Factors Associated With Kidney Disease Progression and Mortality in aReferred CKD Population
Richard A. Hoefield, MRCP,1 Philip A. Kalra, MD,1 Patricia Baker, PhD,2 Beverley Lane, MSc,1
John P. New, FRCP,1 Donal J. O’Donoghue, FRCP,1 Robert N. Foley, MD,3 andRachel J. Middleton, FRCP1
Background: Knowing how kidney disease progresses is important for decision making in patientswith chronic kidney disease (CKD) and for designing clinical services.
Study Design: Prospective cohort study.Setting & Participants: We examined renal function trajectories in CRISIS (Chronic Renal Insuffi-
ciency Standards Implementation Study), in which 1,325 patients with CKD stages 3-5 and mean age of65.1 years were followed up prospectively for a median of 26 months after referral to a regionalnephrology center in the United Kingdom. By protocol, estimated glomerular filtration rate wasdetermined every 12 months.
Predictors: CKD stage defined as estimated glomerular filtration rate �45 (stage 3a), 30-44 (3b),15-29 (4), and �15 (5) mL/min/1.73 m2.
Outcomes: Onset of renal replacement therapy (RRT), death, the composite end point of RRT ordeath, or decreasing CKD stage.
Results: During a median follow-up of 26 months, 13% reached the end point of RRT (5.1 events/100patient-years), 20% died (9.6 deaths/100 patient-years), and 33% reached the combined end point ofRRT or death (14.7 events/100 patient-years). For stage 3a, baseline prevalence and annual probabili-ties of decreasing CKD stage, RRT, and death were 18.0%, 0.41, 0.01, and 0.02, respectively.Corresponding values for stage 3b were 32.5%, 0.22, �0.01, and 0.06; for stage 4, 36.5%, 0.17, 0.03,and 0.10; and for stage 5, 13.2%, zero (by definition), 0.31, and 0.08, respectively. Markov modelprojections suggested a steady decrease for proportions with stages 3a, 3b, and 4; a steady increase fordeath and RRT; and a biphasic pattern for (non-RRT) stage 5, with a plateau in the first 2 years followedby a steady decrease.
Limitations: Single-center observational study.Conclusion: This study suggests that death and RRT are the dominant outcomes in patients referred
hronic kidney disease (CKD) is a signifi-cant public health issue with substantial
conomic implications. Considerable resourcesre diverted to its management in both primarynd secondary care. Stopping the developmentf CKD and treating classic cardiovascular riskactors1-6 is now a major health care priority.
The clinical epidemiology of earlier stage CKDs poorly understood, particularly with regard torogression between stages to end-stage renal
From the 1Vascular Research Group, University ofanchester, Manchester Academic Health Sciences Centre,
alford Royal Hospital Foundation Trust, Salford; 2North-est Institute for Bio-Health Informatics, Manchester Uni-ersity, Manchester, UK; and 3Chronic Disease Researchroup and University of Minnesota, Minneapolis, MN.Received January 28, 2010. Accepted in revised form
une 6, 2010. Originally published online as doi:10.1053/j.
jkd.2010.06.010 on September 9, 2010.
American Journal of Kidney072
isease (ESRD), as well as cardiovascular dis-ase (CVD) and mortality outcomes. It has beenuggested that many patients with CKD do notollow an inexorable course to ESRD, imply-ng that many either die before requiring renaleplacement therapy (RRT) in association withhe increased morbidity and mortality risk en-endered by their CKD or that only a minorityanifest significant progressive kidney dys-
To help health care planners and cliniciansnticipate the clinical needs of patients, we needore detailed information about the long term
utcomes of CKD to determine the best alloca-ion of resources for this population. However,ew studies have investigated clinical outcomesnd kidney disease stage transitions in CKDohorts receiving standard nephrology care.8,9
The objectives of this analysis are to describehe following outcomes of patients in a referredKD cohort: (1) longitudinal evolution of CKDith predictive modeling, (2) risk of death, (3)
isk of requiring RRT, (4) composite end point ofRT or death, and (5) the baseline associationsf death and requirement of RRT.
METHODS
tudyDesign
The Chronic Renal Insufficiency Standards Implementa-ion Study (CRISIS) is an ongoing prospective study initi-ted in 2002 to investigate the evolution and outcomes ofidney disease and its associated comorbid conditions in aeferred population with CKD stages 3-5 not on dialysisherapy.
Renal services in Greater Manchester, UK, are providedy the Greater Manchester Kidney Care Network and di-ided into east and west sectors. Salford Royal HospitalHS Foundation Trust serves the west sector of the network,ith patients having been referred from a community popu-
ation of 1.55 million. A 50% random sample of all patientsith estimated glomerular filtration rate (eGFR) �10-�60L/min/1.73 m2 (determined using the 4-variable Modifica-
ion of Diet in Renal Disease [MDRD] Study equation10)eferred for renal opinion to Salford Royal Hospital renalervices is approached to take part in the study at the initiallinic visit. Recruitment is undertaken in the outpatientepartment, at which time written consent is obtained andligibility is determined. Patients with previous evidence ofRT (dialysis or kidney transplant) are excluded from the
tudy. Ethical approval for this study was obtained fromalford and Trafford Local Research Ethics Committee.
easurements
All participants have kidney function measured at base-ine and annual follow-up visits. Biosamples also are storedt induction into the study. At enrollment into the study andt annual follow-up, demographic, clinical, and laboratoryata are obtained from the electronic patient records, aontinually updated electronic health care system, as well asatient interview and detailed clinical examination.Age, sex, ethnicity, blood pressure, and prescribing data
by medication class) are the demographic variables avail-ble for analysis. Renin-angiotensin blockade was defined asreatment with angiotensin-converting enzyme inhibitorsnd/or angiotensin receptor blockers. Self-reported smoking
tatus was classified as current smokers, ex-smokers, or b
onsmokers. Ex-smokers who had stopped smoking duringhe last year before enrollment were categorized as currentmokers.
Laboratory variables available for analysis included: cre-tinine, hemoglobin, calcium, phosphate, parathyroid hor-one, albumin, C-reactive protein, and cholesterol. Signifi-
ant proteinuria (previously based on 24-hour urinary proteinollections and, more recently, urinary protein-creatinineatio) was defined as 24-hour urinary protein excretion �1/d or urinary protein-creatinine ratio �100 mg/mmol.
Comorbidity variables available for analysis included:ngina, myocardial infarction, coronary angioplasty or stentlacement, coronary artery bypass graft, cardiac arrest, pe-ipheral vascular disease, stroke, or transient ischemic at-ack.
All serum creatinine measurements were performed usingblank rated and compensated Jaffé reaction with a Rocheodular analyzer (www.roche.com). The method was cali-
rated in the laboratory, and in addition to internal qualityontrol, the laboratory participates in the UK National Exter-al Quality Assessment Scheme. Creatinine measurementssed to calculate eGFR were calibrated to be traceable to ansotope-dilution mass spectrometry (IDMS) reference mea-urement procedure. eGFR was calculated using the 4-vari-ble IDMS-traceable MDRD Study equation: eGFR �75(SCr/88.4)�1.154 � age�0.203 � 0.742 (if female) � 1.21if black), where SCr is serum creatinine.
All patients enrolled in CRISIS are treated according totandard renal unit treatment protocols. Current renal unituidelines aim for UK National Institute for Clinical Excel-ence blood pressure targets �130/80 mm Hg in patientsith CKD.11 Renin-angiotensin blockade is recommended
or patients with progressive CKD and/or proteinuria, includ-ng diabetic patients with microalbuminuria. Statin therapyas prescribed when indicated by evidence of previousVD and in all patients with diabetes older than 40 years orith 10-year risk of CVD �20% (according to the 2005
oint British Societies’ guidelines).12
The cohort was divided according to CKD stage using theational Kidney Foundation’s modified Kidney Diseaseutcomes Quality Initiative (K/DOQI) classification ofKD,13 with eGFR of 45-59 mL/min/1.73 m2 representing
tage 3a; 30-44 mL/min/1.73 m2, stage 3b; 15-29 mL/min/.73 m2, stage 4; and �15 mL/min/1.73 m2, stage 5.
The CRISIS database is continually updated. In addition,nternational Classification of Diseases, Tenth Revision causef death is flagged from the Office for National StatisticsONS) on an annual basis. Cause of death was categorized asalignancy, peripheral vascular disease, cerebrovascular,
ardiac, sepsis, renal, and other causes.RRT and death were considered competing risks. RRT
as defined as initiation of dialysis therapy or transplant.he date of the event that occurred first was recorded.
tatistical Analysis
Normally distributed variables are presented as mean �tandard deviation, or median and 25th-75th percentiles forkewed data. Analysis of variance and �2 tests were used toompare continuous and categorical variables across eGFRtages, respectively. Characteristics of the CRISIS cohort at
aseline are presented for all stages (3-5) of the modified
DOQI classification of CKD. A Cox proportional hazardsodel was used to identify variables associated with the end
oints of RRT, death, or the combined end point of RRT oreath. Model variables that met the inclusion criterion of P
0.1 on unadjusted analysis were retained for multivariatenalysis. Candidate variables included age, sex, diabetesellitus, CVD, proteinuria, systolic and diastolic blood
ressure, eGFR, hemoglobin level, serum phosphate level,arathyroid hormone level, cholesterol level, albumin level,moking, and use of renin-angiotensin blockade and statins.esults are described as hazard ratios with 95% confidence
ntervals. Kaplan-Meier estimates of survival function arelotted across the stages of CKD. Log-rank test was used toompare survival distributions between groups. P � 0.05 isonsidered statistically significant. SPSS, version 15.0 (SPSSnc, www.spss.com) was used for these analyses.
A Markov chain analysis using TreeAge Pro 2009 (TreeAgeoftware Inc, www.treeage.com) was used to forecast long-
erm future outcomes in CRISIS. The Markov model waseveloped to predict the trajectory of transition states (GFRr proportion dying or reaching RRT) of the cohort during0 years derived from the baseline prevalence of the popula-ion and actual distribution of the cohort at year 1. In theseodels, RRT and death were considered terminal states.erformance of these models was assessed by comparingctual and predicted proportions in each clinical state inears 2 and 3 after patient enrollment.
RESULTS
Analysis was performed on all 1,325 patientsnrolled in CRISIS between October 1, 2002, andanuary 1, 2008. A total of 76 (5.4%) patients wereost to follow-up during this period, 238 (8%) hadKD stage 3a, 431 (33%) had CKD stage 3b, 481
36%) had CKD stage 4, and 175 (13%) CKDtage 5 at inception into the study.
At baseline, mean age was 65.1 � 14.0 years,44 (63.7%) were men, mean eGFR was 30.9 �3.4 mL/min/1.73 m2, 429 (32%) had diabetes,26 (47%) had evidence of one or more CVDs,17 (31.5%) were current smokers, and 79260%) were receiving renin-angiotensin block-de. Table 1 lists baseline characteristics of theohort stratified by eGFR. The distribution ofrimary diseases was similar across the differentKD cohorts.Median follow-up was 26 months (25th-75th
ercentiles, 14-53). During this period, 172 (13%)eached ESRD requiring RRT (5.1 events/100atient-years), 270 (20%) died (9.6 deaths/100atient-years), and 33% reached the combinednd point of RRT or death (14.7 events/100atient-years).Of the cohort, 17 (1%) participants selected
onservative management of their ESRD. Twelve w
70%) patients in the conservatively managedohort were men, with a mean age of 82.1 years.uring the study period, 14 (82%) conserva-
ively managed patients died, with kidney-elated mortality accounting for 50% of the under-ying cause of death.
Of 270 deaths, cause of death was availableor 241 (89%) patients. CVD (cardiac [70], cere-rovascular [14], and peripheral vascular disease16]) accounted for 100 (41.4%) deaths; sepsis,or 61 (25.3%) deaths; malignancy, for 31 (12.9%)eaths; and kidney-related and other causes ofeath occurred in 28 (11.6%) and 21 (8.7%),espectively.
Figure 1A shows the probability of RRT-freeurvival stratified by eGFR within the CRISISohort. Figure 1B shows the probability of sur-ival, and Fig 1C shows the probability of com-ined RRT/death-free survival according to base-ine eGFR. Kaplan-Meier plots show significantifferences in risks of death and RRT based onGFR in this referred cohort.
Table 2 lists Cox proportional hazards associ-ted with the end points of RRT, death, or theombined end point of RRT or death. Multivari-te associations of RRT included younger age;igher diastolic blood pressure and phosphate lev-ls; lower hemoglobin, serum albumin, and base-ine eGFR levels; and significant proteinuria.
Older age, presence of baseline CVD andiabetes, current smokers, higher serum phos-hate level, and lower baseline eGFR and serumlbumin level were associated with mortality.se of statin therapy was associated with aecreased hazard ratio for death.Multivariate associations of the combined end
oint of RRT or death included older age; currentmokers; presence of baseline CVD; lower hemo-lobin, serum albumin, and baseline eGFR lev-ls; and higher serum phosphate level. Use ofenin-angiotensin blockade and statins was asso-iated with a decreased hazard ratio for theombined end point of RRT or death.
Patients with CKD stage 5 at inception into thetudy were 18 times more likely to reach ESRDequiring RRT than patients with stage 3a CKD.hey also had a 2.5 times increased risk of deathnd a 7 times greater likelihood of reaching theombined end point of death or ESRD thanatients with CKD stage 3a. Similarly, patients
ikely to reach ESRD requiring RRT than thoseith CKD stage 3a, twice as likely to die, and at.5 times increased risk of reaching the com-ined end point of death or ESRD than patientsith CKD stage 3a. Figure 2 shows stage transi-
Note: Values expressed as mean � standard deviation o2 to mL/s/1.73 m2, �0.01667; hemoglobin in g/dL to g/L, �mol/L, �0.3229, albumin in g/dL to g/L, �10, cholester
odium and potassium in mEq/L and mmol/L and PTH in pgAbbreviations: ADPKD, autosomal dominant polycystic
Figure 3 shows results of the Markov chainnalysis with the proportion of total inceptionopulation in CRISIS from year 0 through year 3nd predicted values through year 10. The analy-is maps out the proportion of the original cohortn each category in each year. Figure 3A shows
seline Stratified by eGFR Level
eGFR (mL/min/1.73 m2)
P0-44 (n � 431) 15-29 (n � 481) <15 (n � 175)
36.9 � 4.3 22.3 � 4.2 11.7 � 2.4 —
0.87 � 1.65 1.08 � 1.86 2.23 � 2.77 �0.001
65.1 � 13.1 67.0 � 14.2 64.8 � 13.1 �0.001
65.9 62.6 65.7 0.7
31.3 38.0 34.3 �0.001
99.3 98.5 99.4 0.9
32.3 29.1 34.3 0.1
46.4 51.1 44.0 0.2
47.1 48.0 55.9 0.001
66.1 60.3 46.9 �0.001
65.0 59.0 52.0 0.002
0.00119.2 23.9 26.4
8.6 6.0 9.126.5 28.0 21.5
19.2 16.8 16.56.9 9.7 10.7
19.6 15.7 15.7
138.0 � 20.7 137.8 � 22.7 143.6 � 22.6 0.004
75.3 � 13.3 73.7 � 12.6 76.5 � 12.6 0.003
12.80 � 1.66 12.04 � 1.52 11.35 � 1.60 �0.001
9.22 � 0.44 9.1 � 0.6 9.06 � 0.8 �0.001
3.38 � 0.68 3.81 � 0.77 4.89 � 1.2 �0.001
140.4 � 3.2 140.4 � 3.2 140.3 � 3.1 0.7
4.75 � 0.58 4.92 � 0.74 5.05 � 0.71 �0.001
4.18 � 0.44 4.16 � 0.45 4.12 � 0.46 0.001
9.0 � 17.4 9.9 � 19.7 9.9 � 15.3 0.9
64.0 � 50.2 101.8 � 79.6 156.7 � 137.8 �0.001
66.28 � 46.4 170.15 � 50.27 166.28 � 46.4 0.5
ntage. Conversion factors for units: eGFR in mL/min/1.73lcium in mg/dL to mmol/L, �0.2495, phosphate in mg/dL tog/dL to mmol/L, �0.02586. No conversion necessary ford ng/L.disease; CRP, C-reactive protein; CVD, cardiovascular
lar filtration rate; PTH, parathyroid hormone; SBP, systolic
at Ba
8) 3
7
.0
.0
.1
8
8
1
8
4
.9
.7
.53 1
r perce10, ca
ol in m/mL ankidney
lomeru
he predicted trajectory for patients with stages
3tcppbpdi
shreonma
assmafppiptRmrsc(tst
lHsvtPfic
mRedifferences in risks of death and RRT based on chronickidney disease stage in this referred cohort.
Hoefield et al1076
a and 3b CKD, showing a gradual decrease inhe proportion of patients with moderately de-reased kidney function over time. Figure 3Bredictive plots suggest a steady decrease for theroportion of patients with CKD stage 4 and aiphasic pattern for (non-RRT) stage 5, with alateau in the first 2 years followed by a gradualecrease. Figure 3C shows an analogous steadyncrease for RRT and death over time.
DISCUSSION
This analysis of the CRISIS population de-cribes the significant burden of CKD in terms ofigh mortality and risk of developing ESRDequiring RRT with its inherent considerableconomic implications. In this large cohort ofutpatients with a wide spectrum of CKD underephrology care, a progressive increase in risk ofortality and/or requirement of RRT was associ-
ted with worsening stage of CKD.There currently is a paucity of information to
id effective allocation of renal health care re-ources. The predictive plots of CKD stage tran-ition, death, and RRT outcomes from the Markovodel show a reasonable approximation to the
ctual outcome during the first 36 months ofollow-up in the CRISIS cohort (Fig 3). Therojections suggest a steady decrease for theroportions with stages 3a, 3b, and 4; a steadyncrease for death and RRT; and a biphasicattern for (non-RRT) stage 5, with a plateau inhe first 2 years followed by a steady decrease.esults indicate that patients with eGFR �30L/min/1.73 m2 remain in non-RRT stages for a
elatively short period because of rapid progres-ion to RRT or death. Our data suggest that thelinical needs of those with more advanced CKDstages 4 and 5) may be served best by strategicargeting of resources to increase dialysis provi-ion and improve the infrastructure and capacityo support expansion of transplant services.
Outcomes of patients with advanced CKD inarge referred cohorts are not well described.owever, existing information shows the relation-
hip between eGFR and risk of death and cardio-ascular outcomes in the general population inhe United States and Europe. In the Kaiserermanente general population of Northern Cali-ornia (n � 1,120,295), Go et al2 showed anndependent graded association between de-
Figure 1. Kaplan-Meier curves show (A) renal replace-ent therapy (RRT)-free survival, (B) survival, and (C)RT/death-free survival probabilities according to baselinestimated glomerular filtration rate (eGFR). The plots show
reased eGFR and death, cardiovascular events,
aNvmcip(fidwpc�4
tc
cwChCoda1apt
p3mqG
A
M
D
S
C
R
S
S
D
H
P
P
A
C
C
P
C
mP
ds
CKD Progression and Mortality 1077
nd hospitalization. Results from the Secondational Health and Nutrition Examination Sur-ey (NHANES II) showed that eGFR �70 mL/in/1.73 m2 was associated with a 68% in-
reased risk of all-cause mortality and 50%ncreased risk of cardiovascular mortality com-ared with people with normal kidney functioneGFR �90 mL/min/1.73 m2).14 Similar resultsrom the 3 Cities (3C) Study in France, whichnvolved an elderly cohort, showed an indepen-ent substantially increased risk of mortalityith eGFR �45 mL/min/1.73 m2 compared withatients without CKD, but interestingly, no in-reased risk comparing individuals with eGFR60 mL/min/1.73 m2 and those with eGFR of
5-59 mL/min/1.73 m2.15
The limited data available relating to long-erm risk of death or requirement of RRT in large
Note: Conversion factors for units: hemoglobin in g/L tomol/L to mg/dL, �3.10, albumin in g/L to g/dL, �0.1, cholTH in pg/mL and ng/L.Abbreviations: CI, confidence interval; CKD, chronic k
are have shown that lower eGFR is associatedith increased risk of RRT and death.16-18 In theRIB (Chronic Renal Impairment in Birming-am) Study, a referred cohort of patients withKD stages 3-5 (n � 382) with mean follow-upf 6.3 years, progressive kidney dysfunction evi-enced by a 30% decrease in eGFR was associ-ted with a 3 times increased risk of ESRD and.3 times increased risk of death, although thebsolute risk of death outweighed the risk ofrogression to ESRD, similar to our findings inhe CRISIS cohort.19
Results from CRISIS indicate that our CKDopulation, which had an average eGFR of0.9 � 13.4 mL/min/1.73 m2, has high rates ofortality (9.6 deaths/100 patient-years) and re-
uirement of RRT (5.1 events/100 patient-years).iven that the CRISIS cohort is a referred popu-
ional studies, would be toward underestimationf the true risk of mortality because all patientsave treatment strategies targeted at decreasinghe risk of progressive CKD and improving clini-al outcomes of these high-risk patients. Thesendings contrast appropriately with a recent largebservational study (n � 4,231) of patients withtage 4 CKD under nephrology care (mean eGFR,2 mL/min/1.73 m2; median follow-up, 31onths) that showed a higher RRT risk (143
vents/1,000 patient-years), yet lower mortalityisk (45 deaths/1,000 patient-years), than theRISIS population9; the difference in RRT risk
s explained by the lower overall eGFR in thistudy, and lower mortality, because CRISIS in-luded patients with stage 5 CKD and conserva-ively managed patients.
Younger age, higher blood pressure and pro-einuria, higher phosphate level, and lower base-ine eGFR and hemoglobin levels were associ-ted with requirement of RRT in our cohort.urthermore, older age, CVD, diabetes, currentmokers, lower baseline eGFR and albumin lev-ls, and higher phosphate level, but, interest-ngly, not (as found by others) parathyroid hor-one level, were associated with increased risk
f mortality. These results are consistent withata from other CKD cohorts in North Americand Taiwan.8,9
In concordance with other studies,20-24 our
tudy also shows that absence of significant p
roteinuria and nonsmoking status were associ-ted with lower risk of these main end points,aising the possibility that targeting these modifi-ble risk factors might improve outcomes inatients with CKD.Use of renin-angiotensin blockade and statins
n the analysis were associated with decreasedazard outcomes for the combined end point ofRT or death. However, given the observationalature of CRISIS, we acknowledge that theseesults may have been subject to significant indi-ation bias.
Our study showed that CVD was the predomi-ant cause of death, accounting for 41% of theotal, similar to observations in patients on dialy-is therapy or after kidney transplant.25,26 Pa-ients with diabetes in our cohort had an in-reased risk of mortality, although this was nottatistically significant when analyzed in the com-ined end point (death/RRT) or for RRT alone.his is consistent with the hypothesis that the
ncreased cardiovascular risk engendered by theombination of CKD and diabetes decreaseshe likelihood of many of these patients reachingSRD.27,28
Interestingly, the traditional cardiovascular riskactors of hypertension and hypercholesterol-mia were not associated with increased risk ofortality or the combined end point of RRT orortality in our cohort. This may be explained in
Figure 2. Estimated glo-merular filtration rate (eGFR),renal replacement therapy(RRT), and death transitionsprobabilities within baselineeGFR categories. Transitionstates of GFR, death, or RRTwere derived from the baselineprevalence of the populationand actual distribution of thecohort at year 1. eGFR is givenin mL/min/1.73 m2.
art by the age of the CRISIS population, which
hStinttmwil
w
tobfdi
sdtrr
tf(palsmsti3bfip
CKD Progression and Mortality 1079
as a mean age of 65 years (46% �70 years).everal previous studies in the general popula-
ion have shown that the increased risk of mortal-ty associated with high blood pressure is almostegligible in elderly patients as opposed to pa-ients younger than 65 years.29-31 Furthermore,he detrimental effect of hypercholesterolemiaay be attenuated in patients older than 75 years,ith some studies of non-CKD populations show-
ng increased mortality with lower cholesterolevels.29
In our referred cohort, significant proteinuria
Figure 3. Proportions of to-al inception population in dif-erent clinical states in years 0baseline) through 10. Valuesredicted from Markov chainnalysis are shown in dotted
ines, and actual values arehown in solid lines. The perfor-ance of the model was as-
essed by comparing the ac-ual and predicted proportionsn each clinical state in the first6 months after enrollment. Ab-reviations: GFR, glomerularltration rate; RRT, renal re-lacement therapy.
as independently associated with progression r
o RRT, a finding described repeatedly in previ-us studies of patients with diabetic and nondia-etic kidney disease, in which proteinuria con-ers more rapid rates of progression of kidneyisease and increased cardiovascular mortal-ty.32,33
Our analysis of the CRISIS population haseveral limitations. First, in our study, eGFR waserived from the 4-variable MDRD Study equa-ion,10 with its well-known limitations of accu-acy. Assessing clearance of injected exogenousadioactive markers is the gold standard for accu-
ate measurement of global kidney function; how-
eci
wEpmiwmocwtCnpscdCt
aCwgcCl
thraoiadtCoorOepott
n
dm1
Cl1
rr1
Rcr
dK
Iar
a2
KeC
riK
RfitA
Cnc
t2
tA
Rc
Hoefield et al1080
ver, this method is invasive, expensive, timeonsuming, and not practical for clinical screen-ng in large population-based cohorts.
Second, we included patients in our analysisho selected conservative management of theirSRD, and most subsequently reached the endoint of death, potentially influencing the highortality in our cohort. However, this limitation
s relative because only 1% of our populationas conservatively managed and including themore accurately represents current management
f patients with advanced CKD in secondaryare because they represented 10% of patientsith stage 5 CKD at baseline or who progressed
o this stage during follow-up. In addition, theRISIS cohort represented a large heteroge-eous CKD population receiving standard ne-hrology care in one tertiary referral center. Asuch, our findings may be more generalizable tourrent nephrology practice in the United King-om, as distinct from outcomes of patients withKD enrolled in clinical trials or other interven-
ion studies.In conclusion, our study describes outcomes
nd stage transition of patients with advancedKD in a large referred outpatient cohort whoere managed at a renal center. This study sug-ests that death and RRT are the dominant out-omes in patients referred for management ofKD and that most patients spend comparatively
ittle time in late stages without RRT.CRISIS provides robust estimates of long-
erm outcomes of patients with CKD, which mayave implications for future configuration ofenal services. Further validation no doubt willppear from other similar cohorts under nephrol-gy care, but data from our study may helpmprove the health of these high-risk patients byiding health care planners to more appropriatelyirect resources. The current KDOQI classifica-ion of CKD predominantly focuses on stage ofKD, rather than risk of progression or adverseutcome attached to each patient. We are in favorf refinement of this classification to stratify forisk of progression to RRT and adverse outcome.ur data show that easily obtained clinical param-
ters can assist nephrologists and primary carehysicians in identifying patients at greatest riskf progression to RRT or death, enabling moreargeted treatment and therapeutic intervention
hat may benefit both renal and patient survival. r
ACKNOWLEDGEMENTSSupport: None.Financial Disclosure: The authors declare that they have
o relevant financial interests.
REFERENCES1. Foley RN, Murray AM, Li S, et al. Chronic kidney
isease and the risk for cardiovascular disease, renal replace-ent, and death in the United States Medicare population,
998 to 1999. J Am Soc Nephrol. 2005;16(2):489-495.2. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY.
hronic kidney disease and the risks of death, cardiovascu-ar events, and hospitalization. N Engl J Med. 2004;351(13):296-1305.3. Garg AX, Clark WF, Haynes RB, House AA. Moderate
enal insufficiency and the risk of cardiovascular mortality:esults from the NHANES I. Kidney Int. 2002;61(4):486-1494.4. Muntner P, He J, Hamm L, Loria C, Whelton PK.
enal insufficiency and subsequent death resulting fromardiovascular disease in the United States. J Am Soc Neph-ol. 2002;13(3):745-753.
5. Shlipak MG, Fried LF, Crump C, et al. Cardiovascularisease risk status in elderly persons with renal insufficiency.idney Int. 2002;62(3):997-1004.6. Lash JP, Go AS, Appel LJ, et al. Chronic Renal
nsufficiency Cohort (CRIC) Study: baseline characteristicsnd associations with kidney function. Clin J Am Soc Neph-ol. 2009;4(8):1302-1311.
7. Thorp ML, Eastman L, Smith DH, Johnson ES. Man-ging the burden of chronic kidney disease. Dis Manag.006;9(2):115-121.8. Chiu YL, Chien KL, Lin SL, Chen YM, Tsai TJ, Wu
D. Outcomes of stage 3-5 chronic kidney disease beforend-stage renal disease at a single center in Taiwan. Nephronlin Pract. 2008;109(3):c109-c118.9. Levin A, Djurdjev O, Beaulieu M, Er L. Variability and
isk factors for kidney disease progression and death follow-ng attainment of stage 4 CKD in a referred cohort. Am Jidney Dis. 2008;52(4):661-671.10. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N,
oth D. A more accurate method to estimate glomerularltration rate from serum creatinine: a new prediction equa-
ion. Modification of Diet in Renal Disease Study Group.nn Intern Med. 1999;130(6):461-470.11. National Institute for Health and Clinical Excellence.
12. JBS 2. Joint British Societies’ guidelines on preven-ion of cardiovascular disease in clinical practice. Heart.005;91(suppl 5):v1-52.13. National Kidney Foundation. K/DOQI. Part 4. Defini-
ion and classification of stages of chronic kidney disease.m J Kidney Dis. 2002;39(suppl 1):S46-S75.14. Muntner P, He J, Hamm L, Loria C, Whelton PK.
enal insufficiency and subsequent death resulting fromardiovascular disease in the United States. J Am Soc Neph-
15. Stengel B, Loury P, Froissart M, et al. Long-termutcome in the elderly with chronic kidney disease (CKD) inhe general population [abstract]. Nephrol Dial Transplant.007;22(suppl 6):vi226.16. Mafham MM, Emberson J, Landray M. Meta-analysis
f observational studies relating estimated glomerular filtrationate to risk of death or major cardiovascular events. Programnd abstracts of the American Society of Nephrology Renaleek. November 13-19, 2006; San Diego, CA.17. Feldman HI, Appel LJ, Chertow GM, et al. The Chronic
enal Insufficiency Cohort (CRIC) Study: design and methods.Am Soc Nephrol. 2003;14(7 suppl 2):S148-S153.18. Wheeler DC, Townend JN, Landray MJ. Cardiovascu-
ar risk factors in predialysis patients: baseline data from thehronic Renal Impairment in Birmingham (CRIB) Study.idney Int Suppl. 2003;84:S201-S203.19. Emberson J. Prediction of end-stage renal disease andortality among patients with pre-dialysis chronic kidney dis-
ase: The Chronic Renal Impairment in Birmingham (CRIB)ohort. Paper presented as an abstract at: British Renal Society/enal Association meeting. May 14, 2008; Glasgow, UK.20. Agardh CD, Garcia-Puig J, Charbonnel B, Angelkort
, Barnett AH. Greater reduction of urinary albumin excre-ion in hypertensive type II diabetic patients with incipientephropathy by lisinopril than by nifedipine. J Hum Hyper-ens. 1996;10(3):185-192.
21. Chuahirun T, Wesson DE. Cigarette smoking predictsaster progression of type 2 established diabetic nephropathyespite ACE inhibition. Am J Kidney Dis. 2002;39(2):376-382.
22. Lebovitz HE, Wiegmann TB, Cnaan A, et al. Renalrotective effects of enalapril in hypertensive NIDDM: role ofaseline albuminuria. Kidney Int Suppl. 1994;45:S150-S155.
23. Orth SR, Ritz E, Schrier RW. The renal risks ofmoking. Kidney Int. 1997;51(6):1669-1677.
24. Ruggenenti P, Fassi A, Ilieva AP, et al. Preventingicroalbuminuria in type 2 diabetes. N Engl J Med. 2004;
51(19):1941-1951. r
25. Ansell D, Roderick P, Hodsman A, Ford D, Steen-amp R, Tomson C. UK Renal Registry 11th Annual ReportDecember 2008): chapter 7 survival and causes of death ofK adult patients on renal replacement therapy in 2007:ational and centre-specific analyses. Nephron Clin Pract.009;111(suppl 1):c113-c139.26. Owen WF, Lowrie EG. C-Reactive protein as an
utcome predictor for maintenance hemodialysis patients.idney Int. 1998;54(2):627-636.27. Adler AI, Stevens RJ, Manley SE, Bilous RW, Cull
A, Holman RR. Development and progression of nephrop-thy in type 2 diabetes: the United Kingdom Prospectiveiabetes Study (UKPDS 64). Kidney Int. 2003;63(1):225-32.28. Collins AJ, Li S, Gilbertson DT, Liu J, Chen SC,
erzog CA. Chronic kidney disease and cardiovascularisease in the Medicare population. Kidney Int Suppl. 2003;7:S24-S31.29. Casiglia E, Mazza A, Tikhonoff V, et al. Weak effect
f hypertension and other classic risk factors in the elderlyho have already paid their toll. J Hum Hypertens. 2002;6(1):21-31.30. Kannel WB, D’Agostino RB, Silbershatz H. Blood
ressure and cardiovascular morbidity and mortality rates inhe elderly. Am Heart J. 1997;134(4):758-763.
31. Menotti A, Mulder I, Nissinen A, et al. Cardiovascu-ar risk factors and 10-year all-cause mortality in elderlyuropean male populations; the FINE Study. Finland, Italy,etherlands, Elderly. Eur Heart J. 2001;22(7):573-579.32. Chiurchiu C, Remuzzi G, Ruggenenti P. Angiotensin-
onverting enzyme inhibition and renal protection in nondia-etic patients: the data of the meta-analyses. J Am Socephrol. 2005;16(suppl 1):S58-S63.33. Jafar TH, Stark PC, Schmid CH, et al. Proteinuria as aodifiable risk factor for the progression of non-diabetic
