Shorr Score for MRSA Pneumonia

Understand the Shorr MRSA pneumonia risk score (0–10): predictors, point table, low/intermediate/high risk bands, test characteristics, stewardship context, and bedside limitations for empiric anti-MRSA decisions.

Shorr score for MRSA pneumonia

Bedside integer score (0–10) from Shorr et al. (BMC Infect Dis 2013) for adults presenting with pneumonia who have bacterial infection confirmed by cultures or urinary antigen—informs empiric anti-MRSA decisions alongside guidelines and local data.

Age (years)

Award 1 point if age is under 30 or over 79 (derivation cohort was adults ≥18).

Sex

FemaleMale

Criteria (check all that apply)

Recent hospitalization (≥2 days within the last 90 days) (+2 points)ICU admission on or before index culture (+2 points)Nursing home, SNF, or LTAC exposure within the last 90 days (+1 point)Prior IV antibiotic therapy within the last 30 days (+1 point)Cerebrovascular disease (any), prior to admission (+1 point)Dementia (+1 point)Diabetes mellitus (+1 point only if patient is female, per derivation model)

Disclaimer: Educational decision support only. Antibiotic choices must follow current guidelines, allergies, renal function, microbiology, and institutional protocols. The Shorr score does not replace clinical judgment.

Shorr score — explained

Purpose

The Shorr score estimates the probability that MRSA pneumonia among patients admitted with pneumonia who have documented bacterial evidence (cultures and/or urinary antigen for S. pneumoniae or Legionella) within 48 hours of admission. It was developed to refine the use of empiric anti-MRSA therapy beyond broad “HCAP” labeling.

Scoring (Table 2, Shorr et al. 2013)

Sum integer points (maximum 10):

Variable	Points
Age <30 or >79 years	1
Recent hospitalization (≥2 days in prior 90 days)	2
Nursing home, SNF, or LTAC in prior 90 days	1
Prior IV antibiotics in prior 30 days	1
ICU admission on or before index culture	2
Cerebrovascular disease (any), prior to admission	1
Dementia	1
Female with diabetes mellitus	1

Risk strata (published cohort)

0–1: MRSA prevalence below about 10%; many patients may qualify to avoid empiric anti-MRSA coverage when aligned with stewardship and diagnostics.
2–5: intermediate prevalence band—individualize coverage and diagnostics.
≥6: MRSA prevalence above about 30%; empiric anti-MRSA therapy is commonly considered.

At a threshold of score ≤1 vs >1, reported sensitivity/specificity were ~59%/60%, with negative predictive value ~90% in the 14% baseline MRSA cohort.

Limitations

Retrospective, ICD-9–coded, culture-enriched population—may not generalize to all ED pneumonia presentations.
Moderate discrimination (AUROC ~0.64–0.66); complement with PCR, local antibiograms, and clinical gestalt.
Does not address Gram-negative resistance or non-MRSA pathogens.

Reference

Shorr AF, Myers DE, Huang DB, Nathanson BH, Emons MF, Kollef MH. A risk score for identifying methicillin-resistant Staphylococcus aureus in patients presenting to the hospital with pneumonia. BMC Infect Dis. 2013;13:268. doi:10.1186/1471-2334-13-268

Clinical background

Methicillin-resistant Staphylococcus aureus (MRSA) is an important cause of pneumonia among adults who are hospitalized after presenting from the community. Over the same era, many patients historically met criteria for healthcare-associated pneumonia (HCAP), which was intended to flag risk for drug-resistant organisms. In practice, broad application of HCAP led to widespread empiric anti-MRSA coverage—including in patients with a relatively low likelihood of MRSA—contributing to antibiotic overuse, adverse drug effects, higher costs, and selection pressure without proportional benefit.

The Shorr score is a simple bedside integer rule derived specifically for MRSA among patients with hospital-presenting pneumonia and laboratory evidence of bacterial infection. It does not replace guideline-based care or local antibiograms, but it offers a structured way to stratify risk when clinicians are deciding whether to start or withhold empiric therapy active against MRSA while Gram-negative coverage is addressed separately.

Derivation cohort and scope

The score was developed from a large, multi-hospital, retrospectively identified adult cohort in the United States. Eligible patients had pneumonia on hospital presentation with microbiological confirmation (for example, positive cultures from blood, respiratory secretions, or pleural fluid, and/or urinary antigen testing for select pathogens) obtained within a defined early window after admission. Patients with purely nosocomial pneumonia (such as classic hospital-acquired or ventilator-associated pneumonia by timing definitions) were excluded, reflecting the tool’s intent for community-onset presentations rather than late ICU pneumonias.

Because enrichment with cultures and antigens changes the pre-test probability of identified pathogens compared with all-comers pneumonia, clinicians should interpret the absolute percentages reported in the original paper as most representative of a culture-positive population rather than every emergency department patient with suspected pneumonia.

Score components (0–10 points)

Eight independent predictors are summed. Points were chosen from standardized logistic regression coefficients and rounded to whole numbers for ease of use. Two variables receive two points each; the remainder receive one point when present. The maximum total is 10.

Predictor	Points
Age under 30 years or over 79 years	1
Recent hospitalization (inpatient stay of at least two days within the prior 90 days)	2
Exposure to a nursing home, skilled nursing facility, or long-term acute care hospital within the prior 90 days	1
Prior intravenous antibiotic therapy within the prior 30 days	1
Intensive care unit admission on or before the index culture (a measure of acute severity in the derivation dataset)	2
Cerebrovascular disease of any type documented prior to admission	1
Dementia	1
Female sex with diabetes mellitus (interaction captured as a single risk item in the final model)	1

In clinical documentation, “recent hospitalization” and “ICU level of care” are distinct triggers in the published table; both can be present in the same patient and both contribute their respective point values. Likewise, residential or facility exposure and recent antibiotics are separate healthcare-contact domains and are scored independently when each applies.

Risk bands and epidemiologic anchors

In the published risk strata, total scores from 0 to 1 identified a low-risk group in whom MRSA prevalence fell below roughly ten percent. Scores from 2 to 5 formed an intermediate zone with intermediate MRSA prevalence. Scores of 6 or higher identified a high-risk band in whom MRSA prevalence rose above roughly thirty percent.

These cut points were illustrated consistently across development and internal validation splits, supporting modest generalizability within the study’s design, though external performance will vary by geography, infection prevention practices, colonization pressure, and the aggressiveness of microbiologic testing.

Test characteristics and how to think about negative results

Evaluated dichotomously using a cutoff of scores ≤1 versus >1, the tool had moderate sensitivity and specificity in the derivation/validation reporting, with a negative predictive value that was clinically meaningful in the cohort’s baseline MRSA prevalence. Interpreting negative predictive value requires anchoring to local prevalence: in lower-incidence settings, a “low score” is even more reassuring for ruling out MRSA as a dominant bacterial pathogen, whereas in hyperendemic contexts even low-risk patients may merit expanded diagnostics or narrower but thoughtfully chosen coverage.

Discrimination by the area under the receiver operating characteristic curve was in the mid-sixties—appropriate for a parsimonious eight-variable rule, but clearly not a definitive diagnostic test. Rapid diagnostics such as MRSA nares polymerase chain reaction, lower respiratory samples when feasible, and procalcitonin or other adjuncts (used per local protocol) can refine decisions beyond the score alone.

Practical integration at the bedside

Severity and sepsis pathways may mandate broad therapy or ICU-level monitoring independent of MRSA probability; the score informs anti-MRSA choice, not resuscitation or source control.
Antimicrobial stewardship teams often pair scores like this with automatic pharmacy review, de-escalation triggers when cultures return, and preauthorization criteria for prolonged glycopeptide or lipopeptide courses.
Risk of drug toxicity—nephrotoxicity, infusion-related reactions, drug–drug interactions, and therapeutic drug monitoring burden—provides additional reason to avoid empiric MRSA therapy in well-defined low-risk patients when doing so aligns with guidelines and shared decision-making.
Special populations, including solid organ transplant recipients, prolonged corticosteroid or other immunosuppression, cystic fibrosis, or recent known MRSA infection or colonization, may reasonably fall outside the score’s assumptions even when the raw point total is low.

Limitations and cautions

Retrospective coding of diagnoses and comorbidities, incomplete capture of prior outpatient intravenous antibiotics in administrative data, and colonization mistaken for infection on respiratory cultures can skew apparent performance. Extremes of age, frailty syndromes, dementia, and cerebrovascular disease may function partly as proxies for unmeasured healthcare contact or aspiration risk. Finally, pneumococcal vaccination status, viral co-infection, and institutional outbreak dynamics are not embedded in the original score and may materially change MRSA probability in real time.

Users should treat the Shorr score as decision support: it summarizes epidemiology from a defined patient set, improves transparency when discussing why anti-MRSA therapy is started or held, and should always be reconciled with the individual patient’s trajectory, allied markers of shock, imaging, oxygenation, and institutional pathways for pneumonia management.