Albumin excretion in acute myocardial infarction: a guide for long-term prognosis.

BACKGROUND Albumin excretion rate has been found to be associated with increased risk of mortality in several clinical settings. We assessed the relationship between urinary albumin and 7-year mortality in a cohort of patients with acute myocardial infarction (AMI). METHODS In this prospective study, we examined 505 white patients admitted with AMI to the intensive care unit of 3 hospitals. Main end points were nonearly all-cause and cardiovascular (CV) mortality. Albumin-to-creatinine ratio (ACR) was measured by radioimmunoassay on the first, third, and seventh days after admission. Risk estimates were made using Cox proportional-hazard model and relative odds. Forty patients (7.9%) died early inhospital, and 175 (34.7%) died during the rest of the follow-up (nonearly mortality). RESULTS The ACR measured on the third day predicted the occurrence of 7-year nonearly all-cause and CV mortality. Hazard ratios for ACR > or =0.97 mg/mmol were 3.0 (95% confidence limit 2.2-4.1), P < .0001, for nonearly all-cause mortality and 3.5 (95% confidence limit 2.5-5.0), P < .0001, for CV mortality. Correspondent fully adjusted hazard ratios were 1.9 (95% CI 1.4-2.6), P < .0001, and 2.2 (95% CI 1.5-3.2), P < .0001, respectively. By adding ACR to the 18-variable predictive model, ACR improved significantly both the goodness of fitting of the model for nonearly all-cause (P < .0001) and CV mortality (P < .0001) and the C-statistic value (P < .0001 and P = .002 for nonearly all-cause and CV mortality, respectively). Similar results were obtained for ACR measured on the first day or the seventh day. CONCLUSIONS An early increase of urinary albumin in AMI is a strong independent predictor of long-term adverse clinical outcome. The ACR improved clinical prediction over and above baseline traditional multivariable risk models.