedside. Most ICUs that use APACHE scoring rely on electronic health record integration, dedicated severity-of-illness software, or commercial ICU information systems that pull values directly from the monitoring and laboratory feeds. Manually computing the score from a paper chart is a quality-improvement or audit activity, not a routine bedside task.

How the Score Is Used in the ICU

APACHE II is part of a broader set of severity scores used in critical care. SOFA (Sequential Organ Failure Assessment) is used to track organ dysfunction over time. SAPS II and SAPS III are alternative severity scores used in many European ICUs. The ICNARC model is commonly used in the United Kingdom. APACHE II is most commonly used at admission, while SOFA is repeated daily to track the trajectory of organ failure.

For a single patient, an APACHE II score and its predicted mortality are used to inform, not dictate, clinical conversations. A patient with a very high score and a high predicted mortality may benefit from an explicit goals-of-care discussion with their family, particularly if they have known chronic comorbidities. The score is one input into that conversation, alongside the patient's known wishes, the trajectory of their physiologic measurements, and the response to treatment.

For an ICU as a whole, the standardized mortality ratio (observed mortality divided by APACHE-predicted mortality) is used as a quality indicator. A unit that consistently outperforms the predicted mortality may be providing better-than-average care. A unit that underperforms may need to investigate processes of care. These comparisons require careful adjustment for case mix and are part of formal quality improvement programs in many countries.

For research, APACHE II scores are used to ensure that intervention and control groups in critical care trials are balanced for severity at baseline. Without this kind of adjustment, a trial could appear to show a benefit of an intervention when the apparent effect was driven by sicker patients ending up in one group.

Limitations of APACHE II

APACHE II is a population-level model, not an individual prognostic test. The mortality probability it produces describes what happened, on average, to similar patients in the development cohort. It is not a forecast that this individual patient will die with that exact probability.

The score was developed and validated in adult ICU patients in the 1980s, primarily in U.S. hospitals. It has well-documented limitations when applied outside that context. It is not validated for pediatric patients — the PRISM (Pediatric Risk of Mortality) and PIM (Pediatric Index of Mortality) scores are used in pediatric intensive care instead. It is not validated for cardiac surgery patients, who have specialized scores (such as EuroSCORE for surgical risk and dedicated cardiac ICU systems). It is not validated for burn patients, who are evaluated with burn-specific scores. It tends to overestimate mortality in some specific patient populations, including patients admitted after diabetic ketoacidosis or drug overdose, where physiologic derangement at presentation is severe but recovery rates are high with appropriate treatment.

The score also does not account for many factors that matter clinically, including the response to early treatment (which is captured by repeated SOFA scores instead), the specific etiology of the illness within the broad diagnostic categories, and care goals and limits of treatment. A patient on comfort-only measures who has a high APACHE II score is in a fundamentally different clinical situation than a patient with the same score who is receiving full ICU support.

Like all scoring systems built from data of a particular era, APACHE II reflects the survival expectations of patients in the 1980s, not in 2026. Improvements in mechanical ventilation, sepsis management, and ICU practice generally have shifted observed mortality downwards for many patient categories. APACHE IV and other contemporary models use more recent reference data for this reason. APACHE II is still useful, but the absolute mortality estimates it produces are best understood as relative severity indicators rather than current calibrated probabilities.

Why Public APACHE Calculators Are Inappropriate

An anonymous "calculate your APACHE II" form aimed at patients or family members has several problems that are rarely acknowledged.

The required inputs are not available outside the ICU. Mean arterial pressure, arterial pH, blood gas values, serum chemistry, white blood cell count, and Glasgow Coma Scale are not values the patient or family typically has. Even when discharge paperwork includes some lab results, the values in the paperwork are not the worst values in the first 24 hours of admission and would not produce a valid APACHE score.

The score requires worst-of-24-hours values. A snapshot reading at one point in time underestimates the score because it misses physiologic excursions earlier or later in the day. Conversely, picking the worst-ever value across a long admission overestimates the score by violating the 24-hour window. Computing APACHE II correctly requires structured chart review, not a single web form.

The output is easily misread as a personal forecast. A predicted mortality of 35 percent is a population statistic, not a 35 percent chance that a specific patient will die. A web tool that returns a percentage figure to a worried family member with no clinical context is far more likely to cause distress than to inform any decision.

Clinical decision support is regulated. Software intended to inform diagnosis or treatment of disease can fall under medical device regulation in many jurisdictions (FDA Software as a Medical Device guidance in the United States, the EU Medical Device Regulation in Europe, similar frameworks in the UK, Canada, and Australia). Free public calculators that present clinical scores as personal risk assessments and that are not framed as educational reference often sit in a regulatory grey area at best, and may be unlawful in their target jurisdictions at worst.

It substitutes a number for the conversation that should happen. Patients and families who want to understand the prognosis of someone in the ICU are far better served by talking with the intensive care team. ICU clinicians can explain the trajectory of the patient's specific illness, the response to treatment so far, and the realistic range of outcomes given the full picture, none of which a generic web form can do.

Legitimate Clinical Resources

For accurate clinical information about APACHE II and other severity-of-illness scores, the following sources are widely used and reviewed by clinicians.

MDCalc hosts APACHE II among hundreds of medical calculators. MDCalc is medically reviewed, cites primary literature for each calculator, and is widely used by clinicians for point-of-care reference. The calculators on MDCalc are framed as clinical decision-support tools for healthcare professionals, with explicit context about validation populations, limitations, and the clinical conditions in which the score is appropriate. The site requires the user to confirm they are a healthcare professional for many of its calculators, which is the appropriate framing for a tool of this kind.

The Society of Critical Care Medicine (SCCM) is the principal professional society for critical care in the United States and publishes Critical Care Medicine, the journal in which APACHE II was originally published. SCCM resources include guidelines, education, and the broader context in which severity scores are used.

The European Society of Intensive Care Medicine (ESICM) plays an analogous role in Europe and publishes Intensive Care Medicine.

The original APACHE II publication remains available: Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Critical Care Medicine. 1985;13(10):818-29. The full text and subsequent commentary are accessible through medical libraries and through PubMed.

The UpToDate clinical reference has detailed entries on ICU severity scores and on the prognostic estimation of critical illness, and is the kind of resource clinicians use at the point of care.

For families seeking to understand a loved one's ICU course, the most useful resources are the conversations with the bedside intensive care team and with the dedicated patient and family liaison services that many ICUs now provide. The SCCM also maintains myICUcare.org, a patient-and-family-facing site that explains common ICU procedures, scores, and care processes in plain language without trying to substitute for clinical evaluation.