Non-cardiovascular medication and readmission for heart failure: an observational cohort study.

Background Among heart failure (HF) patients, hospital readmissions are a major concern. The medication taken by a patient may provide information on comorbidities and conditions and may be used as an indicator to identify populations at an increased risk of HF readmission. Aim This study explored the use of non-cardiovascular medication at hospital discharge from the first HF admission in search of indicators of high risk of readmission for HF. Method The study included 22,476 HF patients from the Dutch PHARMO Database Network at their first HF hospitalization. The data was divided into training and validation sets. A Cox regression model with demographics, date of first HF hospital admission and non-cardiovascular medication present at discharge, adjusted for cardiovascular medication, was developed in the training set and subsequently implemented in the validation set. Results The study included 22,476 patients, mean age 76.7 years (range 18-104) and median follow-up time 2.5 years (range 0-15.7 years). During the study period 6,725 (29.9%) patients were readmitted for HF, with a median time-to-readmission of 7 months (range 0-14.3 years). Non-cardiovascular medication associated with a high risk of readmission for HF were identified as indicators of high risk, with no implied causal relationship. Patients prescribed antigout medications presented a 25% increased risk of readmission (HR 1.25, 95%CI 1.09-1.45, P = 0.002). Patients using insulin had an 18% higher risk of readmission versus patients not using insulin (HR 1.18, 95%CI 1.06-1.32, P = 0.002), but not versus patients treated with other blood-glucose-lowering drugs. No association between the risk of readmission and NSAIDs use was observed. Conclusion The results suggest that diabetes is responsible for the higher HF-readmission risk observed in patients prescribed insulin. The observed risk in users of antigout medication should be further investigated. The absence of an association with the use of NSAIDs should be interpreted with caution.