American Journal of Cardiology vol:111 issue:8 pages:1198-1208
Studies estimating the added value of natriuretic peptide levels and electrocardiographic findings beyond all relevant clinical information to identify cardiac dysfunction remain scarce. The aim of this study was to assess the presence of clinically relevant cardiac dysfunction in an unselected population of subjects aged ≥80 years. A cross-sectional analysis using an "intention-to-diagnose" strategy was performed within the BELFRAIL study (n = 567). Baseline B-type natriuretic peptide and N-terminal pro-B-type natriuretic peptide levels were determined and echocardiography was performed at subjects' homes. Logistic regression analysis and classification and regression tree analysis were used as complementary analytic tools. Cardiac dysfunction was present in 17% of subjects without and 31% of subjects with chronic atrial fibrillation (AF) or pacemaker. In subjects without chronic AF or pacemaker, the clinical model showed a C-statistic of 0.79 (95% confidence interval 0.74 to 0.85). The combination of natriuretic peptides with normal results on electrocardiography increased, only marginally, the C-statistic. In subjects with chronic AF or pacemaker, the clinical model showed a very high C-statistic of 0.90 (95% confidence interval 0.82 to 0.98). Classification and regression tree analysis showed that an additional 58 subjects (13%) were correctly classified using natriuretic peptides and electrocardiographic findings among those without chronic AF or pacemaker. Of participants with chronic AF or pacemaker, >90% were correctly classified. In conclusion, in a large population-based sample of patients aged ≥80 years, the clinical model possessed high accuracy to identify cardiac dysfunction in daily practice. Among subjects without chronic AF or pacemaker, a larger number were correctly classified by integrating natriuretic peptides and electrocardiographic findings in the strategy.