Recurrent events analysis for examination of hospitalizations in heart failure: insights from the Enhanced Feedback for Effective Cardiac Treatment (EFFECT) trial

Aims: Hospitalizations often occur multiple times during the disease course of a heart failure (HF) patient. However, repeated hospitalizations have not been explored in a fulsome way in this setting. We investigated the association between patient factors and the risk of hospitalization among patients with HF using an extension of the Cox model for the analysis of recurrent events. Methods and results: We examined hospitalizations and predictors of readmission among newly discharged patients with HF in the Enhanced Feedback For Effective Cardiac Treatment phase 1 (April 1999-March 2001) study with the Prentice-Williams-Peterson model with total time. Of 8948 individuals discharged alive from hospital, 7562 (84.5%) were hospitalized at least once during 15-year follow-up. More than 31 000 hospitalizations were observed. There was a progressive shortening of the interval length between hospitalization episodes. An increasing number of comorbidities (average 2.3 per patient) was associated to an increasing hazard of being readmitted to hospital. Most patient factors associated with the risk of hospitalization have been previously described in the literature. However, the estimates were smaller in comparison to a traditional analysis based on the Cox model. Conclusion: The importance of patient factors for the risk of being admitted to hospital was variable over the course of the disease. Conditions such as diabetes and chronic pulmonary obstructive disease had a sustained association with the rate of hospitalization across all episodes examined. The analysis of recurrent events can explore the longitudinal aspect of HF and the critical issue of hospitalizations in this population.