Objective:

In-hospital mortality prediction models on acute heart failure can be beneficial for decision-making, a situation necessary to evaluate, our goal was to compare predictive and discriminatory capacity of Colombian population.

Method:

A retrospective cohort study in patients with acute heart failure was conducted. The following performance evaluation of in-hospital mortality prediction models were conducted from 2013 to 2015: ADHERE, EHMRG, OPTIMIZE-HF, GWTG-HF and PROTECT. Data was calculated for each model, prediction and discriminatory capacity was evaluated.

Results:

A sample of 776 patients, 56% male, with an average age of 71.5 (standard deviation: 14.3) and with left ventricle ejection fraction rate of 39% was studied. Global mortality was of 6.1%. The area under curve for ADHERE was of 0.56 (95% confidence interval [95% CI]: 0.49-0.64), for EHMRG 0.63 (95% CI: 0.55-0.71), for GWTG-HF 0.63 (95% CI: 0.55-0.70), for OPTIMIZE 0.65 (95% CI: 0.56-0.74) and for PROTECT 0.69 (95% CI: 0.60-0.77).

Conclusions:

The models for predicting in-hospital death in patients with acute heart failure show poor performance, predictability and discrimination in the Colombian population, suggesting the development of mortality prediction scales in patients with acute heart failure specific to our population.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )