Development of Clinical Prediction Rules and Health Services Research in Patients With Heart Failure (Health Services Evaluation in Heart Failure-ESSIC)

DESIGN: multicenter prospective observational cohort study with one year follow-up after inclusion in the study.

SCOPE: Nine hospitals participating in this project: Hospital Universitario Basurto, Santa Marina, Donostia, and Galdakao-Usansolo, Hospital de Antequera, Costa del Sol, Universitario de Canarias, Universitario Parc Taulí y Bellvitge.

SUBJECTS: patients with known chronic heart failure (or de novo diagnosed in the emergency department visit) attended in the emergency services of the participating hospitals prospectively diagnosed with acute or decompensated heart failure and recruited during the first year of the study, including patients admitted or given discharge in the hospital emergency room. The first admission of each patient during the recruitment period will be taken as a reference, and after this episode the follow-up will be carried out during a whole year, collecting all the events that happen related to their illness.

Missing patients: In all patients who meet the selection criteria the investigators will collect data on essential sociodemographic and clinical variables in order to be able to compare the patients lost in the follow-up with the patients who finally participate in the entire study.

Sample size calculation: Predictive model development studies establish that it is necessary to have at least 10 events of the dependent variable of interest (in our case: mortality, major complications, recurrence or re-admissions, separately) for each independent variable included in the multivariate logistic regression model . Given that our intention is to include in the multivariate model a limited but exhaustive number of variables (predictably, no less than 10), the investigators estimate that it will be necessary to have at least 100 events of the dependent variable in the derivation sample (of 1000 patients ) to make sure that the regression model converges properly. Data from our centers indicate that the number of events of the dependent mortality variable would be> 15% of the patients admitted, with the percentages expected from the other parameters with higher results. With all the participating centers and 1 year of recruitment, the investigators hope to recruit around 2000 valid patients (50% for the derivation and 50% in the validation sample) sufficient to meet the stated objectives.