RESUMEN
Long-term outcome prediction in COPD is challenging. We conducted a prospective 5-7-year follow-up study in patients with COPD to determine the association of exacerbation type, discharge levels of inflammatory biomarkers including procalctionin (PCT), C-reactive protein (CRP), white blood cell count (WBC) and plasma proadrenomedullin (ProADM), alone or combined with demographic/clinical characteristics, with long-term all-cause mortality in the COPD setting. The analyzed cohort comprised 469 patients with index hospitalization for pneumonic (n = 252) or non-pneumonic (n = 217) COPD exacerbation. Five-to-seven-year vital status was ascertained via structured phone interviews with patients or their household members/primary care physicians. We investigated predictive accuracy using univariate and multivariate Cox regression models and area under the receiver operating characteristic curve (AUC). After a median [25th-75th percentile] 6.1 [5.6-6.5] years, mortality was 55% (95%CI 50%-59%). Discharge ProADM concentration was strongly associated with 5-7-year non-survival: adjusted hazard ratio (HR)/10-fold increase (95%CI) 10.4 (6.2-17.7). Weaker associations were found for PCT and no significant associations were found for CRP or WBC. Combining ProADM with demographic/clinical variables including age, smoking status, BMI, New York Heart Association dyspnea class, exacerbation type, and comorbidities significantly improved long-term predictive accuracy over that of the demographic/clinical model alone: AUC (95%CI) 0.745 (0.701-0.789) versus 0.727 (0.681-0.772), (p) = .043. In patients hospitalized for COPD exacerbation, discharge ProADM levels appeared to accurately predict 5-7-year all-cause mortality and to improve long-term prognostic accuracy of multidimensional demographic/clinical mortality risk assessment.