ANALYSIS OF COMPETING RISKS IN DECISIONS REGARDING RENAL REPLACEMENT THERAPY FOR OLDER PATIENTS: A NATIONWIDE COHORT STUDY IN JAPAN

In Japan’s super-aged society, the growing number of very old patients with advanced chronic kidney disease (CKD) poses clinical challenges regarding the initiation of renal replacement therapy (RRT). This study aimed to clarify the relationship between RRT initiation and mortality using nationwide real-world data.

We used a nationwide administrative claims and health check-up database from the Japan Medical Data Center (JMDC Inc., Tokyo, Japan). The study included 2,020,203 individuals aged ≥75 years enrolled in the Late-Stage Elderly Health Care System between April 2021 and March 2023. Individuals who died during follow-up were retained, while those who lost eligibility were excluded. As of September 2021, patients receiving RRT were identified via procedure codes, and the prevalence was compared with Japanese Society for Dialysis Therapy (JSDT) data.

To assess one-year RRT initiation, we identified 309,265 individuals who underwent health checkups between July and September 2021 without prior RRT. Among individuals with advanced kidney dysfunction (eGFR < 15 mL/min/1.73 m²), three groups were defined: (1) RRT initiation, (2) death without RRT, and (3) survival without RRT for one year. Frailty was assessed across 15 physical, psychological, and social items, with higher scores indicating greater frailty. Baseline characteristics were compared using the Kruskal–Wallis and chi-square tests (p < 0.01). Cumulative incidence functions (CIFs) were plotted for RRT initiation, accounting for death as a competing risk. Fine–Gray subdistribution hazard models estimated predictors of RRT initiation (p < 0.05).

The database showed high external validity: the overall RRT prevalence (0.74%) was consistent with national JSDT data (0.70%). However, the one-year RRT incidence (0.048%) in the health checkup cohort was lower than national data (0.088%), likely reflecting selection bias from healthier individuals who undergo checkups. A total of 148 individuals initiated RRT within one year. When stratified by baseline kidney function, the incidence was 0.0037% for eGFR > 60, 0.012% for 30–60, 0.47% for 15–30, and 18.5% for <15 mL/min/1.73 m². In those with eGFR < 15 (n = 477), baseline characteristics are shown in Table 1, and competing risks of RRT initiation and death in Figure 1. The one-year cumulative incidence of RRT initiation was 18.5%, and that of pre-RRT mortality was 16.7%. In the Fine–Gray model, older age (subdistribution hazard ratio [sHR] = 0.93) and higher eGFR (sHR = 0.84) significantly reduced the likelihood of RRT initiation, whereas male sex (sHR = 1.85) and hyperuricemia (sHR = 2.27) significantly increased it.

The database of older adults closely reflected national dialysis data in Japan. Advanced age independently predicted non-initiation of RRT, reflecting real-world treatment selection. Mortality was a competing event of similar magnitude to RRT initiation, emphasizing that in very old patients with advanced CKD, decisions regarding RRT are best made through shared decision-making (SDM). SDM should address the high competing risk of mortality and weigh the benefits and burdens of RRT, rather than assuming it as the default pathway. These findings support conservative kidney management(CKM) as a viable and patient-centered option for older adults with advanced CKD.