<

Précis

introduction · 16

<

trends in the u.s. esrd program · 18

<

patient care in the pre-esrd period · 20

<

chronic kidney disease in the general population · 22

<

hospitalization & death by esrd modality · 24

<

esrd care delivery systems · 26

<

chapter summary · 28

WWe used to think that if we knew one, we knew two,

because one and one are two. We are finding that we must

learn a great deal about ‘and.’

ARTHUR EDDINGTON, in Mackay, The Harvest of a Quiet Eye

back¿round on the u.s. esrd program

Préc

is

16

introdu

ctionT

p.1 · Incidentpatient counts, by

modality & data sourceUSRDS incident counts contain all

incident patients identified by the CMS ESRDprogram & the UNOS transplant data. The patient

modality is determined by using one point at initiation, without application of the60-day stable modality rule. CMS-FSD (Facility Survey Data) incident countscontain only first-time dialysis patients started at homes & in-center units. Theydo not include incident transplant patients.

This 2002 Annual Data Report documents the activities ofthe ESRD program in the United States since its legislativeauthorization in 1972, when the ESRD Act gave Medicareentitlement to patients younger than 65 who developESRD and require dialysis or transplantation. More than1,250,000 individuals have been treated since the mid-1970s, including more than 375,000 patients currently inthe program—almost 100,000 of them new to treatmentin 2000. Medicare spending for the ESRD program is ap-proaching $14 billion, while non-Medicare costs are closeto $5.5 billion.

In this Précis we provide information on the size andbreadth of the ESRD program, using the different datasources available from the Centers for Medicare and Med-icaid Services (CMS, formerly HCFA), the ESRD networks,and the USRDS. ESRD patient data are documented in sev-eral ways: on the CMS Medical Evidence form (2728) at theinitiation of treatment, on the census of prevalent patientsreported to the ESRD networks, in the CMS Annual FacilitySurvey (AFS) of dialysis and transplant centers, in trans-plantation information from the United Network for Or-gan Sharing (UNOS), and in Medicare ESRD patient claims.We have spent considerable time this year reconcilingthese different data sources so as to provide the most con-sistent findings possible.

In the first section we present summary statistics on pa-tients in 2000, including incident and prevalent counts andrates. We compare patient populations for the 1992–1996 and 1996–2000 periods, and provide total ESRDcosts for both the Medicare and non-Medicare popu-lations.

The next spread addresses both the care of patientsand their advancing comorbidity as they approachESRD therapy. There has been little information pub-lished on the fate of patients with chronic kidney dis-ease, particularly their likelihood of survival in thefollow-up years and of advancement to ESRD. The Na-tional Kidney Foundation recently estimated that 25million Americans have some degree of chronic kid-ney disease, yet the number of patients beginningESRD treatment each year is approximately 100,000,or 0.4 percent of the population at risk. As a startingpoint for investigations of what happens to individu-

als with chronic kidney disease, we show that these pa-tients are far more likely to die than to be diagnosed withend-stage renal disease, an important reality highlightingthe complexity of these patients.

Following this discussion of chronic kidney disease and theincidence of ESRD we examine trends in rates of hospital-ization and mortality after the initiation of ESRD treatment.We show, in part, that mortality rates for patients of greaterdialysis vintage (time on the therapy) are rising, while sur-vival for patients who have been on dialysis less than threeyears is beginning to improve. These findings are discussedfurther in Chapter Nine.

In the final spread we provide information on recent changesin the renal provider system, focusing on profit status andchain affiliation. Growth in the number of chain-affiliatedunits over the past ten years has been dramatic; in later chap-ters we investigate the changes in costs and clinical param-eters of care associated with this growth.

In the past, methods used to report incident and preva-lent populations, both overall and by modality, havevaried by data source and by the govern-ment organizations providing data.This year, however, we havecollaborated with

17

p.2 · Prevalentpatient counts, by modality

& data sourceUSRDS prevalent counts contain all

prevalent dialysis patients identified by the CMSESRD program. Patient modality is determined by using one

point on December 31 with no 60-day stable modality rule. Prevalent counts fromthe Facility Survey data contain only point prevalent dialysis patients.

p.3 · Transplantcounts, by data source

USRDS counts include CMS &UNOS transplants. These counts represent

the total transplants performed during each calendaryear. Some patients may have more than one transplant in a

year. Facility Survey counts include transplants reported by each certifiedtransplant center.

CMS and the ESRD networks to reconcile the differ-ences in these methods and to therefore provide amore consistent view of the ESRD program.

In Figure p.1 we present information on the incidentpopulation as tracked by the CMS Annual Facility Sur-vey (AFS) and by the USRDS, using the Medical Evi-dence form. Because the USRDS also reviews dialysisclaims within the Medicare system, we count more in-cident dialysis patients than the Facility Survey. For2000, Facility Survey data indicates a total of 91,679incident dialysis patients, compared to 94,022 incidentdialysis patients in the USRDS database. By treatmentmodality, the USRDS database reported 83,635 hemo-dialysis patients, 7,101 peritoneal dialysis patients, and2,170 pre-emptive transplant patients in 2000.

Figure p.2 shows a similar comparison for the preva-lent population. Because we obtain information fromMedicare claims, the USRDS has consistently countedmore dialysis patients than the AFS, particularly be-tween 1994 and 1998. And because claims are notavailable on Medicare HMO and non-Medicare pa-tients, it is more difficult to determine modality forprevalent patients than for those just beginning treat-ment. The AFS data, which include direct census re-ports from the dialysis units, therefore report a highernumber of peritoneal dialysis patients than does theUSRDS.

To report the transplant population, the USRDS adoptsmethods similar to those used in the AFS, and usesinformation from the UNOS registry. Results from thetwo sources are, then, quite close (Figure p.3). TheUSRDS reports that 14,427 renal transplants were per-formed in 2000, while the AFS indicates 14,311.

We have adopted new methods to account for the in-creasing number of patients with Employer HealthGroup Plan (EGHP) coverage. These patients must waitthirty months for Medicare to become their primarypayor, and thus have no Medicare claims during thisperiod. The USRDS previously defined lost-to-followuppatients as those with no claims in a 24-month pe-riod. But because of the increasing number of EGHPpatients falling into this category, we have this yearextended the lost-to-followup period to 36 months.As a result, the number of existing patients now moreclosely matches that of the ESRD network census andthe AFS data.

18

Préc

istren

ds in th

e u.s. es

rd progra

m

D p.a · Summary statistics on reported ESRD therapy in the United States, 2000uring the year 2000, 96,192 new dialy-sis and transplant patients startedESRD treatment (Table p.a). Diabetes

was the primary cause of ESRD in 43.4 per-cent of these patients, while hypertensionand glomerulonephritis were primary diag-noses in 25.5 and 8.4 percent, respectively.These three diagnoses thus accounted for77.4 percent of all new ESRD patients. Theoverall incident rate in 2000, adjusted forage, gender, and race, was 334 new patientsper million population.

The number of patients under ESRD treat-ment on December 31, 2000 was 378,862,including 275,053 dialysis patients and103,809 patients with a functioning trans-plant; this created a prevalent rate of 1,311patients per million population. As re-ported in the CMS Annual Facility Survey(AFS), 14,311 transplants were performedduring 2000. Nineteen percent of theprevalent ESRD population died during2000, a total of 72,342 patients.

The average annual percent change in theprevalent patient census during 1992–1996and 1996–2000 shows that the annualgrowth in the hemodialysis population de-creased from 7.3 to 5.1 percent. The perito-neal dialysis population, which grew anaverage of 3.7 percent per year during theearlier period, declined 6.6 percent per yearbetween 1996 and 2000. And the numberof patients receiving a transplant at the be-ginning of ESRD treatment increased 6.2percent per year from 1992 to 1996. butonly 4.8 percent per year for the 1996–2000period.

Medicare costs for the ESRD program in2000 were $13.82 billion, while non-Medi-care spending accounted for $5.53 billion.Expenditures for the total ESRD programthus totaled $19.35 billion, an increase of6.2 percent between 1999 and 2000. TheHMO Medicare risk population accountsfor approximately one billion dollars ofthe Medicare spending for ESRD.

On a per patient per year basis, the costsfor each ESRD patient increased 2.6 per-cent between 1999 and 2000. After adjust-ments for inflation (using the Bureau ofLabor Statistics inflationary adjustment orthe CMS inflation adjustment for the medi-cal component), however, actual costs ofthe ESRD program per patient per year de-clined between 0.8 and 1.5 percent.

As documented by both the USRDS andthe CMS Annual Facility Survey, the annualpercent change in ESRD patient popula-tions has been decreasing slowly (Figurep.4). Between 1999 and 2000, for example,the USRDS estimates that incident patientcounts grew 5.2 percent, and the AFS datadocuments a 3.2 percent increase. In theprevalent hemodialysis population, theUSRDS shows an increase of 5.3 percent,

A Incident counts: include all known ESRD patients, regardless of any incomplete data on patient characteris-tics and of U.S. residency status.

B Rates are adjusted for age, race, and/or gender using the estimated July 1, 2000 U.S. resident population asthe standard population. All rates are per million population. Rates by age are adjusted for race and gender.Rates by gender are adjusted for race and age. Rates by race are adjusted for age and gender. Rates by dis-ease group and total adjusted rates are adjusted for age, gender, and race. Adjusted rates do not includepatients with other or unknown race. Includes only residents of the 50 states and Washington D.C.

C Patients are classified as receiving dialysis or having a functioning transplant. Those whose treatment mo-dality on December 31 is unknown are assumed to be receiving dialysis. Includes all Medicare and non-Medicare ESRD patients, and patients in the Territories and foreign countries.

D Age is computed at the start of therapy for incidence, on December 31 for point prevalence, at the time oftransplant for transplants, and on the date of death for death.

E Unadjusted total rates include all ESRD patients in the 50 states and Washington D.C.

F From the 2000 CMS Facility Survey.

G Deaths are not counted for patients whose age is unknown.

H Includes patients whose modality is unknown.

Medicare spendingMedicare spending for ESRD in 2000

(billions of dollars)SAF paid claims (Part A & B) 12.37

2% incurred but not reported 0.25

HMO-Medicare risk 1.00

Organ acquisition 0.21

Total Medicare costs 13.82

Non-Medicare spending for ESRD(billions of dollars)

EGHP (MSP) 1.29

Patient obligations 3.06

Non-Medicare patients 1.18

Total non-Medicare costs 5.53Change in Medicare spending (%)from 1999 to 2000

Total 6.2

Per patient year 2.6

Adjusted for inflation -0.8% to -1.5%Medicare spending per patient yearfrom 1996 to 2000

ESRD 46,045

Hemodialysis 54,917

Peritoneal dialysis 46,121

Transplant 17,227

Average annual percent change in rates per millionHD PD Transplant

92-96 96-00 92-96 96-00 92-96 96-00Incident patientsWhite 5.70 7.69 4.57 -3.28 -4.17 11.05Black 6.47 2.99 2.94 -7.64 9.44 -12.21N Am 8.20 -4.07 12.84 -11.55 -3.15 0.68Asian 11.30 1.16 17.52 -10.54 -16.19 11.67DM 10.36 7.15 7.62 -5.04 4.97 -4.50HTN 1.70 5.73 -1.88 -1.78 -7.40 10.30GN 4.27 -0.36 6.08 -7.44 2.85 3.77CK 0.52 2.83 4.35 -3.96 6.21 6.48All 6.20 5.62 4.76 -4.70 -2.43 7.07

Prevalent patientsWhite 6.99 6.51 2.74 -6.04 5.95 4.67Black 7.40 3.55 4.28 -8.08 7.34 5.31N Am 10.95 2.39 4.25 -8.25 7.86 4.57Asian 9.50 3.27 15.75 -5.91 6.37 4.93DM 11.96 8.35 5.98 -5.62 8.64 6.00HTN 5.43 3.96 1.33 -6.65 6.94 4.00GN 5.16 2.52 4.01 -6.89 4.78 4.27CK 2.59 2.20 1.93 -7.66 6.83 5.81All 7.27 5.06 3.70 -6.58 6.22 4.77

Medicare Incidence (A) December 31 Point Prevalence kidney transplants

Patient Adj. Adj. Living ESRDcharacteristics Count Rate (B) Count (C)Rate (B) Dialysis (C) Tx (C) Cadaver donor deaths (G)Age (D)

0-19 1,386 16 6,566 75 2,256 4,310 226 278 13420-44 14,067 125 86,237 779 46,390 39,847 3,008 1,747 4,85145-64 34,223 629 154,961 2,755 106,265 48,696 3,712 1,639 20,64865-74 23,744 1,384 75,184 4,491 65,541 9,643 773 304 21,18175+ 22,768 1,543 55,849 3,842 54,578 1,271 46 15 25,528Unknown 4 65 23 42

White 61,548 254 227,627 943 150,913 76,714 5,025 3,108 46,763Black 27,540 996 121,945 4,241 103,221 18,724 2,157 624 20,611Native American 1,222 716 5,861 3,288 4,605 1,256 120 32 964Asian/Pacific Islander 3,284 393 15,121 1,623 10,699 4,422 345 148 2,000Other/unknown 2,598 8,308 5,615 2,693 118 71 2,004Male 51,573 403 207,516 1,569 145,879 61,637 4,739 2,274 37,714Female 44,611 282 171,307 1,098 129,145 42,162 3,026 1,709 34,515Unknown gender 8 39 29 10 113Primary diagnosis

Diabetes 41,772 145 131,173 456 110,041 21,132 2,266 886 31,153Hypertension 24,566 87 86,739 304 73,631 13,108 1,376 566 19,475Glomerulonephritis 8,102 28 59,056 205 33,182 25,874 1,835 1,040 5,641Cystic kidney disease 2,144 7 16,298 57 7,928 8,370 624 292 1,253Urologic disease 1,630 6 7,387 26 5,202 2,185 145 128 1,223Other known cause 10,444 37 42,578 150 27,688 14,890 936 694 7,407Unknown cause 3,841 13 14,631 51 10,248 4,383 279 169 2,748Missing cause 3,693 11 21,000 62 7,133 13,867 304 208 3,442

All 96,192 334 378,862 1,311 275,053 (H) 103,809 7,765 3,983 72,342Unadjusted rate (E) 337 Total transplants (F)14,311

19

-5

0

5

10

15

20

USRDS

Facility Survey data

Perc

ent

chan

ge

0

2

4

6

8

10

12

-8

-4

0

4

8

12

16

91-9292-93

93-9494-95

95-9696-97

97-9898-99

99-000

2

4

6

8

10

Incident patients: All

Prevalent patients: Hemodialysis

Prevalent patients: Peritoneal dialysis

Prevalent patients: Transplant

1990 1992 1994 1996 1998 2000

Nu

mb

er o

f pat

ien

ts (t

ho

usa

nd

s)

0

1

2

3

4

20

40

60

80

Death

Recovered function

Discontinued dialysis

1990 1992 1994 1996 1998

Cu

mu

lati

ve n

um

ber

of p

atie

nts

(th

ousa

nd

s)

0

10

20

30

40

50

USRDS: previous methods

USRDS: new method

1996 1997 1998 1999 2000

Do

llars

(bill

ion

s)

0

2

4

6

8

10

12

14

Medicare

Non-Medicare

1990 1992 1994 1996 1998 2000

Nu

mb

er o

f pat

ien

ts (t

hou

san

ds)

0

2

4

40

60

80

100

New patients

Patients returning from transplant

Total patients starting or restarting dialysis

p.8 · Trends in Medicare vs. non-Medicare spending

p.6 · Trends in patients ceasing dialysis

Patient losses

p.4 · Annual percent changes in patient counts p.5 · Trends in counts of new dialysis patients

p.7 · Trends in patients lost-to-followup

and the AFS data 6.1 percent. The preva-lent peritoneal dialysis population, in con-trast, decreased 3.8 percent by USRDSmethods and 1.4 percent in the AFS report.Both data sources report a comparable in-crease of 6.0–6.1 percent in the numberof prevalent transplant patients.

The number of incident and prevalent pa-tients does not account for all patientswho receive dialysis therapy. In 2000, forexample, 4,166 patients returned to dialy-sis after a failed renal transplant (Figurep.5). Other patients stop dialysis therapy;the AFS reports that 3,754 patients recov-ered renal function in 2000, while 2,157chose to discontinue dialysis (Figure p.6).

New methods used by the USRDS now al-low us to track 15,221 patients who werepreviously classified as lost-to-followup(Figure p.7).

Medicare expenditures for the ESRD pro-gram rose 26.6 percent between 1996 and2000, while non-Medicare costs for theprogram grew 27.9 percent—an overallgrowth 27.6 percent (Figure p.8).

Figure p.4 incident ESRD patients, from CMS’sannual End-Stage Renal Disease Facility Survey &the USRDS. Figures p.5–6 aggregate patient countsfrom the CMS Facility Survey. Figure p.7 pointprevalent ESRD patients, using a new three-yeargrace period instead of the previous two-year periodfrom the time of ESRD initiation (see Appendix Afor details). Figure p.8 Medicare spending includespaid claims, estimated HMO-risk expenditures, &estimated organ acquisition costs. Paid claims in2000 are inflated for costs incurred but not reported.Non-Medicare spending includes EGHP, estimatedpatient obligations (co-insurance & deductibles), &estimated ESRD costs for non-Medicare patients(seven percent of Medicare costs).

20

Préc

ispatien

t care i

n the pre-

esrd perio

d

I

1995 1996 1997 1998 1999N

um

ber

of p

atie

nts

(th

ous

and

s)

28

32

36

40

44

1995 1996 1997 1998 1999

Perc

ent

of p

atie

nts

10

12

14

16

18

20

22

24

26

28

Medical Evidence form

Claims

0

5

10

15

20

25

1995 1996 1997 1998 1999

All patients

Cu

mu

lati

ve p

erce

nt o

f pat

ien

ts

0

5

10

15

20

25

Months (1st day of month 1=start of ESRD)-24 -18 -12 -6 0

0

5

10

15

20

25

30

Hemodialysis

Peritoneal dialysis

Hem

og

lob

in (g

/dl)

9

10

11

Months (1st day of month 1=start of ESRD)

-24 -18 -12 -6 1 6

EPO

do

se p

er w

eek

(th

ousa

nd

s o

f un

its)

6

8

10

12

14

16

18

Hemoglobin levels

EPO dosing

EPO pre-ESRD

No EPO pre-ESRD

EPO pre-ESRD

No EPO pre-ESRD

1995 1996 1997 1998 1999

Perc

ent

of p

atie

nts

0

20

40

60

80

100

Non-Medicare

Medicare secondary payor

HMO

Fee-for-service

p.9 · Trends in patient counts

Counts of patients 67 & older Pre-ESRD EPO use in patients 67 & older

p.12 · Trends in pre-ESRD EPO use, by data source

p.13 · Patients receiving EPO in the two years pre-ESRD

p.10 · Percent of patients, by insurance type

p.11 · Hemoglobin & EPO use in the pre-ESRD period

nformation on the care of patients in thepre-ESRD period is difficult to obtain, sincemore than half of the patients who initiate

dialysis therapy do not have Medicare astheir primary payor prior to their diagnosisof ESRD. Patients already in the Medicaresystem, however, do have pre-ESRD claims,as well as Medical Evidence form informa-tion. To more carefully document pre-ESRDservices, we studied Medicare patients age67 and older with two years of claims be-fore the initiation of dialysis.

The number of older patients initiating di-alysis increased 34.2 percent between 1995and 1999 (Figure p.9). The percentage ofenrollees covered by Medicare risk HMOprograms nearly doubled, an increase off-set by a 9.8 percent increase in fee-for-ser-vice coverage (Figure p.10).

Mean hemoglobin levels in the two yearsprior to starting dialysis are relatively con-stant at 10.1 g/dl (Figure p.11). After initia-tion, hemoglobins increase from a low of9.9 g/dl to almost 11 g/dl over the first sixmonths. Average EPO doses range from7,000 to 7,600 units per week. Figure p.12shows an increasing trend in erythropoi-etin (EPO) treatment until 1999, when, ac-cording to information on the MedicalEvidence form, 27.9 percent of patients re-ceived treatment. Only 20.7 percent, how-ever, had treatment claims.

Figure p.13 shows the growth in the per-cent of patients receiving EPO throughservices billed to Medicare. A higher per-cent of peritoneal dialysis patients receiveEPO before starting dialysis.

Congestive heart failure and complicationsof infection are the leading causes of hos-pitalization in the pre-ESRD period—notsurprising, since these are major indicationsfor starting ESRD therapy (Figure p.14). Is-chemic heart disease and general renal fail-ure are comparable in the time before ESRD.As expected, vascular access hospitaliza-tions accelerate during the month of dialy-sis initiation.

Dialysis catheters were used most widely in1995; by 2000, however, there was a transi-tion to the use of permanent catheters (Fig-ures p.15–16). For approximately 40 percentof patients there was no evidence of an ac-cess being placed prior to the initiation ofdialysis. Geographically, as few as 34.7 per-cent of patients in some regions receive acatheter, compared to other areas with arate near 50 percent.

Seventy percent of diabetic pre-ESRD pa-tients receive only one HbA1c test, 53 per-cent receive two tests, and only 28 percentreceive four tests in the two years beforeinitiation (Figure p.17). The American Dia-betes Association recommends at least twotests per year. Similar low levels of monitor-

21

0

5

10

15

20

25

30

35

40

Permanent cathetersTemporary cathetersSynthetic graftsFistulas

1995

Months (1st day of month 1=start of ESRD)-12 -6 -1 1 6

Per

cen

t of

pat

ien

ts w

ith

in e

ach

mo

nth

0

2

4

6

8

10

12

14

16

18 2000

0

10

20

30

40

50

60

70Hemoglobin A1c

0

10

20

30

40

50

60

70 Diabetic eye exams

Months (0=beginning of ESRD)-24 -18 -12 -6 0

Cu

mul

ativ

e p

erce

nt

of p

atie

nts

0

10

20

30

40

50

60

70 Lipid testing

1 test

2 tests

3 tests

4+ tests

1 test

2+ tests

1 test

2+ tests

45.5+ (49.1) 42.3 to <45.5 39.8 to <42.3 37.5 to <39.8 below 37.5 (34.7)

11.03+ (13.86) 8.87 to <11.03 7.21 to <8.87 5.72 to <7.21 below 5.72 (4.67)

17.0+ (21.2) 14.1 to <17.0 12.1 to <14.1 10.0 to <12.1 below 10.0 (8.3)

Months (1st day of month 1=start of ESRD)-24 -18 -12 -6 -1 1 6

Cu

mu

lati

ve p

erce

nt o

f pat

ien

ts

0

10

20

30

40

50

60

InfectionVascular accessISHDCHFGeneral renal failure

p.14 · Primary causes of hospitalization in the pre-ESRD period, patients 67 & older

Pre-ESRD vascular access use in patients 67 & older

p.15 · Access placement prior to ESRD p.16 · Access placement at start of ESRD

p.17 · Pre-ESRD diabetic preventive care

Catheters

Fistulas

Grafts

ing occur for diabetic eye exams and lipidtesting. These data indicate that generalmedical care for CKD patients approachingESRD needs considerable improvement.

All figures data from the pre-ESRD CMS files,which contain information on patients 67 & older.

Figures p.9-10 incident ESRD patients. Figure p.11fee-for-service (FFS) patients with EPO claims; EPOdosing is dose per week. Figures p.11 & p.14 inci-dent patients, 1995–1999 combined; claims fromJanuary 1, 1993 to June 30, 2000. Figure p.12 FFSpatients with Medical Evidence forms (2728). Fig-ure p.13 FFS patients; hemodialysis & peritoneal di-alysis identified from the Medical Evidence form. Fig-ure p.14 FFS patients. Figure p.15 HMO patients& patients with Medicare as secondary payor are ex-cluded. Figure p.16 incident patients age 67 or olderat incidence, 1995–1999 combined, by HSA; non-Medicare, Medicare as secondary payor, & HMOpatients are excluded. Figure p.17 diabetic incidentpatients, 1999; HMO patients & patients with Medi-care as secondary payor during 1999 are excluded.

22

Préc

ischr

onic kid

ney dise

ase in t

he gener

al populatio

n

T

Patients without anemia

Patients with anemia

All patients

Percent of patients0 10 20 30 40 50 60 70 80 90 100

ESR

D p

tsin

fo

llow

up

pe

rio

d

Ge

n. M

ed

.p

ts d

uri

ng

en

try

pe

rio

d

ESR

D p

tsin

fo

llow

up

pe

rio

d

Ge

n. M

ed

.p

ts d

uri

ng

en

try

pe

rio

d

ESR

D p

tsin

fo

llow

up

pe

rio

d

Ge

n. M

ed

.p

ts d

uri

ng

en

try

pe

rio

d

0-2930-39

40-4950-59

60-6465-69

70-7475-89

80-8485+

Perc

ent

of p

atie

nts

0

1

2

3

4

5

6

96-9797-98

98-99

Perc

ent

of p

atie

nts

0

3

6

9

12

15

18

96-97 97-98 98-990

2

4

6

8

10Diabetes Chronic kidney disease

18.5+ (19.8) 17.6 to <18.5 16.5 to <17.6 15.0 to <16.5 below 15.0 (14.1)

4.05+ (4.35) 3.74 to <4.05 3.51 to <3.74 3.15 to <3.51 below 3.15 (2.94)

1996-1997

1997-1998

1998-1999

DM/CKD/CHF

DM/CKD

CHF/CKD

CKD only

DM/CHF

CHF only

DM only

None

Diabetic

Non-diabetic

p.18 · ESRD patients, by diagnosis & origins in the general Medicare population

p.19 · CKD in the general Medicare population, by age p.20 · Prevalence of diabetes & chronic kidney disease

p.21 · Geographic variations in prevalence of diabetes & chronic kidney disease

o further study individuals withchronic kidney disease, the USRDS ob-tained data on a five percent sample

of the general Medicare population, ex-cluding individuals with ESRD. From thesepatient claims we can study the distribu-tion of diabetes, chronic kidney disease,congestive heart failure, and anemia.These conditions are defined using meth-ods similar to those used by the NationalCommittee for Quality Assurance (NCQA)to define diabetes.

Figure p.18 shows that a minority of pa-tients—those who carry a diagnosis ofchronic kidney disease, diabetes, or con-gestive heart failure, in any combination—generate the vast majority of ESRD cases,and that this patient distribution is influ-enced by the presence of chronic anemia.This high-risk population merits thoroughattention to diagnosis and treatment.

Figure p.19 shows that the prevalence ofchronic kidney disease (CKD) diagnosesincreases with age, and that it grew be-tween 1996 and 1999. The abrupt changeat age 65 reflects the fact that younger en-rollees are eligible only due to disability,while older participants have no such re-strictions. Figure p.20 demonstrates the in-creasing prevalence of diabetes between1996 and 1999, and the faster growth ofCKD prevalence among diabetics com-pared to non-diabetics.

Figures p.22–23 show that death is a farmore likely outcome than ESRD for all pa-tients, especially those who are older andless healthy. In patients with no diagnosisof CKD the likelihood of death is, in fact,over 100 times greater than that of ESRD.And even in patients age 65 and older withboth CKD and diabetes, the likelihood ofdeath is still almost five times greater thanthat of ESRD.

Finally, Figures p.24–25 compare preva-lence information from two sources: thefive percent general Medicare sample andthe NHANES III study of patients age 65and older. The presence of CKD claims cor-responds to reasonably severe renal insuf-ficiency, as defined in the guidelines ofNational Kidney Foundation’s Dialysis Out-comes Quality Initiative (DOQI). The USRDSestimates that patients with claims docu-mentation of CKD have GFR levels at 32.1ml/min, or late stage three or early stage 4chronic kidney disease in the NKF classifi-cation.

Figure p.18 1997–1998 general Medicare patientscontinuously enrolled in Medicare Part A & Part Bduring 1997–1998 & alive on December 31, 1998.Patients enrolled in an HMO or diagnosed withESRD any time during the two-year period are ex-cluded. Each condition (chronic kidney disease, con-gestive heart failure, diabetes, & anemia) identifiedfrom diagnosis codes: one from Part A inpatient,

Patient complexity in the general Medicare population

Diabetes Chronic kidney disease

23

DM NDM DM NDM DM NDM DM NDM

Perc

ent o

f pat

ien

ts

0

5

10

15

20

25

30

35

Pro

bab

ility

0.00

0.05

0.10

0.15

Diabetes/CKDNon-diabetes/CKD Diabetes/Non-CKD Non-diabetes/Non-CKD

Probability of ESRD: <65 65+

Followup time (months)

0 2 4 6 8 10 12 14 16 18 20 22 240.0

0.1

0.2

0.3

0.4

0 2 4 6 8 10 12 14 16 18 20 22 24

Probability of death

Diabetes Congestiveheart failure

Chronickidney dis.

Per

cen

t of

pat

ien

ts

0

2

4

6

8

10

12

14

<65

65+

NHANES

General Medicare

Probability of ESRDCKD (40,250 patients) Non-CKD

(1,246,710 patients)

Probability of deathCKD Non-CKD

All 0-<15 30-<45 60-<7515-<30 45-<60 75-<90

Hem

og

lob

in (g

/dl)

10

11

12

13

14

15

GFR groups (NHANES)

0-<15 30-<45 60-<7515-<30 45-<60 75-<90

90+

Nu

mb

er o

f pat

ien

ts (m

illio

ns)

0

2

4

6

8

10 Patient counts Hemoglobin

p.22 · ESRD & death in the followup period

p.24 · Identified diseases, by data source p.25 · Glomerular filtration rate & hemoglobin levels (NHANES III data)

Development of ESRD & mortality in the general Medicare population

p.23 · Life table estimates for the probability of developing ESRD & of death

skilled nursing, or home health claims; two from PartA outpatient claims; &/or two from Part B claims.Because of inconsistencies in anemia coding, all codesfrom 280 to 285 are included. Figures p.19–20 di-agnosis of CKD during the two-year entry period;general Medicare patients who, during any two con-secutive calendar years 1996–1999, were continu-ously enrolled in Medicare Part A & Part B. Patientsenrolled in an HMO or diagnosed with ESRD anytime during the two-year entry period are excluded.Age calculated on the first day of the entry period.Figure p.21 percent of patients; general Medicarepatients age 65 or older, 1998–1999 combined, whowere continually enrolled in Medicare Part A & PartB & alive on January 1, 1998. Patients enrolled inan HMO or diagnosed with ESRD any time duringthe study period are excluded. Adjusted for age, gen-der, & race. Figures p.22–23 general Medicare pa-tients, 1996–1997 combined, continuously enrolledin Medicare Part A & Part B & alive on December31, 1997. Patients enrolled in an HMO or diagnosedwith ESRD are excluded. Figures p.24–25 patientsage 65 & older, continuously enrolled in MedicarePart A & Part B during the cohort year & alive onDecember 31 of that year; one-year general Medi-care cohorts spanning 1996–1999. Patients enrolledin an HMO or diagnosed with ESRD any time dur-ing the cohort year are excluded.

Figures p.24–25 Using diagnosis codes from theMedicare five percent data, we identified 1.8 percentof patients as having CKD. By applying this percentto the histogram of eGFR in NHANES III data,1.8percent of patients are below a eGFR cutoff, corre-sponding to an eGFR value of 32.1 ml/min. An esti-mate of the eGFR level identified in the Medicarefive percent data from CKD codes is therefore 32.1ml/min.

24

Préc

ishospit

alization

& dea

th by esrd

modality

C

Cause of hospitalization

Ad

mis

sio

ns

per

1,0

00 p

atie

nt

year

s at

ris

k

0

300

600

900

1,200

1,500

1,800

2,100<3 years ESRD

All-cause CHF ISHD Oth. CV Infect. Other0

300

600

900

1,200

1,500

1,800

2,1003+ years ESRD

<3 years ESRD

3+ years ESRD

Ad

mis

sion

s p

er 1

,000

pat

ien

t ye

ars

at r

isk

0

400

800

1,900

2,000

2,100

2,200

1996 1997 1998 1999 20000

400

800

1,800

1,900

2,000

2,100

Dialysis

Hemodialysis

Peritoneal dialysis

Transplant

Dialysis

Hemodialysis

Peritoneal dialysis

Transplant

2,270+ (2,530) 1,990 to <2,270 1,790 to <1,990 1,580 to <1,790 below 1,580 (1,410)

908+ (1,105) 775 to <908 679 to <775 594 to <679below 594 (517)

16.10+ (19.68) 13.53 to <16.10 11.35 to <13.53 9.25 to <11.35 below 9.25 (6.95)

6.81+ (11.91) 4.52 to <6.81 2.96 to <4.52 1.44 to <2.96below 1.44 (0.88)

Hemodialysis

Peritoneal dialysis

Transplant

p.26 · Hospital admissions, by vintage & modality

p.28 · Geographic variations in hospital admission rates, by modality

p.29 · Geographic variations in hospital days, by modality

p.27 · Trends in admissions, by vintage & modalityompared to patients who have been ondialysis for three or more years, thosewith shorter vintages have slightly

higher hospitalization rates, even withoutconsidering hospitalization at the initiationof therapy (Figure p.26). Hemodialysis pa-tients have the highest rates of hospitaliza-tion, while rates for transplant patients area great deal lower.

Rates for hemodialysis patients have fluc-tuated over the past five years, but do notshow a trend (Figure p.27). Rates for peri-toneal dialysis patients do show a markeddecline, but this trend should be inter-preted with caution, since the peritonealdialysis population has been shrinking,and the lower rates may only reflect pa-tient selection.

There is wide geographic variation in thehospitalization rates for transplant pa-tients, and rates in the upper quintile aretwice as high as those in the lower (Fig-ures p.28–29). The reasons for this largevariation are unclear.

Figures p.30–31 present death rates for theperiod prevalent population during 1998–2000, again divided into younger andolder vintages. Figure p.30 shows that, inpatients of younger vintage, death ratesdue to infection or ischemic heart diseaseare comparable between hemodialysisand peritoneal dialysis patients. Deathsdue to congestive heart failure are morecommon in peritoneal dialysis patients,while those due to other cardiac diseasesare occur more frequently in the hemodi-alysis population. In patients of older vin-tage, those on peritoneal dialysis havehigher rates of death due to infection, andlower rates due to other causes. The deathrates for all transplant patients, regardlessof vintage, are dramatically lower thanthose for dialysis patients.

There are striking differences in death ratetrends by patient vintage (Figure p.31). Ex-cept for a recent slight increase, possiblydue to the inclusion of patients withgreater comorbidity, the rate for patientsof younger vintage has been decliningsteadily. That for patients who have beentreated for longer periods of time, in con-trast, has been increasing. This distinctionmay be attributed to longer survival ofnew patients, who subsequently die of car-diovascular disease. It may also indicate animprovement in the care of newer patientsthat is not shared by those who have re-ceived therapy for longer periods. Furtherinvestigation of death rates by patient vin-tage is needed in order to ensure that pa-tient selection bias does not complicatethe results.

Figure p.32 shows a reasonably flat deathrate for hemodialysis patients in the firstyear after ESRD, possibly due to a combi-

Dialysis Transplant

TransplantDialysis

25

37.1+ (39.1) 35.6 to <37.1 34.2 to <35.6 33.1 to <34.2 below 33.1 (31.2)

268+ (282) 253 to <268 242 to <253 223 to <242 below 223 (208)

CHF ISHD Oth. card. Infect. Other0

20

40

60

80

100

Cause of death

Dea

ths

per

1,0

00 p

atie

nt y

ears

at r

isk

Dea

ths

per

1,0

00 p

atie

nt y

ears

at r

isk

0

20

40

60

80

100 <3 years ESRD

3+ years ESRD0

25

200

210

220

230

240

250

260<3 years ESRD

3+ years ESRD

1990 1992 1994 1996 1998 20000

25

200

210

220

230

240

250

260

Hemodialysis

Peritoneal dialysis

Transplant

DialysisHemodialysisPeritoneal dialysisTransplant

1990 1992 1994 1996 1998 2000

Dea

ths

per

1,0

00 p

atie

nt

year

s at

risk

0

40

160

180

200

220

240

260

280First-year Second-year

1990 1992 1994 1996 19980

40

160

180

200

220

240

260

280

DialysisHemodialysisPeritoneal dialysisTransplant

p.32 · First- & second-year mortality rates, by modality

p.33 · Geographic variations in mortality rates, by modality

p.30 · Mortality rates, by vintage & modality p.31 · Mortality rate trends, by vintage & modality

Dialysis Transplant

nation of improving care and increasingcomplexity of the patients who entertreatment. Those who survive to the sec-ond year show a distinct decline in thedeath rate, possibly reflecting improveddialysis and anemia therapy. Most striking,however, is the death rate for peritonealdialysis patients, which during the firstyear is significantly lower than the rate forhemodialysis patients, yet increases to acomparable level in the second year.

Any analyses which include patients of dif-ferent vintages require careful attention,since the most dramatic improvements insurvival have occurred in patients relativelynew to dialysis therapy. These improve-ments fade from view as patients of oldervintages are included. A similar situation isnoted in Chapter Nine in the comparisonof death rates in incident and prevalentpatients.

The data here suggest that significantprogress has been made in reducingdeath rates, but that older vintage patientsappear not to have benefited from theseimprovements. This suggests that morecareful attention is needed in the care ofpatients who have been on dialysis forlonger periods of time.

The highest death rates for dialysis pa-tients occur primarily in the northern halfof the country (Figure p.33). Rates for trans-plant patients, while not showing as cleara pattern, are highest in many of the re-gions in which rates for dialysis patientsare low. Between the lowest and the high-est quintiles rates differ 36 percent for pa-tients on dialysis, and 25 percent fortransplant patients.

Figure p.26 period prevalent ESRD patients, 2000.Figure p.27 period prevalent ESRD patients. Fig-ure p.28 hospital admissions per 1,000 patient yearsat risk, period prevalent ESRD patients, 2000, byHSA, unadjusted. Figure p.29 hospital days per pa-tient year at risk, period prevalent ESRD patients,2000, by HSA, unadjusted. Figure p.30 periodprevalent ESRD patients, 1998–2000 combined, un-adjusted. Figure p.31 period prevalent ESRD pa-tients, unadjusted. Figure p.32 incident ESRD pa-tients, adjusted for age, gender, race, & primary di-agnosis. Figure p.33 deaths per 1,000 patient yearsat risk, period prevalent ESRD patients, 2000, byHSA, unadjusted.

26

Préc

isesrd

care delivery

system

s B

Nu

mb

er o

f pat

ien

ts (t

ho

usa

nd

s)N

um

ber

of p

atie

nts

(th

ou

san

ds)

Nu

mb

er o

f pat

ien

ts (t

ho

usa

nd

s)

0

50

100

150

200

250

Num

ber

of u

nit

s

0

500

1,000

1,500

2,000

2,500

3,000

3,500

0

50

100

150

200

250

Num

ber

of u

nit

s

0

500

1,000

1,500

2,000

2,500

3,000

3,500

1990 1992 1994 1996 1998 20000

50

100

150

200

1990 1992 1994 1996 1998 2000

Num

ber

of u

nit

s

0

500

1,000

1,500

2,000

2,500

For-profit & non-profit units

Hospital-based & freestanding units

Chain-affiliated & non-chain units

Hospital-based

Freestanding

Freestanding

Hospital-based

Chain-affiliatedChain-affiliated

Non-chain

Non-chain

For-profit

Non-profitNon-profit

For-profit

77.7+ (95.0) 68.3 to <77.7 58.3 to <68.3 29.2 to <58.3 below 29.2 (10.5)

77.7+ (90.5) 68.3 to <77.7 58.3 to <68.3 29.2 to <58.3 below 29.2 (12.8)

p.34 · Trends in patient & unit counts, by unit characteristics

Growth in the numbers of patients & units

p.35 · Geographic variations in the percent of patients treated in chain-affiliated units

oth the provider delivery system andthe cost of caring for ESRD patientscontinue to change significantly. More

patients, for instance, are being treated infor-profit units, while the number of pa-tients in non-profit units has fallen (Figurep.34). These growth patterns are compa-rable to those seen in freestanding versushospital-based units, reflecting the factthat freestanding units are predominantlyfor-profit. Also, the number of patients inchain-affiliated units, as determined fromCMS’s annual End-Stage Renal Disease Fa-cility Survey and Independent Renal Dialy-sis Facilities Cost Reports, is increasing,while the number of patients in non-affili-ated units has declined. This is expected,as dialysis units are steadily being ac-quired into large chains.

The results of the change in reimburse-ment policy can be seen in growing out-patient expenditures for dialysis patients,and decreasing expenditures for inpatientservices (Figures p.36–37). For transplantrecipients, inpatient expenditures havealso declined slightly overall.

For patients with a functioning graft, in-patient expenditures per patient year haveremained unchanged. Overall inpatientexpenditures, however, have increased,due to the growing number of patients.Physician/supplier payments per patientyear have also grown.

In terms of total expenditures, the largestincreases in costs over the last six yearshave been related to inpatient, outpatient,and physician services. Further informa-tion on providers and the economics ofthe ESRD treatment program is providedin Chapters Eleven and Twelve.

Figures p.34–35 dialysis patients; data obtainedfrom CMS’s annual End-Stage Renal Disease Facil-ity Survey, CMS’s Independent Renal Dialysis Fa-cilities Cost Reports, & the CDC’s National Surveil-lance of Dialysis-Associated Diseases in the UnitedStates. Figures p.36–37 period prevalent ESRDpatients, HCFA model; transplant data include anestimate of organ acquisition costs ($25,000 pertransplant). Decreased home health expenditures in1999 were caused by changes in Medicare reimburse-ment policies for home health agencies. Figure p.37patients with Medicare as secondary payor are ex-cluded.

1996 2000

27

95 96 97 98 99 00 95 96 97 98 99 00 95 96 97 98 99 00 95 96 97 98 99 00

95 96 97 98 99 00 95 96 97 98 99 00 95 96 97 98 99 00 95 96 97 98 99 00

Tota

l exp

end

itu

res

(mill

ion

s o

f do

llars

)

0

50

100

150

200

250

300

350

1,000

2,000

3,000

4,000

5,000

0

10

20

30

40

50

60

70

80

90

500

520

540

560

580Dialysis Functioning graftTransplant Graft failure

0

50

100

150

200

250

300

350

0

5

10

15

20

25

50

55

60

65

70

75

Inpatient

Outpatient

Physician/supplier

Home health

Skilled nursing

Hospice

Inpatient

Outpatient

Home health

Inpatient

Outpatient

Home health

Skilled nursing Hospice

Skilled nursing Hospice

Inpatient

Outpatient

Skilled nursing Hospice

Home health

Physician/supplier

Physician/supplier

Physician/supplier

Tota

l exp

end

itur

es p

er p

atie

nt y

ear a

t ris

k (t

ho

usa

nd

s of

do

llars

)

0

5

10

15

20

25

0

2

4

6

8

10

12

14

60

65

70

75Dialysis Functioning graftTransplant Graft failure

0

1

2

3

4

5

6

0

2

4

6

8

10

12

26

28

30

Inpatient

Outpatient

Physician/supplier

Home health Skilled nursing

Hospice

Inpatient

Outpatient

Home health

Inpatient

Outpatient

Home health

Skilled nursing Hospice

Skilled nursing Hospice

Inpatient

Outpatient

Skilled nursing Hospice

Home health

Physician/supplier

Physician/supplier

Physician/supplier

p.36 · Trends in total Medicare ESRD program expenditures

p.37 · Trends in Medicare ESRD program expenditures per patient year at risk

28

Préc

ischapter s

ummary

Patient populations & analytical methods♦ Figures p.9–16 include ESRD patients from 1995–1999 who were age 67

or older at the time of incidence. Claims-based analyses (Figures p.11–16)

include only those patients who were fee-for-service during the 30-month

study period (non-Medicare, Medicare as secondary payor, and HMO risk

patients are excluded). EPO and hemoglobin data are obtained from Part

A and B claims. Cause of hospitalization (Figure p.14) is determined from

ICD-9-CM diagnosis codes for inpatient claims.

♦ Study cohorts for Figures p.18–23 are derived from the 1996–1999 gen-

eral Medicare files, and include patients continuously enrolled in both

Part A and Part B in 1996–1997, 1997–1998, or 1998–1999, alive on the last

day of the observation periods, and residing in the 50 United States or

the District of Columbia. Patients enrolled in a managed care program

(HMO), with Medicare as a secondary payor (MSP), or with a diagnosis of

ESRD any time in 1996–1997, 1997–1998, or 1998–1999 are excluded.

♦ Figures p.26–27 show hospitalization rates per 1,000 patient years at risk

for period prevalent ESRD patients of different vintages. Vintage is de-

fined as the time from the first ESRD service date until January 1 of the

year for prevalent patients, or, for incident patients, as less than one year.

Calculations of unadjusted rates in Figures p.26–29 follow methods used

in the morbidity and hospitalization section, discussed under Chapter

Six and Section E of Appendix A.

Conclusions♦ Data from the Medicare Facility Survey form and the USRDS database show

comparable numbers of patients, 96,000 patients new to ESRD therapy

and 378,000 prevalent patients. As documented in the CMS Facility Sur-

vey data, the total number of transplants in 2000 was 14,311.

♦ The number of patients treated with hemodialysis increased 5.6 percent

per year over the last five years, while the number of peritoneal dialysis

patients declined 4.7 percent each year.

♦ Total Medicare expenditures for the ESRD program in 2000 were $13.82

billion, while non-Medicare costs totaled $5.53 billion. After adjustments

for inflation, the actual cost of the Medicare ESRD program declined be-

tween 0.8 and 1.5 percent.

♦ Approximately 20 percent of patients age 67 and older receive EPO, and

these patients have an average hemoglobin of 10 g/dl. The percentage of

patients receiving EPO has increased over the last six years.

♦ Congestive heart failure and infectious complications occur frequently

as patients with chronic kidney disease advance towards ESRD.

Maps: National means & patient populationsFigure number p.16 p.16 p.16 p.21 p.21 p.28 p.28 p.29 p.29 p.33

Caths. Grafts Fistulas DM CKD Dialysis Tx Dialysis Tx Dialysis

Overall value for all patients 39.3 14.7 8.6 18.0 3.8 1996 807 14.2 5.3 236.5

Total patients 56,053 20,932 12,292 1,096,779 1,096,779 223,274 24,185 223,285 24,186 327,010

Overall value for patients mapped 39.6 14.9 8.7 17.8 3.8 1996 805 14.2 5.3 236.2

Patients dropped due to missing HSA/state 760 100 108 14,497 14,497 3,592 274 3,592 274 5,351

Figure number p.33 p.35 p.35

Tx 1996 2000

Overall value for all patients 34.5 43.4 62.7

Total patients 97,627 7,223 7,223

Overall value for patients mapped 35.1 43.4 62.7

Patients dropped due to missing HSA/state 94,878 180 180

♦ The percent of patients using temporary or permanent catheters is sig-

nificant, and there is geographic variation in the use of these accesses.

♦ Preventive health care measures in the diabetic population are signifi-

cantly underutilized during the period preceding ESRD. Rates of glyce-

mic control monitoring, diabetic eye exams, and lipid testing for these

patients are far below those recommended by the American Diabetes

Association or the American Heart Association.

♦ Three percent of the general Medicare population accounts for almost

two-thirds of the Medicare population with ESRD.

♦ Eighteen percent of Medicare patients have diabetes. Rates of chronic kid-

ney disease in diabetics are almost four times greater than among non-

diabetics.

♦ In the Medicare population the likelihood of death is five to 100 times

greater than the likelihood of developing ESRD.

♦ The estimated glomerular filtration rate in the elderly Medicare popula-

tion is 58 ml/min, and 32.1 ml/min in individuals with chronic kidney dis-

ease. This suggests that Medicare claims data document only those

individuals who have Stage 4 or higher chronic kidney disease.

♦ Hospitalization rates have declined among peritoneal dialysis patients

over the last five years, and have remained relatively stable in the hemo-

dialysis population. Hospitalization rates for transplant patients are far

lower than in the dialysis population.

♦ While overall death rates among prevalent dialysis patients have de-

creased over the past decade, rates for those who have been on dialysis

three or more years have grown. First-year death rates in the incident

population have remained fairly stable, and second-year rates have de-

clined for both hemodialysis and peritoneal dialysis patients.

♦ Patterns in the death rates suggest that survival has improved for patients

new to dialysis therapy, but that patients who have been on ESRD for

longer periods have not experienced the survival benefits associated with

improved dialysis therapy and anemia treatment.

♦ The ESRD provider system has changed dramatically over the last ten years,

and the majority of units are now for-profit.

♦ The number of patients cared for by non-profit units has declined since

the mid-1990s, and the number of patients treated in chain-affiliated units

is now almost 60 percent greater than the number receiving care in non-

chain units.

♦ Dialysis costs per patient year have been fairly stable for the physician/

supplier and inpatient components, while costs for outpatient services

have increased.