Atrial natriuretic peptide levels in the prediction of congestive heart failure risk in frail elderly.

Abstract

To develop a noninvasive clinical predictive model for acute congestive heart failure (CHF) in a frail elderly cohort using bedside clinical assessment (medical history and physical examination) and venous atrial natriuretic peptide (ANP) levels. One-year prospective blinded cohort study. Life care facility. Three hundred thirty-one frail, elderly volunteers free of acute illness at study entry (mean +/- SD age, 88 +/- 7 years; 23% male, 77% female). Clinical episodes of CHF with confirmation of acute pulmonary edema by chest roentgenogram. Fifteen percent of the elderly cohort developed at least one episode of CHF during the 1-year follow-up period. Those developing CHF had significantly higher mean +/- SE ANP values at study entry: 493 +/- 55 vs 207 +/- 15 pmol/L. The risk for development of CHF rose progressively with increasing ANP levels at study entry. In multivariate analysis, only two independent variables significantly predicted CHF: ANP value greater than 200 pmol/L (adjusted odds ratio [OR], 7.9; 95% confidence interval [CI], 3.2 to 19.2) and history of CHF in the previous year (adjusted OR, 7.0; 95% CI, 2.9 to 17). Stratifying the cohort by these two variables results in three CHF risk groups: 55% of the population at 3% annual risk of CHF, 37% of the population at 20% to 24% annual risk of CHF, and 8% of the population at 66% annual risk of CHF. This simple clinical prediction model identifies elderly subjects at risk for CHF and allows appropriate focusing of medical resources for prevention, early detection, and treatment of this highly morbid clinical syndrome.