Pneumologie 2018; 72(S 01): S8
DOI: 10.1055/s-0037-1619137
Sektion 4 – Infektiologie und Tuberkulose
Freie Vorträge – Titel: Neue und praxisrelevante Forschungsergebnisse aus der pneumologischen Infektiologie
Georg Thieme Verlag KG Stuttgart · New York

Prediction of 28 day mortality and need for ICU treatment in the PROGRESS Study by CAP severity scores

P Creutz
1   Med. Klinik mit Schwerpunkt Infektiologie und Pneumologie, Charité-Universitätsmedizin Berlin; Campus Benjamin Franklin
,
P Ahnert
2   Institut für Medizinische Informatik, Statistik und Epidemiologie, Mediziniche Fakultät, Universität Leipzig
,
H Kirsten
2   Institut für Medizinische Informatik, Statistik und Epidemiologie, Mediziniche Fakultät, Universität Leipzig
,
M Kiehntopf
3   Jena University Hospital
,
M Bauer
3   Jena University Hospital
,
U Völker
4   University Medicine Greifswald
,
T Chakraborty
5   University Hospital Gießen
,
M Scholz
6   Universität Leipzig
,
M Löffler
6   Universität Leipzig
,
N Suttorp
7   Charité-Universitätsmedizin Berlin
› Author Affiliations
Further Information

Publication History

Publication Date:
21 February 2018 (online)

Also available at
 

Purpose:

Community acquired pneumonia (CAP) is a frequent infectious disease worldwide. Optimized risk stratification during course of disease could improve management of CAP patients and reduce mortality levels: Understanding prognosis can assist physicians in the initial site of treatment decision – home vs. hospital or intensive care unit and can be used to estimate and communicate expected outcomes to patients during course of disease. For scientific work standardized risk stratification of CAP is relevant for selecting patient subgroups for clinical trials.

Patients and methods:

With time series data of 1,532 patients enrolled in the multi-center PROGRESS study, we compared CAP severity scores regarding their performance in correctly predicting a subsequent severe course of CAP (death or need for intensive care) or death within 28 days alone. Considered scores were CRB-65, CURB-65, PSI, SIRS-Score, SOFA, qSOFA, Halm criteria, SCAP, SMART-COP, and the two laboratory parameters CRP and PCT. Comparisons were performed by receiver operating characteristics or precision recall analysis using R. PROGRESS Patients were younger than the overall German population of hospitalized CAP patients (median of 59 vs. 73 years) with lower in-hospital mortality (2.3% vs. 13.9%).

Results:

In our PROGRESS data, several scores obtained at the time of hospitalization were able to predict 28 day mortality with AUCs > 80% (SOFA, PSI, SCAP, CURB-65, SMART-COP, decreasing AUC) with SOFA reaching 87% closely followed by PSI. For prediction of need for later intensive care treatment, the situation is similar but highest AUC is lower with 79% (SOFA, SCAP, SMART-COP, CURB-65, Halm, PSI, decreasing AUC). PCT and CRP did not perform well for both endpoints with AUCs < 60%.

Conclusion:

SOFA and closely related scores are suitable for predicting 28 day mortality and need for ICU admission in the PROGRESS cohort, but with limited specificity, especially in the severity range that is characterized by SOFA scores from 5 to 10. Here, biomarkers further improving sensitivity and specificity would be highly desirable for clinical decision-making.