Sequential organ failure assessment score is an excellent operationalization of disease severity of adult patients with hospitalized community acquired pneumonia - results from the prospective observational PROGRESS study

Abstract

Background: CAP (Community acquired pneumonia) is frequent, with a high mortality rate and a high burden on health care systems. Development of predictive biomarkers, new therapeutic concepts, and epidemiologic research require a valid, reproducible, and quantitative measure describing CAP severity.

Methods: Using time series data of 1532 patients enrolled in the PROGRESS study, we compared putative measures of CAP severity for their utility as an operationalization. Comparison was based on ability to correctly identify patients with an objectively severe state of disease (death or need for intensive care with at least one of the following: substantial respiratory support, treatment with catecholamines, or dialysis). We considered IDSA/ATS minor criteria, CRB-65, CURB-65, Halm criteria, qSOFA, PSI, SCAP, SIRS-Score, SMART-COP, and SOFA.

Results: SOFA significantly outperformed other scores in correctly identifying a severe state of disease at the day of enrollment (AUC = 0.948), mainly caused by higher discriminative power at higher score values. Runners-up were the sum of IDSA/ATS minor criteria (AUC = 0.916) and SCAP (AUC = 0.868). SOFA performed similarly well on subsequent study days (all AUC > 0.9) and across age groups. In univariate and multivariate analysis, age, sex, and pack-years significantly contributed to higher SOFA values whereas antibiosis before hospitalization predicted lower SOFA.

Conclusions: SOFA score can serve as an excellent operationalization of CAP severity and is proposed as endpoint for biomarker and therapeutic studies.

Trial registration: clinicaltrials.gov NCT02782013 , May 25, 2016, retrospectively registered.

Keywords: Biomarker; Clinical epidemiology; Infectious disease; Lung disease; Prospective clinical study; Severity score.

Fig. 1

Study sites, hospitals in Germany (N = 54) and Austria (N = 2) participating in the study (Size and color of circles indicate the number of patients collected by the corresponding site)

Fig. 2

Flow chart of study procedures

Fig. 3

Primary Endpoint (PE) in the PROGRESS study. a Distribution of observed PE states across study events (adm = admission to hospital, d0 = enrollment, d1 = study visit 1, d2 = study visit 2, d3 = study visit 3, d4 = study visit 4, d4+ = time between study visit 4 and 28d follow-up. b Frequencies of specific treatments qualifying for the PE

Fig. 4

Performance of scores regarding PE prediction. a Receiver operating characteristics for severity scores at enrollment. b Percentage of patients with PE in dependence on severity scores: Severity scores were rescaled to the unit interval for this purpose. For SOFA we pooled scores > 10, for Halm > 5 and for SMART-COP > 8 in order to deal with sparsely filled score classes. For SCAP, quintiles were used

Fig. 5

Time series data of SOFA. a ROC analysis of SOFA for different study time points: Diagnostic power is similar for all time points; b Contribution of SOFA sub-scores for day of enrollment and study visits. As expected, the pulmonary SOFA sub-score has the largest impact, which even increases during the course of therapy. Displayed numbers refer to the percentage of the pulmonary SOFA sub-score