Trends in prevalence of patient case-mix adjusters used in the Medicare dialysis payment system
Objectives. The Medicare End-Stage Renal Disease Prospective Payment System (PPS) used data from 2006-08 to set weights for each case-mix adjuster that is part of the bundled payment formula. The details of the population case-mix were not made public, and little is known about consistency of case-mix over time. This study estimated the prevalence of case-mix adjusters during 2006-2008and analyzed changes in case-mix prevalence from 2000-2008.
Methods: Cross-sectional cohort study using United States Renal Data System data for Medicare dialysis patients. Three 3-year cohorts (2000-02, 2003-05, 2006-08) were analyzed for changes over time in case-mix prevalence.
Results: Double-digit trends were observed in many case-mix categories between 2000-02 and 2006-08. Large dialysis declines were observed in prevalence of patients with low BMI, pericarditis, new to dialysis, and ages 18-44. Large increases were observed in chronic co-morbidities, pneumonia and age cohort 80+.
Conclusions: Substantial changes in case-mix adjuster prevalence suggest the PPS payment formula should be regularly updated.
In January 2011, the Centers for Medicare and Medicaid Services launched the End-Stage Renal Disease prospective payment system for dialysis-related services provided to Medicare beneficiaries. Under the PPS, the payment-per-dialysis treatment has been expanded to include services that previously were reimbursed separately, and payments are now case-mix adjusted to account for the variable costs of care for patients of different ages and body size and for the presence of certain co-morbidities. 1
The ESRD PPS as implemented in 2011 used patient case-mix from the 2006–08 Medicare fee-for-service population to set weights for each patient-level case-mix adjuster in the payment formula. The prevalence of each case-mix adjuster underlying the payment weights instituted for the PPS was not made public, either in the Proposed Rule issued in September 2009, 2or in the Final Rule, 1and concerns were raised by several provider organizations that the assumed prevalence, especially for chronic and acute co-morbidities, was higher than dialysis facilities would be able to report for billing purposes. 3–6
Further, little is known about how stable the prevalence of these case-mix adjusters has been over time, either before the period used to set payment weights or since. While CMS promised to monitor the prevalence of co-morbidity diagnoses recognized for payment adjustment as compared to the prevalence of these categories used in the initial rate setting 1and to monitor co-morbidities eligible for payment adjustment to determine if the co-morbidity adjustments need to be refined in future rulemaking, 7no evidence of such monitoring and/or recognition of the need for adjustments to the payment weights has been forthcoming as of this time.
However, evidence can be gleaned from the Annual Data Reports (ADRs) published by the United States Renal Data Service (USRDS) that trends in the prevalence of at least some of the PPS patient-level case-mix adjuster categories has occurred. An analysis of data found in the 2014 ADR 8showing 10-year trends in the prevalence of patients by age cohort and the percent of patients new to dialysis illustrates that the percentage of patients in these case-mix adjuster categories has indeed changed over time. For example, the percent of patients in the high-payment multiplier age cohort 20-44 declined by 16.5% from 2003-12, from 15.8% of dialysis patients to 13.2% (see Figure 1). Similarly, the highest payment multiplier is applied to patients new to dialysis (incident), a category that had a 15.8% decline from 2003-12 (see Figure 1).
While CMS has updated some components of the PPS since implementation in 2011, including annual adjustments to the outlier formula, a recalibration of the base rate to account for decreased drug utilization in 2014 and modifications to the ESRD market basket and the wage portion of the base rate for 2015, no updates have been made to the patient case-mix payment adjusters. Disconnects between actual patient case-mix and that assumed in the payment formula may have resulted in reduced payments of 1-2% to dialysis providers in the initial years of the PPS .9To the extent that the payment adjusters do not accurately reflect the costs of care for patients of varying illness and cost burden, access to care may be compromised for patients with high-cost characteristics, in areas where such characteristics are more concentrated, or in less competitive markets where facilities may have greater discretion over which patients to accept for treatment. 10,11
In its comment letter to CMS on the 2014 PPS Proposed Rule, the Medicare Payment Advisory Commission (MedPAC) called for an analysis of the PPS’s case-mix adjusters, to ensure the accuracy of the ESRD PPS. 12This issue was also recognized in the American Taxpayer Relief Act of 2012 (ATRA), which required the Secretary of Health and Human Services to conduct an analysis of the case-mix payment adjustments and make appropriate revisions by no later than January 1, 2016. 13In order to meet this timetable, the analysis will need to be completed in time for the release of the 2016 PPS proposed rule sometime in July 2015. CMS has indicated that it will conduct this analysis as part of the 2016 rule-making process. 14
Therefore, given the trends observed in the USRDS data, the lack of information about the prevalence of the case-mix adjusters used for payments, and the widespread recognition of the need for monitoring and analysis of the PPS case-mix adjusters, a timely review is needed. As part of this process, this study was designed to analyze the changes in patient-level case-mix adjuster prevalence in the Medicare Fee-For-Service (FFS) population in the years leading up to the PPS implementation (2000–2008). The primary goal of the study was to provide population-level estimates of, and trends in, prevalence of case-mix factors that were to be applied to payments under the ESRD PPS, up to the time period used to set the PPS adjusters. It was hoped that such an analysis would lend insights into what the baseline prevalence of each case-mix adjuster used by the PPS payment formula actually was, and how case-mix prevalence in this population has changed over time.
This was a retrospective, cross-sectional study of Medicare FFS patients usingUSRDS claims and eligibility data from the Standard Analytical Files (SAFs) 15for 2000–2008. Data were split into 3-year cohorts: 2000–2002, 2003–2005, and 2006–2008. All Medicare FFS patients aged 18+ who were on dialysis during the 9-year study period were included. The prevalence of a PPS patient-level case-mix adjuster was counted for those calendar months in which the patient was on dialysis and covered by Medicare as the primary payer for more than half of the month. Note that the 90-day waiting period for a new dialysis patient was not included. Thus, the first 3 months for a patient who began dialysis but was not already covered by Medicare were not included in this analysis.
The percentage of all patients and of all patient months within each 3-year cohort was calculated for each case-mix adjuster. Percent change in case-mix adjuster prevalence was then calculated between 3-year cohorts. No statistical significance tests were performed on the differences in case-mix prevalence between the 3-year cohorts because:
a) This was a population analysis of the entire Medicare FFS patient population for the years 2000-2008. No statistical inferences were made for any other population.
b) The very large cohort sizes (over 400,000 patients in each cohort) would show statistically significant differences that may not be meaningful.
Case-mix prevalence calculations
We conducted two separate analyses based on the following prevalence definitions:
- Percentage of patients: The denominator is the number of patients who received dialysis and were covered by Medicare for at least one month in a given 3-year cohort.
- Percent of patient months: The denominator is the total number of months in which patients received dialysis and were covered by Medicare in a given 3-year cohort.
Case-mix adjuster definitions replicated the PPS reimbursement definitions as closely as possible. Age was defined at first date of ESRD only. For patient prevalence, age was reported for the median month of dialysis treatment during each cohort period. For example, a patient treated with dialysis from Feb 2007 to June 2008 (17 months total) had a median month of dialysis treatment of Oct 2007 (ninth month), and thus the age in 2007 was reported. Patient age was grouped into PPS age cohorts (18–44, 45–59, 60–69, 70–79, 80+). Body surface area (BSA) and Body Mass Index (BMI) were calculated from height and weight measured at the time of first ESRD only.
Formulas applied included:
- BSA = 0.007184 xheight(centimeters)0.725 xweight(kilograms)0.425
- BMI = Weight(kilograms)/ Height(meters)2 BMI was grouped as <18.5 vs. ≥18.5
- Duration of Renal Replacement Therapy (RRT) <= 4 months (new to dialysis): For each month in which a patient was receiving dialysis and covered by Medicare, we indicated whether it occurred within the first 4 months in which dialysis was first received for at least half of the month.
- Acute and chronic conditions: We determined acute and chronic conditions from ICD-9-CM codes in Medicare claims. The list of codes used was the list published by CMS for use in PPS reimbursement as of October 1, 2011. 16
We searched all institutional claims (Medicare Part A) and Physician/Supplier claims (Medicare Part B) from 2000 to 2008. For acute conditions (pericarditis, bacterial pneumonia, gastrointestinal bleeding), once a claim was found, a patient was considered to have the acute condition for that month and three subsequent months, per the PPS reimbursement rules. Any claim within these four months was considered to be part of the same acute episode. For chronic conditions (hereditary hemolytic and sickle cell anemia, monoclonal gammopathy, myelodysplastic syndrome), once a claim was found, a patient was considered to have the condition for that and all subsequent months.
For hereditary hemolytic and sickle cell anemia, we also used the primary cause of renal failure (from the Medical Evidence (MEDEVID) SAF file) to determine whether the condition was present. Patients who had a diagnosis of “28260 Sickle cell disease/anemia” or “28269 Sickle cell trait and other sickle cell (HbS/Hb other)” were considered to have the condition for all dialysis months. For myelodysplastic syndrome, the ICD-9-CM codes originally listed for this analysis were created in October 2011. Prior to 2007, the code 238.7 was the only code used. Therefore, we included 238.7 in our search for this condition.
Validity and reliability of estimates
Two methods were used to validate the study results. First, prevalence estimates were compared to those for a 2002-2004 Medicare patient cohort published in a 2008 report to Congress, 17which used very similar, although not identical, case-mix category definitions to those ultimately promulgated by CMS in the 2011 PPS Final Rule and subsequent regulations. Second, an estimated mean patient payment multiplier was calculated for the 2006-08 cohort by running the prevalence estimates through a validated PPS payment multiplier model.18This was then compared to the estimated mean patient payment multiplier published in the 2011 PPS Final Rule for the same time period.
Per the 2011 Final Rule, the standardization adjustment including all facility and patient factors was .9407, based on CMS’ analysis of 2006-08 claims and cost reports.1 Payment adjustments were therefore expected to total 1/.9407 = 1.063 times the PPS base rate (about $14-15 per treatment). This standardization adjustment included the estimated effect of the facility low-volume adjuster (LVA). In order to estimate the patient-only case-mix multiplier, we estimated the facility-level adjustments for the LVA to be worth 0.3% of payments, using data from the 2011 Final Rule Provider-level Impact file.19 Backing out the 0.3% from the estimated mean case-mix multiplier leaves 6% for patient-level adjusters.
The number of patients and number of patient-months in each 3-year cohort are shown below:
2000–2002: Patients = 408,208; Patient months = 7,626,998
2003–2005: Patients = 454,965; Patient months = 8,679,756
2006–2008: Patients = 481,134; Patient months = 9,362,698
Table 1 shows the prevalence of each patient-level case-mix adjuster in each of the three 3-year cohorts, as well as the prevalence reported for 2002-2004 in the Leavitt Report. 17 The estimated average patient payment multiplier using the 2006–08 prevalence estimates from this study was 1.072, which is higher than the 1.060 mean patient multiplier estimated from the 2011 PPS Final Rule.
Trends in patient case-mix
Large percentage changes were observed in the prevalence of many case-mix categories between 2000–02 and 2006–08. Large increases were observed in all chronic comorbidities [myelodysplastic syndrome (+169%), monoclonal gammopathy (+140%), hemolytic and sickle cell anemias (+94.4%)] as well as in prevalence of bacterial pneumonia (+26.7%) and patients aged 80+ (+16.1%) (see Figure 2).
Large declines were observed in prevalence of low BMI patients (–33.4%), prevalence of pericarditis (–23.4%), patients in the first 4 months of dialysis (–13.0%), and patients aged 18-44 (–12.5%) (see Figure 3).
There were large changes in patient-level case-mix adjuster prevalence in the Medicare ESRD population between 2000–02 and 2006–08. Important demographic changes included:
- New dialysis patients as a percent of total declined, with fewer patients in the high payment adjuster age cohort 18–44, and increasing prevalence of chronic comorbidities. Further, there is evidence from the USRDS that at least some of these trends have continued after 2008 (see Figure 1). Such changes in case-mix prevalence have implications for payment. For instance, lower prevalence of patient categories with high payment adjusters may result in lower than expected payments program-wide, because dollars set aside for these adjusters are not paid out.
Alternatively, and as has happened with other prospective payment systems in the past, there is the potential that the financial incentives to “upcode” patient case-mix could influence dialysis provider behavior such that reported case-mix prevalence increases for some conditions.
Under a prospective payment system, there is a direct link between payments and the medical record. This gives providers incentives to increase the accuracy and completeness of medical record documentation and coding. 20For example, after Medicare’s Inpatient PPS for hospitals (IPPS) was implemented, the cumulative formal update from the beginning of the IPPS through fiscal year 1988 totaled less than 8%, but the cumulative increase in payments due to case-mix change for the same period was over 20% .20In the 2011 PPS Final Rule, CMS noted that, “Historically, there has not been a financial incentive for ESRD facilities to document the presence of co-morbidities. We believe that by including co-morbidity adjustments under the ESRD PPS, ESRD facilities will implement more active processes for gathering diagnostic information.” 1
An equitable payment system should encourage more accurate coding but not allow coding changes to generate additional payments unrelated to patient need. 20However, under the ESRD PPS there is no real opportunity for upcoding as there was under Medicare’s IPPS, since the ESRD case-mix is limited to only a few diagnosis-related (versus procedure-related) conditions plus age, and body height and body weight measures. More specifically, the design and selection of case-mix factors for the ESRD PPS explicitly sought to avoid adjusters that could be outcomes of or influenced by the dialysis treatment process or that would create adverse incentives, e.g., extra payments for the treatment of access site infections. 17 The highest payment adjuster category––a patient new to dialysis––is not coded by the provider. Age can be verified from other documentation. The only real opportunities for upcoding are in the diagnosis codes for acute and chronic conditions. However, this study examined changes in the prevalence of comorbidities in the years prior to the implementation of the ESRD PPS and did not rely on dialysis facility coding of these conditions. As such, any observed increase in prevalence of diagnosis-related conditions is most likely a measure of real case-mix changes versus medical documentation and coding practice changes.
While some trend in ESRD patient case-mix may have occurred due to better dialysis facility coding starting in 2011, this study did not use dialysis facility coding. Further, dialysis facilities face serious difficulties in capturing the necessary documentation in order to be reimbursed appropriately for patients with acute and chronic comorbidities for which there is a payment adjuster. An analysis by the Moran Company in 2011 found that the probability of identification of a co?morbid condition increases with access to specialty physician and hospital claims from the prior year for acute conditions and claims from multiple past years for chronic conditions. 21The physicians who most often diagnose these conditions are oncologists, hematologists, primary care physicians, hospital physicians, gastroenterologists, and cardiologists and these specialists are often uncooperative in providing diagnostic records to dialysis facilities, citing HIPAA restrictions. 22 As a result, the prevalence of diagnosis-based case-mix adjusters as measured from dialysis claims post-2011 is unlikely to be higher than what was observed from all provider-type claims used in the setting of the case-mix adjusters.
This study had a few limitations that should be noted. This was a cross-sectional population analysis; therefore the effect of changes in the patients included in each 3-year cohort on changes in case-mix adjuster prevalence was not examined. Secondly, the study period ended with 2008 data. Further changes between 2008 and the implementation date of the PPS (2011) or beyond were not determined. Finally, due to data limitations, there were some slight methodological differences in case-mix prevalence calculations in this study from how Medicare calculates case-mix for reimbursement. This may account for the slight difference in average patient payment multiplier estimated by this study as compared to the computation published by Medicare, and for some discrepancies in prevalence estimates reported in this study vs. the 2008 Leavitt Report.17
Substantial changes in the prevalence of case-mix adjusters were observed over time since 2000. The PPS payment formula may benefit from being regularly updated to reflect case-mix changes, in order to ensure that the payments designed to compensate dialysis facilities for higher cost cases are appropriately paid out. As part of any payment system update, Medicare should consider publishing the prevalence of each case-mix adjuster used to calculate reimbursement factors, in order to improve the transparency in the rate-setting process. -by Christopher S. Hollenbeak, PhD; Robert J. Rubin, MD; Spiros Tzivelekis, MSc; Mark Stephens, BA
JMS has received consulting fees from Amgen Inc. related to this study. CH received funds from Prima Health Analytics for this study. ST is an employee of Amgen Inc., and owns stock or stock options in Amgen Inc. RR is a consultant to Amgen Inc. but received no fees related to this manuscript. The authors wish to thank Eric Shaefer, The Pennsylvania State University College of Medicine, who provided data programming support for this study.
1. Centers for Medicare & Medicaid Services (US). ESRD prospective payment system final rule, regulation number CMS-1418-F. 2010. Available at: http://edocket.access.gpo.gov/2010/pdf/2010-18466.pdf. Accessed August 12, 2010.
2. Centers for Medicare & Medicaid Services (US). ESRD prospective payment system proposed rule, regulation number CMS-1418-P. 2009. Available at: http://www.gpo.gov/fdsys/pkg/FR-2009-09-29/pdf/E9-22486.pdf. Accessed July 14, 2014.
3. Dialysis Clinic, Inc. Comment Letter on 2011 ESRD PPS Proposed Rule. 2009. Available at: http://www.dciinc.org/fckeditor/file/Dialysis%20Clinic%20Inc%20Proposed%20Bundling%20Comments.pdf. Accessed July 11, 2014.
4. Kidney Care Partners. Comment Letter on 2011 ESRD PPS Proposed Rule. 2009. Available at: http://kidneycarepartners.com/files/KCP_ESRD_PPS_Proposed_Rule_Comment_Letter_FINAL.pdf. Accessed July 11, 2014.
5. Mayne TJ, Burgess M, Weldon J. Finding the case mix adjusters in the bundle. Nephrol. News Issues 2010;24(2):34, 37-38, 40.
6. Mayne TJ, Jonckheere J, Zumwalt L, Krishnan M. Case mix adjusters continue to be difficult to detect. 2011. Available at: http://www.davitaclinicalresearch.com/Userfiles/pdfs/ASN-2011/ASN_Moran_v7final.pdf. Accessed June 20, 2013.
7. Centers for Medicare & Medicaid Services (US). ESRD prospective payment system 2012 proposed rule, Regulation CMS-1577-P. 2011. Available at: http://www.gpo.gov/fdsys/pkg/FR-2011-07-08/pdf/2011-16874.pdf. Accessed July 9, 2014.
8. United States Renal Data System. USRDS 2014 Annual Data Report: Atlas of Chronic Kidney Disease and End-Stage Renal Disease in the United States. Bethesda, Md.: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases.; 2014. Available at: http://www.usrds.org/2014/view/Default.aspx. Accessed November 18, 2014.
9. Kidney Care Partners. Comment Letter on 2014 ESRD PPS Proposed Rule. 2013. Available at: http://www.kidneycarepartners.com/filemanager/files/KCP%20Final%20CY%202014%20PPS%20PY%202016%20QIP%20Letter.pdf. Accessed September 5, 2013.
10. Farley DO. Competition under fixed prices: Effects on patient selection and service strategies by hemodialysis providers. Med. Care Res. Rev. 1996;53(3):330-349. doi:10.1177/107755879605300307.
11. Hirth RA, Held PJ, Orzol SM, Dor A. Practice patterns, case mix, Medicare payment policy, and dialysis facility costs. Health Serv. Res. 1999;33(6):1567-1592.
12. Medicare Payment Advisory Commission. Comment Letter on 2014 ESRD PPS Proposed Rule. 2013. Available at: http://www.medpac.gov/documents/08302013_MedPAC_ESRD_COMMENT.pdf. Accessed September 5, 2013.
13. Government Printing Office (US). American Taxpayer Relief Act of 2012.; 2013:115-116. Available at: http://www.gpo.gov/fdsys/pkg/BILLS-112hr8eas/pdf/BILLS-112hr8eas.pdf. Accessed January 29, 2013.
14. Centers for Medicare & Medicaid Services. End-Stage Renal Disease Prospective Payment System, Quality Incentive Program 2015 Proposed Rule. Washington D.C: Department of Health and Human Services; 2014. Available at: http://www.gpo.gov/fdsys/pkg/FR-2014-07-11/pdf/2014-15840.pdf. Accessed July 11, 2014.
15. United States Renal Data System. USRDS 2009 Researcher’s Guide to the USRDS Database. Bethesda, Md.: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases.; 2010.
16. Centers for Medicare & Medicaid Services (US). ESRD Prospective Payment System (ESRD PPS) Co-morbidity Categories Diagnostic Codes Effective 10-1-2011. 2011. Available at: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/ESRDpayment/downloads/ESRDPPSCo-morbidityCategoriesDiagnosticCodesEffective10-1-2011.pdf. Accessed June 7, 2013.
17. Leavitt MO. A Design for a Bundled End Stage Renal Disease Prospective Payment System. Congress; 2008. Available at: http://www.cms.gov/Medicare/End-Stage-Renal-Disease/ESRDGeneralInformation/downloads/ESRDReportToCongress.pdf. Accessed September 9, 2014.
18. Stephens M, Maione-Downing B, Neumann PJ, Rubin RJ, Tzivelekis S. The Relative importance of patient case-mix to payments under the ESRD Prospective Payment System (PPS). 2014. Available at: https://ww5.aievolution.com/nkf1401/files/content/abstracts/9963/0417_Stephens.pdf. Accessed April 16, 2014.
19. Centers for Medicare & Medicaid Services (US). CY 2011 ESRD PPS Final Rule Provider Level Impact File. 2010. Available at: http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/ESRDpayment/Downloads/CY_2011_Final_ESRD_PPS_Facility_Level_Impact_File.zip. Accessed February 21, 2013.
20. Steinwald B, Dummit LA. Hospital case-mix change: sicker patients or DRG creep? Health Aff. (Millwood) 1989;8(2):35-47. doi:10.1377/hlthaff.8.2.35.
21. The Moran Company. Review of 5% Random Sample of Medicare SAF File for 2007, 2008 Industry Claims. Outpatient Claims Include Rural Health Centers, Federally?Qualified Health Centers, Rehab Facilities, and Psychiatric Facilities. Available upon request; 2011.
22. Kidney Care Partners. Comment Letter on 2012 ESRD PPS Proposed Rule. 2011. Available at: http://kidneycarepartners.com/files/WASHINGTON-5189440-v1-KCP_PPS_CY_2012_QIP_PY_2013_2014_Comment_Letter_%28FINAL%29.PDF. Accessed September 5, 2013.