Reducing hospital readmissions: analysing the accuracy of predictors and efficacy of interventions

Background/Aims Hospital admissions represent a substantial burden, both to patients and healthcare systems. This study aimed to assess the accuracy of predictors of readmissions (defined as unplanned rehospitalisations within 30 days of a patient's index discharge) and the efficacy of interventions to reduce readmission rates. Methods Data from 201 329 patients who had a diagnosis of acute myocardial infarction, heart failure or pneumonia at their index visit were extracted from the database of a regional healthcare system in the US. The predictors of readmission assessed were the patient's risk of unplanned readmission score and whether they were frequent healthcare service users and/or belonged to a high-risk group, as classified by the healthcare system. Interventions included competition of a follow-up appointment, discussion of medications with the patient and implementation of the ‘meds-to-beds’ programme. Data were analysed with logistic regression models to assess the relationship of readmissions with the predictors and interventions. Results Frequent healthcare service user status, high-risk group status and risk of unplanned readmission score were all found to be accurate predictors of readmission. For example, frequent healthcare service users were 12 times more likely to be readmitted. Of the seven interventions investigated, only the completion of a follow-up appointment had a significant relationship with readmission rates, although the meds-to-beds intervention was also found to have some benefits for high-risk patients. Conclusions Identification of individuals who are at high risk of a readmission should be prioritised by healthcare systems to ensure that appropriate interventions to prevent readmission can be implemented both during and after discharge from the index visit. Follow-up appointments are a particularly important part of this strategy.