#3054 The hospital frailty risk score lacks service planning and decision-making utility in end stage kidney disease

Abstract

Abstract Background and Aims Survival time for patients commencing dialysis is poor (median 2.0 years in our data) and treatment is invasive. While the age of patients commencing dialysis has increased significantly, the survival advantage diminishes with increasing comorbidity and frailty. The Hospital Frailty Risk Score (HFRS) is a recently-described score derived from ICD-10 codes which has been shown to correlate with outcomes in patients aged >75 years admitted to acute hospital settings. As part of a wider initiative to improve service planning and patient decision-making in the renal setting, we investigated the utility of the HFRS in modelling the survival time of dialysis patients with chronic kidney disease (CKD) both prior to reaching end-stage kidney disease (ESKD), and at dialysis start date. Method A secondary care dataset was obtained of all patients who attended pre-dialysis clinic at a single renal unit in the north of England between 2006–2019 (n = 2,573). Patients choosing dialysis and conservative management were included in the cohort; 1,081 eventually underwent dialysis including 329 patients who were aged ≥70 years when they commenced dialysis. 62.9% were male. Retrospective survival analysis was conducted using the HFRS prior to, and at, dialysis start date and estimates compared using the log-rank test. Results HFRS on dialysis start date was associated with survival time in patients aged ≥70 years (Low frailty Median = 2.5 years; CI = [2.1–3.0 years] vs. Intermediate frailty Median = 1.5 years; CI = [1.0–2.0 years] vs. High frailty Median = 1.3 years; CI = [0.2–2.2 years]; p = 0.005). No association was found between HFRS and survival in the same patients (aged ≥70 on dialysis start date), at any of 3 timepoints prior to dialysis: eGFR level of 20, 15, or 12 (p = 0.18, p = 0.63, p = 0.38 respectively). However, only 150/329 patients (46%) had HFRS recorded prior to eGFR level so sample size was limited. 60 patients’ HFRS had also worsened by dialysis start date (8 had improved). Additional analyses including all age groups found a weaker association between HFRS on dialysis start date and survival (Low frailty Median = 2.1 years; CI = [1.9–2.3 years] vs. Intermediate frailty Median = 1.7 years; CI = [1.5–2.0 years] vs. High frailty Median = 2.1 years; CI = [1.3–2.4 years]; p = 0.04). HFRS and survival were associated at eGFR 15 for all ages (p = 0.02) but not eGFR 20 or 12. Conclusion We identified issues with the HFRS and its implementation that present challenges to its utility for treatment decision-making or service planning in a renal setting. While the HFRS has previously been shown to be predictive of 30-day outcomes in an acute admissions unit, it performed poorly in our cohort of patients with CKD, a long-term condition where HFRS may worsen or improve between treatment choice and dialysis start. Although HFRS did predict outcome from dialysis start in patients aged ≥70 years, this is of limited use given the abundance of validated tools that predict outcome at ESKD. More work is needed to refine the HFRS for pre-ESKD decision-making in a renal setting.