Development and validation of a risk score to predict community-acquired pneumonia occurrence in the adult population

Abstract

Background

Community-acquired pneumonia (CAP) preventive strategies can benefit from a quantification of individual CAP risk. This study develops and validates a CAP Risk Score (CAP-RS) for the adult population to predict CAP occurrence in the next five years.

Methods

The development phase was as follows: a population-based case-control study to identify potential CAP risk factors for inclusion in the CAP-RS after weighting according to odds ratios; development of a numerical scoring system for weighted risk factors; and establishment of cut-off points to discriminate between different risk levels. The validation phase consisted of a population-based case-control study and a retrospective cohort study (with 47 836 adults aged ≥18 years corresponding to three Maresme (Barcelona) primary care centres) followed up over a five-year period (2015–2019).

Results

786 new CAP cases were identified. 15 factors were included in the CAP-RS. Risk was higher in subjects with CAP than without CAP (4.5 vs 1.9; p < 0.001), and the association (OR) between the CAP-RS and the occurrence of CAP increased as the CAP-RS value increased. AUC-ROC was 0.67 (p < 0.001). Cut-offs were established at <1, <5, and <10 points as best discriminating between risk groups. Annual CAP incidence was 1.9, 3.1, 6.2, and 12.4 new cases/10³ inhabitants for the no, moderately, severely, and very severely increased risk groups, respectively. Significant differences in CAP-free survival were observed between the four CAR-RS categories.

Conclusions

The 15-item CAP-RS, which stratifies risk with good validity, can aid in the design and implementation of preventive CAP strategies for adult populations.