Reducing repeated hospitalizations of patients with chronic conditions is a policy objective for improving system efficiency. We test the hypothesis that the risk of readmission is associated with the timing and intensity of primary care follow-up after discharge, focusing on patients hospitalized for heart failure in France. We propose a discrete-time model which takes into account that primary care treatments have a lagged and cumulative effect on readmission risk, and an instrumental variable approach, exploiting geographical differences in availability of generalists. We show that the early consultations with a GP after discharge can reduce the 28-day readmission risk by almost 50%, and that patients with higher ambulatory care utilization have smaller odds of readmission. Furthermore, geographical disparities in primary care affect indirectly the readmission risk. These results suggest that interventions which strengthen communication between hospitals and generalists are elemental for reducing readmissions and for developing effective strategies at the hospital level, it is also necessary to consider primary care resources that are available to patients.
Keywords: Discrete-time model; Hospital; Instrumental variable; Primary care; Readmissions.