Highlight
- The rising incidence of Legionnaires’ disease necessitates improved diagnostic tools for community-acquired pneumonia (CAP).
- The Fiumefreddo Score for Ruling Out Legionella Pneumonia showed high sensitivity but limited specificity and practical challenges due to infrequent LDH measurement.
- The updated SwissLEGIO score simplifies predictors to routinely available variables while improving specificity without compromising sensitivity.
- Implementing the SwissLEGIO score could reduce unnecessary Legionella-specific microbiologic testing by up to 52%, optimizing resource utilization and patient care.
Study Background
Legionnaires’ disease (LD), a severe form of pneumonia caused by Legionella bacteria, accounts for an increasing proportion of community-acquired pneumonia (CAP) cases reported in Europe and the United States. Accurate and timely diagnosis is crucial for guiding appropriate antimicrobial therapy and instituting public health measures to contain outbreaks. However, Legionella diagnosis remains challenging due to variable clinical presentations and limitations in microbiological testing availability and costs. This has prompted efforts to develop clinical prediction tools to identify patients at elevated risk of LD, who warrant targeted diagnostic testing.
Among these, a Fiumefreddo Score for Ruling Out Legionella Pneumonia integrating clinical and laboratory variables has been previously proposed, encompassing fever, absence or dry cough, hyponatremia, elevated C-reactive protein (CRP), elevated lactate dehydrogenase (LDH), and thrombocytopenia. Yet, its external validation and applicability in routine clinical workflows have been uncertain, particularly concerning the availability of some biomarkers like LDH.
Study Design
This multicenter study by Bigler et al., conducted within the SwissLEGIO Hospital Network, aimed to externally validate and refine the aforementioned prediction score. The investigators analyzed data collected prospectively from August 2022 to March 2024, including 196 confirmed LD cases and 196 matched Legionella test-negative CAP controls across 20 Swiss hospitals.
Patients were matched based on factors such as demographics and clinical settings. The availability and predictive performance of the original six score variables were assessed in routine care contexts. To enhance usability and diagnostic performance, the dataset was split into development and validation cohorts, facilitating model simplification and recalibration.
Key Findings
The original six-predictor score demonstrated high sensitivity (91%; 95% confidence interval [CI]: 86–96%), which is critical for ruling out LD and minimizing missed cases. However, its specificity was limited at 35% (95% CI: 28–42%), leading to potential over-testing. Notably, LDH—a laboratory marker of tissue damage—was measured infrequently in routine practice, reducing the original score’s feasibility. Moreover, platelet count, reflecting thrombocytopenia, showed poor discriminatory capability.
In response, the investigators developed the simplified SwissLEGIO score employing five predictors: fever above 38°C, serum sodium below 133 mmol/L (hyponatremia), CRP exceeding 180 mg/L, absence or dry cough, and prior treatment with β-lactam antibiotics. This streamlined model maintained high sensitivity between 88% and 92% and enhanced specificity to a range of 46% to 58% at a cut-off score of two or more.
This improvement translates into clinically meaningful implications. In settings with a presumed Legionella prevalence of 4% among CAP presentations, applying the SwissLEGIO score to rule out LD in patients scoring less than two could reduce Legionella-specific microbiologic testing by approximately 36% to 52%. Thus, the tool offers a pragmatic balance between sensitivity and specificity, optimizing diagnostic stewardship.
Expert Commentary
The refinement of clinical prediction tools for LD diagnosis addresses a vital need to reduce unnecessary antibiotic use and laboratory testing while preserving patient safety. The SwissLEGIO score’s reliance on commonly available clinical and laboratory data enhances its real-world feasibility. Its improved specificity compared to the original model may alleviate healthcare burdens, especially in resource-constrained environments.
Nonetheless, some limitations warrant consideration. The study population was exclusively Swiss, potentially affecting generalizability. Variations in microbiological testing practices and Legionella prevalence globally may influence predictive performance. Furthermore, while the score guides when to conduct Legionella testing, it does not replace definitive microbiological diagnosis and clinical judgment.
Future research could validate the SwissLEGIO score in diverse geographic settings and integrate it within electronic health records to facilitate rapid, automated decision support. Additionally, exploring the score’s utility in outpatient settings remains an open question.
Conclusion
The SwissLEGIO score represents a clinically valuable, user-friendly screening tool for ruling out Legionella infection in hospitalized CAP patients. By focusing on five easily obtainable predictors, it preserves high sensitivity while improving specificity, thus enabling targeted microbiological testing. Adoption of the SwissLEGIO score in clinical workflows could streamline diagnostics, reduce unnecessary laboratory investigations, and enhance patient management strategies amid the rising burden of Legionnaires’ disease.
Funding
The study was conducted by the SwissLEGIO Hospital Network with no reported external funding sources.
References
Bigler M, Dräger S, Zacher F, Hattendorf J, Mäusezahl D, Albrich WC; SwissLEGIO Hospital Network. Multicenter validation and update of a Legionella prediction score to guide microbiological testing in community-acquired pneumonia. Int J Infect Dis. 2026 Sep;170:108955. doi: 10.1016/j.ijid.2026.108955. Epub 2026 Jul 3. PMID: 42398699.
