The Long-term Pneumonia Mortality Index. An International Multicenter Derivation and Validation Study for Patients with Community-Acquired Pneumonia

Rationale Long-term mortality after hospital discharge in community-acquired pneumonia (CAP) is a neglected issue. A validated, accurate score predicting mortality is needed. Objective To derive and validate the Long-term Pneumonia Mortality Index (L-PMI). Methods L-PMI derivation used data from a Spanish multicenter prospective study on CAP with a 1-year follow-up. A logistic regression model for predicting 1-year mortality risk was developed. Performance of the model was assessed by estimating the area under the receiver operator characteristic curve. The model was externally validated using two independent multicenter CAP cohorts from Germany and the USA, and one COVID-19 pneumonia cohort from Spain. We then derived a score from the model to ease its application. Measurements and Main Results The long-term mortality rate after discharge was 6.3, 4.4, 17.4 and 3.6% in the derivation, CAPNETZ (only 6-months follow-up), USA and COVID-19 cohorts, respectively. The L-PMI included: age, smoking history, nursing home resident, Charlson comorbidity index, CURB-65 score, use of non-invasive or invasive mechanical ventilation, and in-hospital cardiovascular events. The L-PMI showed high discrimination for mortality in the derivation cohort: AUC 0.82 (95% confidence interval [0.78, 0.85]). In the validation cohorts, the AUC ranged from 0.78 (0.73, 0.83) to 0.75 (0.73, 0.77) for CAPNETZ and USA cohorts, respectively, and 0.88 (0.84, 0.93) for COVID-19 cohort. L-PMI allows classification of patients according to their mortality risk into low, intermediate, and high-risk groups. Conclusions The L-PMI is a novel clinical prediction score that identifies pneumonia patients at risk of mortality up to 1 year after discharge.