Development and Validation of a Simple-to-Use Nomogram for Predicting In-Hospital Mortality in Patients With Acute Heart Failure Undergoing Continuous Renal Replacement Therapy

Abstract

Background: Patients with acute heart failure (AHF) who require continuous renal replacement therapy (CRRT) have a high risk of in-hospital mortality. It is clinically important to screen high-risk patients using a model or scoring system. This study aimed to develop and validate a simple-to-use nomogram consisting of independent prognostic variables for the prediction of in-hospital mortality in patients with AHF undergoing CRRT.Methods: We collected clinical data for 121 patients with a diagnosis of AHF who underwent CRRT in an AHF unit between September 2011 and August 2020 and from 105 patients in the medical information mart for intensive care III (MIMIC-III) database. The nomogram model was created using a visual processing logistic regression model and verified using the standard method.Results: Patient age, days after admission, lactic acid level, blood glucose concentration, and diastolic blood pressure were the significant prognostic factors in the logistic regression analyses and were included in our model (named D-GLAD) as predictors. The resulting model containing the above-mentioned five factors had good discrimination ability in both the training group (C-index, 0.829) and the validation group (C-index, 0.740). The calibration and clinical effectiveness showed the nomogram to be accurate for the prediction of in-hospital mortality in both the training and validation cohort when compared with other models. The in-hospital mortality rates in the low-risk, moderate-risk, and high-risk groups were 14.46, 40.74, and 71.91%, respectively.Conclusion: The nomogram allowed the optimal prediction of in-hospital mortality in adults with AHF undergoing CRRT. Using this simple-to-use model, the in-hospital mortality risk can be determined for an individual patient and could be useful for the early identification of high-risk patients. An online version of the D-GLAD model can be accessed at https://ahfcrrt-d-glad.shinyapps.io/DynNomapp/.Clinical Trial Registration:www.ClinicalTrials.gov, identifier: NCT0751838.