How Pneumonia Severity Scores Help Decide Who Needs the Hospital

When a person arrives with pneumonia, the first question is not which antibiotic but where care should happen: home, a hospital ward, or the intensive care unit. Two tools shape that decision, the Pneumonia Severity Index (PSI) and CURB-65, and both were built to predict one thing, the risk of death within 30 days. The 2019 ATS/IDSA guideline recommends using a validated rule of this kind, preferentially the PSI, yet it is explicit that a low predicted mortality is not the same as a safe discharge. A prognosis and a triage decision are related questions, not identical ones.

What the two scores actually measure

The Pneumonia Severity Index, published by Fine and colleagues in 1997, sorts adults into five risk classes using about twenty variables: age, sex, nursing-home residence, coexisting illnesses such as cancer, liver disease, heart failure, and kidney disease, plus vital signs and laboratory values including blood pressure, respiratory rate, oxygenation, blood urea nitrogen, sodium, and arterial pH. Classes I through III carry a low 30-day mortality, class IV an intermediate risk, and class V the highest. The rule was derived from more than fourteen thousand inpatients and validated in tens of thousands more, which is why the guideline treats it as a reference standard.

CURB-65, derived by Lim and colleagues in 2003, trades detail for speed. It awards one point each for Confusion, elevated blood Urea, Respiratory rate of 30 or more, low Blood pressure, and age 65 or older. A score of 0 or 1 marks a low-risk group, 2 an intermediate group, and 3 to 5 a high-risk group. Because a clinician can total it at the bedside without waiting for a full metabolic panel, it is popular in emergency departments, and a further-simplified version, CRB-65, drops the urea entirely.

Why a mortality prediction is not a triage decision

Here is the distinction that gets lost. Both scores were engineered to answer one question: how likely is this patient to be dead in 30 days? That is a prognostic question. The decision a clinician actually faces at the front door is different: is it safe to treat this person at home, or does something about their condition require the resources of a ward or an ICU? Those questions overlap, but they are not the same, and the gap between them is where judgment lives.

Consider a healthy 30-year-old with pneumonia and an oxygen saturation of 88 percent. Age drives so much of the PSI that this patient can land in a low-risk class, yet hypoxemia of that degree usually needs supplemental oxygen and monitoring that home cannot provide. The score sees a low probability of death; the clinician sees a person who needs the hospital today. The reverse also happens. An older adult with a frightening-looking score may be stable, improving, and better served at home with support than exposed to the delirium and hospital-acquired infection risk of an admission.

Death is only one of several reasons to admit. Inability to keep down oral antibiotics or fluids, uncontrolled pain, a decompensating chronic illness, no one at home to help, or unstable blood pressure can each justify hospitalization regardless of what a low class or a CURB-65 of 1 suggests. A score cannot see any of that, which is why treating it as the whole answer is a category error.

What the 2019 ATS/IDSA guideline actually frames

The 2019 community-acquired pneumonia guideline from the American Thoracic Society and the Infectious Diseases Society of America gives a specific, and often misquoted, instruction. It makes a strong recommendation to determine the site of care using a validated clinical prediction rule for prognosis, preferentially the PSI over CURB-65, on moderate-quality evidence. The stated reason is that, compared with CURB-65, the PSI identifies a larger share of patients as low risk and has higher discriminative power for predicting mortality.

Two qualifiers in that recommendation matter as much as the recommendation itself. First, the guideline says the rule should be used together with clinical judgment, not in place of it, precisely because a prognostic score can miss the admission drivers described above. Second, the panel recommends starting empiric antibiotics in adults with confirmed pneumonia regardless of the initial serum procalcitonin level, so that biomarker is not the tool for deciding whether to treat. For the sickest patients, the guideline points to a separate set of criteria for severe pneumonia and direct ICU admission, including septic shock requiring vasopressors and respiratory failure requiring mechanical ventilation, rather than to a PSI class or a CURB-65 total.

How to read a score without over-trusting it

The most defensible way to use these tools is as a floor, not a verdict. A low score is permission to consider outpatient care, not a command to discharge, and a clinician can and should override it upward when oxygenation, oral intake, comorbidity, or circumstances demand. The scores are least reliable at exactly the edges where they are used most, which is why the guideline pairs them with judgment rather than replacing it. This article is educational and is not medical advice; decisions about pneumonia care belong to a patient and the clinician evaluating them in person.