Critical Care Medicine: ICU Utilization

Recognizing the Critically Ill Patient

There are no commonly accepted criteria for admission to the intensive care unit (ICU). In practice, clinicians identify patients who require intense resources or are at risk for deterioration by recognizing signs and symptoms of clinical instability — hypotension, hypoxemia, arrhythmias, and changes in mental status. Initial resuscitation is often fast-paced and occurs at the extremes of physiologic deterioration, before a detailed picture of the patient’s chronic medical problems is available, so ongoing bedside assessment remains essential even as monitoring technology refines the evaluation.

Several scoring systems — the Acute Physiology and Chronic Health Evaluation (APACHE), the Sequential Organ Failure Assessment (SOFA), and the Simplified Acute Physiology Score (SAPS) — were designed to classify the severity of illness of patients admitted to the ICU. They combine vital signs with risk factors such as chronic disease, emergent surgery, and immunosuppression to estimate mortality risk. Although these scores are rarely, if ever, used as admission criteria, they are used to compare disease severity, track progression, and benchmark outcomes. Disease-specific scoring systems can also triage patients on the basis of their risk for deterioration or death — for example, the simplified Pulmonary Embolism Severity Index for pulmonary embolism, the quick SOFA (qSOFA) for sepsis, and the Pneumonia Severity Index or CURB-65 for predicting mortality in community-acquired pneumonia.

Grading Severity of Illness: SOFA, qSOFA, and APACHE

The SOFA score grades dysfunction in six organ systems (Table 1), each scored from 0 (normal) to 4 (severe dysfunction). Because it can be measured repeatedly — typically daily — a rising SOFA score tracks worsening organ failure and correlates with increasing mortality; under the current sepsis consensus definition, an acute increase of 2 or more points from baseline in the setting of suspected or documented infection identifies sepsis.

A simplified bedside variant, the quick SOFA (qSOFA), identifies patients with suspected infection who are at risk for poor outcomes when at least two of three criteria are present: respiratory rate of 22/min or higher, altered mental status, or systolic blood pressure of 100 mm Hg or lower. qSOFA was developed to screen for risk of poor outcomes from sepsis in out-of-hospital, emergency department, and hospital ward settings — it was not developed, and is not currently recommended, as a diagnostic screening tool for sepsis itself.

APACHE II remains the most commonly used severity-of-illness scoring system in North America. It integrates age, type of ICU admission (after elective surgery versus nonsurgical or after emergency surgery), chronic health problems, and 12 physiologic variables — the worst values for each during the first 24 hours after admission — to derive a predicted hospital mortality rate; updated versions (APACHE III and APACHE IV) have also been published.

Electronic medical record databases are increasingly being harnessed to provide warning and predictive systems built from routinely collected data — vital signs, age, and current or trending clinical information — to identify patients at risk for deterioration. These predictive models are evolving with the application of machine learning and artificial intelligence, and there is growing interest in real-time EHR-derived scores that alert providers to patients at risk for sepsis or poor outcomes; yet their application remains at local levels and is not widespread.

ICU Triage and Admission Decisions

Determining admission to a critical care bed is more complicated than simply assessing how sick a patient is: it is determined in a significant way by hospital and unit policies and by the resources unique to each institution. No standard admission model exists, although international societies have published guidelines, and hospitals have developed institution-specific protocols to better manage patients who require life-support technology.

Allocation decisions also intersect with goals of care. The Task Force on Ethics of the Society of Critical Care Medicine has judged it ethically sound to withhold or withdraw life-sustaining care when a patient or the patient’s surrogate requests it, or when the physician judges that the goals of therapy are not achievable; withholding and withdrawing are considered fundamentally similar acts. An informed patient’s wishes regarding life-sustaining therapy should be respected, and physicians are not ethically obligated to provide treatment they deem futile — although arrangements may be made to transfer the patient’s care to another provider. When withholding or withdrawal of care is being considered, critical care providers should meet regularly with the patient or surrogate to discuss prognosis, and an independent hospital ethics service can help navigate complex decisions.

Rapid Response Systems

To improve early management and resuscitation of patients who are deteriorating or at risk for deterioration, many health care systems have developed rapid response teams (RRTs) or medical emergency response teams. Their aims are to recognize deteriorating patients promptly, trigger early evaluation and management, and move the patient to a higher level of care. A key element of the system is the ability of any member of the health care team — or a family member — to trigger the team. Systematic reviews and meta-analyses demonstrate a variable decrease in cardiac and respiratory arrests, unexpected deterioration, and mortality in adults.

Organization of Critical Care

An ICU is equipped with technology that allows continuous patient monitoring and delivery of life-sustaining interventions such as mechanical ventilation and emergency extracorporeal organ support. Hospital units designated as ICUs typically have the sickest patients and require nurse-to-patient ratios of 1:1 or 1:2, whereas a progressive care unit (also called an intermediate, transitional, or step-down unit) may have a nurse-to-patient ratio as high as 1:5, reflecting patients with lower acuity.

Continuous monitoring pairs intermittent “gold-standard” measurements with noninvasive surrogates. Arterial blood gas analysis measures pH, PaO₂, and PaCO₂ directly but is episodic and painful, whereas pulse oximetry — which exploits the differing light-absorption properties of oxyhemoglobin and deoxyhemoglobin — provides a continuous estimate of arterial oxygen saturation. One caveat: pulse oximetry can overestimate oxygen saturation in patients with darker skin, so correlation with arterial blood gas is essential whenever precise oxygenation matters.

Critical care units are generally described along two axes — open versus closed, and low-intensity versus high-intensity staffing (Table 2).

In an open unit, patients are managed by their primary hospital team, which may or may not include a critical care consultant; patients in a closed unit are managed primarily by the critical care team. Low-intensity units are open units. Although high-intensity units can be open or closed, the critical care team is present throughout the day, providing consultation. Data supporting high-intensity and closed models with continuous staffing of ICUs by intensivists have been conflicting; recent data, however, show a mortality benefit with the closed ICU.

The nature of the intensive care unit requires an interprofessional and multidisciplinary team. The key features that allow the best team dynamics are explicit recognition of the roles and responsibilities of team members, the use of shared communication models, and an environment of psychological safety.