Abstract P184: Optimal Renal Function Indices During Heart Failure Hospitalization to Predict In-Hospital and Post-Discharge Mortality

Abstract

BACKGROUND: Abnormal baseline renal function and worsening renal function (WRF) during heart failure (HF) hospitalization are associated with worse outcomes. However, which measure of renal function is the most prognostic is unknown. We sought to determine the renal function measure most predictive of in-hospital and post-discharge mortality. METHODS: This was a retrospective cohort study of patients with HF who received care from a large health system between Jan 1, 2000 and June 30, 2008. Patients with preexisting end-stage renal disease were excluded. Blood urea nitrogen (BUN), creatinine (Cr), and MDRD estimated glomerular filtration rate (eGFR) values at admission, and changes during hospitalization were tested for their association with in-hospital and one year post-discharge mortality. Logistic regression and conditional receiver operating curves for each index were used to determine the association with mortality by comparing the area under curve (AUC). RESULTS: Among 7394 patients meeting the inclusion criteria, 204 died in-hospital, and 1652 were discharged but died within 1 year. Baseline BUN was the strongest correlate for both in-hospital and post-discharge mortality (AUC= 0.724 and 0.656, respectively; p<0.001 vs. Cr or eGFR). BUN thresholds of 31 and 29 mg/dL were the most predictive with 4.6 and 3.0 fold increased mortality, respectively. After adjustment for baseline BUN, in-hospital change in Cr and BUN performed similarly and superior to eGFR for in-hospital death (model AUC 0.812; p<0.001 vs. eGFR). For post-discharge death, maximal change performed better than net change, and Cr, BUN, eGFR each performed similarly (model AUC=0.661). Standard and optimized WRF thresholds are shown in the table . CONCLUSIONS: Among hospitalized HF patients baseline BUN is the renal index most strongly associated with mortality. Common WRF definitions are suboptimal for predicting mortality, and the optimal thresholds depend on baseline renal function. Optimized WRF creatinine definition vs. traditional thresholds for predicting mortality (N=7394) BUN Strata ΔCreatinine cutpoint Sensitivity In-hospital death Specificity In-hospital death Odds Ratio 0 < BUN ≤ 15 0.2 0.3 0.5 94.4 (17/18) 72.2 (13/18) 50.0 (9/18) 48.9 (822/1680) 67.7 (1137/1680) 88.5 (1487/1680) 16.29 5.44 7.70 16 ≤ BUN ≤ 22 0.3 0.5 0.6 62.5 (20/32) 59.4 (19/32) 59.4 (19/32) 63.5 (1279/2014) 82.2 (1656/2014) 88.0 (1772/2014) 2.90 6.76 10.70 23 ≤ BUN ≤ 32 0.3 0.5 0.7 63.9 (23/36) 50.0 (18/36) 44.4 (16/36) 60.0 (1039/1731) 79.4 (1375/1731) 89.3 (1546/1731) 2.66 3.86 6.69 33 ≤ BUN < 200 0.3 0.5 0.8 67.8 (80/118) 60.2 (71/118) 54.9 (65/118) 56.9 (1005/1765) 74.2 (1310/1765) 85.8 (1514/1765) 2.78 4.35 7.40 BUN Strata ΔCreatinine cutpoint Sensitivity Post-discharge death Specificity Post-discharge death Odds ratio 0 < BUN ≤ 15 0.3 0.5 83.0 (181/218) 93.6 (204/218) 18.4 (206/1118) 7.4 (83/1118) 1.11 1.17 16 ≤ BUN ≤ 22 0.3 0.5 42.6 (142/333) 22.5 (75/333) 65.4 (902/1380) 83.7 (1155/1380) 1.40 1.49 23 ≤ BUN ≤ 32 0.1 0.3 0.5 73.3 (302/412) 56.6 (233/412) 20.9 (86/412) 33.0 (362/1096) 46.9 (514/1096) 80.1(878/1096) 1.35 1.15 1.06 33 ≤ BUN < 200 0.1 0.3 0.5 69.2 (477/689) 46.0 (317/689) 28.0 (193/689) 37.0 (331/895) 58.4 (523/895) 74.2 (664/895) 1.32 1.20 1.19