2010 USRDS Annual Data Report
This Chapter
Download
Download PDF
*corresponding data in Excel included
Search This Page
Search All
Translate

Appendices

Sections this chapter: 

In this appendix we present details on the USRDS database, its standardized working datasets and specialized code definitions, and our common data processing practices. We also describe the statistical methods used in this ADR. The Researcher's guide to the USRDS database, available online, provides additional information about the database and Standard Analysis Files.

Data Sources

The USRDS maintains a stand-alone database with data on diagnoses and demographic characteristics of ESRD patients, along with biochemical data, dialysis claims, and information on treatment and payor histories, hospitalization events, deaths, physician/supplier services, and providers.

REMIS/REBUS/PMMIS DATABASE

The major source of ESRD patient information for the USRDS is the Renal Beneficiary and Utilization System (REBUS) of the Centers for Medicare and Medicaid Services (CMS, formerly HCFA), adopted in 1995 as the On-Line Transaction Processing system from the previous Program Management and Medical Information System (PMMIS) database. The REBUS/PMMIS database contains demographic, diagnosis, and treatment history information for all Medicare beneficiaries with ESRD. The database has also been expanded to include non-Medicare patients, as discussed later in this appendix. Having advanced its database technology, CMS migrated the REBUS database into an Oracle relational database in the fall of 2003, including all patients who were alive and had ESRD as of January 1, 1995, or who were incident after this date. This database is known as the Renal Management Information System (REMIS).

CMS updates the REMIS/REBUS/PMMIS database on a regular basis, using the Medicare Enrollment Database (EDB), Medicare inpatient and outpatient claims, the Organ Procurement and Transplantation Network (OPTN) transplant database, ESRD Medical Evidence forms (2728) provided by the ESRD networks, and ESRD Death Notification forms (2746) obtained from renal providers, as well as the Standard Information Management System (SIMS) database of the ESRD networks. CMS has also established data integrity rules to ensure accurate identification of patients in the SIMS and CMS databases. Each ESRD patient is now identified with a unique patient identification number common to both databases, ensuring that data on all patients are consistently managed over time.

CMS MEDICARE ENROLLMENT DATABASE

The Medicare Enrollment Database (EDB) is the designated repository of all Medicare beneficiary enrollment and entitlement data, and provides current and historical information on residence, Medicare as secondary payor (MSP) and employer group health plan (EGHP) status, and Health Insurance Claim/Beneficiary Identification Code (HIC/BIC) cross-referencing.

ESRD MEDICAL EVIDENCE FORM (CMS 2728)

The ESRD Medical Evidence (ME) form is used to register patients at the onset of ESRD, and must be submitted by dialysis or transplant providers within 45 days of initiation. The form establishes Medicare eligibility for individuals previously not Medicare beneficiaries, reclassifies previously eligible beneficiaries as ESRD patients, and provides demographic and diagnostic information on all new patients. The CMS, USRDS, and renal research communities rely on the form to ascertain patient demographics, primary diagnosis, comorbidities, and biochemical test results at the time of ESRD initiation. Before 1995, units were required to file the ME form only for Medicare-eligible patients. Since the 1995 revision, however, providers are required to complete the form for all new ESRD patients.

The third major revision of the ME form, released in May, 2005, was intended to remedy several shortcomings found in the 1995 form and its earlier version. Key additions target pre-ESRD care and vascular access use, and additional new fields collect information on glycosylated hemoglobin and lipid testing, on the frequency of hemodialysis sessions, and on whether patients are informed of transplant options.

This form is the only source of information about the cause of a patient's ESRD. Because the list of diseases has been revised, the USRDS stores the codes from each version so that detail is not lost through conversion of one set of codes to the other.

ESRD DEATH NOTIFICATION FORM (CMS 2746)

The ESRD Death Notification form is used to report the death of ESRD patients. According to CMS policy, this form must be submitted by dialysis or transplant providers within 30 days of a patient's death, and provides the date and causes of death (primary and secondary), reasons for discontinuation of renal replacement therapy, if applicable, and evidence of hospice care prior to death. It is the primary source of death information for CMS and the USRDS, identifying more than 99 percent of deaths. The USRDS also utilizes the Social Security Administration's (SSA) Death Master File as a supplemental data source for ascertaining death in a small group of lost-to-follow-up ESRD patients; this file, however, identifies only all-cause deaths.

OPTN TRANSPLANT DATABASE

In the early 1980s CMS began collecting data on all Medicare kidney transplants. In 1988, the United Network of Organ Sharing (now OPTN) was created to provide a national system for allocating donor organs. OPTN also began collecting data on all transplants. These two efforts were consolidated in 1994, and OPTN became the single source of data on transplant donors and recipients.

The CMS and OPTN transplant data files overlap for 1988–1993, and some patients with ME forms indicating transplant as the initial modality are not included in either file. To resolve conflicts among the three sources, the USRDS adopts the following procedure:

OPTN transplants are accepted into the database.
CMS transplants before 1988 are accepted.
CMS transplants from 1988 to 1993 are accepted if there is no OPTN transplant record for that patient within 30 days of the CMS transplant.
Transplants indicated on ME forms are accepted if there is no previously accepted record of a transplant for that patient within 30 days of the date listed on the ME form.

CMS STANDARD ANALYTICAL FILES (SAFs)

These files contain billing data from final action claims, submitted by Medicare beneficiaries with ESRD, in which all adjustments are resolved. For inpatient/outpatient institutional claims we use the following data: inpatient, 100 percent SAF; outpatient, 100 percent SAF; home health agency (HHA), 100 percent SAF; hospice, 100 percent SAF; and skilled nursing facility (SNF), 100 percent SAF. For physician/supplier claims, we use: physician/supplier, 100 percent SAF; and durable medical equipment (DME), 100 percent SAF.

CMS SAFs are updated each quarter through June of the next year, when the annual files are finalized. Datasets for the current year are created six months into the year and updated quarterly until finalized at 18 months, after which they are not updated to include late arriving claims. Annual files are thus approximately 98 percent complete. The USRDS 2010 ADR includes all claims up to December 31, 2008. Patient-specific demographic and diagnosis information, however, includes data as recent as October, 2009.

Inpatient transplant and outpatient dialysis claims records are used to identify new ESRD patients for whom no ME form has been filed. These patients, primarily non-Medicare patients, or beneficiaries who develop ESRD while on Medicare because of age or disability, will eventually be entered into the REMIS/REBUS/PMMIS — and hence the USRDS — database through the claims records. For patients without ME forms these claims are the only reliable information from which to determine first ESRD service dates. These paid claims records are, however, only a supplement to, rather than a replacement of, other sources of information on incidence and prevalence.

The problem of timely identification has lessened since the revision of the ME form in April 1995, and the amended ESRD entitlement policy that now requires the form to be submitted for all ESRD patients regardless of insurance and eligibility status.

CMS 5 percent STANDARD ANALYTICAL FILES (SAFs)

These files contain billing data from final action claims submitted by Medicare beneficiaries, in which all adjustments have been resolved. The claims data are selected randomly from general Medicare claims (final action claims) using five combinations of the last two digits of the CMS Health Insurance Claims (HIC) number: 05, 20, 45, 70, and 95. Since the same two-digit numbers are used each year, one should expect to see the same beneficiaries in these annual datasets. These claims are categorized into the inpatient (IP), outpatient (OP), home health agency (HHA), hospice (HS), skilled nursing facility (SNF), physician/supplier (PB), and durable medical equipment (DME) SAFs.

The files are updated each quarter through June of the next year, when annual files are finalized. Datasets for the current year are created six months into the year and updated quarterly until finalized at 18 months, after which they are not updated to include late arriving claims. Annual files are thus approximately 98 percent complete. The USRDS 2010 ADR includes all claims up to December 31, 2008.

STANDARD INFORMATION MANAGEMENT SYSTEM (SIMS) DATABASE (ESRD NETWORKS)

The USRDS continues to collaborate with CMS and the ESRD networks to address data tracking issues relating to non-Medicare ESRD patients. Past ADRs have documented the lack of consistent Medicare claims data among these patients. Working solely with data from the ME form, the USRDS could establish the first ESRD service date, but could not generate a more detailed treatment history. With the integration of the SIMS event data into the USRDS database, however, we can now address issues in the non-Medicare ESRD population such as the large and growing number of lost-to-follow-up patients, and look as well at patients for whom there previously were no data on initial modality or death. This data integration is detailed in the section on data management and preparation.

CMS DIALYSIS FACILITY COMPARE DATA

The USRDS uses the CMS Dialysis Facility Compare data to define chain and ownership information for each renal facility. Prior to the 2003 ADR, similar data were extracted from the Independent Renal Facility Cost Report (CMS 265-94).

ESRD CLINICAL PERFORMANCE MEASURES PROJECT

CMS developed its ESRD Clinical Performance Measures Project (CPM, formerly the ESRD Core Indicators Project) to collect information on the quality of care provided to dialysis patients. The data originate from data collection forms completed by staff at primary care facilities, and focus on dialysis adequacy measures, anemia management, and vascular access. Additional clinical parameters such as albumin are available as well. These data have been collected annually since 1994, using a random sample of adult (age 18 and older) patients alive and on dialysis at the end of each calendar year; on average, roughly 8,500 adult in-center hemodialysis patients and 1,500 peritoneal dialysis patients are surveyed each year. Data collection for all hemodialysis patients age 12–17 was begun in 2000. Collection was then expanded in 2002 to all in-center hemodialysis patients younger than 18, and in 2005 to all peritoneal dialysis patients of this age. The USRDS Coordinating Center, in collaboration with CMS, is now making these ESRD CPM data available to the general research community.

MEDICARE CURRENT BENEFICIARY SURVEY (MCBS)

The Medicare Current Beneficiary Survey is a longitudinal survey of a nationally representative sample of aged, disabled, and institutionalized Medicare beneficiaries. The MCBS contains information on the health status, health care use and expenditures, drug prescriptions, health insurance coverage, and socioeconomic and demographic characteristics of the entire spectrum of Medicare beneficiaries. Data are made available by CMS in two datasets: Access to Care (1992–2007), and Cost and Use (1992–2006), with the 2007 and 2006 files, respectively, the latest updates for the 2010 ADR.

In the fall of 1991, the MCBS began to be conducted three times per calendar year (winter, summer, and fall), and in 1994 a sample rotation scheme was introduced. Survey participants are kept in the sample for four years, with approximately one-third rolling off, and with new participants added each fall to keep the overall sample size at approximately 12,000 each calendar year.

CMS PRESCRIPTION DRUG EVENT (PDE) FILE

In December 2003, Congress passed the Medicare Prescription Drug, Improvement, and Modernization Act (MMA), amending the Social Security Act by adding Part D under Title XVIII. With this new Part D coverage, health plans must submit a summary record called the prescription drug event (PDE) record to CMS whenever a Medicare beneficiary fills a prescription. The PDE record contains 37 data elements; the USRDS receives PDE records with 30 data elements and excluding a few non-critical fields. Each drug is identified by a National Drug Index (NDC) code; the record also contains prescription dosing information, drug costs above and below the out-of-pocket threshold, other true out-of-pocket (TrOOP) amounts, plan paid amounts, and low-income cost-sharing subsidy amounts.

Due to delays in the availability of the data, only the 2006 and 2007 PDE files were available for the 2010 ADR. The USRDS will, however, include both 2008 and 2009 PDE data in its 2011 ADR.

THOMSON Reuters MARKETSCAN DATA

The Thomson Reuters MarketScan Commercial Claims and Encounters Database includes specific health services records for employees and their dependents in a selection of large employers, health plans, and government and public organizations. The database includes nine files: Annual Enrollment Summary Table, Enrollment Detail Table, Inpatient Admissions Table, Inpatient Services Table, Outpatient Services Table, Outpatient Pharmaceutical Claims Table, Facility (Inpatient and Outpatient) Header Table, Aggregated Populations Table, and the Red Book (prescription drug information by National Drug Code). The strength of this database lies in the quality of its cost information, where claims data include actual paid dollars and net payments by the insurer.

The MarketScan database links billing and encounter data to detailed patient demographic and enrollment information across sites and types of providers, and over time from 1999 to 2008, and includes commercial health data from approximately 100 payors. About 80 percent of those covered are self-insured. Each year the database contains health data for about 10.5 million people. For details about the MarketScan data, please visit www.usrds.org.

INGENIX i3 DATA

The Ingenix i3 database is a commercial and non-capitated health plan database covering employees from multiple employers within a single insurer. In addition to the usual service encounter and drug data, similar to that of the MarketScan database, this database also includes laboratory data, allowing for comparisons between claims-based and lab-based definitions of diseases. In order to protect the discount structure of its business, the billing data of this single insurer discloses only charged dollars without actual paid amounts or the portion paid by the insurer.

The Ingenix i3 database links billing and encounter data to detailed demographic and enrollment information of individual employees from 2000 to 2008, and contains health data for approximately 14 million people annually. For details about what is contained in the Ingenix i3 data, please visit our website at www.usrds.org.

NATIONAL HEALTH & NUTRITION EXAMINATION SURVEY (NHANES)

NHANES is a series of health examination surveys conducted by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention (CDC). Begun in 1960, NHANES is designed to monitor the health and nutritional status of the non-institutionalized civilian population in the United States. NHANES III was conducted in two phases between 1988 and 1994. In 1999, NHANES became a continuous annual survey to allow annual estimates, with release of public-use data files every two years. Both NHANES III and NHANES 1999–2006 were nationally representative cross-sectional surveys and used a complex, stratified, multistage probability cluster sampling design that included selection of primary sampling units (counties), household segments within the counties, and sample persons from selected households. Survey participants were interviewed in their homes and/or received standardized medical examinations in mobile examination centers. Both surveys over-sampled African Americans, Mexican Americans, and individuals age 60 or older to improve the estimates for these subgroups.

ANNUAL FACILITY SURVEY (AFS)

Independent ESRD patient counts are available not only from the CMS ESRD database, but also from CMS's Annual Facility Survey (CMS 2744), which all Medicare-certified dialysis units must complete at the end of each year. The AFS reports counts of patients being treated at the end of the year, new ESRD patients starting treatment during the year, and patients dying during the year. Both Medicare and non-Medicare end-of-year patients are counted. While AFS files do not carry patient-specific demographic and diagnosis data, they provide independent patient counts used to complement the CMS patient-specific records. Starting with the 2005 AFS, CMS stopped posting data from these surveys on the web. And beginning with the 2007 ADR, the USRDS has extracted the relevant facility survey data directly from the SIMS database.

CDC SURVEILLANCE

The CDC used its National Surveillance of Dialysis-Associated Diseases to collect data from U.S. dialysis facilities on patient and staff counts, membrane types, reuse practices, water treatment, therapy, vascular access use, antibiotic use, hepatitis vaccination and conversion rates, and the incidence of HIV, AIDS, and tuberculosis. No data are patient-specific. The CDC did not conduct a survey in 1998, and terminated this program after 2002.

UNITED STATES CENSUS

In rate calculations throughout this year's ADR we use data from the 2000 U.S. Census, and also incorporate CDC population estimates by race. Our methods are described on later in this appendix.

Data management & Preparation

Our main computer system is based on a VMS cluster running Alpha EV6 processors. We currently maintain three nodes in the cluster: three 4-CPU (i.e. Alpha EV6 processor) servers, each with 16-GB RAM memory. Through the HP Advanced Server System, we map VMS directories to network shares accessible to Windows clients as mapped network drives. The Alpha EV6s are connected to 30 terabytes of RAID-5 (Redundant Array of Independent Disks, level 5) disk farms, which are managed by three interconnecting high-speed disk controllers via Fibre Channel. All data in disk farms are independently accessible through Alpha server nodes.

We use SAS database management system and development tools as our core database technology platform; this differs from the Oracle RDBMS system used by the previous contractor only in physical data allocation and management. All information in the earlier system was integrated into the new database, and its continuity and completeness are maintained.

DATA LOADING & CLEANING

Data files come to the USRDS in IBM 3490 and 3490e cartridges/CD-ROMs with EBCDIC, ASCII, or SAS formats. Due to increased awareness of and concerns over data security and patient privacy protection, in 2008 CMS began delivering most of the USRDS requested data via a dedicated and secured T1 line connection. CMS has also instituted data encryption procedures on all out-bound data regardless of file format and transportation medium. Once loaded and decrypted, files are converted into SAS datasets for processing, and a series of data verification steps is completed to ensure data quality and integrity before updating the USRDS database.

DATABASE UPDATES

For this ADR, patient demographic and diagnosis data are updated through October, 2009, and Medicare inpatient/outpatient and physician/supplier claims through December 31, 2008.

ESRD PATIENT DETERMINATION

A person is identified as having ESRD when a physician certifies the disease on the CMS ME form, or when there is other evidence of chronic dialysis or a kidney transplant. Patients with acute kidney failure who are on dialysis for days or weeks, but who then recover kidney function, are excluded from the database if their ME forms have not been submitted. Patients who die soon after kidney failure without receiving dialysis are sometimes missed.

The ESRD First Service Date (FSD) is the single most important data element in the USRDS database, and each patient must, at a minimum, have a valid FSD. This date is used to determine the incident year of each new patient and the first year in which the patient is counted as prevalent. The date 90 days after the FSD is used as the starting point for most survival analyses.

The FSD is derived by taking the earliest of the date of the start of dialysis for chronic kidney failure, as reported on the ME form; the date of a kidney transplant, as reported on a CMS or OPTN transplant form, an ME form, or a hospital inpatient claim; or the date of the first Medicare dialysis claim. Most FSDs are obtained from the ME form. In the absence of this form, the date of the first Medicare dialysis claim or transplant usually supplies the FSD. In the few cases in which the date of the earliest dialysis claim precedes the first dialysis date reported on the ME form, the earliest claim date is used as the FSD. However, starting with the 2007 ADR, a patient entering into the ESRD program after December 31, 1994, has his or her FSD defined solely by the regular dialysis start date or the preemptive transplant date, whichever is earliest, on the ME form. This new method of determining the FSD aligns more closely to the methods used by CMS. After careful monitoring and repeated comparative analyses of the traditional USRDS method to the new ME method, the USRDS began applying the ME method to incident patients entering into the ESRD program on or after January 1, 1995.

MEDICARE & NON-MEDICARE ('ZZ') PATIENTS

Beneficiaries are enrolled in Medicare based on criteria defined in Title XVIII of the Social Security Act of 1965, and in subsequent amendments to the act. A person in one of these four categories is eligible to apply for Medicare: age 65 and over, disabled, ESRD program, and Railroad Retirement Board (RRB).

Most ESRD patients are eligible to apply for Medicare as their primary insurance payor. Some, however, are not immediately eligible for Medicare coverage because of their employment status and insurance benefits. These patients are usually covered by employer group health plans (EGHPs), and must wait 30–33 months before becoming eligible to have Medicare as their primary payor. Some of these patients, particularly new patients since 1995, have FSDs established by ME forms, but have no dialysis claims or hospitalization events in the CMS claims database. In the REBUS/PMMIS database all non-Medicare ESRD patients are assigned a code of 'ZZ' in the two-character Beneficiary Identification Code field. CMS does not generally include these patients in the datasets released to researchers.

The USRDS recognizes that 'ZZ' patients are true ESRD patients, and should be included in patient counts for incidence, prevalence, and modality. Calculations of standardized mortality ratios, standardized hospitalization ratios, and standardized transplantation ratios, however, should not include these patients because of the small number of claims available in the first 30–33 months after their first ESRD service. Furthermore, it may not be possible to link 'ZZ' patients to their ESRD Death Notification forms or the OPTN transplant data, or to determine comorbidity or inpatient/outpatient and physician/supplier services. Because such data are limited, event rates that include these patients must be assessed with caution.

We continue to include 'ZZ' patients in the mortality rate calculations of the ADR. We are collaborating with CMS and other interested researchers to establish a consistent approach to managing the data for these patients. The USRDS, in working with CMS, has been able to resolve most of the 'ZZ' patients since the release of the ESRD Patient Database, REMIS, in the fall of 2003. According to our most recent assessment — performed during the production of the 2007 ADR — we have determined that at least 99 percent of 'ZZ' patients have been resolved due to significant advancements in the REMIS/REBUS database system.

DEATH DATE DETERMINATION

After the ESRD First Service Date, the date of death is the most critical piece of information in the ESRD database. Death dates are obtained from several sources, including the CMS Medicare Enrollment Database, CMS forms 2746 (ESRD Death Notification form) and 2728 (ESRD Medical Evidence form), the OPTN transplant follow-up form, the ESRD Network SIMS database, and the Social Security Death Master File. Because multiple sources report death information for the same patient, one patient may have several reported dates. The USRDS therefore uses an algorithm to determine the date of death. EDB information is given first priority, and, in the absence of an EDB death date, other sources are evaluated in the following order: form 2746, form 2728, SIMS data, the transplant follow-up form, and, if no other death date is available, the Death Master file.

Lost-to-follow-up methodology

The USRDS uses all available data to create a treatment history for each patient in the database, including all modality events, their duration, and the renal providers involved in each patient's care.

Gaps frequently exist in the billing data upon which modality periods are based. The USRDS assumes that a modality continues until death or the next modality-determining event. A patient with a functioning transplant is assumed to maintain it unless a transplant failure or death notification is encountered in the data. In the absence of a death notification, dialysis claims, or other confirmation of a continuing modality, a dialysis modality, in contrast, is assumed to continue for only 365 days from the date of the last claim. After this period the patient is declared lost-to-follow-up until the occurrence of a dialysis claim or transplant event.

Because Medicare may be the secondary payor for up to the first 30–33 months of ESRD, delaying the submission of Medicare dialysis claims, lost-to-follow-up categorization cannot begin until the end of the third year after the start of ESRD service. This "first three-year rule" is particularly important for non-Medicare patients, who may be followed for up to three years with limited event or mortality data. These patients would contribute dialysis or transplant days to the denominator of rate calculations, but only questionable event data to the numerator. In comparison to the two-year rule used in the 2001 ADR, this three-year rule significantly reduces the number of lost-to-follow-up patients in the prevalent population.

A number of events can result in a lack of dialysis data and eventual reclassification of a patient as lost-to-follow-up:

The patient may have recovered renal function (RRF) and no longer have ESRD. For a valid patient classification, this event must occur within 180 days of the FSD, and the RRF period must persist for at least 90 days.
The patient may have left the country.
The patient may receive dialysis covered by a payor other than Medicare, or have received a transplant not paid for by Medicare or reported to OPTN.
The patient may be enrolled in a Medicare HMO, so that Medicare dialysis claims are not generated even though the patient is eligible for Medicare coverage.
The patient's death may not have been reported to the Social Security Administration or to CMS.

INTEGRATION OF THE USRDS, SIMS, & REMIS DATABASES

We have worked to reconcile ESRD patients in the SIMS, REMIS, and USRDS databases. We have analyzed each database for duplicate records, consolidated these records, and integrated the databases. Data were then re-analyzed for duplicates, which were themselves consolidated. This consolidation of patients is an ongoing collaborative effort between the ESRD Networks, CMS, and the USRDS.

Treatment histories compiled by the USRDS rely on Medicare dialysis billing records, which contain no information on dialysis therapy or modality changes in non-Medicare patients. Beginning with the 2003 ADR, we incorporate treatment-specific information from the ESRD Networks' SIMS event database to improve the tracking of these patients in the USRDS database, and of patients who are considered lost-to-follow-up. Efforts to integrate the USRDS, SIMS, and REMIS databases continue to pay dividends in reducing the number of lost-to-follow-up patients.

We continue to take a conservative approach to incorporating SIMS Event History data into the USRDS treatment history; as we learn more about the data, we may expand this approach. We currently make the following updates on an annual basis:

The USRDS database is updated with mortality data from the SIMS event database. The database is updated for each incident patient whose initial modality is listed as "unknown dialysis," and for whom the SIMS database lists a known dialytic modality within 90 days of the established first ESRD service date.
Data on non-Medicare "lost-to-follow-up" patients are substituted with available SIMS treatment information.
Since the 2007 ADR we have included the RRF event in the modality sequence, reducing lost-to-follow-up episodes for prevalent patients. This event is now established in our database only if it occurs within the first 180 days of the FSD and lasts for at least 90 days, a definition more conservative than that in the SIMS event database.

60-DAY STABLE MODALITY RULE: TREATMENT HISTORY

This rule requires that a modality continue for at least 60 days before it is considered a primary or switched modality. It is used to construct a patient's modality sequence, or treatment history, so that incident and prevalent patients are known to have stable and established modalities. Starting with the 2003 ADR, all descriptive data in the incident, prevalent, and modality sections are based on incident and prevalent cohorts produced from the modality sequence without using this rule. In analyses of patient outcomes such as hospitalization and mortality, in contrast, this rule is applied.

90-DAY RULE: OUTCOMES ANALYSES

This rule defines each patient's start date, for data analyses, as day 91 of ESRD. Allowing outcomes to be compared among all ESRD patients at a stable and logical point in time, it is used primarily to calculate survival rates and compare outcomes by modality at several points in time. Use of the rule overcomes the difficulties of examining data from the first three months of ESRD service (an unstable time for new patients as renal providers try to determine the best treatment modality), and from in-center hemodialysis patients younger than 65 and not disabled, who cannot bill Medicare for their dialysis treatments and hospitalizations until 90 days after the first ESRD service date. Patients on peritoneal dialysis or home dialysis, or with transplant as the first modality, can bill immediately.

SERUM ALBUMIN DATA

The ME form reports albumin level along with the test's lower limit, which indicates the testing method: bromcresol purple or bromcresol green, with lower limits of 3.2 and 3.5 g/dl, respectively.

In producing the 2004 ADR we found that, in 1995–2003, almost 50 percent of forms contained lower limit values equal to "zero," while another 25 percent reported values other than the expected 3.2 and 3.5 g/dl. Only 25 percent (n=173,000) of incident patients had legitimate lower limit values. Further analyses, however, showed that these patients are a representative cohort sample, with similar demographic distributions by age, gender, race, and cause of ESRD to those of the overall ESRD population. For all figures in the 2005 and later ADRs which present serum albumin data from the ME form, we therefore include only those incident patients with both an albumin lower limit of 3.2 or 3.5 g/dl and an albumin value.

Database Definitions

MODALITIES

The USRDS and the CMS ESRD group have worked extensively on methods of categorizing patients by ESRD modality. While the ME form is the primary source of data on modality at ESRD initiation, the modality it indicates may be temporary, as patients often change to a new one in the first 90 days, and it can be difficult to track modality during this time. Patients age 65 and older have Medicare claims in the first 90 days; these claims contain revenue codes designating modality. Patients younger than 65 and in employer group health plans (EGHPs) or Medicare risk programs, however, have no such claims. Modality may thus not be determined until Medicare becomes the primary payor at day 91 or, for EGHP patients, at 30–33 months after the first ESRD service date. These limitations influence our ability to determine a patient's exact modality at any one point in time.

Of particular concern are patients categorized as having an unstable modality (i.e. on a modality for fewer than 60 consecutive days) in the first 90 days, and who are therefore not recognized as being hemodialysis or peritoneal dialysis patients. These patients tend to have higher death and hospitalization rates, and unless they are identified and assigned to modalities, interpretations of modality-specific outcomes should be viewed with caution. These patients are included in the "all ESRD" category, which provides a more complete view of mortality and hospitalization with the least biasing of the data.

As mentioned earlier, a new modality/event — recovered renal function — was introduced in the 2007 ADR. This event can be established only if it occurs within first 180 days of the FSD and if the RRF period persists for at least 90 days. The RRF event is similar to the lost-to-follow-up event in that patients with an RRF event will not be included in the prevalent populations for outcomes analyses. However, as with lost-to-follow-up events, we keep them in the modality sequence so that subsequent renal failure episodes can be tracked closely and in a timely manner.

Individual analyses categorize modalities in different ways; these are defined in the methods sections for each chapter.

PAYORS

Information on payors is obtained from the CMS Medicare Enrollment Database. We also examine Medicare outpatient claims to identify patients for whom the EDB does not indicate Medicare as primary payor (MPP), but who have at least three consecutive months of dialysis treatment covered by Medicare; these patients are also designated as having MPP coverage. From these two data sources we construct a payor sequence file to provide payor history, and, starting with the 2003 ADR, we use this file to identify Medicare eligibility status and other payors.

The construction of this file is similar to that of the treatment history file. Payor status is maintained for each ESRD patient from the first ESRD service date until death or the end of the study period. Payor data are used to categorize a patient as MPP, MSP with EGHP, MSP non-EGHP, Medicare Advantage (Medicare + Choice), Medicaid, or a combination of payers. With this approach, the USRDS is now able to apply payor status information in all outcome analyses using the "as-treated" model (see the discussion of Chapter Eleven).

PRIMARY CAUSE OF RENAL FAILURE

Information on the primary cause of renal failure is obtained directly from the ME form. For the ADR we use eight categories, with ICD-9-CM codes as follows:

diabetes: 250.00 and 250.01
hypertension: 403.9, 440.1, and 593.81
glomerulonephritis: 580.0, 580.4, 582.0, 582.1, 582.9, 583.1, 583.2, 583.4, and 583.81
cystic kidney: 753.13, 753.14, and 753.16
other urologic: 223.0, 223.9, 590.0, 592.0, 592.9, and 599.6
other cause: all other ICD-9-CM codes covered in the list of primary causes on the ME form, with the exception of 799.9
unknown cause: 799.9 and ICD-9-CM codes not covered in the list of primary causes on the ME form
missing cause: no ICD-9-CM code listed

RACE & ETHNICITY

Data on patient race and ethnicity are obtained from the ME form, the CMS Medicare Enrollment Database, and the REMIS/REBUS identification file. Because they are addressed in separate questions on the ME form, racial and ethnic categories can overlap.

Patient ethnicity became a required field on the 1995 revised ME form; because data for 1995 are incomplete, information on Hispanic patients is presented starting in 1996. The non-Hispanic category includes all non-Hispanics and patients with unknown ethnicity.

Because of the small number of ESRD patients of some races, as well as the construction of the U.S. census data, we concentrate on white, African American, Native American (including Alaskan Native), and Asian (including Pacific Islander) populations. Data on patients of other races will be presented as their numbers increase.

EGHP COHORT

As mentioned, EGHP data in this year's ADR are derived from the MarketScan and Ingenix I3 databases. To examine the demographic segment not represented by Medicare, we use enrollment information to construct yearly cohorts of enrollees younger than 65. To ensure that we select enrollees with the potential to generate claims evidence appropriate to the analytical demands, rules for inclusion also include 12 months of continuous coverage in a commercial fee-for-service plan, and, for medication analyses, continuous prescription drug coverage. Comorbidities are identified using claims. Patients with at least one inpatient claim or at least two outpatient claims during the period of interest and with a diagnosis code of a particular comorbidity are identified as having that comorbidity.

ESRD COHORT IN THE EGHP POPULATION

As the MarketScan and I3 databases provide no identifiable data elements, we cannot link them directly to the USRDS ESRD registry. To identify ESRD patients we thus use a process similar to that of the registry. Transplant patients are identified by evidence of a kidney transplant procedure or an adverse graft event, and chronic dialysis patients by evidence of continuous history of dialysis therapy, with at least three consecutive months of dialysis service and with dialysis service claims in at least 70 percent of treatment months. Treatment months are defined from the first dialysis claim to the earliest of kidney transplant, death, or end of enrollment. Both inpatient and outpatient claims are evaluated for evidence of dialysis service history.

The first ESRD service date is set to the earliest of the first dialysis service date or the transplant date. If neither is available, the start of enrollment is used. Incidence is defined by a first ESRD service date at least 60 days after the start of enrollment.

prιcis

For Figure p.1 we identify chronic kidney disease (CKD), congestive heart failure (CHF), and diabetes in patients from the 5 percent Medicare sample using methods described for Chapter Eleven; these methods are also used to determine diabetic status and CHF in the ESRD population. Costs for the "cost year" are determined for the entire calendar year for patients who have fee-for-service coverage and Medicare as primary payor. Because this analysis combines the ESRD cohort with the 5 percent Medicare sample, ESRD patients in the 5 percent sample are excluded.

Methods for the portion of Table p.a that addresses Medicare spending are addressed in the discussion of Chapter Eleven.

Total transplant counts shown in Table p.a include all transplants performed in 2008, as reported by the OPTN. Transplants of unknown donor type are excluded from by-donor counts. New waiting list counts include all patients added to the waiting list for a kidney-alone or a kidney-pancreas transplant in 2008; patients added at multiple centers are counted once. The total N on the waiting list includes all patients listed for a kidney-alone or kidney-pancreas transplant as of December 31, 2008, regardless of when they first listed. If patients are added to the list in early 2008 and removed from the list before the end of the year, it is possible for a group to have more new patients than existing patients. Median time on the list is shown for patients on the list on December 31, 2008.

Rates in Figures p.4–5 and p.7–8 are adjusted for age, gender, and race.

Figure p.6 shows point prevalent wait list counts for those listed for a kidney-alone transplant, and displays the median time to transplant, as well as the 25th and 75th percentiles for time to transplant, among patients transplanted during the given year. Time to transplant is computed using Kaplan-Meier methodology.

Quality of Care

Most information on this spread is obtained from the ME form. For Figure p.13, an archived PMMIS quarterly dialysis record is used to track transfusions in ESRD patients before 1991. The percentage of hemodialysis patients receiving transfusions is calculated as the number receiving at least one transfusion in a given quarter divided by the number with at least one dialysis record in that quarter. Since the archived data are current only to the third quarter of 1995, we emulate this method, using Medicare claims generated by ESRD facilities, to update the data.

Hospitalization & Mortality

Figure p.16 shows the percent change since 1993 in admission rates for period prevalent ESRD patients. Included patients have Medicare as a primary payor and are residents of the 50 states, the District of Columbia, Puerto Rico, and the Territories. Patients with AIDS as a primary or secondary cause of death are excluded, as are patients with missing age or gender information. Methods generally follow those described for the prevalent patient cohorts in Chapter Six and Reference Section G. Rates are adjusted for age, gender, race, and primary diagnosis using the model-based adjustment method. The reference cohort includes period prevalent ESRD patients, 2005. Principal ICD-9-CM diagnosis codes for cardiovascular and infectious hospitalizations are listed in the discussion of Figure 6.2. New dialysis access codes for peritoneal dialysis patients appeared in late 1998; dialysis access values are therefore shown for peritoneal dialysis patients as a change since 1999 rather than 1993. For peritoneal dialysis patients, dialysis access hospitalizations are those defined as "pure" inpatient vascular/dialysis access events, as described for Tables G.11–15. For hemodialysis patients, vascular access hospitalizations include "pure" inpatient vascular access events, and vascular access for hemodialysis patients excludes codes specific to peritoneal dialysis catheters (996.56, 996.68, and V56.2).

Figure p.17 illustrates trends in mortality rates by vintage for period prevalent dialysis patients alive on renal replacement therapy on January 1, with a first service date at least 90 days prior to the beginning of the year, and reaching day 91 of ESRD treatment during the year. Patients with unknown age or gender, or of a race other than white, African American, Native American, or Asian, are excluded. Patients are followed from January 1 until death, transplantation, or the end of the year, and all-cause mortality rates are adjusted for age, gender, race, and primary diagnosis using generalized mixed models. The reference population consists of 2005 prevalent dialysis patients, and adjusted mortalities across vintages are comparable.

Figure p.18 presents adjusted first-, second–third, and fourth–fifth-year mortality rates, by modality, for incident dialysis and first transplant patients. Patients are followed from day 91 until death or December 31, 2008. Dialysis patients are also censored at transplant. Rates are computed from the Cox model using the model-based adjustment method, described later in this appendix, and adjusted for age, gender, race, and primary cause of ESRD. The reference population consists of 2005 incident ESRD patients, and these rates are comparable across modalities.

Figure p.19 presents one-year survival for 2007 incident hemodialysis, peritoneal dialysis, and propensity-matched hemodialysis patients using the Kaplan-Meier method.

ESRD expenditures

Methods used for Figures p.20–28 are described in the text for Chapter Eleven and in the figure captions.

healthy people 2010

Targets in this chapter come directly or are estimated from published HP2010 objectives on CKD, diabetes, and immunizations.

Objective 4.1 Incident rates in Figures hp.2–3, hp.4 (first graph), and hp.26, and in Table hp.a, are calculated using the methods described for Chapter Two, later in this appendix. Rates of diabetes in the general population (second graph in Figure hp.4) are obtained from the CDC's Behavioral Risk Factor Surveillance System, at www.cdc.gov/brfss. Rates in Figures hp.2 and hp.4 are adjusted for age, gender, and race. In Figure hp.3 and Table hp.a, rates by age are adjusted for gender and race, rates by gender are adjusted for age and race, and rates by race and ethnicity are adjusted for age and gender.

Objective 4.2 The cohort includes period prevalent ESRD patients, 1991–2008. Cause-specific cardiovascular mortality is defined using CMS codes 27, 31, and 32 (congestive heart failure); 26 (atherosclerotic heart disease); 02 and 23 (myocardial infarction); and 01, 04, 25, 28–30, and 36–37 (other cardiovascular disease). Age is calculated for point prevalent patients as of January 1, and for incident patients as of the first ESRD service date. We exclude patients with unknown age, gender, or race, and those with an age calculated to be less than zero. Rates are estimated as the number of patients who die from cardiovascular disease in each year per 1,000 patient years at risk.

Objective 4.3 Figure hp.8 and Table hp.c use data from the newest version of the ME form. The cohort for Figures hp.9–10 includes incident ESRD patients, age 67 and older at initiation. Albumin and lipid tests are identified from Medicare claims during the two-year period prior to ESRD.

Objective 4.4 For Figures hp.12–13, the calculation of placement rates follows methods used in Chapter Five. Data from the CMS ESRD Clinical Performance Measures Project are used for Table hp.d (ESRD CPM year 2008) and Figures hp.11 and hp.32 (ESRD CPM years 1999–2008); included patients are those whose date of dialysis initiation, according to the CPM data, occurs in the same year as the data collection, and the access type represents the access used during the last quarter of the year. To obtain consistent information on race and ethnicity, patients included in the CPM dataset are matched to those in the ESRD database using UID numbers.

Objective 4.5 The cohort for Figures hp.14–15 and hp.33 and for Table hp.e includes patients younger than 70 in 1991–2007. Percentages are calculated as the number of patients placed on the deceased donor organ wait list or receiving a deceased donor transplant within one year of initiation, divided by the number of patients without a living donor available (i.e., patients receiving a living donor transplant are excluded), and are estimated using the Kaplan-Meier method. Note that this method differs from those used in previous ADRs, which showed the percentage of point prevalent dialysis patients on the wait list as of December 31 of the given year.

Objective 4.6 The cohort here includes patients from 1991–2005 who are younger than 70 at ESRD certification. Patients are followed for three years, from ESRD certification until the first of death, transplant, or censoring at three years post-transplant. Percentages are calculated using the Kaplan-Meier methodology.

Objective 4.7 Incident rates of ESRD due to diabetes are calculated using the methods described for Chapter Two. Rates in Figure hp.18 are adjusted for age, gender, and race, those in Figure hp.19 are adjusted for gender and race, and those in Figure hp.20 are adjusted for age and gender. In Table hp.g, rates by age are adjusted for gender and race, rates by gender are adjusted for age and race, and rates by race and ethnicity are adjusted for age and gender.

Objective 4.8 Methods and codes used to determine rates of glycosylated hemoglobin (A1c) testing and eye examinations are taken from HEDIS 2008 specifications (HEDIS 2008, an NCQA program, is used to monitor the performance of managed health care plans). CPT codes 83036 and 83037 are used to identify diabetic glycosylated hemoglobin testing (A1c; claims made within 30 days of the last claim for each patient are excluded, and at least two A1c claims must be counted). Codes used to identify diabetic eye examinations are as follows: CPT codes, 92002, 92004, 92012, 92014, 92018, 92019, 92225, 92226, 92230, 92235, 92240, 92250, 92260, 67101, 67105, 67107, 67108, 67110, 67112, 67141, 67145, 67208, 67210, 67218, 67227, 67228, 67028, 67030, 67031, 67036, 67038, 67039, 67041, 67042, 67043, 67113, 67121, 67221, 67228, S0625, S0620, S0621, and S3000; ICD-9-CM procedure codes, 14.1–14.5, 14.9, 95.02, 95.03, 95.04, 95.11, 95.12, and 95.16; and ICD-9-CM diagnosis code V72.0. Lipid testing is identified through CPT codes 80061, 82465, 83715–83721, 84478, 83700, 83701, and 83704. The general Medicare population includes patients diagnosed with CKD and diabetes in each year, continuously enrolled in the Medicare inpatient/outpatient and physician/supplier program during the whole year, and age 65 or older at the beginning of the year. Testing is tracked during each year. Patients are excluded if they are enrolled in a managed care program (HMO), acquire Medicare as secondary payor, are diagnosed with ESRD during the year, have a missing date of birth, or do not live in the 50 states, the District of Columbia, Puerto Rico, or the Territories. Racial and ethnic categories are mutually exclusive. Methods of defining CKD and diabetes are described in the CKD volume.

Figure hp.36 illustrates diabetic preventive care among pre-ESRD patients, by network. The cohort includes 2008 incident ESRD patients age 67 or older at initiation, with diabetes one year prior to the start of ESRD, and with Medicare inpatient/outpatient and physician/supplier coverage during that year. Eye examinations are tracked in the two years prior to ESRD, while lipid and A1c testing are tracked in the one year prior.

Figures hp.21–22 and Table hp.h also show ACEI/ARB use for diabetic CKD patients in the Medicare population, using Cost and Use data from the Medicare Current Beneficiary Survey (MCBS). To ensure that we obtain information on all therapy received by each person during each study year, included patients are continuously enrolled in the Medicare inpatient/outpatient and physician/supplier program during the entire year, survive until the end of the year, have a completed survey, are not enrolled in an HMO, and do not have ESRD; they also reside in the 50 states or the District of Columbia and are community-dwelling respondents. Comorbidities, including CKD and diabetes, are defined from the claims, using the same method used with the 5 percent data. Drug use information is obtained from the MCBS Cost and Use data file "Prescribed Medicine Events," and SUDAAN (Research Triangle Institute, Research Triangle Park, NC) is used to analyze all data.

Objective 5.11 The cohort for Figures hp.23–25 and Table hp.i is similar to that used for Objective 4.8, but includes all diabetic patients instead of diabetic CKD patients. CPT codes for urinary microalbumin measurement are also identified from HEDIS 2008 specifications, and include 82042, 82043, 82044, and 84156. Testing is tracked during each year. Figure hp.37 illustrates urinary microalbumin measurements in the pre-ESRD population by ESRD network; the cohort is the same as that used in Figure hp.36.

Objective 14.29 The cohort for influenza vaccinations includes all ESRD patients initiating therapy at least 90 days prior to September 1 of each year and alive on December 31. For pneumococcal pneumonia vaccinations, the cohort includes all ESRD patients initiating therapy at least 90 days before January 1 of the graphed time period and alive on December 31. Patients not residing in the 50 states, the District of Columbia, Puerto Rico, or the Territories are excluded, as are those who have a missing date of birth, who have ESRD for fewer than 90 days prior to the start of the reporting interval, or who are lost-to-follow-up during the study period. Influenza vaccinations are tracked between September 1 and December 31 of each year, while pneumococcal pneumonia vaccinations are tracked during the time periods graphed. Patients in both analyses have Medicare inpatient/outpatient and physician/supplier coverage during the study periods. All ages are calculated at the end of the graphed time period. Influenza vaccinations are identified by CPT codes 90724, 90657, 90658, 90659, and 90660, and HCPCS code G0008; pneumococcal vaccinations are identified through CPT codes 90669 and 90732, and HCPCS codes J6065 and G0009.

Chapter One Emerging Issues  Top

Mortality & hospitalization

Figures 1.1–14 and 1.22–24 present cause-specific infectious admission rates among dialysis patients. Patients included in Figures 1.1–12 and 1.22–24 have Medicare as a primary payor and are residents of the 50 states, the District of Columbia, Puerto Rico, and the Territories. The maps in Figures 1.13–14 include only residents of the 50 states and the District of Columbia. As in the hospitalization analyses (Chapter Six), patients with missing data for age or gender, or with AIDS as a primary or secondary cause of death, are excluded.

Principal ICD-9-CM diagnosis codes are used to define categories of cause-specific infectious admissions. Vascular access infections in Figures 1.1, 1.6, 1.11, and 1.14, and 1.23 are identified by ICD-9-CM diagnosis codes listed later in this appendix for Figure 6.4. Admissions for all-cause infection in Figures 1.2–5, 1.10, 1.13, and 1.22 are identified by the principal ICD-9-CM diagnosis codes listed in the discussion of Figure 6.2. Codes for pneumonia and bacteremia/septicemia in Figures 1.7–8, 1.12, and 1.24 are listed for Figure 6.4. Codes for urinary tract infection in Figure 1.9 include 590–590.9, 595–595.4, 597–597.89, 598, 599.0, 601–601.9, 604–604.9, 607.1–2, 608.0, 608.4, 616.1, 616.3–4, and 616.8.

In Figures 1.1–9, cohorts include incident hemodialysis patients age 20 and older, 1991–2007. Since in-center hemodialysis patients who are younger than 65 and not disabled cannot bill for hospitalizations until 90 days after ESRD initiation, the 90-day rule is applied to patients age 20–64. Patients are required to survive the first 90 days after hemodialysis initiation, and are followed for admissions after day 90. For patients age 65 and older, however, modality is determined on the first ESRD service date, and to avoid counting the admission at initiation, follow-up begins the day after initiation. Data cleaning and counting of admissions and time at risk for admissions generally follow that described for Reference Section G; here, however, incident patients are followed during intervals in the first year of hemodialysis rather than in prevalent years. Censoring occurs at death, three days prior to transplant, end of payor status, one year after initiation, or December 31, 2008, and also at loss to follow-up for those younger than 65. Adjusted rates are calculated with a model-based adjustment method and an interval Poisson model. For Figures 1.1–9, the reference cohort includes incident hemodialysis patients in 2005 who are alive at day 90 after initiation. Figures 1.6–9 are presented by interval after initiation and adjusted for age, gender, race, and primary diagnosis. Figures 1.1–5 are presented by interval and a demographic factor and adjusted for the other three respective factors.

Figures 1.10–12 show admission rates for all-cause infection, vascular access infection, and bacteremia/septicemia in the first year of hemodialysis, and include patients incident in 2006–2007. These figures are restricted to patients age 65 and older so that admissions data are available in the first three months after initiation. Included patients have a Medical Evidence form indicating access type used on the first outpatient dialysis. The category "catheter/maturing internal access" includes patients with a catheter in addition to a maturing arteriovenous fistula or maturing graft, while "catheter only" includes those with a catheter and without a maturing internal access. Follow-up begins the day after initiation, and patients are censored at death, three days prior to transplant, end of payor status, one year after initiation, or December 31, 2008. Rates are presented by initial access type and race and adjusted for age, gender, and primary diagnosis. A model-based adjustment method is used, with incident patients in 2005, age 65 and older, as the reference cohort. Figures 1.13–14 display state-level infectious admission rates among 2006–2007 incident hemodialysis patients age 65 and older. Methods generally follow those described for 1.10–12, except that these state-level rates are unadjusted and restricted to the first six months after hemodialysis initiation.

Figures 1.22–24 include prevalent hemodialysis patients age 20 and older in both the CPM and USRDS ESRD Medicare data. Included patients reached day 90 following initiation on or before October 1, 2007. Follow-up begins on January 1, 2008, and censoring occurs at death, three days prior to transplant, modality change, end of payor status, placement of a different type of access, or December 31, 2008. Vascular access is defined as the most recent type used according to the CPM data. Dialysis vintage is computed as the time from the first ESRD service date to January 1, 2008. Rates are presented by race, vintage, and access type, and adjusted for age, gender, and primary diagnosis. The model-based adjustment method is used, and includes prevalent patients in the current and previous two years, weighted by 1, ½, and 1/4, respectively. The reference cohort includes 2005 prevalent hemodialysis patients in both the CPM and USRDS databases.

Outpatient Antibiotic Use

Figures 1.15–33 (excluding 1.22–24) use outpatient claims to identify the use of IV antibiotics and the evidence of bacterial cultures in the hemodialysis population. CPT codes used to identify IV antibiotics include J3370 (vancomycin); J0690 (cefazolin); J0713, J0692, and J0696 (broad spectrum cephalosporins); J1580, J3260, J0278, and J1840 (aminoglycosides); and J1956 (aevofloxacin), as well as an extensive list of other codes which account for less than 3 percent of IV antibiotics used. Part D data is used to identify prescriptions for antibiotics. HCPCs codes for bacterial cultures include 87040–87077, 87081, and 87086–87088, where 87040 represents a blood culture.

Figures 1.15–19 include incident hemodialysis patients initiating during 2007 and surviving on hemodialysis at least six months. Figures 1.20–21 include January 1, 2007 point prevalent patients who survive and remain on hemodialysis at least 30 days, and who are also in the 2008 CPM data. Access represents the "current access" as of the CPM data collection period (October through December, 2007). Kaplan-Meier methods are used to calculate cumulative probabilities.

Figures 1.25–27 include incident hemodialysis patients initiating during 2006 or the first five months of 2007 and surviving on hemodialysis for at least six months.

Figures 1.28–33 include incident hemodialysis patients initiating in 2006 or 2007 and dying before the end of 2007. IV and antibiotic claims prior to death are identified as described above. Vascular access for patients in Figures 1.30–33 represents the access at initiation as obtained from the ME form.

Chapter Two Incidence & Prevalence  Top

Here and throughout the ADR, the USRDS generally reports point prevalence — the type of prevalence used throughout most of the book — as of December 31, while period prevalence is reported for a calendar year. Annual period prevalent data thus consist both of patients who have the disease at the end of the year and those who have the disease during the year and die before the year's end. Because the USRDS treats successful transplantation as a therapy rather than as a "recovery" from ESRD, patients with a functioning transplant are counted as prevalent patients.

Because data are available only for patients whose ESRD therapy is reported to CMS, patients who die of ESRD before receiving treatment or whose therapy is not reported to CMS are not included in the database. We therefore qualify the terms incidence and prevalence as incidence and prevalence of reported ESRD. Some ESRD registries use the term "acceptance into ESRD therapy." We believe, however, that "incidence of reported ESRD therapy" is more precise, because "acceptance" implies that remaining patients are rejected, when they may simply not be identified as ESRD cases or may not be reported to CMS. Beginning with the 1992 ADR, lost-to-follow-up patients are not included in the point prevalent counts; they are, however, reported in Table B.1 of the Reference Tables.

Rate adjustments in this chapter are as follows: overall rates (including those in the maps) are adjusted for age, gender, and race; rates by age are adjusted for gender and race; rates by race or ethnicity are adjusted for age and gender; and rates by primary diagnosis are adjusted for age, gender, and race.

Reference section A

The Reference Tables present parallel sets of counts and rates for incidence (Section A) and December 31 point prevalence (Section B). Section B also presents annual period prevalent counts and counts of lost-to-follow-up patients. Because the U.S. population figures (shown in Reference Section M) used in the ADR include only residents of the 50 states and the District of Columbia, tables also focus on patients from these areas. Exceptions are Tables A.1, A.6, A.8, and A.10, all of which present data specific to patients in Puerto Rico and the Territories, or include these patients in the patient population. Age is computed as of the beginning of ESRD therapy.

Rates in Table A.9 are calculated using the model-based method (described in the Statistical Methods section later in this appendix), and adjusted for age, race, and gender, with the 2005 national population as reference.

Reference section B

With the exception of Tables B.1, B.6, B.8, and B.10, these tables focus on patients in the 50 states and the District of Columbia. Age is calculated as of December 31. Table B.9 is constructed similarly to Table A.9.

Chapter Three Patient Characteristics  Top

Data used in this chapter are obtained from the ME form.

Figure 3.1 includes 2008 incident hemodialysis patients with ME forms. Access type and primary cause are identified from the ME form, and data exclude patients with unknown access type.

Figure 3.5 includes incident hemodialysis patients who have valid EPO claims during each of the first four months after initiation.

Figures 3.14–18 and Table 3.c focus on transplant as a post-dialysis modality, and are limited to patients with a revised edition ME form. Figure 3.14 details whether patients initiating dialysis in 2008 were informed of their transplant options. Figure 3.15 shows the proportion of new dialysis patients placed on the waiting list or receiving a transplant within a year of initiation; the denominator includes 2007 incident dialysis patients age 18–69. In Figure 3.16, with the same cohort, we use the Kaplan-Meier method to estimate the probability of being placed on the waiting list or receiving a transplant within a year of initiating dialysis. Follow-up is censored at death. Figure 3.17 shows geographic differences in the proportion of white and African American patients placed on the waiting list or transplanted within a year of initiating dialysis, among 2006–2007 incident dialysis patients age 18–69. Table 3.c is limited to 2007 incident dialysis patients age 18–69, and Figure 3.18 to 2008 incident dialysis patients not informed of their transplant options.

Chapter Four Treatment Modalities  Top

Chapter Four and the associated reference tables describe the treatment modalities of all known ESRD patients, both Medicare and non-Medicare, who are not classified as lost-to-follow-up or having recovered renal function (RRF). The RRF event, introduced in the 2007 ADR, is defined as an event that occurs within the first 180 days of ESRD initiation and lasts for at least 90 days. By definition, patients classified as having RRF post-initiation are included in the incident counts. Unless noted otherwise, incident and point prevalent cohorts without the 60-day stable modality rule are used in the analyses. Treatment modalities are defined as follows:
center hemodialysis: hemodialysis treatment received at a dialysis center
center self-hemodialysis: hemodialysis administered by the patient at a dialysis center; a category usually combined with center hemodialysis
home hemodialysis: hemodialysis administered by the patient at home; cannot always be reliably identified in the database
CAPD: continuous ambulatory peritoneal dialysis; usually combined with CCPD
CCPD: continuous cycling peritoneal dialysis; usually combined with CAPD
peritoneal dialysis: analyses typically consist of CAPD and CCPD only, unless stated otherwise
other peritoneal dialysis: primarily intermittent peritoneal dialysis (IPD), a small category except among very young children; usually combined with unknown dialysis and uncertain dialysis to form an other/unknown dialysis category
uncertain dialysis: a period in which the dialysis type is unknown or multiple modalities occur but none last 60 days; usually combined with other peritoneal dialysis and unknown dialysis to form an other/unknown dialysis category
unknown dialysis: a period in which the dialysis modality is not known (e.g. when dialysis sessions are performed in a hospital); usually combined with other peritoneal dialysis (IPD) and uncertain dialysis to form an other/unknown dialysis category
renal transplantation: a functioning graft from either a living donor (a blood relative or other living person) or a deceased donor
death: a category not appearing in the year-end modality tables, which report only living patients, but used as an outcome (e.g. in tables showing living patients followed for a period of time for their modality treatment histor

Data on modality and provider characteristics are presented in Figures 4.3 and 4.6. For a description of the provider data used in these figures, please see the discussion of Chapter Ten. All provider-related figures include only dialysis patients. Figures 4.4 and 4.7 show modality and payor information, while Table c and Figures 4.9–11 provide a closer look at the demographic and geographic variations of home hemodialysis patients. New in the 2010 ADR, Tables 4.d–e and Figures 4.12–13 provide information on Part D enrollment, by modality. Enrollment is defined through continuous coverage, during each year of interest, by a standard Part D plan, a managed care organization Part D plan, or an employer-sponsored Part D plan. Due to the delay in obtaining 2008 Part D event data, only 2006–2007 data are shown in this chapter. We plan to present both 2008 and 2009 data in the 2011 ADR.

Rates by age are adjusted for gender and race, rates by gender are adjusted for age and race, rates by race and ethnicity are adjusted for age and gender, and rates by primary diagnosis are adjusted for age, gender, and race.

Reference Section D

Reference Section D is divided into four parts. The first, Tables D.1–11 and D.15–16, provides counts and percentages — by demographics, geographic location, and treatment modality — of incident and prevalent patients alive at the end of each year. Age is computed as of the start of ESRD for incident patients, and as of December 31 for point prevalent patients.

Table D.12 shows modality at day 90 and at two years after first service for all incident Medicare patients beginning renal replacement therapy from 2004 to 2006. The 90-day rule is used to exclude patients who die during the first 90 days of ESRD, and age is computed as of the first ESRD service date.

The third section, Tables D.13–14, presents counts of prevalent patients alive at the end of each year, by ESRD exposure time and modality. Table D.13 shows counts by the number of years of ESRD, while Table D.14 presents counts by the number of years on the end-of-year treatment modality. For the duration of ESRD exposure, zero should be read as less than one year, one as at least one full year but less than two, and so on.

The fourth section, Tables D.17–24, presents counts of incident and prevalent patients alive at the end of selected years (i.e. 2000, 2004, 2008), by demographic characteristics, payor category, and treatment modality. Again, age is computed as of the start of ESRD for incident patients, and as of December 31 for point prevalent patients. The definitions of payor categories can be found under "database definitions" at the beginning of this appendix.

Chapter Five Clinical Care & Preventive Health  Top

In Figure 5.1, for both Kt/V measurements, 2008 ESRD CPM data are used to calculate a mean Kt/V value for each patient from the 1–3 values present for each, and the percent of patients with a mean Kt/V over a certain threshold is determined. For prevalent hemodialysis patients in 2008, each patient's URR is obtained from the G-modifier attached to CPT code 90999, with a revenue code of 821 or 825. Each measurement is categorized into one of five ranges, and the median URR is calculated; for patients whose median lies between two ranges, we assign a weight of 0.5 to each. Information on new hemodialysis patients with an arteriovenous fistula as the first access is calculated as described for Figure hp.11. Hemoglobin levels are calculated for EPO-treated, 2008 prevalent hemodialysis patients, using available EPO claims during the year. EPO claims with a dose per administration of less than 500 or greater than 80,000 units, or with a hematocrit value less than 10 percent or greater than 50, are omitted. For each patient a yearly mean hemoglobin is calculated as the mean of all hematocrit values divided by three. Data on albumin are obtained for incident hemodialysis patients in 2008 who have a valid result on their ME form; those with a lower limit equal to zero are omitted. Data for influenza, pneumococcal pneumonia, and hepatitis B vaccinations are from Figures 5.14–16.

Anemia Treatment

Figure 5.2 presents the monthly distribution of patients by mean hemoglobin group, with each month containing all patients with at least one valid EPO claim during the month. The hemoglobin is calculated as the reported hematocrit value divided by three. Figure 5.3 shows the mean hemoglobin, by month, for prevalent dialysis patients with EPO claims, along with the monthly EPO dose per week for patients with 20 or fewer administrations per month. A patient's time at risk including only those days in which he or she is not in an inpatient hospital setting.

Figures 5.4–8 include data from all incident dialysis patients with an EPO claim in the first 30 days of ESRD therapy, and at least one EPO claim during each of the following six months. EPO claims with a dose per administration of less than 500 units or more than 80,000 units are omitted, as are those with an average dose per day (calculated as the total EPO units on the claim divided by the number of days spanned by the claim) of less than 100 units or greater than 10,000 units. For 2008, patients are incident prior to June 1, to allow them to have six months of EPO and/or iron claims after their incident date. For graphs by starting hemoglobin, patients are included only if they have a hematocrit listed on the ME form, and their starting hemoglobin is determined from this value. In Figure 5.4, a mean hemoglobin is calculated for each patient from claims during the month, and the average of these values is then calculated for each month. For Figure 5.5, the mean EPO dose per week is adjusted by only including days during a month in which a patient is not in an inpatient hospital setting, so that the mean EPO dose represents outpatient dosing only. Access type is defined using the ME form. Figure 5.9 includes the same type of patients as in Figures 5.4–.8, except that patients are required to have at least one EPO claim during each of the following twelve months; probabilities are calculated using the Kaplan-Meier method.

Preventive Care

Figures 5.10–13 present data on diabetic preventive care. ESRD patients without Medicare inpatient/outpatient and physician/supplier coverage during the entire study period are omitted from these analyses, as are general Medicare patients enrolled in an HMO or diagnosed with ESRD during the study period. Also omitted are those who do not reside in the 50 states, the District of Columbia, Puerto Rico, or the Territories; who have a missing date of birth; who do not survive the entire reporting period; who have ESRD for fewer than 90 days prior to the start of the reporting interval; or who are lost-to-follow-up during the study period.

Age is generally calculated at the end of the study period. Methods and codes used to determine rates of diabetic glycosylated hemoglobin (A1c) testing, lipid testing, and eye examinations are described in the methods for the HP2010 chapter, Objective 4.8. Patients are defined as having diabetes either through medical claims (one inpatient/outpatient, two physician/supplier, two outpatient, or one physician/supplier and one outpatient), or through a listing of diabetes on the ME form as the primary cause of ESRD or as a comorbid condition. ICD-9-CM diagnosis codes used to define diabetes are 250, 357.2, 362.0x, and 366.41. Comprehensive diabetic care includes at least four A1c tests, at least two lipids tests, and at least one eye exam. A1c and lipid tests are at least 30 days apart.

The ESRD population includes patients initiating therapy at least 90 days prior to January 1 of the first year of each study period and with diabetes in the first year. Testing is tracked in the second year of each study period, and tests are at least 30 days apart.

Figures 5.14–16 show rates of influenza, pneumococcal pneumonia, and hepatitis B vaccinations for prevalent ESRD patients by modality, age, race/ethnicity, and time period. Cohorts and methods for Figures 5.14–15 are the same as those described for Objective 14.29 in the HP2010 chapter, while the cohort for Figure 5.16 includes ESRD patients initiating therapy at least 90 days before January 1 of each year and alive on December 31. Patients without Medicare inpatient/outpatient and physician/supplier coverage during the entire year are omitted, as are those who do not reside in the 50 states, the District of Columbia, Puerto Rico, or the Territories; who have a missing date of birth; or who are lost-to-follow-up during the year. Age is generally calculated at the end of the study period. Hepatitis B vaccinations are tracked in each year and identified through CPT codes 90636, 90740, 90743–90744, 90748, 90731, 90723, and G0010.

Figures 5.17–23 include prevalent hemodialysis patients who are in both the USRDS and ESRD CPM databases, and whose day 91 begins prior to October 1 of the prevalent year. The access represents the current access being used, according to the CPM data. Claims are searched during the following calendar year for events and complications. Figure 5.23 includes incident peritoneal dialysis patients from the USRDS database. For Figures 5.20–23, complication rates are calculated as the number of events (from Medicare claims) divided by the time at risk, which is censored at death, change in modality, change in payment status, or the placement of a different type of access. Vascular access codes are listed in the methods for Chapter Eleven.

Medicare Part D use

Figure 5.24 shows Part D enrollment among dialysis, transplant, CKD, and non-CKD Medicare enrollees in 2007. Each cohort includes patients alive on January 1, 2007, and alive and with Medicare inpatient/outpatient and physician/supplier coverage during all of 2007. For the dialysis and transplant cohorts, modality assignment is identified on January 1, 2007, and is uninterrupted during all of 2007. For the CKD and non-CKD cohorts, patients are age 66 or older on January 1, 2007. CKD status is identified from Medicare claims during 2006. Part D enrollment is defined through continuous coverage during 2007 by a standard Part D plan, a managed care organization Part D plan, or an employer-sponsored Part D plan. Figure 5.25 shows Part D enrollment in these cohorts, with stratification by age and race.

Using the same cohorts, Figures 5.26–28 summarize use of the Part D benefit among patients continuously enrolled in Part D during 2007. Low income subsidy (LIS) is defined by any degree of subsidization during all of 2007.

Cumulative medication use in Figure 5.28 is estimated by taking all Part D claims in 2007, linking National Drug Codes on claims to generic names in the 2007 edition of Red Book, and counting the number of unique generic names. Part D claims for products in the Red Book therapeutic classes of "device and non-drug items," "diabetes mellitus supplies," and "diagnostic agents" are excluded.

Figures 5.29–40 illustrate, by age and race, prescription drug use in patients continuously enrolled in Part D during 2007. Cohorts are the same as those used in Figure 5.24. Drug use is defined by at least one applicable prescription fill during 2007. For Figures 5.31–32, other lipid lowering agents include cholesterol absorption inhibitors, niacin, and omega-3 fatty acids.

Chapter Six Morbidity & Mortality  Top

Hospitalization

Methods used to examine hospitalization in prevalent patients generally echo those used for the tables in Reference Section G (described below). Inclusion and exclusion criteria are generally the same, as are the methods for counting hospital admissions and days, and defining the follow-up time at risk. One difference is the exclusion in Reference Section G of patients of races that are unknown or other than white, African American, Native American, or Asian; these patients are included in the Chapter Six figures.

New to this year's ADR, hospitalization data exclude inpatient stays for the purpose of rehabilitation therapy. Inpatient rehabilitation claims are identified by provider numbers; numbers for inpatient rehabilitation facilities include values 3025–3099 in the 3rd through 6th positions or "R" or "T" in the 3rd position.

Inpatient institutional claims are used for the analyses, and methods for cleaning claims follow those described for Section G. Adjusted rates are calculated using the model-based adjustment method on the observed category-specific rates. This method is described further in the discussion of Section G, and in the statistical methods section later in this appendix.

Methods for rates in Figures 6.2–3 follow those described for Reference Section G. In Figure 6.2, methods include those described for Figure p.16 in the Prιcis. Principal ICD-9-CM diagnosis codes are used to identify cardiovascular and infectious admissions. The cardiovascular category consists of codes 276.6, 394–398.99, 401–405, 410–420, 421.9, 422.90, 422.99, 423–438, and 440–459, while infection is indicated by codes 001–139, 254.1, 320–326, 331.81, 372–372.39, 373.0–373.2, 382–382.4, 383.0, 386.33, 386.35, 388.60, 390–393, 421–421.1, 422.0, 422.91–422.93, 460–466, 472–474.0, 475–476.1, 478.21–478.24, 478.29, 480–490, 491.1, 494, 510–511, 513.0, 518.6, 519.01, 522.5, 522.7, 527.3, 528.3, 540–542, 566–567.9, 569.5, 572–572.1, 573.1–573.3, 575–575.12, 590–590.9, 595–595.4, 597–597.89, 598, 599.0, 601–601.9, 604–604.9, 607.1, 607.2, 608.0, 608.4, 611.0, 614–616.1, 616.3–616.4, 616.8, 670, 680–686.9, 706.0, 711–711.9, 730–730.3, 730.8–730.9, 790.7–790.8, 996.60–996.69, 997.62, 998.5, and 999.3. Figure 6.3 presents adjusted rates of total hospital admissions and days per patient year. Prevalent ESRD patients are included, and rates are adjusted for age, gender, race, and primary diagnosis with the 2005 ESRD cohort used as the reference.

Figure 6.4 shows adjusted admission rates for principal diagnoses among prevalent ESRD patients. Again, rates are adjusted for age, gender, race, and primary diagnosis, with ESRD patients in 2005 used as the reference cohort. Principal ICD-9-CM codes are as follows: pneumonia, 480–486 and 487.0; bacteremia/septicemia, 038.0–038.9 and 790.7; cellulitis, 682; vascular access infection (hemodialysis patients only), 996.62; and peritonitis (peritoneal dialysis patients only), 567.

Table 6.a presents adjusted admission rates and relative risks of admission among adult (age 20 and older) period prevalent hemodialysis patients. Principal ICD-9-CM diagnosis codes are used to identify cardiovascular and infectious admissions, and are listed in the discussion of Figure 6.2. Vascular access admissions are defined as "pure" inpatient vascular access events, as described for vascular/dialysis access events in Tables G.11–15; vascular access, however, excludes the codes specific to peritoneal dialysis catheters (996.56, 996.68, and V56.2). Rates and relative risks are adjusted for age, gender, race, and primary ESRD diagnosis, while values presented by one factor are adjusted for the other three. Adjusted relative risks are calculated with a Poisson model, with reference groups as listed. For adjusted rates, hemodialysis patients in 2005 are used as the reference cohort. Values by age, gender, race, and primary diagnosis are shown for 2007–2008 prevalent hemodialysis patients.

Figure 6.8 presents unadjusted rates of hospital admissions in 2007 hemodialysis, peritoneal dialysis, and matched hemodialysis patients, using Medicare claims.

Figures 6.11–13 show rates by age, adjusted for gender, race, and primary diagnosis, using the model-based adjustment method. They include period prevalent dialysis (Figures 6.11–12) and hemodialysis (Figure 6.13) patients age 20 and older, with the 2005 dialysis cohort as the reference. Figure 6.11 presents adjusted rates of cause-specific hospital admissions per patient year. The categories for cardiovascular disease and infection are defined by the codes listed for Figure 6.2; the infection codes for Figure 6.11 exclude those due to an internal device. The principal ICD-9-CM diagnosis codes used for infection due to internal device (related to a vascular access device or peritoneal dialysis catheter) are 996.62 and 996.68. At the end of 1998 a new ICD-9-CM code was added for infections due to internal devices in peritoneal dialysis patients; data prior to this date are omitted.

Figure 6.12 shows adjusted event rates for inpatient coronary revascularization. Patients are followed until the first coronary revascularization event, and are censored at the earliest of death, three days prior to transplant, or the end of the calendar year. Events are identified from inpatient and physician/supplier claims occurring within a hospital stay. The following ICD-9-CM procedure and CPT codes are used to identify events: angioplasty, procedure codes 00.66, 36.01, 36.02, and 36.05, and CPT codes 92982, 92984, 92995, and 92996; coronary stents, procedure code 36.06 and CPT codes 92980–92981; and bypass, procedure codes 36.1x and CPT codes 33510–33523, 33533–33536. Rates for coronary stents are shown starting in 1997, due to the release of CPT codes.

Figure 6.13 displays adjusted vascular access placement rates for period prevalent adult hemodialysis patients. These are not hospital admission rates, but procedure rates for vascular access placements in an inpatient setting. Vascular access placements are obtained from CPT codes on physician/supplier claims, and are restricted to those occurring in the hospital (during an inpatient stay or emergency room visit). Categories include catheters, fistulas, and grafts; the CPT codes used to define them are found in Table a.a later in this appendix. The category for all vascular access placements includes the CPT codes for all of the above categories. Methods are also used to exclude vascular access used for purposes other than dialysis. Catheter placement codes that are not specific for dialysis are included only if they are accompanied by an ICD-9-CM renal diagnosis code. Also, rates for catheter and all vascular access placements exclude patients with specific chemotherapy or parenteral nutrition claims during the year. Inpatient/outpatient institutional, physician/supplier, and durable medical equipment claims indicate chemotherapy (CPT codes 96408, 96410, and 96412) or parenteral nutrition (CPT codes B4164–B5200, B9004, B9006, and B9999).

Mortality

Patient cohorts for all mortality figures here include both Medicare and non–Medicare patients living in the 50 states, the District of Columbia, Puerto Rico, and the Territories.

Figure 6.1 shows trends in mortality rates, by modality, for incident ESRD patients, 1980–2007. The population groups include all-ESRD, hemodialysis, CAPD/CCPD, and first transplant (known deceased and living donors only). Adjusted first-, second-, third-, fourth-, and fifth-year mortality rates for incident cohorts — including all-ESRD, hemodialysis, CAPD/CCPD, and first transplant patients — are computed from the Cox model using the model-based adjustment method, described later in this appendix. Mortality rates for all patients are adjusted for age, gender, race, and primary diagnosis. The reference population for adjusted rates consists of 2005 incident ESRD patients.

Figure 6.5 shows all-cause mortality, by age, for 2008 prevalent ESRD, dialysis, transplant, and general Medicare patients, calculated using generalized mixed models, and adjusted for gender and race. Medicare patients from 2008 are used as the reference cohort.

Figure 6.6 illustrates trends in mortality rates, by patient vintage, for period prevalent dialysis patients alive on renal replacement therapy on January 1, with a first service date at least 90 days prior to the beginning of the year, and reaching day 91 of ESRD treatment during the year. Patients with unknown age or gender, or of a race other than white, African American, Native American, or Asian, are excluded. Patients are followed from January 1 until death, transplantation, or the end of the year, and all-cause rates are adjusted for age, gender, race, and primary diagnosis using generalized mixed models. The reference population consists of 2005 prevalent dialysis patients, and adjusted mortalities are comparable across vintages.

Figure 6.7 presents five-year survival, by modality, for 1994–1998 and 1999–2003 incident patients, with modality defined on the first ESRD service date. Transplant is defined as the first transplant in the incident year. Patients with unknown age, gender, or primary diagnosis, and those with a listed age greater than 110, are excluded, as are dialysis patients who die or receive a transplant in the first 90 days. Dialysis patients are followed from day 91 until death, transplantation, or the end of 2008, while transplant patients are followed from the first transplant date until death or the end of 2008. All probabilities are adjusted for age, gender, and race; overall probabilities are also adjusted for primary diagnosis. The reference population consists of 2005 incident ESRD patients, and adjusted probabilities are comparable across modalities.

Table 6.b and Figures 6.9–10 present one-year survival for 2007 incident hemodialysis, peritoneal dialysis, and propensity-matched hemodialysis patients using the Kaplan-Meier method. Comorbidities are determined from the ME form. Log-rank tests are used to compare survival in Figure 6.10.

Figure 6.14 displays adjusted all-cause and cause-specific mortality in incident dialysis patients, 1991–2007, residing in the 50 states, the District of Columbia, Puerto Rico, or the Territories. Patients with unknown age, gender, or primary diagnosis are excluded, as are those with a listed age greater than 110. Patients are followed from the first service date up to one year, and censored at transplant or recovery of kidney function. Overall rates are adjusted for age, gender, race, and primary diagnosis. The reference population consists of 2005 incident dialysis patients, and adjusted rates can be compared across years and causes of mortality.

Figures 6.15–16 display adjusted mortality due to cardiovascular disease and infection, respectively, by age. Populations are the same as in Figure 6.14. Rates by age are adjusted for gender, race, and primary diagnosis, and the reference population consists of 2005 incident dialysis patients.

Reference section G

Hospitalization reference tables present adjusted total admission and hospital day rates, by year, 1993–2008. They begin in 1993 because Medicare inpatient claims are available beginning in 1991, and the model-based adjustment method uses data from the current and previous two years to obtain the predicted rates. (This method is further discussed later in this section and in the statistical methods section at the end of this appendix.)

Because hospitalization data for non-Medicare patients may be incomplete, analyses in this section include only patients with Medicare as their primary payor. Hospitalization data are obtained from institutional inpatient claims. As in Chapter Six, hospitalization data in Reference Section G also now exclude inpatient stays for the purpose of rehabilitation therapy.

Tables G.1–15 include dialysis and transplant patients on their modality for at least 60 days, reaching day 91 of ESRD by the end of the year, and residing in the 50 states, the District of Columbia, Puerto Rico, and the Territories. Excluded are patients with AIDS as a primary or secondary cause of death; patients with missing values for age, gender, or race; and patients of races that are unknown or other than white, African American, Native American, or Asian. Age is determined on January 1 of each year. Patients are also classified according to their primary cause of ESRD, in which the "other" category includes patients with missing data or causes other than diabetes, hypertension, or glomerulonephritis.

Patients are classified by modality at the beginning of the year:

all dialysis: patients on hemodialysis, CAPD/CCPD, or dialysis of an unknown type, as well as those on more than one modality in the past 60 days
hemodialysis: patients on hemodialysis for at least 60 days as of the start of the period at risk
CAPD/CCPD: patients on CAPD/CCPD for at least 60 days as of the start of the period at risk transplant: patients with a functioning transplant, and who received the transplant less than three years prior to the start of the period at risk
all-ESRD: all patients
To limit the contribution of patient years at risk from patients who do not have Medicare coverage but do have Medicare as a secondary payor or HMO coverage, and who therefore have incomplete hospitalization data, cohorts include only patients with Medicare Parts A and B coverage at the start of follow-up. The follow-up period is censored when payor status changes to a patient no longer having Medicare Parts A and B coverage or Medicare as a primary payor.

For patients in the all-dialysis, hemodialysis, and peritoneal dialysis categories, the period at risk for all hospitalization analyses is from January 1 or day 91 of ESRD until the earliest of death, three days prior to transplant, end of Medicare Parts A and B coverage, or December 31. Modality change is considered a censoring event only in the case of a change from dialysis to transplant. For dialysis patients in the all-ESRD category, in contrast, the analysis period is censored only at death, end of Medicare Parts A and B coverage, or December 31 of the year; a modality change is not used as a censoring event. For transplant patients in the all-ESRD and transplant categories, the period is censored at the earliest of death, three years after the transplant date, end of Medicare Parts A and B coverage, or December 31 of the year. The censoring of transplant patients at three years following the transplant is necessary because Medicare eligibility may be lost and hospitalization data may be incomplete for these patients.

Time at risk is calculated differently for hospital days and total admissions. Since a hospitalized patient remains at risk for additional hospital days, rates for hospital days include hospital days in the time at risk. Since a currently hospitalized patient is not, however, at risk for new admissions, hospital days for each year are subtracted from the time at risk for total admissions. In the case of hospitalizations in which admission occurs the same day as discharge, zero days are subtracted from the time at risk for total admissions. When bridge hospitalizations span the start of the analysis period, only the days within the period are subtracted from the time at risk for total admissions.

All admissions and hospital days during the analysis period are included, respectively, in the total admissions and hospital days for each year. An admission for a hospitalization that occurs before and spans the start of the analysis period is excluded from the total admissions for that period, and only the hospitalization days within the period are counted in the total days for hospital day rates. The minimum length of stay is one day, and hospitalizations with an admission and discharge on the same day, as well as those with a discharge the day after admission, are both counted as one day.

As in previous ADRs, all overlapping and only certain adjacent hospitalizations are combined, due to the fact that many adjacent claims may actually be legitimate separate hospitalizations. Specifically, hospitalizations with an admission on the same day or the day after a previous discharge are combined only when there is a discharge transfer code or indication of an interim claim. In the case of two hospitalizations combined into one, the principal diagnosis and procedure codes are retained from the first of the two hospitalizations, with the combined hospitalization extending from the first admission date to the last discharge date.

The methodology for computing adjusted total admission and hospital day rates uses the model-based adjustment method (discussed in the statistical methods section). Predicted rates for each subgroup combination of age, gender, race, primary diagnosis, and year are obtained using a model with the Poisson assumption. For prevalent patient cohorts, this model uses data from the current and previous two years, with respective weights of 1, Ό, and ?. Adjusted rates are then calculated using the direct adjustment method, with all 2005 ESRD patients as the reference cohort.

Tables G.11–15 show inpatient utilization in period prevalent ESRD patients. Methods — including modality definitions, inclusion criteria, data cleaning, follow-up time definitions, and rate calculations — generally follow those previously described for the total admission rates in Tables G.1–5, but some differences do exist. While patients of races other than white, African American, Native American, or Asian are excluded from G.1–5, they are included in G.11–15, except where rates are given by race. Rates are unadjusted and reflect total admissions per 100 patient years for 2000–2002, 2003–2005, and 2006–2008 (pooled) prevalent patients. While the rates for all causes are computed similarly to the unadjusted rates in G.1–5, the other nine cause-specific categories only include admissions for specific diseases. Vascular access and peritoneal dialysis access hospitalizations are those classified as "pure" inpatient vascular/dialysis access events. Such access events are defined as admissions with a specified ICD-9-CM principal diagnosis code, or an ICD-9-CM principal procedure code in conjunction with a certain DRG code. Codes are listed later in this appendix in Table a.b. If an admission does not qualify as vascular/dialysis access, it is classified by the principal diagnosis code into one of eight other mutually exclusive groups. Categories and ICD-9-CM codes are as follows: circulatory diseases, 390–459; digestive diseases, 520–579; genitourinary diseases, 580–629; endocrine and metabolic diseases, 240–279; respiratory diseases, 460–519; infectious diseases, 001–139; and cancer, 140–172, 174–208, 230–231, and 233–234. Hospitalizations that do not fall under any of these categories are counted under all others.

Supplementary tables providing additional rates and counts are available on our website and CD-ROM. Tables G.1.1–5.1 present adjusted rates similar to those shown in G.1–5, but include more patient subgroups. Additional tables (G.1.2–5.2) display the counts of the total admissions, patient years at risk, and total patients that are used to calculate the total admission rates. Standard errors of the rates in Tables G.1–10 and G.1.1–5.1 are also available.

Reference Section H

Cohorts for tables in Section H include both Medicare and non–Medicare patients living in the 50 states, the District of Columbia, Puerto Rico, and the Territories.

Cohorts in Tables H.1–12 include both incident and prevalent patients. Incident cohorts are limited to patients who reach day 91 of ESRD treatment during the year, while prevalent cohorts include patients alive on renal replacement therapy on January 1 and whose first service date is at least 90 days prior to the beginning of the year. Because calculations include only one year of follow-up, a prevalent patient surviving to the end of the year contributes one year at risk, while a prevalent patient dying during the year contributes less than one year. Since the calculation for incident patients begins on day 91 of ESRD, most patients contribute less than one year at risk; a full year is contributed only if day 91 of ESRD is January 1 and the patient survives to the end of the year. Patients considered lost-to-follow-up at the beginning of the year are excluded. The period at risk is not censored at the start of a lost-to-follow-up period, however; if a patient enters the lost-to-follow-up category during a calendar year, he or she remains in the death rate computation until the end of that year. Patient cohort populations often overlap. Patients with a functioning transplant on the start date, for example, are included in the all-ESRD and functioning transplant categories, while patients on dialysis are defined as both all-ESRD and all-dialysis. A patient in the all-dialysis category may also be reported in one of two subgroups — hemodialysis or CAPD/CCPD — if he or she has been on that modality for at least the previous 60 days. Dialysis patients not on hemodialysis or CAPD/CCPD, or on that modality for fewer than 60 days, are included only in the all-ESRD and all-dialysis categories.

Tables H.1, H.2, and H.2.1 present mortality data for all ESRD patients. Total deaths are presented in Table H.1. Overall unadjusted and adjusted annual mortality rates by age, gender, race/ethnicity, primary diagnosis, and vintage are presented in Table H.2. The unadjusted mortality rates are calculated by dividing total patient deaths in a category — male, for example — by total follow-up time in the same category. For the adjusted rates, generalized mixed models are used to calculate the smoothed rates; these methods are described in the statistical methods section later in this appendix. After obtaining smoothed rates from the generalized mixed models, direct adjustment methods are used. Overall mortality rates are adjusted for age, gender, race, primary diagnosis, and vintage, while rates for each individual category are adjusted for the remaining four. The reference population includes 2005 prevalent ESRD patients. Table H.2.1 presents unadjusted mortality rates by patient age, gender, race, and primary diagnosis for 2008 prevalent ESRD patients; rates are smoothed using a generalized mixed model.

The same methods are used for Tables H.3, H.4, and H.4.1 (dialysis); H.5 (dialysis patients, never on transplant waitlist); H.6 (dialysis patients on transplant waitlist); H.7 (dialysis patients, returned to dialysis from transplant); H.8 and H.8.1 (hemodialysis); H.9 and H.9.1 (CAPD/CCPD); and H.10 and H.10.1 (transplant).

Reference section I

These tables, which include only incident cohorts, present patient counts and survival probabilities. All causes of death are included, as are all non-Medicare patients and patients living in the 50 states, the District of Columbia, Puerto Rico, and the Territories. Patients with unknown gender or age, or whose listed age is greater than 110, are excluded.

Patient selection criteria are the same for both unadjusted and adjusted survival probabilities. All new ESRD patients who have a first ESRD service date between January 1, 1980, and December 31, 2007, are included in the analysis. These patients are followed until December 31, 2008, with a maximum follow-up time of 24 years and a minimum of one year.

Results are reported for the following groups:

all ESRD: all ESRD patients beginning renal replacement therapy in a calendar year and surviving beyond day 90; patients are censored only at the end of follow-up dialysis only: all dialysis patients starting renal replacement therapy in a calendar year, surviving beyond day 90, and not receiving a transplant by day 91; patients are censored at transplant or the end of follow-up
hemodialysis only: all hemodialysis patients starting renal replacement therapy in a calendar year, surviving beyond day 90, and not receiving a transplant by day 91; patients are censored at transplant or the end of follow-up
peritoneal dialysis only: all peritoneal dialysis patients starting renal replacement therapy in a calendar year, surviving beyond day 90, and not receiving a transplant by day 91; patients are censored at transplant or the end of follow-up
transplant: patients with a functioning transplant at the start of the period and who have had the transplant for at least 60 days; the period at risk is censored at the end of the year Unadjusted patient survival probabilities are estimated using the Kaplan-Meier method, while the Cox model and the model-based adjustment method are used for adjusted probabilities.
To limit imprecision due to small cell sizes, adjusted probabilities use aggregate categories for age, gender, race, and primary diagnosis. For each cohort, a probability presented for one variable is adjusted for the remaining three. Overall probabilities for all patients are adjusted for each of the four variables, as described later in the statistical methods section. The reference population consists of 2005 incident ESRD patients.

Chapter Seven Transplantation  Top

Figure 7.1 presents an overview of the transplant population. Panel 1 juxtaposes the growing rate of ESRD with the falling rate of transplantation in patients age 20 and older at transplant, 1988–2008. Most adult-only figures are limited to patients age 18 and older, but this figure is limited to age 20 and older because census population data are provided in five-year increments. Panel 2 summarizes the wait list, showing, by prior transplant status, the number of patients age 20 and older on the OPTN kidney or kidney-pancreas wait list on December 31 of the year, and the median wait time for a deceased-donor kidney. Patients with overlapping listings at more than one center are counted once. Median wait time is plotted only when the Kaplan-Meier median is observed, and is thus missing for patients listed more recently. Panel 3 presents transplant counts for patients 20 and older, by donor type, obtained through a combination of OPTN and CMS data.

Transplant Wait List

Figures 7.2 and 7.4 show the number and distribution of adult (age 18 and older) patients on the OPTN kidney or kidney-pancreas wait list on December 31 of the year. Because patients may list at multiple transplant centers, Figure 7.2 shows the number of unique patients and the proportion of patients listed at multiple centers, by status (active/inactive). Distributions by age, race, blood type, and panel reactive antibody (PRA), shown in Figure 7.4, are based on first listings only. Age is determined as of December 31 of the given year. PRA is the maximum recorded value.

In 2003, OPTN started the expanded criteria donor (ECD) program, to allow patients to indicate their willingness to accept a kidney from a "marginal" donor. Figure 7.3 shows state-level differences in the proportion of wait-listed patients willing to accept an ECD kidney, and is limited to patients first listed for a kidney-alone transplant in 2007–2008. In this figure, "state" refers to the location of the transplant center at which a patient is listed.

Figure 7.5 presents observed and projected median wait times in patients age 18 and older at transplant and listed for a first-time, kidney-only transplant. Median wait times are estimated for each year using the Kaplan-Meier methodology. Years for which the median is observed are plotted, while for cases in which a subgroup has fewer than 15 patients the median is not plotted and is left as unknown. For more recent years in which the median has not yet been observed — i.e., more than 50 percent of the patients listed in that year have yet to be transplanted — the median time is estimated using a linear regression model, and plotted with a dotted line. The regression analysis considers all years for which the median is observed, excluding cells with fewer than 15 patients, as described above. A regression line is estimated using the year of transplantation as an independent variable. To improve the fit of this line, a quadratic term for the year of transplantation is included in the model. Predicted medians are then estimated from the resulting regression line.

Figure 7.6 shows median wait times, by state, for adults receiving a deceased donor kidney during 2008. Wait time is calculated as the transplant date minus the date the patient is added to the kidney or kidney-pancreas wait list, not necessarily the date he or she is first listed at the center where the transplant is performed. Figure 7.7 shows projected median wait times, by state, for adult patients listed for a deceased donor kidney transplant in 2008. Projections are estimated using the same methods as in Figure 7.5. In both maps, state is that of the patient's listing, not of his or her primary residence.

Figure 7.8 illustrates changes over time in the percent of listed patients who receive a living donor transplant within one year of listing. Patients include all first time listings from 1991–2007, including pediatric patients. Kaplan-Meier methods are used to estimate proportions, and data are censored at removal from the list, deceased donor transplant, or death. Figure 7.9 reports proportions of patients who receive a deceased donor transplant within three years of listing, by age and blood type. Pediatric patients are those age 0–17 at listing, while adults are 18 and older at listing; data by blood type are limited to adults. Kaplan-Meier methods are used to estimate proportions, with censoring at removal from the list, living donor transplantation, or death.

Figure 7.10 illustrates three-year outcomes for adult patients first listed in 2005, and five-year outcomes for those first listed in 2003. Patient outcomes are classified into five groups: 1) received a deceased donor transplant, 2) received a living donor transplant, 3) died awaiting a transplant, 4) removed from the list prior to transplantation, or 5) still waiting.

Figure 7.11 shows adjusted one-year mortality, by state of residence, for January 1, 2008 point prevalent wait-list patients. A Poisson regression is used to estimate rates, adjusted for age, gender, white/non-white race, willingness to accept an ECD donor, and time on the list prior to 2007. Patients are followed for up to one year.

Figure 7.12 shows the likelihood of adult patients dying while awaiting transplant in the first through fifth year after listing, looking at those first listed in 1991–2007. The likelihood of dying is estimated from Cox proportional hazard models, adjusted for listing year, age, gender, race, primary diagnosis, and PRA level at listing; the 2005 period prevalent cohort is used as reference. Patients are censored at removal from the list and end of follow-up.

Donation & Transplantation

Figure 7.13 shows rate of organ donation per million population by age, gender, and race. A deceased donor is counted once, regardless of how many organs are transplanted. Figure 7.14 presents unadjusted donation rates per 1,000 deaths, by state. Population and death count estimates for the year from July 1, 2007, to July 1, 2008, are obtained from the U.S. Census Bureau. Figure 7.15 presents living donor transplant counts by donor relation, with data on living donor relations obtained from the OPTN.

Figure 7.16 shows the growth in the percentage of adult transplants using kidneys from ECDs and non-heart beating donors (DCDs), 1991–2008. Factors defining ECD status are listed on the OPTN Deceased Donor Registration form, and patients here include only first-time, kidney-only recipients of a deceased donor kidney. ECDs are age 60 or older, or age 50–59 with two or more of the following: death attributed to CVA, history of hypertension, or creatinine greater than 1.5 mg/dl. DCD status became available from the OPTN in 1993.

Figure 7.17 displays the relationship between a deceased donor kidney's ECD/SCD (standard criteria donor) status and the Kidney Donor Risk Index (KDRI) score for all first-time kidney-only transplants in 2008. The KDRI was published in 2009 by Rao, et al. as a means to more finely grade the quality of deceased donor kidneys, and is based on donor factors such as age, weight, race, and health history, and on transplant factors such as cold ischemia time and the number of HLA mismatches. Analysis here is limited to transplants in adult recipients with ABO-compatible donors and no history of any prior organ transplant. Percentages of transplants by ECD/SCD status are plotted within deciles of the KDRI in 2008.

Transplant rates in Figure 7.18, by age, gender, race, and primary diagnosis, show trends in adult transplants per 100 ESRD patient years, 1991–2008. Rates presented by one variable are adjusted for the remaining three using a Poisson regression model. Figures 7.19–20 use an identical method, but are limited to deceased and living donor transplants, respectively.

Figure 7.21 shows transplant rates per 100 dialysis patient years, by state, in 2008. Rates are estimated from a Poisson regression, adjusting for age, gender, race, and primary cause of renal failure, then standardized to the age, gender, race, and primary cause of renal disease makeup of the national population of dialysis patients incident in 2008. The state is the recipient's last known state of residence, not necessarily the state where the transplant was performed.

Transplant Discharge & follow-up

Figure 7.22 reports the percentage of patients with evidence of delayed graft function (defined by a need for dialysis in the first week after transplantation), by donor type and ECD and DCD status, as reported to the OPTN. Figure 7.23 shows median length of stay for transplant hospitalizations by donor type, ECD and DCD status, and delayed graft function. For both figures, each year's cohort includes all adult patients whose grafts were functioning upon discharge. DCD status was unavailable until 1993, and not reported prior to that year. And Figure 7.24 shows the percentage of adult transplants with primary non-function, defined as kidney failure within seven days of transplantation.

Figure 7.25 presents the one-year cumulative incidence of acute rejections in adult, first-time, kidney-alone transplant patients discharged from the transplant hospitalization with a functioning graft. A patient is assumed to have acute rejection if OPTN data collection forms note 1) acute rejection episodes, 2) that medications were given for acute rejection, or that 3) acute rejection was the primary or secondary cause of graft failure. Biopsy-proven status was available starting in 1991 on the OPTN Transplant Recipient Registration, which identifies early rejection; it was not, however, added to the Transplant Recipient Follow-up form until April, 2003. Rejections that are a primary or contributing cause of graft failure are assumed to be biopsy-proven, while rejections identified by treatment status are not. Cumulative incidence is estimated using Kaplan-Meier methods, censored at death or graft failure.

Figure 7.26–27 illustrate the distribution of eGFR by donor type, at discharge and at 12 months post-transplant. Figure 7.26 includes adult patients discharged with a functioning graft within 15 days of transplantation, while Figure 7.27 includes patients alive with graft function 12 months post-transplant. GFR is estimated using the CKD-EPI equation.

Immunosuppression

Figure 7.28 presents data on immunosuppressive medications used in adult recipients at the time of transplantation, as reported on the OPTN Immunosuppression Treatment form. All such medications are indicated on the form as maintenance immunosuppression. The mycophenolate data includes mycophenolate mofetil and mycophonelate sodium, and the mTOR inhibitors include sirolimus and everoliumus. Data on mTOR inhibitors and steroids are also shown at one year post-transplant. Figure 7.29 highlights changes in the use of induction antibodies over the last decade, with data shown for first-time, kidney-alone transplants in 1998, 2003, and 2008.

Figures 7.30–33 address data related to KDIGO guidelines for post-transplant monitoring and prophylaxis. Figure 7.30 shows antibiotic use in the first three months post-transplant, Figure 7.32 illustrates use of cardiovascular medications in the first six months post-transplant, and Figure 7.33 presents data on the use of medications for the control of diabetes and hyperlipidemia during this same period. The cohort for these figures includes adult patients transplanted between January 1 and June 30, 2007, who remain alive with function six months post-transplant, and who have Medicare Parts A, B, and D coverage during this period. Medication use is defined by at least one prescription fill during the six months post-transplant. In Figure 7.33, other lipid lowering agents include bile acid sequestrants, cholesterol absorption inhibitors, fibrates, niacin, and omega-3 fatty acids; other anti-diabetes agents include alphaglucosidase inhibitors, biguanides, dipeptidyl peptidase-4 inhibitors, incretin mimetics, meglitinides, and amylin analogs.

Figure 7.31 displays the percent of patients with Medicare claims for influenza vaccinations, lipid testing, and CBC panels. The cohort is limited to adult patients with Medicare coverage, transplanted in 1991–2007, and discharged alive with graft function. To avoid counting inpatient procedures done as part of the transplant hospitalization, claims are searched from one day after the discharge date to one year post-transplant. Percentages are estimated using Kaplan-Meier methods, with censoring at graft failure, death, or loss of Medicare coverage. HCPCS codes for testing are as follows: influenza vaccination, 90724, 90657, 90658, 90659, 90660, and G0008; lipid panel, 80061, 82465, 83715, 83716, 83717, 83718, 83719, 83720, 83721, and 84478; and CBC panel, 85025, 85027, 80050, and 80055.

Transplant Outcomes

Figures 7.34–35 present five- and ten-year graft survival, as well as conditional half-lives, for adult recipients of kidneys from deceased and living donors. All estimates are made from Cox proportional hazards models, adjusted for transplant year, age, gender, race, and primary diagnosis, and based on the population's average survival curves, rather than on curves of the average patient in the population. Estimates of conditional half-lives are conditional on first-year graft survival, and estimated from the cumulative hazard between years one and two. Conditional half-lives are interpreted as the estimated median survival of grafts surviving the first year, while half-lives are interpreted as the estimated median survival of all grafts.

Figure 7.36 presents first-year and second-year post-transplant hospital admission rates for adult Medicare patients receiving their first kidney-alone transplant in 2006. Data are collected from Medicare claims occurring within two years of discharge from the transplant hospitalization, and exclude the hospitalization itself. Admission rates are censored at graft failure, loss of Medicare coverage, or December 31, 2008. Statistical methods for computing admission rates are similar to those described for Reference Section G, but cohorts are constructed differently. Instead of computing rates in point prevalent patients within a given year, we define the cohort based on the transplant year, and examine hospital claims up to a year post-transplant for first-year data and two years post-transplant for second-year data. Figure 7.37 illustrates the primary cause of hospitalization for cardiovascular problems and infection in the first and second years post-transplant in Medicare patients with their first kidney-alone transplant in 2004–2006.

Figure 7.38 presents data on the three-year incidence of post-transplant lymphoproliferative disorder (PTLD). The patient population includes first-time, kidney-only transplant recipients, 2001–2005. PTLD is identified from the OPTN Post-Transplant Malignancy form and the Transplant Recipient Follow-Up form. Figure 7.39 shows three-year cumulative incidence of malignancies other than PTLD, using data from OPTN data sources and Medicare claims. The cumulative incidence curve based on OPTN data uses the same population and methods as Figure 7.38, while the curve based on Medicare claims is limited to patients with Medicare primary coverage. PTLD and cancer indicated as cause of graft failure or death are included, under the assumption that they occurred prior to the graft failure or death they precipitated. Three-year incidence is estimated from a Cox proportional hazards model, adjusting for age, gender, race, Hispanic ethnicity, primary cause of renal failure, year of transplantation, donor type, hepatitis B and C serology, education level, employment status, time on dialysis, donor age, donor gender, donor race, HLA mismatches, recipient-donor body surface area matching, body mass index, panel reactive antibodies, cytomegalovirus matching, baseline maintenance immunosuppression (cyclosporin, neoral, tacrolimus, rapamycin, azathioprine, mycophenolate mofetil), and anti-lymphocyte receptor antibody use (IL-2, other), and estimated from the population average curve rather than the curve of the average patient. Events are censored at graft failure and death.

Figure 7.40 presents the three-year cumulative incidence of new onset diabetes following transplant, looking at Medicare patients transplanted during 2001–2005. To identify de novo post-transplant diabetes, the cohort is limited to patients with six months of Medicare primary payor coverage prior to transplantation; patients with claims for diabetes during this period are omitted. Cumulative incidence in the three years following the transplant is estimated using a Cox proportional hazards model, as above. Events are censored at graft failure, death, or loss of Medicare coverage.

Figure 7.41 shows the three-year cumulative incidence of post-transplant infection with BK virus. The study population, data sources, and analytic methods are similar to those used for 7.38. Pediatric patients are those 0–17 years old at the time of transplant, while adults are patients 18 and older at transplant.

Figure 7.42 shows the average eGFR by year post-transplant and by donor type among adult patients receiving a kidney-alone first transplant in 2002–2003. Annual serum creatinine levels from discharge to five years post-transplant come from the OPTN Transplant Recipient Follow-up form through year-end 2008, and eGFR is estimated using the CKD-EPI equation. We assume an eGFR of 0 for patients who remain alive and resume dialysis.

In Figure 7.43, we show the relationship between one-year eGFR and five-year graft failure. The cohort includes first-time, adult, kidney-alone transplant patients, 1999–2003, alive with function one year post-transplant. Patients are followed until graft failure, death, or five years post-transplant; the outcome is all-cause graft failure. Using a Cox proportional hazard model, we estimate the hazard ratio for graft failure by one-year eGFR group, adjusting for transplant year, age, gender, race, donor time, pre-transplant time on dialysis, recipient body mass index, primary cause of ESRD, number of HLA mismatches, PRA at transplant, primary payor, donor age, donor/recipient CMV matching status, and delayed graft function. GFR is estimated using the CKD-EPI equation.

In Figure 7.44 we present the rate of return to dialysis/preemptive retransplantation, the rate of death with a functioning graft, and the rate of any graft failure, which includes failure due to death. Rates are limited to adult patients, and estimated from a Poisson regression, adjusting for age, gender, and race.

Figure 7.45 displays causes of death for adult patients transplanted in 2004–2008 who subsequently die with a functioning graft. Causes of death are ascertained from OPTN transplant follow-up data, or, if unknown, from the ESRD Death Notification form.

Figures 7.46–51 examine the transfusion status of transplant and waiting list patients and its relationship to waiting time, PRA levels, and outcomes. Figure 7.46 includes all first-time, kidney-only transplant patients, 1991–2008, and displays the proportion of patients with history of any pre-transplant transfusion. Transfusion data come from the OPTN Transplant Recipient Registration, and indicate whether or not a transfusion occurred at some time prior to transplant; the timing and number of transfusions is not recorded. Figure 7.47 uses the subset of patients who received deceased donor transplants, and reports the median months from listing to transplant by pre-transplant transfusion status. Patients with no reported no transfusion history or an unknown history are classified as "not transfused."

Figure 7.48 presents the three-year cumulative incidence of transfusion among wait-listed patients by PRA level at listing. The cohort is limited to wait-list patients with primary Medicare coverage, and transfusion data come from Medicare claims. Incidence is estimated using Kaplan-Meier methods, with censoring at transplant, death, removal from the waiting list, or loss of Medicare coverage. Figure 7.49 shows the distribution of listed patients by PRA and number of years after listing for adults listed between 1998 and 2003. The PRA level is the last known PRA value at each post-listing year. The denominator for each post-listing year includes all patients who remain listed at that time. A patient who receives a transplant 18 months after listing, for example, will contribute to denominators at listing and at year one, but not years two through five.

Figure 7.50 shows the effect of pre-transplant transfusion on PRA elevations at transplant. The cohort includes first-time, kidney-only transplant recipients age 18 and older, transplanted in 2004–2008. Logistic regression models are used to predict the probability of PRA above six different cut points, adjusting for age, race, time on dialysis, primary cause of ESRD, blood type, education, employment, BMI, and comorbid conditions. Separate models are run for men, nulliparous women, and parous women, and odds ratios are plotted for positive versus negative transfusion history, along with 95 percent confidence intervals. Figure 7.51 shows the relationship between pre-transplant transfusion and rates of death and transplant. The patient population includes patients age 18 and older with Medicare primary coverage, listed for a kidney transplant in 2001–2005. Hazard ratios come from a Cox proportional hazard model, where transfusion is a time-dependent covariate, and transfusion information comes from Medicare claims. Other adjusters include year of listing, age, gender, race, ethnicity, cause of ESRD, blood type, BMI, pre-transplant time on dialysis, education, dialysis type, and comorbid conditions as noted on the ME form.

Reference section E

Tables E.1–5 present measures regarding the wait list for renal transplantation. Wait list data prior to 1988 are not shown; the OPTN wait list began in earnest in 1987. Table E.1 presents counts of patients newly added to the wait list for a kidney or kidney-pancreas transplant on December 31 of the given year. Patients listed at multiple transplant centers are counted only once. Table E.2 presents wait times, defined as the median time in days from first listing to transplant among patients listed for a kidney-alone transplant, and estimated with the Kaplan-Meier method. Table E.3 presents counts of patients on the wait list on December 31 of the given year, regardless of when the first listing occurred. Table E.4 includes point prevalent dialysis patients on December 31 of the given year. Table E.5 presents the percentage of patients wait-listed or receiving a transplant within one year of ESRD initiation; patients receiving a transplant from a living donor are excluded from the measure in the first half of the table and included in the second half. Percentages are calculated using the Kaplan-Meier methodology. This measure is modeled after Healthy People 2010 Objective 4.5.

Transplant counts are presented in Tables E.6–8. All known transplant events are included unless specified in the footnote, and all counts include non-Medicare patients. Table E.8 illustrates the distribution of transplanted patients by donor type and PRA level, determined from the OPTN Recipient Histocompatibility form, and shows as well a cross-tabulation of recipients and donors in terms of CMV antibody status, hepatitis C antibody status, and Epstein-Barr antibody status at the time of transplantation. A recipient/donor is considered positive for any of these antibodies if any applicable OPTN data source indicates positive. "Unknown" status is applied when no applicable data fields indicate "positive" or "negative." Cold ischemia time (in hours) is reported for deceased donor transplants only, and is taken from the OPTN Transplant Recipient Registration form.

Transplant rates per 100 dialysis patient years are shown in Table E.9. All hemodialysis patients, peritoneal dialysis (CAPD/CCPD) patients, and patients on an unknown form of dialysis are included, as are all non-Medicare patients. A patient's dialysis days are counted from the beginning of the specified year, or day one of ESRD dialysis therapy if treatment begins mid-year, until the first of transplant, death, or the end of the year. Patients lost to follow-up in a given year are not censored at the lost-to-follow-up date, but are followed until the end of the calendar year. Dialysis time for patients returning from transplant is counted. Transplant rates are calculated as the number of transplant events divided by the total number of dialysis patient years for each year.

Reference section F

This section presents probabilities of graft survival and graft failure necessitating dialysis or retransplantation, by donor type, age, gender, race, ethnicity, primary diagnosis, and transplant number. Data are presented for outcomes at 90 days, one year, two years, three years, five years, and ten years post-transplant. In previous ADRs, "graft failure necessitating dialysis or retransplantation" was referred to as "death-censored graft failure." Due to confusion regarding terminology, we have renamed this outcome. This section now seeks to address two major issues: the probability of graft survival at various times post-transplant, and the probability that a patient will return to dialysis or require retransplantation at various times post-transplant. Patients are followed from the transplant date to graft failure, death, or the end of the follow-up period (December 31, 2008). In the analysis of graft survival, death is considered a graft failure. In the analysis of graft failure necessitating dialysis or retransplantation, patients are followed until graft failure (excluding death), and patient follow-up is censored at death. To produce a standard patient cohort, patients with unknown age or gender are omitted. Unknown age is defined as a missing age at transplant, or an age calculated to be less than zero or greater than or equal to 100. Patients are also excluded if their first ESRD service date is prior to 1977.

Unadjusted survival probabilities are estimated using the Kaplan-Meier methodology, while the Cox proportional hazards model is used for adjusted probabilities. Probabilities are adjusted for age, gender, race, primary diagnosis, and first versus subsequent transplant, and standardized to 2005 patient characteristics.

Chapter Eight Pediatric esrd  Top

Information on pediatric patients is a subset of ESRD patient data used throughout the ADR; methods used for most figures are therefore the same as those described in the related chapter discussions.

Preventive care

Figures 8.11–12 show rates of preventive healthcare in pediatric ESRD patients by modality and race. Methods and codes used to determine vaccination rates are similar to those described for Chapter Five. All patients are age 0–19 at the beginning of each study period; reside in the 50 states, the District of Columbia, Puerto Rico, or the Territories; and have Medicare inpatient/outpatient and physician/supplier coverage for the entire period.

For influenza vaccinations, the cohort includes patients starting ESRD therapy at least 90 days prior to September 1 and alive on December 31 of each year; rates are calculated for patients vaccinated in the last four months of each year. For pneumococcal pneumonia vaccinations, the cohort includes prevalent patients initiating therapy at least 90 days prior to January 1 of the first year of each two-year period and alive on December 31 of the second year; rates are calculated for patients receiving one vaccination in each period. Years 2005–2008 are grouped in Figures 8.11, and 2005–2006 and 2007–2008 are grouped in Figure 8.12.

Hospitalization & survival

Figures 8.8–10 and 8.13–15 show admission rates among pediatric ESRD patients. Patients have Medicare as their primary payor and are residents of the 50 states, the District of Columbia, Puerto Rico, and the Territories. Patients with AIDS as a primary or secondary cause of death, and those with missing age or gender information, are excluded.

Figures 8.8–10 include period prevalent ESRD patients age 0–19 during pooled years 2005–2008. Age is determined on January 1 of each year. Cohorts and admission rate calculations follow those described for Reference Section G. Rates in Figures 8.8–10 are unadjusted. Principal ICD-9-CM codes for bacteremia/septicemia and pneumonia are listed under the discussion of Figure 6.4; those for respiratory infection exclude pneumonia and are as follows: 460–466, 472–474.0x, 475–476.1, 478.21–478.24, 487.1–487.8, 488–490, 491.1, 494, 510–511, 513.0, 518.6, and 519.01.

Figures 8.13–15 present adjusted admission rates in the first year of dialysis among incident dialysis patients age 0–19 in 2000–2007. Since in-center hemodialysis patients who are younger than 65 and not disabled cannot bill for hospitalizations until 90 days after ESRD initiation, the 90-day rule is applied. Patients are required to survive the first 90 days after dialysis initiation, and are followed for admissions for up to one year. Data cleaning and counting of admissions and time at risk for admissions generally follow that described for Reference Section G; here, however, incident patients are followed during intervals following day 90 rather than during prevalent years. Censoring occurs at death, loss to follow-up, three days prior to transplant, end of payor status, December 31, 2008, or after one year. Rates by age are adjusted for gender, race, and primary diagnosis, and those by race are adjusted for age, gender, and primary diagnosis. Adjusted rates are calculated with a model-based adjustment method and an interval Poisson model. The reference cohort includes incident dialysis patients age 0–19 in 2004–2005. Principal ICD-9-CM diagnosis codes used for cardiovascular and infectious hospitalizations are listed in the discussion of Figure 6.2.

Figure 8.16–18 present adjusted all-cause and cause-specific mortality in the first months of dialysis, by age, for incident dialysis patients younger than 20. Patients are followed from the day of ESRD onset until December 31, 2008, and censored at loss to follow-up or transplantation. Rates are adjusted for gender, race, and primary diagnosis of ESRD. Incident dialysis patients younger than 20, 2004–2005, are used as the reference cohort.

Figure 8.19 presents five-year survival for 1994–1998 and 1999–2003 incident dialysis patients, age 0–19. Patients with unknown age, gender, or primary diagnosis are excluded, as are dialysis patients who die or receive a transplant in the first 90 days. Patients are followed from day 91 until death, transplant, or the end of 2008. Probabilities by age are adjusted for gender, race, and primary diagnosis; probabilities by race are adjusted for age, gender, and primary diagnosis; overall probabilities are adjusted for age, gender, race, and primary diagnosis. The reference population consists of 2004–2005 incident pediatric ESRD patients.

Chapter Nine Special Studies  Top

Cardiovascular Special Studies

Figures 9.2–3 describe the use of implantable cardioverter defibrillators (ICDs) or cardiac resynchronization therapy with defibrillator (CRT-D) in ESRD patients. Annual study cohorts include period prevalent Medicare hemodialysis, peritoneal dialysis, and transplant patients followed from either January 1 (for point prevalent patients) or ESRD day 90 (for incident patients) until the earliest of death, end of Medicare as primary payor status, modality change, or December 31 of the year. Device implantation is identified from an inpatient or outpatient facility claim with ICD-9-CM procedure codes 37.94 (ICD) or 00.51 (CRT-D, for claims after January 1, 2002).

Figures 9.4–5 describe the demographics and comorbidity of patients age 20 or older who received their first ICD or CRT-D between 1999 and 2008. Comorbid conditions are defined from the ME form and from Medicare claims submitted during the one year prior to device implantation. Figure 9.6 shows all-cause survival after ICD or CRT-D implantation, with stratification by indication (primary or secondary prevention), in patients age 20 or older who received their first ICD or CRT-D between 1999 and 2008. Secondary prevention is indicated by ICD-9-CM diagnosis codes 427.1 (paroxysmal ventricular tachycardia), 427.4, or 427.5 during the hospitalization for device implantation. The absence of such diagnoses indicates primary prevention. Patients are followed from the date of first device implantation to the earliest of death, modality change, date of second device implantation, three years after implantation, or December 31, 2008.

Table 9.a describes prescription drug therapy in Medicare dialysis patients with their first diagnosis for cardiovascular disease (CVD) or receiving their first treatment for CVD in 2007. Index events for CVD include acute myocardial infarction (AMI), atrial fibrillation (AF), cerebrovascular accident/transient ischemic attack (CVA/TIA), congestive heart failure (CHF), and peripheral arterial disease (PAD), while index events for CVD treatment include percutaneous coronary interventions (PCI), coronary artery bypass graft surgery (CABG), and use of an implantable cardioverter defibrillator (ICD) or cardiac resynchronization therapy with defibrillator (CRT-D).

For each of the eight index events, a study cohort is identified from the 2007 Medicare ESRD database. Patients have the index event during 2007, are continuously enrolled in Medicare inpatient/outpatient and physician/supplier coverage during the one year preceding the date of the index event, and are not enrolled in an HMO during that year. The one-year period preceding the date of the index event is the baseline period. Patients with a pre-existing condition of the index event are also identified during the baseline period, but are not excluded in the analysis for Table 9.a.

Using the method employed to identify patients with CKD, we identify those with pre-existing AMI, AF, CVA/TIA, or CHF during the baseline period. ICD/CRT-D is defined through ICD-9-CM procedure codes in inpatient/outpatient facility claims, and PCI and CABG are identified through both ICD-9-CM procedure codes (in inpatient/outpatient claims) and CPT codes (in physician/supplier claims). PAD is defined through either diagnosis codes or procedure codes; if by diagnosis codes, we use the standard method; if by procedure codes, we employ the method used for PCI and CABG. AMI, AF, CVA/TIA, CHF, PAD, first PCI and CABG surgery, and the first implantation of ICD/CRT-D are defined on the date of the first appearance of a diagnosis or procedure code in the 2007 claims.

With the exception of AMI, the data sources and methods used to define each event are the same as those used in defining the pre-existing condition at baseline. The AMI event is defined as the first appearance of the diagnosis code on an inpatient claim.

The same codes are used to define AF, PAD, PCI, CABG, and ICD/CRT-D as pre-existing conditions at baseline and as an event in 2007, while different codes are used for CHF, CVA/TIA and AMI:

AF: 427.3 (ICD-9-CM diagnosis code)
AMI, 410 and 412 for condition at baseline; 410, 410.x0, and 410.x1 for event (ICD-9-CM diagnosis codes)
CHF: 398.91, 422.xx, 425.x, 428.xx, 402.x1, 404.x1, 404.x3, and V42.1 for condition at baseline (ICD-9-CM diagnosis codes); 398.91, 425.x, 428.xx, 402.x1, 404.x1, and 404.x3 for event (ICD-9-CM diagnosis codes)
CVA/TIA: 430–438 for condition at baseline; 430–437 for event (ICD-9-CM diagnosis codes) PAD: 440–444, 447, and 557 (ICD-9-CM diagnosis codes); 84.0, 84.1, 84.91, 39.25, 39.26, and 39.29 (ICD-9-CM procedure codes); 24900, 24920, 25900, 25905, 25920, 25927, 27295, 27590, 27591, 27592, 27598, 27880, 27881, 27882, 27888, 27889, 28800, 28805, 34900, 35131, 35132, 35141, 35142, 35151, 35152, 34051, 34151, 34201, 34203, 34800–34834, 35081–35103, 35331, 35341, 35351, 35355, 35361, 35363, 35371, 35372, 35381, 35450, 35452, 35454, 35456, 35459, 35470, 35471, 35472, 35473, 35474, 35480, 35481, 35482, 35483, 35485, 35490, 35491, 35492, 35493, 35495, 35521, 35531, 35533, 35541, 35546, 35548, 35549, 35551, 35556, 35558, 35563, 35565, 35566, 35571, 35583, 35585, 35587, 35621, 35623, 35646, 35647, 35651, 35654, 35656, 35661, 35663, 35665, 35666, and 35671 (CPT codes)
CABG surgery: 36.1x (ICD-9-CM procedure codes); 33510–33523 and 33533–33536 (CPT codes) PCI: 00.66, 36.01, 36.02, 36.05, and 36.06 (ICD-9-CM procedure codes); 92980–92982, 92984, and 92995–92996 (CPT codes)
ICD: 37.94 (ICD-9-CM procedure code)
CRT-D: 00.51 (ICD-9-CM procedure code)
Table 9.a and Figures 9.7–10 include Medicare enrollees with a CVD event between January 1, 2007, and November 30, 2007, discharged within two weeks of the date of the index event (if hospitalized at the time of the event), remaining outside the hospital at one month after the date of the index event, and carrying continuous Medicare Part D coverage during the interval from one month before to one month after the date of the index event; use of a particular drug is defined by at least one filling of a prescription for the drug during this interval. Drugs are identified from National Drug Codes included on Part D claims, and linked with the 2007 edition of Red Book. In Figure 9.7, other drugs include digoxin, eplerenone, and spironolactone. In Figure 9.9, patients are followed from one month after the date of the index event until the earlier of death or one year after the date of the index event. In Figure 9.10, patients are followed from one month after the date of the index event until the earliest of hospitalization, death, or one year after the date of the index event. Patients with a pre-existing condition of the index event are excluded in Figures 9.9–10.

Figure 9.11 presents rates of rehospitalization and rehospitalization or death in Medicare dialysis patients with their first CVD diagnosis or treatment in 2007. Patients with a pre-existing condition of the index event are excluded.

In Table 9.b we summarize patient distribution by body mass index (BMI) in Medicare dialysis patients with their first CVD diagnosis or treatment in 2008. Cohort construction is identical to that of Table 9.a, and patients with a pre-existing condition are excluded. BMI is estimated from the mean of measurements on outpatient dialysis facility claims submitted between October 1 and December 31, 2007.

Figure 9.12 shows the adjusted relative risks of incident CHF in 2008. Cohort construction is identical to that of Table 9.a; patients with a pre-existing condition are excluded. BMI is defined as in Table 9.b. Patients are followed form January 1, 2008, until the earliest of a CHF event, death, change in modality, or December 31, 2008. Relative risks are derived from a Cox proportional hazards model, with adjustment for age, gender, race, diabetes, and dialytic modality. Figure 9.13 shows unadjusted survival in the cohort, by BMI, following the date of the CHF event. Patients are followed from the CHF event until the earliest of death or one year after the date of the CHF event. And Figure 9.14 presents adjusted relative risks of death in the cohort, following the date of the CHF event. Follow-up is identical to that described for Figure 9.13. Relative risks are derived from a Cox proportional hazards model, with adjustment for age, gender, race, diabetes, and dialytic modality.

Table 9.c displays demographic and BMI distributions for Medicare ESRD patients undergoing bariatric surgery in 1999–2008. Events are defined by inpatient facility claims with (1) a procedure code for bariatric surgery, (2) a diagnosis-related group (DRG) for obesity surgery or a diagnosis code for obesity, and (3) the absence of any ICD-9-CM diagnosis code for cancer of the digestive tract or peritoneum, or inflammatory bowel disease. Codes used to identify patients are as follows:

bariatric surgery (procedure codes): 43.5, 43.6, 43.7, 43.89, 44.31, 44.38, 44.39, 44.68, 44.69, 44.95, 44.96, 44.97, 45.50, 45.51, 45.62, 45.90, and 45.91.
obesity surgery (DRGs): 288 (before September 30, 2008) and 619, 620, and 621 (after October 1, 2008)
obesity (DRGs): 278.x and V85.4
cancer of the digestive tract or peritoneum (DRGs): 150–159
inflammatory bowel disease (DRGs): 555–558.
Events are retained if the patient has Medicare as primary payor at the time of surgery.

In Table 9.c, BMI is measured at the time of ESRD initiation, diabetes is defined from the ME form and from Medicare claims during the one year preceding surgery, and modality is defined at the time of surgery. Figure 9.16 shows the distribution of bariatric surgery types: adjustable band (procedure codes 44.95, 44.96, 44.97), gastric bypass (44.31, 44.38, 44.39), gastroplasty (44.68, 44.69), and malabsorptive (43.5, 43.6, 43.7, 43.89, 45.50, 45.51, 45.62, 45.90, and 45.91). Figure 9.17 illustrates survival following bariatric surgery; patients are followed until the earliest of death, three years after the time of surgery, or December 31, 2008. Figure 9.18 shows mean BMI, by month, during the two years following bariatric surgery, looking at dialysis patients undergoing bariatric surgery between 2005 and 2008. BMI is ascertained from outpatient dialysis facility claims.

Figure 9.19 presents geographic variations in BMI in 2005 and 2008 dialysis patients. Study cohorts include point prevalent dialysis patients, alive and with Medicare inpatient/outpatient and physician/supplier coverage on January 1 of the year, with follow-up until the earliest of death, modality change, cessation of Medicare coverage, or December 31 of the year. The BMI of each patient is estimated from the mean of all BMI measurements on outpatient dialysis facility claims submitted during follow-up.

Rehabilitation & Quality of Life Special Studies

Data summarized in this section include 319,564 incident dialysis patients for whom the 2005 version of the ME form was completed between January 1, 2005, and September 30, 2008. The study population is derived from the 379,035 patients in the PATIENTS 2009 SAF for whom the ME form was completed. Patients are excluded if their first ESRD service date does not match the date of first regular dialysis, if their dialysis start date is prior to January 1, 2005 or more than 62 days from the date of signature on the ME form, if they are wait-listed or transplanted prior to dialysis initiation, or if their age information is missing.

Figures 9.20–24 use data from Question 26 on the ME form, while Figures 9.25–30 use data from Question 27. The ME form is also the source of all demographic information.

Medical insurance coverage categories correspond to those in Question 12 on the ME form, as follows:

Private: employer group health insurance
Medicare: Medicare, Medicare Advantage
Other: DVA, other
Medicaid: Medicaid
None: none
Figures 9.31–32 present Kaplan-Meier estimates of wait-listed patients, by whether or not they are informed of transplant options, and by selected reasons for not being informed. In the Kaplan-Meier estimation, death and kidney transplantation prior to wait-list are treated as censoring.

Chapter Ten Nutrition special studies  Top

Throughout the atlas and in Reference Section J, we define a chain-affiliated unit as one of a group of 20 or more freestanding dialysis units which are owned or operated by a corporation at the end of a year. The affiliation category of "small dialysis organization," or SDO, includes all organizations meeting our definition of a chain but having 20 or more and fewer than 100 units.

In previous years, chain affiliation was determined from the "Provider Name" field of the CMS Annual Facility Survey and the "Chain Organization Name" field of the CMS Independent Renal Facility Cost Report. Currently, however, it is determined solely from the "Chain Name" field of the CMS patient-accessible, web-based Dialysis Facility Compare database (DFC). If there is a major reorganization of chains, such as the purchase of Gambro by DaVita, the USRDS will insure that these transactions are represented in the ADR.

Data are obtained from the Facility Survey (1988–2007), the Cost Report (Form 265-94, 1994–2000), the DFC database (2001 to the present), and the CDC National Surveillance of Dialysis-Associated Diseases in the United States (1988–2002, excluding 1998, when the CDC did not conduct a survey). The CDC discontinued the National Surveillance of Dialysis-Associated Diseases after 2002. In 2008, there were 5,501 facilities in the Facility Survey.

A facility's hospital-based or freestanding status is determined from the third and fourth digits of the provider number assigned to each dialysis unit by CMS. For years prior to 2002, we determine facility profit status through the ownership type field on the CMS survey. In the 2002 CMS survey the profit status variable was dropped, so for that and subsequent years we use the profit status field of the DFC database. There are, however, a small number of facilities in the CMS survey that are not in the DFC database; these facilities have an unknown profit status, and are omitted from any graph showing profit status.

For provider-specific analyses, unless otherwise noted, the dialysis provider for individual patients is assigned as follows: for prevalent studies, the patient is assigned to the facility providing dialysis services at the prevalent date, as determined from the treatment history. For incident analyses, the patient is assigned to the facility providing dialysis services at the incident date, as determined from the treatment history. In either case, if provider data are unavailable from the patient's treatment history, the patient is assigned to "unknown provider" or excluded, depending on the analysis.

Figure 10.5 includes period prevalent dialysis patients in 2003 and 2008. Data for mean hemoglobin include only patients with valid EPO claims. A mean is calculated for each patient from all valid claims during the year, and chain affiliation is defined at the final patient claim of the year.

Figure 10.6 illustrates IV iron use, by dialysis unit affiliation and product type. The cohort consists of patients initiating ESRD therapy at least 90 days prior to the start of 2008, and receiving dialysis on December 31, 2007. All patients survive, continue dialysis, and carry Medicare as primary payor during all of 2008. Iron use is indicated by inpatient/outpatient claims with HCPCS codes J1750, J1755–J1756, J1760, J1770, J1780, and J2915–J2916. For iron use, chain affiliation is defined at the beginning of follow-up.

Figures 10.7–8 include data from all incident dialysis patients with an EPO claim in the first 30 days of ESRD therapy, and at least one EPO claim during each of the following six months. Figure 10.9 includes point prevalent dialysis patients in 2003 and 2008, with a first service date 90 days prior to January 1 of each year and alive through the end of the year. Rates represent patients with one or more transfusions within the year. In the case of an overlap in transfusion dates, only one event is used. If both inpatient and outpatient claims indicate a transfusion event and have the same "from" date, the inpatient claim is used; if inpatient and outpatient claims partially overlap, the claim with the earliest date is used; and if one or more short-period claims indicating a transfusion are within a long-period claim indicating a transfusion, the long period claim is used.

Figures 10.10–12 include incident hemodialysis patients in 2008, and show the cumulative probability of testing in the first three months of dialysis, by unit affiliation. Tests are identified from outpatient and physician/supplier claims during the year, using the following HCPCS codes: serum ferritin, 82728; transferrin saturation, 83550, 83540, and 84466; and parathyroid hormone testing, 83970.

Figures 10.13–15 use the same cohort as Figure 5.10, here for 2007–2008; Figure 10.16 uses the cohort from Figure 5.14, here for 2005 and 2008; Figure 10.17 uses the 2003–2004 and 2007–2008 cohorts from Figure 5.15; and Figure 10.18 uses the cohort from Figure 5.16, here for 2005 and 2008. All are limited to dialysis patients.

Figures 10.19–21 use the Model 1 (as-treated actuarial model) methods described for Chapter Eleven. Costs for clinical services (Figures 10.19–20) are taken from outpatient facility claims for period prevalent dialysis patients, and expressed as per person per month costs. Costs for preventive care (Figure 10.21) are obtained from the outpatient facility claims as well as physician/supplier claims, and are identified using CPT codes: A1c testing, 83036 and 83037; lipid testing, 80061, 82465, 83715–83721, and 84478; diabetic eye examinations, 67101, 67105, 67107, 67108, 67110, 67112, 67141, 67145, 67208, 67210, 67218, 67227, 67228, 92002, 92004, 92012, 92014, 92018, 92019, 92225, 92226, 92230, 92235, 92240, 92250, and 92260; influenza vaccinations, 90645, 90646, 90647, 90648, 90565, 90656, 90657, 90658, 90659, 90660, 90724, 90737, 90661–90663 and G0008; and pneumococcal pneumonia vaccinations, 90669, 90732, J6065, and G0009. Comprehensive diabetic monitoring includes at least four A1c tests, at least two lipid tests, and at least one diabetic eye examination per year.

Figures 10.22–25 present rates of hospitalization for vascular access infection. The cohort includes 2007 incident hemodialysis patients with Medicare as primary payor. Catheter access and internal access (arteriovenous graft or arteriovenous fistula) are identified from the ME form. Patients are followed from the date of initiation until the earliest of death, modality change, payor change, loss to follow-up, or one year. Hospitalization for vascular access infection is identified from inpatient claims by ICD-9-CM princial diagnosis code 99662.

Figures 10.26–29 show rates of hospitalization for fluid overload, heart failure, and pulmonary edema. The overall cohort includes 2008 point prevalent hemodialysis patients with Medicare as primary payor. Patients are followed from January 1, 2008, to the earliest of death, modality change, payor change, loss to follow-up, or December 31, 2008. The initial cohort for "extra dialysis" is the same, but the first four weeks of 2008 are used as the entry period to identify patients receiving extra dialysis, defined as more than three dialysis sessions in at least three of the four weeks. These patients are followed from the day after the entry period to the earliest of death, modality change, payor change date, loss to follow-up, or December 31, 2008. Hospitalization for fluid overload, heart failure, or pulmonary edema is identified using the method of Arneson, et al.

Figures 10.30–37 compare mortality and hospitalization among dialysis provider types, chains, and regions, using standardized mortality ratios (SMRs) and standardized hospitalization ratios (SHRs). Both are estimated by the traditional SMR calculation method. A patient's dialysis provider is defined on January 1, 2008. Patients are followed from January 1, 2008, to the first of death, transplant, or December 31, 2008. Patients dying of AIDS are excluded; those dying of drug overdose (street drugs) or an accident not related to treatment are censored at the date of death. SMR calculations include all January 1, 2008, point prevalent hemodialysis patients, while SHR calculations include only hemodialysis patients with Medicare as primary payor, and use the number of hospital admissions as the endpoint. Both SMRs and SHRs are adjusted for age, gender, race, primary diagnosis, and vintage, with the 2008 point prevalent hemodialysis patients as the reference cohort for the SMR calculations, and Medicare patients used for the SHR data.

Chapter Eleven Costs of esrd  Top

The majority of the economic analyses in this year's ADR use the as-treated model, described in detail later in this section.

Payor Sequence

The payor sequence is similar in concept to the USRDS treatment history. Payor status is tracked for each ESRD patient from the first ESRD service date until death or the end of the study period. Data from the Medicare Enrollment Database, as well as dialysis claims information, are used to categorize payor status as Medicare primary payor (MPP), Medicare secondary payor (MSP), Medicare Advantage (HMO), or non-Medicare. The claims database contains data only for MPP and MSP patients, so economic analyses are restricted to these categories. In addition, since it is impossible to determine the complete cost of care for ESRD patients with MSP coverage, most analyses exclude patients during the periods when they have this coverage.

Table p.a in the Prιcis summarizes data on the costs of ESRD treatment. Total 2008 Medicare spending is calculated from the claims data, and includes all paid claims for ESRD patients in the USRDS database. Cost aggregation for each patient begins at the first ESRD service date. Total 2008 Medicare spending is inflated by 2 percent to account for incomplete claims, and organ acquisition costs are estimated with the same methods used in the 1999 ADR (pages 149–150). HMO costs are estimated using the total HMO months for 2008 (obtained from the CMS managed care organization file) in conjunction with the 2008 AAPCC rate.

Non-Medicare EGHP spending is estimated by separately computing the per year at-risk costs for EGHP and non-EGHP patients, then multiplying the difference by the EGHP years at risk for 2008. Patient obligations are estimated as the difference between Medicare allowable and net payment amounts. Non-Medicare patient spending is estimated as the number of patient months at risk for non-Medicare patients (determined from the USRDS payor sequence) multiplied by the AAPCC rate.

Changes in Medicare spending from 2007 to 2008 are obtained from Table K.2, without the 2 percent adjustment for late claims. Calculations of per person per year (PPPY) at-risk costs are based on patients for whom Medicare is the primary payor during the study period (Table K.e), again using non-inflated results. The range for inflation-adjusted costs is calculated using the overall Consumer Price Index (3.8 percent) and Medical Consumer Price Index (4.2 percent).

Data on costs for vascular access physician/supplier services (Figures 11.16–17) are obtained directly from the physician/supplier Standard Analytical Files (SAF), and do not include facility charges. Physician/supplier vascular access procedures and costs are identified through CPT codes (Table a.b). Because some CPT codes are not specific to an ESRD access (e.g., central venous catheter, radiological procedures), our selection process requires that certain CPT codes be accompanied by a renal-related diagnosis code for inclusion in the analysis (these codes are identified with an asterisk in Table a.b). PPPY total vascular access costs (Figure 11.18) are obtained from event-based analyses, and include both physician/supplier costs as described above, and facility costs that can be attributed to vascular access services. Facility costs are difficult to identify. For inpatient facility costs, vascular access procedures in the inpatient setting (identified from the physician/supplier SAF) are matched with inpatient claims, and all procedures performed during a given inpatient stay (admission date through discharge date) are considered a single vascular access event. Because vascular access procedures are often performed when a patient is hospitalized for another reason, costs for inpatient facilities are included in the analysis only if the cause of hospitalization can be reasonably attributed to vascular access, using DRG and ICD-9-CM principal procedure codes, or ICD-9-CM principal diagnosis codes (Table a.b). Such hospitalizations are labeled "pure" inpatient vascular access events.

For outpatient facility costs, physician/supplier claims with vascular access procedures performed in the outpatient setting are linked to outpatient claims, using service dates and CPT codes. These costs are included in the analysis only if a matching CPT code is found on both physician/supplier and outpatient claims. Once again, all procedures and costs for the entire matching outpatient claim are considered part of a single vascular access event. Since the CPT code is not a required element on outpatient claims, not all outpatient facility costs for vascular access can be identified. Events that can be identified in the outpatient claims are labeled "pure" outpatient vascular access events.

Although vascular access procedures can be identified from claims data, it is not possible to determine with certainty the type of access being used for dialysis at any given time. In order to compare overall and vascular access costs by type of access, data are analyzed for the hemodialysis cohort from the CMS ESRD Clinical Performance Measures Project for 1999–2008. The CPM project collects data annually on a random sample of hemodialysis and peritoneal dialysis patients, including the type of vascular access being used for hemodialysis at the time of data collection. The CPM data for hemodialysis patients are collected from October through December of the year prior to the cohort year (e.g., CPM data were collected from October through December, 2007, for the 2008 cohort). For Figures 11.18–19 we classify patients by the vascular access in use at the time of the CPM data collection, and aggregate costs for the following calendar year, with follow-up until the earliest of death, transplant, modality change, or the end of the calendar year. This analysis is limited to patients with Medicare as primary payor.

Figure 11.20 includes period prevalent hemodialysis patients with Medicare as primary payor, not enrolled in Medicare Advantage, and residing in the United States. Inpatient hospital stays during calendar year 2008 are identified as being for a vascular access infection if the principal diagnosis is 996.62, "Infection of Internal Device." The total payment amount from these inpatient stays is used to calculate a PPPY cost.

Figures 11.21–27 describe PPPY costs for items billed in the outpatient SAFs, particularly injectable drugs, for period prevalent dialysis patients with Medicare as primary payor. And Figures 11.28–33 detail PPPY costs billed in the inpatient SAF for this population, focusing on infectious hospitalizations. The causes of these hospitalizations are determined from the principal ICD-9-CM codes.

Figures 11.34–46 present PPPY costs for the services described in Figures 11.28–33, by modality and race. Modalities are determined using model 1 (as-treated actuarial model) methodology, as described below. Data are also presented for a subset of hemodialysis patients who are matched to peritoneal dialysis patients, using a propensity score technique. In the cohort of dialysis patients, we first estimate the propensity for peritoneal dialysis prescription by fitting a logistic model of dialytic modality, with age, race (white, black, other), gender, primary cause of ESRD (diabetes, hypertension, glomerulonephritis, cystic kidney disease, other known, unknown), cumulative ESRD duration, and seven diagnosed comorbid conditions (cardiovascular disease, hypertension, diabetes, COPD or tobacco use, cancer, alcohol or drug dependence, and in need of assistance) as predictors. Age and ESRD duration are parameterized with quadratic polynomials. The propensity for peritoneal dialysis prescription is defined as the estimated probability of peritoneal dialysis as dialytic modality. We then assemble a matched cohort by matching to each peritoneal dialysis patient with propensity p a hemodialysis patient with propensity q, such that |p – q| is minimized, and we use a greedy matching algorithm.

Figures 11.47–59 present cost data for the Medicare Part D prescription drug benefit. Data are currently available only for calendar years 2006 (the first year of the benefit) and 2007. Costs are estimated net pay, calculated as the sum of the plan payment amount and the low income subsidy; they do not include out-of-pocket expenditures. Figure 11.47 and Table 11.a include all Part D costs for ESRD patients, starting on January 1, 2006 (or the first ESRD service date if after this date), regardless of payor status. Figures 11.48–59 include December 31, 2006 point prevalent ESRD patients enrolled in Part D for all of 2007. Costs are estimated net pay (expressed as per person per month), are presented separately for dialysis and transplant patients, and are restricted to drugs in the specific categories addressed in each figure; combination drugs which cross categories (e.g., a beta blocker with a lipid lowering agent) are not included, unless specifically noted in the figure caption.

Reference section K: Medicare claims data

Cost information in this section is derived from Medicare inpatient/outpatient and physician/supplier claims data in the CMS SAFs, which are created annually six months after the end of each calendar year. The data for 2004–2008 are comprised of approximately 43 million institutional claims for hospital inpatient and outpatient facilities, outpatient dialysis facilities, skilled nursing facilities, hospice facilities, and home health agencies, as well as over 370 million line items from physician/supplier claims. Claims data are obtained for all patient identification numbers in the USRDS database, and the Renal Management Information System (REMIS) is used to gather all CMS ID numbers under which patients may have claims. The claims data are then merged with patient demographic data and modality information in the USRDS database.

The economic analyses for this section focus on two amounts found in the claims data: the claim payment amount, which is the amount of the payment made from the Medicare trust fund for the services covered by the claim record; and the pass-through per diem amount, which applies to inpatient claims and reimburses the provider for capital-related costs, direct medical education costs, and kidney acquisition costs.

Payment categories

Medicare payments are broken into several categories, as shown in Table a.c. Estimates of costs from the Outpatient SAF are derived for the individual services provided. Since actual payment amounts are provided only for the entire claim, cost estimates for dialysis, EPO, iron, and so forth are calculated from the claim-level "Total Charge," the payment amount, and the revenue line-level "Total Charge," as follows: payment (line) = [total charge (line) / total charge (claim)] * payment (claim). In August of 2000 CMS added to the Outpatient SAF a field containing line item payment amounts. According to CMS documentation, the total of these payments may not equal the total paid amount for the claim. In such cases, each line item cost is discounted by the ratio of the sum of line item payment amounts to the total paid amount for the claim. Since complete data on line item payments are available starting with the 2001 Outpatient SAF, the estimates for outpatient payment categories are taken directly from the claims data for calendar years 2001–2008, with adjustments as noted.

Model 1: as-treated actuarial model

In an as-treated model patients are first classified by their modality at entry into the analysis, and retain that classification until a modality change. When a change is encountered in the data, the beginning modality is censored at the change date plus 60 days, and a new observation with the new modality is created. The first 60 days after a change are attributed to the previous modality to account for any carryover effects. If the change is from dialysis to transplant, however, the modality is censored, and the transplant modality begins on the date of the transplant hospital admission. In the case of changes involving only a change from one type of dialysis to another, the new modality must last at least 60 days in order to be counted. Aggregation of Medicare payments is done on an as-treated basis, attributing all payments to the patient's modality at the time of the claim.

In Section K of the Reference Tables we classify patients into four as-treated modality categories: hemodialysis, CAPD/CCPD, other dialysis, and transplant. The "other dialysis" category includes cases in which the dialysis modality is unknown or is not hemodialysis or CAPD/CCPD, while the transplant category includes patients who have a functioning graft at the start of the period or who receive a transplant during the period. Some tables also include categories for all dialysis (hemodialysis, CAPD/CCPD, and other dialysis) and all ESRD (all-dialysis and transplant).

The study spans the 18 years from January 1, 1991, to December 31, 2008, and ESRD patients prevalent on January 1, 1991, or incident at any time during the period are potentially eligible for inclusion. The initial study start date for a given patient is defined as the latest of January 1, 1991, the first ESRD service date in the USRDS database for that patient, or the earliest Medicare eligibility date from the payor sequence. Because it is impossible to characterize the total cost of their care, patients for whom Medicare is the secondary payor at any time during the study period are classified as MSP for the duration of the MSP status in the payor sequence. If the payor status changes to Medicare as primary payor, a new sequence begins at the change date. Patients who are non-Medicare or enrolled in a Medicare Advantage program are excluded until their payor status changes to Medicare (either as primary or secondary payor). Patients classified as MSP are included in Tables K.1–3, and are excluded for the rest of the tables in Section K.

For each modality period, Medicare payments are aggregated from the modality start date until the earliest of death, transplant, modality change, loss to follow-up, or December 31, 2008. Patients incurring no inpatient/outpatient or physician/supplier Medicare costs for the entire period are excluded, and Medicare payment amounts are linearly prorated for claims that span the start or end date of a modality period or of the study itself.

To express costs as dollars per year at risk, total costs during the follow-up period are divided by the length of the period. Costs per patient year at risk are calculated by patient category, and stratified by age, gender, race, modality, and diabetic status. Diabetic status is based on the primary diagnosis, as recorded on the Medical Evidence form. A patient with a non-diabetic cause of renal failure may have diabetes, but the disease is not judged to be the cause of ESRD. Patient age is calculated at the study start date, and patients with a missing date of birth are excluded from the analysis.

Model 2: categorical calendar year model

This model, described in the HCFA (now CMS) research report on ESRD (1993–1995), is used for Figure 11.12, as well as Reference Tables K.9–12. With this method, patients are classified into four mutually exclusive treatment groups:

dialysis: ESRD patients who are on dialysis for the entire calendar year, or for that part of the year in which they are alive, ESRD, and Medicare entitled.
transplant: ESRD patients with a kidney transplant during the calendar year.
functioning graft: ESRD patients with a functioning graft for the entire calendar year, or for that part of the year in which they are alive, ESRD, and Medicare entitled.
graft failure: ESRD patients who have had a transplant, but return to dialysis due to loss of graft function during the calendar year; patients with a graft failure and a transplant in the same calendar year are classified in the transplant category.

EGHP patients

Several figures in the Prιcis and Chapter Eleven include data for EGHP patients. Patients in the MarketScan database who are identified as having ESRD, are under 65 years of age, and do not have evidence of Medicare payments (either as primary or secondary payor) are included in these analyses. Medicare payments are identified in the MarkestScan database, and patients are excluded on the basis of these payments in order to obtain a more accurate estimate of ESRD costs in the private sector. The payment amounts presented are the net payments and do not include deductibles and copayments.

Chapter Twelve International comparisons  Top

The international data for this ADR have been collected from the following sources, using the data form at the end of this section:

the Sociedad Argentina de Nefrologνa (SAN) and Instituto Nacional Central Unico Coordinador de Ablaciσn e Implante (INCUCAI; Marinovich et al.)
the Australian and New Zealand Dialysis and Transplant Registry (ANZDATA)
the Austria OEDTR
the Bangladesh Renal Registry
the French-Speaking Belgium ESRD Registry, Bruxelles
Nederlandstalige Belgische Vereniging voor Nefrologie (NBVN)
Clinical Center University of Sarajevo, Bosnia, and Herzegovina
Sociedade Brasileira de Nefrologia and Associacao Brasileira de Transplante de Orgaos the Canadian Organ Replacement Register (CORR
) the Chilean Renal Registry
the Croatian Society of Nephrology, Dialysis, and Transplantation
the Czech Dialysis Registry
the Danish Society of Nephrology
the ERA-EDTA Registry
the Finnish Registry for Kidney Diseases
the French Renal Epidemiology and Information Network (REIN) Registry
the Hellenic Renal Registry, Greece
the Hong Kong Renal Registry
the Department of Transplantation and Surgery in Hungary
the Landspitali University Hospital, Iceland
the Israeli Renal Registry
the Jalisco State Dialysis and Transplant Registry, Mexico
the Japanese Society of Dialysis Therapy
the Korean Society of Nephrology ESRD registry
Registre Nιphrologique du Grand Duchι de Luxembourg
the National Renal Registry, Malaysia
Instituto Mexicano De Trasplantes, Cuernavaca Morelos, Mexico
the Netherlands Dialysis Registry
the Norwegian National Hospital
the Philippines Renal Disease Registry Project
the Polish Renal Registry
the Romanian Renal Registry
the Society of Dialysis, Russia
the Scottish Renal Registry
the Registro Espaρol de Enfermos Renales and Organizaciσn Nacional de Trasplantes, Spain
the Swedish Renal Registry
the Taiwan Society of Nephrology
the Thailand Renal Replacement Therapy Registry and the Nephrology Society of Thailand
the Turkish Society of Nephrology
the UK Renal Registry
the Uruguayan Dialysis Registry and Uruguayan Registry of Renal Transplantation
the U.S. Census Bureau International Database
Thank you to all who provided data for this year's ADR. We are especially grateful to Dr. Kitty Jager and Anneke Kramer at the ERA-EDTA Registry for their help in coordinating much of the European data presented in this chapter. Data for some countries do not represent 100 percent of the ESRD population; interpretation of changes in incident and prevalent rates must therefore be performed with caution. Notations are made in the captions for those countries reporting data only for dialysis patients. Data from Belgium do not include patients younger than 20, and data from England, Wales and N. Ireland do not include patients younger than 18.

To contribute data from your country's registry, please complete the International Data Collection Form and return it to the USRDS at usrds@usrds.org.

ESRD in Asian & indigenous populations

Figures 12.9–11 examine the incidence of ESRD in the U.S. Asian population and in Japan and Taiwan. Unadjusted rates are expressed as the number of incident ESRD patients on dialysis per million population. U.S. data are obtained from the ME form, while data from Japan and Taiwan are provided by the Japanese Society of Dialysis Therapy and the Taiwan Society of Nephrology.

In Figures 12.12–14, the Canadian indigenous population includes First Nation, Metis, and Inuit; data are provided by the Canadian Organ Replacement Registry. Data for Australia and New Zealand are provided by the Australian and New Zealand Dialysis and Transplant Registry (ANZDATA). Since the general population for the indigenous populations in Canada and New Zealand is available only for selected census years, the population for years without census data is estimated by the formula, P t+n = P t (1+r) n. It is assumed that the general population increases or decreases over time with a consistent rate (r) based on the existing total population (Pt).

Vascular access

Tables L.1–3 include period prevalent hemodialysis patients, 1999–2007, who have Medicare as their primary payer. Placements are identified from Medicare claims, and rates represent the total number of events divided by the time at risk. Follow-up time is censored at death, change in modality, change in payor status, or the end of the prevalent year. Tables L.4–6 include January 1, 2007 point prevalent hemodialysis patients. Vintage represents the amount of time between the first service date and January 1, 2007.

Tables L.7–14 include point prevalent hemodialysis patients with Medicare as their primary payor who are also in the ESRD CPM report for the corresponding year. Current access is determined from the CPM data as the access used at the time of the most recent data collection, i.e., during the months of October, November, and December of the year prior to the prevalent year. Complications and intervention events are obtained from claims during the time at risk in the prevalent year, which is censored at death, change in modality, change in payor status, or a claim for the placement of a different hemodialysis vascular access. Patients who have a placement claim after the time of the CPM data collection but prior to the start of the prevalent year are excluded.

Tables L.14–15 include point prevalent peritoneal dialysis patients with Medicare as primary payor. Complications and intervention events are obtained from claims during the time at risk in the prevalent year, which is censored at death, change in modality, change in payor status, or a claim for the placement of a hemodialysis vascular access.

Census populations

The 2000 U.S. Census, available in 2002, introduced a new race category with additional groupings. Estimates for 1990–1999 were back-calculated based on the actual 2000 census. Later data, however, include racial groups that do not coincide with those in the ESRD data. For rate calculations throughout the ADR we thus use the CDC's Bridged Race Dataset, which estimates white, African American, Native American, and Asian populations. The data and methods for these estimates are available at http://tinyurl.com/28kpp9j.

Statistical methods

Methods for calculating rates

The calculation of observed rates is straightforward, with some rates based on counts and others on follow-up time. The ESRD incident rate in 2002, for example, is the observed incident count divided by the 2002 population size and, if the unit is per million population, multiplied by one million; the 2002 death rate for prevalent ESRD patients is the number of deaths in 2002 divided by the total follow-up time (patient years) in 2002 of the 2002 prevalent patients, and, if the unit is per thousand patient years, multiplied by one thousand. Standard errors of estimated rates are based on the assumption of the data; the observed count has a Poisson or binomial distribution. The count-based rate describes the proportion of having "event" and the time-based rate tells how often the "event" happens when the "event" rate is invariant over time.

Model-based rates

Some patient groups may be very small, and their observed rates therefore unstable. If follow-up time is considered, the hazard of an event may change over time. A model-based method can improve the stability of these estimates and incorporate changes of hazard over time. In this ADR, for example, we have used the generalized linear mixed Poisson model to estimate prevalent patient mortality rates for Reference Section H.

Measurement unit for rates

Both raw and model-based rates are calculated per unit of population (such as per 1,000 patients) or per unit of follow-up time (such as per 1,000 patient years). Calculating rates per unit of follow-up time can account for varying lengths of follow-up among patients. Patient years are calculated as the total number of years, or fractions of a year, of follow-up time for a group of patients.

Take, for example, a calculation of 1997 first hospitalization rates for two groups of patients, all receiving dialysis therapy on January 1, 1997. Group A consists of three patients: Patient 1 had a first hospitalization on March 31, 1997; Patient 2 was hospitalized on June 30, 1997; and Patient 3 was on dialysis through December 31, 1997, with no hospitalizations. Group B also has three patients: Patient 4 was first hospitalized on December 31, 1997; Patient 5 was hospitalized on September 30, 1997; and Patient 6 was on hemodialysis the entire year, with no hospitalizations through December 31, 1997.

Patients 1 to 6 contribute 0.25, 0.5, 1.0, 1.0, 0.75, and 1.0 patient years at risk, respectively. The first hospitalization rate per thousand patients is 667 for both groups in 1997. But the first hospitalization rate per thousand patient years at risk is 1,143 for Group A and 727 for Group B (calculated as [2 total events / 1.75 total patient years at risk] x 1,000 for Group A and [2 total events / 2.75 patient years at risk] x 1,000 for Group B). The resulting rate is lower for Group B because of the longer total follow-up time.

Rates per unit of population may be influenced by the proportion of patients who are followed for only a fraction of a year. The event rate per unit of population is likely to be lower, for example, in a group of patients followed for only one month until censoring than in a group whose patients are each followed for up to a full year. Rates per unit of follow-up time at risk, in contrast, count only the actual time that a patient is at risk for the event.

Methods for adjusting rates

Because each cohort contains a different patient mix, unadjusted event rates may not be comparable across cohorts. Adjusted analyses make results comparable by reporting rates that would have arisen had each cohort contained patients with the same distribution of confounders — such as age, gender, race, and primary diagnosis — as the reference population.

Direct adjustment

There are several rate adjustment methods, but only the direct method allows rates to be compared (Pickle LW, White AA). Here the adjusted rate is derived by applying the observed category-specific rates to a single standard population, i.e. the rate is a weighted average of the observed category-specific rates, using as weights the proportion of each category in the reference population. Categories are defined by the adjusting variables. For example, if a rate is adjusted for race and gender and there are three race groups (white, African American, and other) and two gender groups, there are six categories: white males, white females, African American males, African American females, males of other races, and females of other races.

Suppose we try to compare state-level incident rates in 2001 after removing the difference caused by race. To do this, we need to calculate the adjusted incident rate, adjusted for race, for each state. Because racial distributions in each state are quite different, we use as reference the national population — here, the population at the end of 2001 — with five race groups (white, African American, Native American, Asian/Pacific Islander, and other).

Assuming the incident rate of state A in 2001 is 173 per million population, and the race-specific rates and national populations are as shown in the following table, the adjusted incident rate of state A with the national population as reference is (153 x 75.1%) + (250 x 12.3%) + (303 x 0.9%) + (174 x 3.6%) + (220 x 8%) = 158.73 per million population. This means that if state A had the same racial distribution as the entire country, its incident rate would be 158.73 instead of 173. If state B had an adjusted incident rate of 205, we could say that state B had a higher incident rate than state A if they both had the same racial distribution as the whole country.

Incident rate National of State A population (%) White 153 75.1
African American 250 12.3
Native American 303 0.9
Asian/Pacific Islander 174 3.6
Other 220 8.0

This method is used to produce some adjusted incident and prevalent rates in Chapters Two and Three and in Reference Sections A and B, as well as in the model-based adjustment method.

Model-based adjustment

Under some circumstances there are disadvantages to the direct adjustment method. Suppose we are calculating mortality rates for a set of groups, and adjusting for potential confounding variables. If one category in a group has only a few patients or deaths, its estimated mortality rate will be unstable, likely making the adjusted rate unstable as well. In addition, if one includes category no patients, the method is not valid for calculating an adjusted mortality rate for the group. An attractive alternative is a model-based approach, in which we find a good model to calculate category-specific estimated rates for each group and then calculate direct adjusted rates using these estimates with a given reference population. This method can also be extended to adjustments with continuous adjusting variables (Liu et al., 2006). There is, unfortunately, no straightforward way here to calculate standard errors of the adjusted rates for some models; the bootstrap approach works well, but is time consuming.

In this ADR we use model-based adjustments to calculate adjusted mortality rates; adjusted survival probabilities based on the Cox regression model; adjusted hospitalization rates and state-level adjusted incident and prevalent rates using the Poisson model; adjusted HSA-level incident and prevalent rates based on the Bayesian spatial hierarchical model, and some other rates, described in the text on the individual figures.

Survival probabilities & mortality rates

Unadjusted survival probabilities

In this ADR, unadjusted survival probabilities are calculated using the Kaplan-Meier method, and corresponding standard errors are calculated with Greenwood's formula (Kalbfleisch JD, Prentice RL). Survival probabilities in Reference Section I are expressed as percentages from 0 to 100. The mortality/event rate in the period of (0,t) is calculated by –ln(Survivor at time t). This event rate will be the same as that estimated by event time divided by follow-up time after adjustment of the unit if the event rate is a constant over time.

Survival probability with competing risks

When competing risks exist, the estimate of the cumulative incidence function of a specific cause may be biased if the other competing risks are ignored. If we have K competing risks, the cumulative incidence function of cause k, k=1, 2, , K, at time t, Ik(t), is defined as the probability of failing from cause k before time t (including time t), Prob(T=t, D=k). Then
Ik(t) = ?0t?k(s)S(s)ds
where ?k(s) is the hazard of event from cause k at time s and S(s) is the survival probability at time s. If we have failing time t1, t2, , tm, the cumulative incidence function of cause k at time t is estimated by

^Ik(t) = ?^?k(tj)S^(tj-1)

j:tj=t

where ^?k(tj)=dkj /nj, S^(tj–1) is the Kaplan-Meier estimate of survival at time tj–1, dkj is the number of patients failing from cause k at time tj, and nj is the number of patients at risk at prior time tj (Putter et al.).

Adjusted survival probabilities

Adjusted survival probabilities are reported in Reference Sections G and I, with age, gender, race, and primary diagnosis used as adjusting risk factors. The model-based adjustment method is used, with survival probabilities predicted from the Cox regression model (Kalbfleisch JD, Prentice RL). This process yields estimates of probabilities that would have arisen in each year if the patients had had the same attributes as the reference population. Since the probabilities in each table are adjusted to the same reference set of patient attributes, any remaining differences among cohorts and years are due to factors other than age, gender, race, and primary diagnosis. The adjusted mortality rates for incident cohorts in Reference Section H are calculated using similar methods.

Generalized linear models

Generalized linear mixed model for mortality rates

We use the generalized linear mixed model with log link and Poisson distribution to calculate mortality and first transplant rates for prevalent patients. While rates are reported for a year, data from the previous two years with different weights are also used to improve the stability of the estimates. The generalized linear mixed model is used as well for SMR calculations, described later in this section.

The generalized linear mixed model, which considers both fixed and random effects, is implemented using the SAS macro GLIMMIX. Rates for the intersections of age, gender, race, and diagnosis are estimated using the log linear equation Log (rate) = (fixed effects) + (random effect). Fixed effects include year, age, gender, race, and primary diagnosis, and all two-way interactions among age, gender, race, and primary diagnosis. Assumed to be independently and identically distributed with a normal distribution, the random effect is the four-way interaction of age, gender, race, and primary diagnosis. Age is used as a categorical variable in main effect and four-way interactions, and as a continuous variable in the two-way interactions.

For tables with mortality rates for both intersecting and marginal groups we have used a single model to calculate all rates in each table. The marginal rates are simply the weighted averages of the estimated, cross-classified rates, with cell-specific patient years as weights. For this approach the use of a single model means that GLIMMIX cannot give the standard errors for some of these estimated rates; the bootstrap method is therefore used instead.

The adjusted mortality rates for prevalent cohorts in Section H are calculated using the direct adjustment method based on the category-specific mortality rates from the generalized linear mixed models.

Generalized linear model for hospitalization rates

In this ADR, hospitalization reference tables present rates of total admissions and hospital days. We use a generalized linear model with log link and Poisson distribution; the model includes age, gender, race, primary diagnosis, and their two-way interactions.

To stabilize the estimates, three years of data are used with different weights. Year is also included in the model as a covariate. The adjusted hospitalization rates are calculated using the direct adjustment method based on the category-specific admission rate from the generalized linear models.

Standardized mortality ratios

The standardized mortality ratio (SMR) compares the mortality of a group of patients relative to a specific norm, or reference, after adjusting for some important risk factors. For example, the state-level SMR is used to compare mortality in prevalent dialysis patients — after adjusting for age, gender, race, primary diagnosis, and ESRD vintage — in each state using the national dialysis population in the corresponding year as the reference. An SMR of 1.05 for a state indicates that patients in this state have a risk of death approximately five percent higher than that of patients in the reference population of all U.S. dialysis patients.

Traditional method of SMR calculation

The traditional approach used to calculate unit-specific SMRs is straightforward: produce unit-specific expected death counts and compute the "observed/expected" ratio. There are two methods of producing unit-specific expected death counts. In the indirect method, the expected death count is the weighted sum of category-specific death rates in the reference population, and the weights are the category-specific total follow-up times in the units. In the model-based method, a statistical model is employed to estimate the category-specific death rate for the reference population, and the indirect method is then used to produce the expected death count for each unit based on the estimates of category-specific death rates of the reference population from the model.

Expected remaining lifetimes

The expected remaining lifetime for a patient group is the average of the remaining life expectancies for the patients in that group. Some patients will live longer than, and some will live less than, the average. Although the average cannot be known until all patients in the cohort have died, the expected remaining lifetime can be projected by assuming that patients in the cohort will die at the same rates as those observed among groups of recently prevalent ESRD patients.

For a subgroup of ESRD patients of a particular age, the expected remaining lifetime is calculated using a survival function, estimated for the group. Let S(A) denote the survival function of patients at time A. Among patients alive at age A, the probability of surviving X more years is S(X|A) = S(A+X)/S(A). For a given starting age A, the expected remaining lifetime is then equal to the area under the curve of S(X|A) plotted versus X. Because few patients live beyond 100, this area is truncated at the upper age limit A + X = 100.

Half-lives

Conditional half-life

The conditional half-life is conditional on having survived a given period of length T0 without the event, the point at which 50 percent of patients who survived the given period remain alive. In other words, it is the median remaining lifetime conditional on surviving a given period T0.

The conditional half-life is estimated using the Kaplan-Meier method if the median survival time falls in the duration of follow-up. Otherwise, the conditional half-life is estimated as the following:

Estimate the survival probabilities S(T0) and S(T1) using the Kaplan-Meier method from the data available, where T0< T1 and T1 is within the follow-up
΅ = T1–T0
(ln[S(T0)]–ln[S(T1)])
the estimate of the conditional half-life = ΅·ln(2)
This method can be used only when the hazard is a constant after T0 and T1 is chosen to be big enough to obtain a stable estimate of ln(S(T0))-ln(S(T1)).

Adjusted conditional half-life

When comparing the conditional half-time of different groups, the difference of conditional half-lives may partially reflect the difference of patient characteristics among groups. To remove the part of the difference attributable to patient characteristics, the adjusted conditional half-life can be estimated. For example, if patient age, race, and gender are different among groups and are factors for survival, we can calculate the adjusted conditional half-life by adjusting for age, race, and gender with a given reference population. The method for estimating the adjusted conditional half-life is the same as described above for the unadjusted conditional half-life estimate, with the exception of step 1. Usually a Cox regression model is fit for each group, with age, race, and gender as the covariates. The log survival at time T0 and T1 is calculated from the Cox model estimates for each cross-sectional subgroup of age*race*gender. The weighted average of the ln(survival)s over the subgroups at each time point is then calculated, with the patient proportion of each subgroup in the reference population as the weight. The ln(S(T0)) and the ln(S(T1)) in the first step above are replaced by the corresponding weighted averages.

Mapping methods

Mapping is an important tool for assessing environmental determinants and illustrating spatial patterns and temporal trends. Geographic resolution is enhanced by mapping at the level of small regions, but this can increase data instability. The use of smoothing methods, however, can help stabilize data and show geographic patterns while still maintaining geographic resolution.

Much of disease mapping within the ADR is by Health Service Area (HSA), an approach we continue to adopt from the Atlas of United States Mortality (Centers for Disease Control and Prevention). Remaining maps are by state or census division. Each HSA is a group of counties described by the CDC authors as "an area that is relatively self-contained with respect to hospital care." The methods described here have been used for all HSA-level maps in the ADR. Because the distribution of age, gender, and race in a population can affect incident and prevalent ESRD rates, we have included maps in which data are adjusted for these variables as well as smoothed. Maps by state and census division are not smoothed.

In many figures, data ranges have been standardized to invite comparisons across years, modalities, or patient characteristics. In remaining maps, HSAs are divided into quintiles.

Throughout the ADR, data in maps and graphs are unadjusted unless otherwise noted. HSA-level information is mapped according to the patient's residence (with the exception of some maps of organ donation rates in Chapter Seven). Because of area size and limitations in the mapping software, data for Puerto Rico and the U.S. Territories are not included in the maps.

Methods for smoothing & adjusting map data

To smooth map data we use a Bayesian spatial hierarchical model (Waller et al.). This method is a statistical approach that uses the log linear model (Poisson regression model) to fit the incident counts of the regions. The region effects, as random effects, follow the Conditional Autoregressive (CAR) Normal distribution, and the precision of the effects has a Gamma distribution. The model smooths the incident counts by borrowing information for each HSA from its neighbors through the relationship defined by CAR; neighbors, in our definition, are HSAs sharing a boundary. Smoothed incident rates are obtained by dividing the predicted counts by the corresponding population sizes. For adjusted maps, an almost non-informative prior is assigned to fixed effects of age, gender, and race with the Bayesian model. Adjusted incident rates are calculated using the model-based adjustment method based on the predicted values from the Bayesian spatial hierarchical model, with the national population as reference.

This model is also used to smooth prevalent rates and calculate some percentages. To smooth maps of mean hemoglobin, eGFRs, and creatinine levels, the model is extended to assume that the means have a normal distribution.

Special studies & data collection forms

The USRDS website includes complete copies of the CMS Medical Evidence (2728) and Death Notification forms (2746); the OPTN Transplant Candidate Registration form, Kidney Transplant Recipient Registration form and Kidney Transplant Recipient Follow-up form; and forms used for data collection in USRDS Special Studies.

Esrd networks  Top

Network 1

Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont
ESRD Network of New England, Inc.
30 Hazel Terrace
Woodbridge, CT 06525
phone 203.387.9332
fax 203.389.9902

Network 5

Virginia, West Virginia,
Maryland, District of Columbia
Mid-Atlantic Renal Coalition
1527 Huguenot Road
Midlothian, VA 23113
phone 804.794.3757
fax 804.794.3793

Network 9/10

Illinois, Indiana, Kentucky, Ohio
The Renal Network, Inc.
911 East 86th Street, Suite 202
Indianapolis, IN 46240
phone 317.255.8265
fax 317.255.8291

Network 14

Texas
ESRD Network of Texas, Inc.
4040 McEwen, Suite 350
Dallas, TX 75244
phone 972.503.3215
fax 972.503.3219

Network 18

Southern California
Southern California Renal
Disease Council, Inc.
6255 Sunset Boulevard, Suite 2211
Los Angeles, CA 90028
phone 323.962.2020
fax 323.962.2891

Esrd networks

Network 2

New York
CKD Network for New York
1979 Marcus Avenue, Suite 105
Lake Success, NY 11042-1002
phone 516.209.5578
fax 516.326.8929

Network 6

Georgia, North Carolina,

South Carolina

Southeastern Kidney Council, Inc.

1000 St. Albans Drive, Suite 270

Raleigh, NC 27609
phone 919.855.0882
fax 919.855.0753

Network 11

Minnesota, Michigan, North Dakota, South Dakota, Wisconsin
Renal Network of the Upper Midwest, Inc.
1360 Energy Park Drive, Suite 200
St. Paul, MN 55108
phone 651.644.9877
fax 651.644.9853

Network 15

Arizona, Colorado, Nevada,
New Mexico, Utah, Wyoming
Intermountain ESRD Network, Inc.
165 S. Union Boulevard, Suite 466
Lakewood, CO 80228
phone 303.831.8818
fax 303.860.8392

Network 1 Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont

Network 2 New York

Network 3

New Jersey, Puerto Rico, Virgin Islands
Cranbury Gates Office Park
109 S. Main Street, Suite 21
Cranbury, NJ 08512–3174
phone 609.490.0310
fax 609.490.0835

Network 4

Delaware, Pennsylvania
ESRD Network 4, Inc.
40 24th Street, Suite 410
Pittsburgh, PA 15222
phone 412.325.2250
fax 412.325.1811

Network 5 Maryland, Virginia, Washington D.C., West Virginia

Network 6 Georgia, North Carolina, South Carolina

Network 7

Florida
FMQAI: The Florida ESRD Network
5201 W Kennedy Blvd, Suite 900
Tampa, FL 33609
phone 813.383.1530
fax 813.354.1514

Network 8

Alabama, Mississippi, Tennessee
1755 Lelia Drive, Suite 400
Jackson, MS 39216
phone 601.936.9260
fax 601.932.4446

Network 9 Indiana, Kentucky, Ohio

Network 10 Illinois

Network 11 Michigan, Minnesota, North Dakota, South Dakota, Wisconsin

Network 12

Iowa, Kansas, Missouri, Nebraska
ESRD Network #12
7306 NW Tiffany Springs Pkwy
Suite 230
Kansas City, MO 64153
phone 816.880.9990
fax 816.880.9088

Network 13

Arkansas, Louisiana, Oklahoma
ESRD Network Organization #13
4200 Perimeter Center Drive, Suite 102
Oklahoma City, OK 73112-2314
phone 405.942.6000
fax 405.942.6884

Network 14 Texas

Network 15 Arizona, Colorado, Nevada, New Mexico, Utah, Wyoming

Network 16

Alaska, Idaho, Montana, Oregon, Washington
Northwest Renal Network
4702 42nd Avenue, SW
Seattle, WA 98116
phone 206.923.0714
fax 206.923.0716

Network 17

American Samoa, Guam, Mariana Islands, Hawaii, Northern California
Western Pacific Renal Network, LLC
505 San Marin Drive, Bldg A, Suite 300
Novato, CA 94945
phone .415.897.2400
fax 415.897.2422

Network 18 Southern California

CPT codes for vascular access & peritoneal dialysis access services

Complication 34101*, 35190*, 35321*, 35458*, 35460*, 35475*, 35476*, 35484*, 35875*, 35876*, 35900*, 35903*, 35910*, 36005*, 36145, 36534*, 36535*, 36550*, 36575*, 36580*, 36581*, 36584*, 36589*,36593, 36596*, 36597*, 36815, 36831, 36832, 36833, 36834*, 36838, 36860, 36861, 36870, 37190*, 37201*, 37205*, 37206*, 37207*, 37208*, 37607, 49422, 75790, 75820*, 75860*, 75896*, 75960*, 75962*, 75978*, 75998*, 76937*, 77001, 00532*, 01784*, 01844*, 90939, 90940, G0159, G0392, G0393, and M0900

Hemodialysis catheter placement 36011*, 36488*, 36489*, 36490*, 36491*, 36533*, 36555*, 36556*, 36557*, 36558*, 36565*, and 36800

Peritoneal dialysis catheter placement 49419, 49420, and 49421

Synthetic graft placement 36830

Fistula placement 36818, 36819, 36820, 36821, and 36825

Other placement 36810, 36835

*Requires accompanying renal diagnosis code for inclusion.

This list is comprehensive and includes codes that are now obsolete, but that were in use at some point during the study period (1991–2008).

DRG codesa: prior to October 1, 2007

112 Percutaneous cardiovascular procedure 120 Other circulatory system OR procedure 315 Other kidney and urinary tract OR procedure 442 Other OR procedure for injuries with complication 443 Other OR procedure for injuries without complication 478 Other vascular procedure with complication 479 Other vascular procedure without complication

DRG codesa: after September 30, 2007

252 Other vascular procedures with Major complicating conditions (MCC) 264 Other circulatory system O.R. procedures 673 Other kidney & urinary tract procedures with MCC 674 Other kidney & urinary tract procedures with CC 675 Other kidney & urinary tract procedures without CC/MCC 907 Other O.R. procedures for injuries with MCC 908 Other O.R. procedures for injuries with CC 909 Other O.R. procedures for injuries without CC/Medicare

ICD-9-CM procedure codes (a)

38.95 Venous catheterization for renal dialysis 39.27 Arteriovenostomy for renal dialysis 39.42 Revision of arteriovenous shunt for renal dialysis 39.43 Removal of arteriovenous shunt for renal dialysis 39.93 Placement of vessel-to-vessel cannula 39.94 Replacement of vessel-to-vessel cannula 86.07 Placement of totally implantable vascular access device

ICD-9-CM diagnosis codes (b)

996.1 Mechanical complication of vascular device, implant, graft 996.56 Mechanical complication due to peritoneal dialysis catheter 996.62 Infectious complication of vascular device, implant, graft 996.68 Infectious complication due to peritoneal dialysis catheter 996.73 Other complication due to renal dialysis device, implant, graft V56.1 Fitting and adjustment of extracorporeal dialysis catheter V56.2 Fitting and adjustment of peritoneal dialysis catheter

a DRG and procedure codes are used in conjunction to define inpatient pure vascular access events (both must be present)

b the presence of any of these diagnosis codes as the "Principal Diagnosis Code" is sufficient to define an inpatient pure vascular access or peritoneal dialysis access event

DRG & ICD-9-CM codes for vascular access & peritoneal dialysis access services

Medicare categories of payment & basis for categorizing claim

Total Sum of all payments

Total inpatient Sum of all payments originating from the inpatient SAF, including pass-throughs

Medical DRG Inpatient SAF, DRG

Surgical DRG Inpatient SAF, DRG

Transplant DRG Inpatient SAF, DRG 302 & 512

Other DRG Inpatient SAF, DRG not included in the above categories

Non-transplant pass-throughs Inpatient SAF, DRG not 302 or 512, calculated from per diem and covered days

Transplant pass-throughs Inpatient SAF, DRG 302, calculated from per diem and covered days

Total outpatient Sum of all payments originating from the Outpatient SAF

Outpatient hemodialysis Outpatient SAF, hemodialysis revenue codes

Outpatient peritoneal dialysis Outpatient SAF, peritoneal dialysis revenue codes

Outpatient other dialysis Outpatient SAF, dialysis revenue codes other than HD or PD

Outpatient ESA Outpatient SAF, revenue codes and/or HCPCS code

Outpatient vitamin D hormones Outpatient SAF, revenue and HCPCS codes

Outpatient iron Outpatient SAF, revenue and HCPCS codes

Outpatient other injectables Outpatient SAF, revenue and HCPCS codes

Radiology Outpatient SAF, revenue and/or CPT codes

Pharmacy Outpatient SAF, revenue codes

Ambulance Outpatient SAF, revenue codes

Laboratory/pathology Outpatient SAF, revenue and/or CPT codes

Outpatient other Outpatient SAF, does not qualify for any other cost category

Skilled nursing facility Skilled nursing facility SAF

Home health agency Home health SAF

Hospice Hospice SAF

Total physician/supplier Sum of physician/supplier payments

Transplant surgery Physician/supplier SAF, CPT codes

Inpatient surgery Physician/supplier SAF, CPT, and place of service codes

Outpatient surgery Physician/supplier SAF, CPT and place of service codes

E&M nephrologist inpatient Physician/supplier SAF, CPT, place of service and specialty codes

E&M nephrologist outpatient Physician/supplier SAF, CPT, place of service and specialty codes

E&M non-nephrologist inpatient Physician/supplier SAF, CPT, place of service and specialty codes

E&M non-nephrologist outpatient Physician/supplier SAF, CPT, place of service and specialty code

Dialysis capitation Physician/supplier SAF, CPT and/or type of service codes

Inpatient dialysis Physician/supplier SAF, CPT codes

Home dialysis Physician/supplier SAF, HCPCS and place of service codes

Vascular access Physician/supplier SAF, CPT codes

Peritoneal access Physician/supplier SAF, CPT codes

Physician/supplier ESA Physician/supplier SAF, HCPCS codes

Physician/supplier iron Physician/supplier SAF, HCPCS codes

Immunosuppressive drugs Physician/supplier SAF, HCPCS codes

Durable medical equipment Physician/supplier SAF, HCPCS codes

Physician/supplier radiology Physician/supplier SAF, CPT and specialty codes

Physician/supplier lab/pathology Physician/supplier SAF, CPT codes

Physician/supplier ambulance Physician/supplier SAF, HCPCS and place of service codes

Other physician/supplier Physician/supplier SAF, does not qualify for any other category

E&M: Evaluation and management

Bibliography  Top

Analytic and reporting guidelines: the Third National Health and Nutrition Examination Survey, NHANES III (1988–1994). http://www.cdc.gov/nchs/data/nhanes/nhanes3/nh3gui.pdf. Accessed October, 2002.

Arias E. United States life tables, 200. National vital statistics reports 57 (14). Hyattsville, Maryland: National Center for Health Statistics, 2009.

Arneson TJ, Liu J, Qiu Y, Gilbertson DT, Foley RN, Collins AJ. Hospital treatment for fluid overload in the Medicare hemodialysis population. Clin J Am Soc Nephrol 5: 1054–1063, 2010.

Centers for Disease Control and Prevention. NHANES 1999–2000 data files, available at the National Center for Health Statistics,
http://www.cdc.gov/nchs/about/major/nhanes/nhanes99_00.htm.

Centers for Disease Control and Prevention. NHANES 2001–2002 Data Files, available at National Center for Health Statistics,
http://www.cdc.gov/nchs/about/major/nhanes/nhanes01-02.htm.

Centers for Disease Control and Prevention (CDC). State-specific incidence of diabetes among adults — participating states, 1995–1997 and 2005–2007. MMWR Morb Mortal Wkly Rep. 2008 Oct 31; 57(43): 1169–73.

Centers for Disease Control and Prevention. The third National Health and Nutrition Examination Survey (NHANES III 1988–94) reference manuals and reports [CD-ROM]. Bethesda MD: National Center for Health Statistics, 2004.

Chan KE, Lazarus JM, Thadhani R, Hakim RM. Warfarin use associates with increased risk for stroke in hemodialysis patients with atrial fibrillation. J Am Soc Nephrol. 2009 Oct;20(10):2223-33. Epub 2009 Aug 27.

Cohen LM, Ruthazer R, Moss AH, Germain MJ. Predicting six-month mortality for patients who are on maintenance hemodialysis. Clin J Am Soc Nephrol. 2010;5(1):72–79.

Foley RN, Murray AM, Li S, Herzog XA, McBean AM, Eggers PW, and Collins AJ. Chronic kidney disease and the risk for cardiovascular disease, renal replacement, and death in the United States Medicare population, 1998 to 1999. J Am Soc Nephrol, 2005; 16:489–95.

Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med. 2004 Sep 23; 351(13): 1296–305.

Herzog CA, Li S, Weinhandl ED, Strief JW, Collins AJ, Gilbertson DT. Survival of dialysis patients after cardiac arrest and the impact of implantable cardioverter defibrillators. Kidney Int 2005 Aug; 68(2): 818–25.

Kalbfleisch JD, Prentice RL. The statistical analysis of failure time data. New York: Wiley, 1980.

Kaplan EL, Meier P. Nonparametric estimation from incomplete observation. J Amer Stat Assoc 1958; 3: 457–481.

Kurella M, Covinsky KE, Collins AJ, Chertow GM. Octogenarians and nonagenarians starting dialysis in the United States. Ann Intern Med. 2007 Feb 6;146(3): 177–83.

Kurella Tamura M, Covinsky KE, Chertow GM, Yaffe K, Landefeld CS, McCulloch CE. Functional status of elderly adults before and after initiation of dialysis. N Engl J Med. 2009 Oct 15;361(16): 1539–47.

Levey AS, Berg RL, Gassman JJ, Hall PM, Walker WG. Creatinine filtration, secretion and excretion during progressive renal disease. Modification of Diet in Renal Disease (MDRD) Study Group. Kidney Int Suppl 1989 Nov 27: S73–80.

Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999; 130 (6): 461–70.

Levey AS, Coresh J, Greene T, Stevens LA, Zhang YL, Hendriksen S, Kusek JW, Van Lente F; Chronic Kidney Disease Epidemiology Collaboration. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med 2006 Aug 15; 145(4): 247–54.

Liu J, Louis TA, Pan W, Ma JZ, Collins AJ. Methods for estimating and interpreting provider-specific standardized mortality ratios. J Health Serv and Outcomes Res Methodology Sept 2003; 4 (3): 135–149.

Liu J, Louis TA, Pan W, Ma JZ, Collins AJ. State-level adjusted ESRD incident rates: use of observed versus model-predicted category-specific rates. Kidney Int. 2006 Apr; 69(8): 1459–63.

Marinovich S, Lavorato C, Celia E, Bisignano L, Soratti M, Hansen Krogh D, Tagliafichi V, Moriρigo C, Rosa Diez G and Fernandez V. Registro Argentino de Diαlisis Crσnica 2008. Sociedad Argentina de Nefrologνa (SAN) and Instituto Nacional Central Unico Coordinador de Ablaciσn e Implante (INCUCAI). Buenos Aires, Argentina. 2010.

Mungoile M, Pickle LW, Simonson KH. Application of a weighted head-banging algorithm to mortality data maps. Statistics in Medicine 1999; 18, 3201–3209.

National Committee for Quality Assurance. HEDIS 2008 technical specifications. Washington

DC: National Committee for Quality Assurance; 2007.

National Kidney Foundation K/DOQI Clinical Practice Guidelines for Cardiovascular Disease in Dialysis Patients. AJKD 2005; 45: S1–S154 (Suppl 3).

NHANES Analytic Guidelines.
http://www.cdc.gov/nchs/data/nhanes/nhanes_general_guidelines_june_04.pdf.

Pickle LW, Mungiole M, Jones GK, White AA. Atlas of United States Mortality. Hyattsville (MD): National Center for Health Statistics; 1996.

Pickle LW, White AA. Effects of the choice of age-adjustment method on maps of death rates. Statistics in Med 1995; 14, 615–627.

Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: Computing risks and multi-state model. Statistics in Medicine, 26 (2007), 2389–2430.

Rao PS, Schaubel DE, Guidinger MK et al. A Comprehensive Risk Quantification Score for Deceased Donor Kidneys: The Kidney Donor Risk Index. Transplantation 2009;88(2):231–236.

Research Triangle Institute (2004). SUDAAN language manual, Release 9.0 Research Triangle park. NC: Research Triangle Institute.

SAS Institute Inc. SAS system for mixed models, Cary (NC): SAS Institute Inc.; 1996.

SAS Institute Inc. SAS/ETS User's Guide. Cary (NC): SAS Institute Inc.; 1993.

Schwartz GJ, et al. A simple estimate of glomerular filtration rate in children derived from body length and plasma creatinine. Pediatrics. 1976 Aug; 58(2): 259–63

Singh AK, Szczech L, Tang KL, Barnhart H, Sapp S, Wolfson M, Reddan D; CHOIR Investigators. Correction of anemia with epoetin alfa in chronic kidney disease. N Engl J Med 2006, Nov 16; 355 (20): 2085–98.

Stevens LA, Coresh J, Schmid CH, Feldman HI, Froissart M, Kusek J, Rossert J, Van LF, Bruce RD, III, Zhang YL, Greene T, Levey AS: Estimating GFR using serum cystatin C alone and in combination with serum creatinine: a pooled analysis of 3,418 individuals with CKD. Am J Kidney Dis 51: 395–406, 2008.

Tonelli M, Wiebe N, Culleton B, House A, Rabbat C, Fok M, McAlister F, Garg AX. Chronic kidney disease and mortality risk: a systematic review. J Am Soc Nephrol. 2006 Jul; 17 (7): 2034–47.

Waller LA, Carlin BP, Xia H, Gelfand AE. Hierarchical spatio-temporal mapping of disease rates. J Amer Stat Assoc 1997; 92, 607–617.