P013. BNP as a Predictor of Diuretic Nonadherence in 30-Day Heart Failure Readmissions

Autor: Henry Viejo, Rebecca Meraz, Katie Frank, Nancy Vish, Kathryn Osteen
Rok vydání: 2021
Předmět:
Zdroj: Heart & Lung. 50:567
ISSN: 0147-9563
DOI: 10.1016/j.hrtlng.2021.03.068
Popis: Background Diuretic treatment effectively treats heart failure (HF), lowers brain natriuretic peptide (BNP), and can reduce hospital readmissions, yet approximately 50% of HF patients do not take medications as prescribed. Additionally, 25% of HF hospital readmissions are thought to be preventable and associated with imperfect adherence to prescribed medications. One barrier to solving this problem is the lack of an objective marker for identifying or gauging diuretic-taking habits in the HF patient. Better understanding of the impact of imperfect diuretic adherence on BNP provides insight into the significance of taking diuretics as prescribed to 30-day HF readmissions and the usefulness of BNP in identifying patients at-risk for diuretic nonadherence. Objective The purpose of this retrospective, exploratory investigation was to examine the association between diuretic nonadherence and BNP in those with a 30-day HF readmission and further explore the usefulness of BNP as an objective marker for diuretic adherence. Methods A retrospective review of electronic health records from 10 Texas hospitals between the years of 2014 and 2018 was conducted. We examined the association of nonadherence to BNP with three BNP-related variables: change in BNP from index hospitalization to readmission, percent change over time, and readmission BNP. We investigated the relationship of each with nonadherence and patient characteristics by performing backwards stepwise logistic regression using Bayesian information criterion (BIC) for model selection. The discrimination ability of these models was evaluated by calculating area under the curves (AUCs) and testing for differences among them with the DeLong method. Results Of the study sample (N=405), the medical records of 124 (31%) patients explicitly documented diuretic nonadherence at readmission. In final BIC-selected logistic regression models for predicting nonadherence in patients with a 30-day HF readmission, BNP-related variables were the only remaining predictors and were significantly associated with nonadherence; other variables concerning patient characteristics were excluded. The discrimination ability of these models was fairly high (between 0.64 and 0.67), and none had a significantly superior AUC than the others. Conclusions Results suggest that variables relating to patient characteristics are not as important as those relating to BNP for predicting diuretic nonadherence. Additionally, more investigation is warranted to determine if one of the three BNP-related variables we considered is a superior objective marker of diuretic nonadherence. BNP could be useful in identifying patients at risk for diuretic nonadherence and a hospital readmission. Since medication-taking habits are modifiable, finding new ways to gauge diuretic adherence could reduce hospital readmissions and improve HF outcomes.
Databáze: OpenAIRE
Externí odkaz: