Abstract: TH-PO716
Appraisal of Kidney-Related Parameters in Predictive Models of Heart Failure
Session Information
- Hypertension and CVD: Epidemiology, Risk Factors
November 07, 2019 | Location: Exhibit Hall, Walter E. Washington Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Hypertension and CVD
- 1402 Hypertension and CVD: Clinical, Outcomes, and Trials
Authors
- Chamarthi, Gajapathiraju, University of Florida, Gainesville, Florida, United States
- De Jesus, Eddy J., University of Florida, Gainesville, Florida, United States
- Kazory, Amir, University of Florida, Gainesville, Florida, United States
Background
The clinical course of patients with heart failure (HF) is known to be variable. Insights into factors that relate to subsequent adverse outcomes may help identify those patients in need of more intense monitoring and therapy. Several models have been developed for HF that simultaneously take into account multiple factors to refine their predictive ability. The clinical relevance of kidney-related parameters has long been recognized in this setting. However, no study so far has explored how consistent they have appeared in various models. We sought to appraise kidney-related parameters (KRP) in contemporary models of HF.
Methods
Articles cited in PubMed database using keywords “heart failure”, “prediction”, and “model” were searched. Available data from clinical trials performed between January 1995 and December 2018 were included. The studies were selected if they prognosticated outcomes in HF population through a predictive model that consisted of at least 2 factors. Pertinent data on KRP (e.g. serum creatinine, blood urea nitrogen [BUN], and serum sodium level) were extracted and reviewed.
Results
A total of 15 studies with 82,706 participants were included, of which 5 were validated in a HF cohort different from the model derivation cohort. They consisted of a variety of HF populations (e.g. acute, chronic, carrying mechanical circulatory device) and the median number of included parameters was 7. There was substantial variation across models in the reporting of the KRP as well as the studied outcomes. While no study included estimated glomerular filtration rate, serum creatinine and BUN were included in only 6 and 4 studies respectively. Similarly, 4 and 7 models contained data on serum sodium level and blood pressure respectively. Serum uric acid and history of kidney disease were each included in only 1 study.
Conclusion
We found that available models for prediction of HF outcomes do not consistently include KRP while generally portending high prognostic ability. Development of these models is based on multivariate regression methods to define the proportional significance and coefficients of the prognostic variables. Therefore, this finding supports the notion that, contrary to conventional thinking, the impact of KRP on the outcomes of patients with HF may be confounded or modulated by other covariates (e.g. congestion) as the emerging data have implied.