OPTIMIZING REFERRAL TO RENAL CARE MANAGEMENT PROGRAM THROUGH USE OF A PREDICTIVE MODEL FOR TRANSITION TO DIALYSIS IN A MEDICARE ADVANTAGE POPULATION

Purpose: Chronic Kidney Disease (CKD) is a complex condition and evidence demonstrates that renal care management (RCM) benefits individuals with late-stage CKD, especially around the time of transition to dialysis.  Traditionally, individuals are identified for RCM via an estimated glomerular filtration rate (eGFR) of <20mL/min/1.73m², but many individuals are still not identified in a timely manner, sometimes due to lack of an eGFR value.  Predictive modeling (PM) has enabled more effective referral to many disease management programs; hence a PM was developed to identify individuals at high risk of transitioning to dialysis who could benefit from RCM. 

Method: A broad range of data sources, such as administrative medical, pharmacy, lab, and consumer data, were leveraged to build a dialysis PM for a Medicare Advantage population with a large national insurer.   Individuals with highest risk based on the model were referred to RCM between Mar and Oct 2014.  Using administrative claims as of Jan. 31, 2015, the following outcomes were assessed: Annualized transition to dialysis rate (defined as number of individuals transitioned to dialysis divided by the total person-year since referral), time to dialysis since referral, and program enrollment rate.  These outcomes were compared for individuals identified by the PM (PM group) against individuals identified by other referral sources (e.g., eGFR only, nurse referral) (standard group) during the same time period.  Statistical comparisons were made with chi-square and unpaired t-tests as appropriate.

Result: During the study period, 947 individuals (mean age 71 years, 54% male) were referred in the PM group; 7,607 individuals (mean age 75 years, 43% male) were identified in the standard group.  The annualized transition to dialysis rate was higher for the PM group compared to the standard group (0.21 vs. 0.11, p<0.001), and this was also the case for time to dialysis (120 days vs. 92 days, p<0.001).  There was no difference in the program enrollment rates (35% for the PM group vs. 37% for the standard group, p=0.103).

Conclusion: Individuals with CKD who are at high risk of transitioning to dialysis can be identified by a dialysis PM, regardless of the availability of an eGFR.  These individuals have a higher transition rate, a longer window for intervention, and similar enrollment rate as individuals identified from other referral sources.