BACKGROUND: Nephrologists need to better understand the impact of their decisions about long-term treatment strategies. Healthcare planning requires the anticipation of demand. Indicators from ESRD registries are especially difficult to interpret when the underlying dynamic process is not well understood. Therefore, we have developed a statistical tool to study the course of incident ESRD patient cohorts over time and to quantify, by simulations, the impact of various expected changes or new strategies. METHODS: Based on the data from 67 258 ESRD adult patients, we first estimated transition rates between 10 different modalities of treatment ('compartments') with a multistate model. In a second step, we predicted the number of patients in each compartment at each time point for a cohort of 1000 patients for 180 months after the onset of renal replacement therapy (RRT). We tested two scenarios to illustrate the possibility of simulating policy changes. RESULTS: Increased use of non-assisted automated peritoneal dialysis (PD) (from 7.7 to 19.2% at RRT onset) will not substantially influence the proportion of total RRT time in PD for patients aged 18-44 without diabetes. Improving access to kidney transplants from cadaveric donors for patients aged 45-69 with diabetes will increase the 15-year restricted mean lifetime by 5 months and the time spent with a functioning graft (34 versus 23%). CONCLUSIONS: A model based on patients' treatment trajectories can improve the description and understanding of RRT as a dynamic phenomenon. Its use for simulation may help professionals and decision-makers to optimize renal organization and care.
BACKGROUND: Nephrologists need to better understand the impact of their decisions about long-term treatment strategies. Healthcare planning requires the anticipation of demand. Indicators from ESRD registries are especially difficult to interpret when the underlying dynamic process is not well understood. Therefore, we have developed a statistical tool to study the course of incident ESRDpatient cohorts over time and to quantify, by simulations, the impact of various expected changes or new strategies. METHODS: Based on the data from 67 258 ESRD adult patients, we first estimated transition rates between 10 different modalities of treatment ('compartments') with a multistate model. In a second step, we predicted the number of patients in each compartment at each time point for a cohort of 1000 patients for 180 months after the onset of renal replacement therapy (RRT). We tested two scenarios to illustrate the possibility of simulating policy changes. RESULTS: Increased use of non-assisted automated peritoneal dialysis (PD) (from 7.7 to 19.2% at RRT onset) will not substantially influence the proportion of total RRT time in PD for patients aged 18-44 without diabetes. Improving access to kidney transplants from cadaveric donors for patients aged 45-69 with diabetes will increase the 15-year restricted mean lifetime by 5 months and the time spent with a functioning graft (34 versus 23%). CONCLUSIONS: A model based on patients' treatment trajectories can improve the description and understanding of RRT as a dynamic phenomenon. Its use for simulation may help professionals and decision-makers to optimize renal organization and care.
Authors: Mohsen Yaghoubi; Sonya Cressman; Louisa Edwards; Steven Shechter; Mary M Doyle-Waters; Paul Keown; Ruth Sapir-Pichhadze; Stirling Bryan Journal: Appl Health Econ Health Policy Date: 2022-08-09 Impact factor: 3.686