Provider Monitoring and Pay‐for‐Performance When Multiple Providers Affect Outcomes: An Application to Renal Dialysis

Abstract

Objective. To characterize the influence of dialysis facilities and nephrologists on resource use and patient outcomes in the dialysis population and to illustrate how such information can be used to inform payment system design. Data Sources. Medicare claims for all hemodialysis patients for whom Medicare was the primary payer in 2004, combined with the Medicare Enrollment Database and the CMS Medical Evidence Form (CMS Form 2728), which is completed at onset of renal replacement therapy. Study Design. Resource use (mainly drugs and laboratory tests) per dialysis session and two clinical outcomes (achieving targets for anemia management and dose of dialysis) were modeled at the patient level with random effects for nephrologist and dialysis facility, controlling for patient characteristics. Results. For each measure, both the physician and the facility had significant effects. However, facilities were more influential than physicians, as measured by the standard deviation of the random effects. Conclusions. The success of tools such as P4P and provider profiling relies upon the identification of providers most able to enhance efficiency and quality. This paper demonstrates a method for determining the extent to which variation in health care costs and quality of care can be attributed to physicians and institutional providers. Because variation in quality and cost attributable to facilities is consistently larger than that attributable to physicians, if provider profiling or financial incentives are targeted to only one type of provider, the facility appears to be the appropriate locus.Key Words. Pay-for-performance, monitoring, dialysis Private and public payers are focusing on measuring and rewarding quality and efficiency in health care (Milgate and Cheng 2006;Rosenthal et al. 2006). Such efforts include ''pay-for-performance'' (P4P) systems that reward mear Health Research and Educational Trust

The Initial Impact of Medicare's New Prospective Payment System for Kidney Dialysis

“…This conclusion is consistent with prior work on dialysis providers. 7,8 For some, but not all, of the medications added to the bundled payment system, providers began to shift use during the last quarter under the old payment system. At that point, the coming payment changes were both well known and imminent, suggesting that these early changes may have been anticipatory.…”

Discussion on “Time-dynamic Profiling with Application to Hospital Readmission Among Patients on Dialysis,” by Jason P. Estes, Danh V. Nguyen, Yanjun Chen, Lorien S. Dalrymple, Connie M. Rhee, Kamyar Kalantar-Zadeh, and Damla Senturk

“…In order to account for this, He et al (2013) included hospital as a random effect and found that this resulted in a substantial improvement in fit. It should be noted that several previous studies have suggested an influence of multiple types of providers on treatment practices and outcomes among ESRD patients (Hirth et al, ; Hirth et al, ; Turenne et al, ). Estes et al includes a random effect for patients, which seems particularly appropriate and important when patients are followed over a long period of time.…”

“…In order to account for this, He et al (2013) included hospital as a random effect and found that this resulted in a substantial improvement in fit. It should be noted that several previous studies have suggested an influence of multiple types of providers on treatment practices and outcomes among ESRD patients (Hirth et al, 2009;Hirth et al, 2010;Turenne et al, 2010 In effect, we can define a standardized ratio measure with respect to any distribution of the covariates that we wish, so that we could define the SDRR for facility i with reference to the patient population in facility i . This would be estimated with…”

Provider‐Hospital “Fit” and Patient Outcomes: Evidence from Massachusetts Cardiac Surgeons, 2002–2004

“…Our use of hierarchical models extends prior use of random effects for physicians (Bronskill et al 2002; Pietz et al 2002; Ferraris et al 2008), for hospitals (Hannan et al 2003; Peterson et al 2004; Carey et al 2008; Marcin et al 2008), or for both (Austin, Tu, and Alter 2003; Hirth et al 2009). Our approach hinges on two‐way crossed mixed models with random effects for hospital, surgeon, and the hospital × surgeon interaction.…”