Predictors of 12-month worsening renal function after heart failure hospitalisation

Abstract Background Worsening renal function (WRF) is associated with worse outcomes in heart failure (HF)¹. Most studies of WRF predictors use in-hospital changes in creatinine, not changes over longer time periods. In previous work we have found WRF over 12 months is associated with worse HF outcomes. Purpose Assess the prevalence of WRF over 12 months after HF hospitalisation and evaluate predictors of 12-month WRF. Methods All patients hospitalised with a HF diagnosis during 2013/2014 in Northland/Auckland regions of New Zealand were recorded. Data linkages allowed for identification of demographic data, comorbidities (based on discharge codes and medications), medication dispensing, blood test results, and mortality. Creatinine measured around discharge and 12-months post discharge were identified. Patients with eGFR = 15 at discharge or who died within 12 months were excluded. WRF was defined as a 25% increase in serum creatinine between discharge and 12-month measurements. Multivariable logistic regression was used to assess WRF predictors, adjusting for discharge creatinine to account for baseline differences. Predictors were selected a-priori for availability and known relevance to HF outcomes and were age, sex, ethnicity, NZ deprivation index, ischaemic aetiology, previous HF hospitalisation, diabetes, multimorbidity (presence of 2 or more other chronic conditions), loop diuretic, ACE/ARB, and MRA use, discharge sodium and haemoglobin. Patients were followed for 4 years from index hospitalisation for re-hospitalisation and all-cause mortality. Results Of 6531 HF hospitalisation patients, discharge and 12-month creatinine were available for 3686 patients. 147 were excluded due to eGFR = 15, leaving 3539 patients available. The median age was 75, with 45% female (see Table 1 for cohort characteristics). WRF occurred in 924 (26%) of the cohort. 530 (57%) of the WRF group died over the 3 years following the initial 12-month follow-up period, compared to 1119 (43%) of the non-WRF group. Independent predictors of WRF were diabetes (adjusted odds ratio aOR 1.34; 95% CI 1.12-1.60), loop diuretic use (1.28; 1.07-1.53), abnormal haemoglobin (1.56; 1.31-1.86), ischaemic aetiology (1.17; 1.00-1.38), multimorbidity (1.24; 1.03-1.49), 2-3 prior HF hospitalisations (1.27; 1.05-1.54), age/10 years (1.08; 1.01-1.16), and log(discharge creatinine) (0.31; 0.24-0.39). See Figure 1 for aOR for all variables. Conclusions One quarter of the cohort experienced WRF in 12 months after discharge. Diabetes, age and loop diuretic use are known independent predictors of WRF over shorter time periods. Ischaemic aetiology, multimorbidity, prior HF hospitalisation and haemoglobin were additional independent predictors of WRF. Higher discharge creatinine was associated with a lower risk for WRF, which may have multiple explanations. Further research is required to determine the best WRF definition over longer periods and its relationship to HF outcomes.Table 1:Cohort Characteristics Figure 1:Predictors of WRF from model