Interaction between GFR and Risk Factors for Morbidity and Mortality in African Americans with CKD

Transcription

1 Article Interaction between GFR and Risk Factors for Morbidity and Mortality in African Americans with CKD Kevin F. Erickson,* Janice Lea, and William M. McClellan Summary Background and objectives The African American Study of Kidney Disease Trial identified risk factors for CKD progression and suggested that GFR level may modify the association between these risk factors and CKD progression or death. Design, setting, participants, & measurements Enrollment in the African American Study of Kidney Disease Trial occurred between June of 1995 and September of 2001, with median follow-up of 48.6 months. Among 1094 patients with hypertensive kidney disease in the trial, this study tested whether the association between six previously identified risk factors for CKD progression (or death) and a composite clinical outcome (progression of CKD, ESRD, or death) depends on level of GFR. Multivariate Cox regression was used to control for other baseline risk factors. Results After controlling for baseline risk factors, only proteinuria was more closely associated with the composite clinical outcome at lower levels of GFR (P value for interaction term=0.002); increased hazards of the clinical composite outcome associated with a doubling of proteinuria ranged from 30% (95% confidence interval=21% 39%) with a GFR of 50 to 55% (95% confidence interval=40% 72%) with a GFR of 25. Conclusions The magnitude of the association between proteinuria and CKD progression, ESRD, or death in the African American Study of Kidney Disease Trial cohort depends on the level of GFR; proteinuria is a stronger independent predictor of the composite clinical outcome at lower levels of GFR. This finding reinforces that African Americans with proteinuria and lower GFR represent a population at particularly high risk for adverse outcomes. Clin J Am Soc Nephrol 8: 75 81, doi: /CJN Introduction Patients with CKD progress to ESRD at varying rates (1). African Americans with CKD suffer from particularly high rates of progression to ESRD (2). In 2009, progression to ESRD in the United States was 3.5 times higher in African Americans than Caucasians (3). Data from the African American Study of Hypertension and Kidney Disease (AASK) Trial show that the average rate of decline in GFR accelerated as patients progressed to ESRD (4), whereas a study in a more racially diverse population found a constant rate of decline (5). Like all patients with CKD, African Americans with CKD also suffer from higher rates of infection, cardiovascular disease, and mortality compared with the non-ckd population (3,6). Understanding why African Americans are at a higher risk for CKD progression can help to (1) directtherapy to those patients who will benefit most, (2) appropriately counsel patients about disease prognosis, and (3) identify important modifiable risk factors. Higher prevalence of several well described risk factors for CKD progression and development of ESRD in African Americans such as proteinuria, hypertension, lower socioeconomic status, coronary disease, and high cholesterol accounts for some, but not all, of the elevated risk of ESRD (7). Among African Americans with CKD and hypertension followed in the AASK Trial, a number of risk factors associated with CKD progression or death have been identified (8,9). One study from the AASK cohort suggested that a doubling of proteinuria may be more closely associated with CKD progression in patients with advanced CKD compared with patients with mild CKD (8). Although the interaction between proteinuria, severity of CKD, and CKD progression could help to explain the increased risk of CKD progression observed in African Americans, it is unknown whether this interaction persists after accounting for other known risk factors for CKD progression. We used data from the AASK cohort to test the hypothesis that GFR level modifies the relationship between proteinuria and CKD progression or death after controlling for other known CKD risk factors. Additionally, we examine whether GFR level modifies the relationship between other previously described risk factors for CKD progression or death in the AASK cohort. *Division of Nephrology, Stanford University School of Medicine, Palo Alto, California; Stanford University Center for Primary Care and Outcomes Research, Stanford University, Stanford, California; Division of Nephrology, Emory University School of Medicine, Atlanta, Georgia; and Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia Correspondence: Dr. Kevin Erickson, 780 Welch Road, Suite 106, Palo Alto, CA kevine1@stanford.edu Vol 8 January, 2013 Copyright 2013 by the American Society of Nephrology 75

2 76 Clinical Journal of the American Society of Nephrology Materials and Methods Participants and Study Design This study is a secondary analysis of the main AASK Trial cohort, which has been described previously (4). Briefly, the AASK Trial was a multicenter randomized clinical trial comparing the effect of strict BP control (mean arterial pressure [MAP],92 or,125/75 mmhg) with normal BP control (MAP= or 140/90 mmhg) and three antihypertensive regimens. Enrollment in the AASK Trial occurred between June of 1995 and September of Patients were randomized to receive ramipril, metoprolol, or amlodipine. Other drugs were added stepwise to achieve the BP goals. Patients (n=1094) were self-identified African Americans who had hypertension and were ages years with GFR between 20 and 65 ml/min per 1.73 m 2 and no other identified causes of renal insufficiency. Exclusion criteria included (1) diastolic BP,95 mmhg; (2) known history of diabetes mellitus (fasting blood glucose.140 mg/dl or random glucose.200 mg/dl); (3) urinary protein to creatinine ratio (UP/Cr).2.5; (4) accelerated or malignant hypertension within 6 months; (5) secondary hypertension; (6) serious systemic disease; (7) clinical congestive heart failure; and (8) specific indication for or contraindication to a study drug or study procedure. Data Collection During the AASK Trial, baseline information was collected for demographics and lifestyle habits. A central laboratory measured baseline serum and urinary chemistry levels. GFR was assessed by I (125) -iothalamate clearance two times during baseline, at 3 and 6 months, and every 6 months thereafter during follow-up. GFR was expressed after standardization for body surface area in units of milliliters per minute per 1.73 m 2. Urinary protein excretion was expressed as UP/Cr from a 24-hour urine collection. Three consecutive seated BP measurements were taken using a Hawksley random zero sphygmomanometer after at least 5 minutes rest, with the mean of the two last readings recorded. The baseline BPs were the BP measurements obtained at the screening visits before randomization. Outcome Variables The main outcome measure in this study was a clinical composite given by the time from randomization to (1) a GFR event, defined by a confirmed decline in GFR by either 50% or 25 ml/min per 1.73 m 2 from the mean of two baseline GFR measurements, (2) occurrence of ESRD, or (3) death. A secondary renal outcome was a composite given by the time from randomization to either (1) a GFR event or (2) occurrence of ESRD. Predictor Variables Risk factors for CKD progression or death assessed in this study included risk factors previously linked to progression or death in epidemiologic studies from the AASK Trial cohort. These factors included smoking history, sex, age, fasting blood glucose, serum phosphate, MAP, serum albumin, urine sodium, LDL cholesterol, triglycerides, history of heart disease, hematocrit, urine protein, urine urea nitrogen, body mass index, and high uric acid (.8 mg/dl). Heart disease was included as a composite of either a history of heart disease or left ventricular hypertrophy. Because of a skewed distribution and based on a previously described association between doubling of UP/Cr and hazard of CKD progression, we used the log transformation of UP/Cr as a predictor in the analysis (8). We selected six risk factors to test for an interaction with GFR in their association with clinical composite outcomes. These factors were selected aprioribased on pathophysiologic plausibility and the strength of their association with outcomes described in previous analyses of the AASK cohort. They included proteinuria, smoking history, heart disease, male sex, hematocrit, and serum phosphorus. Baseline Characteristics To describe variation in differences in baseline characteristics and outcomes by level of GFR, the dataset was separated (for descriptive purposes) according to high and low GFR (high and low GFR defined by stages 3a CKD [GFR$45] and stages 3b or 4 CKD [GFR,45], respectively). Baseline Kaplan Meier curves were plotted after stratifying the population by GFR level and proteinuria, with the median value of used for stratifying proteinuria. Statistical Analyses In a multivariate analysis, time to death, GFR decline, or ESRD was modeled using Cox regression as a function of known CKD risk factors and selected interaction terms. P values based on the 95% confidence intervals (CIs) for interaction term coefficients were used as a measure of statistical significance, and a Bonferroni correction was used to test for statistical significance in the setting of multiple comparisons (threshold for statistical significance was 0.008). Before their inclusion in the model, all variables were assessed for linearity in their association with hazard of clinical composite outcome (10). Because of nonlinear associations with the hazard of clinical composite outcome, GFR was included in the model after a log transformation, whereas serum albumin and body mass index were divided into dichotomous variables based on their median values. Because of substantial amounts of missing values for triglycerides and left ventricular hypertrophy (a component of the heart disease composite), we undertook multiple imputation to estimate missing values (11,12). The results after multiple imputation were not substantively different in estimated interaction terms compared with the results from the complete dataset. Baseline characteristics were compared among participants with and without missing data, and they were different only in sex (56.7% of participants with missing data were male versus 43.3% of patients without missing data were male) and MAP (113 mmhg in patients with missing data versus 115 mmhg in patients without missing data). Each variable was tested for violation of the proportionate hazards assumption. A separate analysis corrected for observed violations without significantly changing the results. A separate analysis using a secondary renal composite outcome assessed the interaction between GFR and the six

3 Clin J Am Soc Nephrol 8: 75 81, January, 2013 GFR and Risk Factors for Clinical Outcomes, Erickson et al. 77 selected risk factors in determining time to GFR decline or ESRD. In this analysis, patients were censored at death. As in the primary analysis, GFR and proteinuria were modeled using log transformations, and albumin and body mass index were separated into dichotomous variables. This project was approved by the institutional review board of Emory University School of Medicine. Results Population Characteristics There were 1094 patients included in the analysis; 603 (55.1%) patients had a GFR$45 ml/min per 1.73 m 2, whereas 491 (44.9%) patients had a GFR,45 ml/min per 1.73 m 2. The clinical composite outcome (death, GFR decline, or ESRD) was more common in patients with lower baseline GFR, with 44.8% of patients with a GFR,45 ml/ min per 1.73 m 2 experiencing the composite outcome compared with 19.9% of patients with GFR$45 ml/min per 1.73 m 2. The renal composite outcome (GFR decline or ESRD) was also more common in patients with lower baseline GFR (36.0% in patients with lower GFR; 13.8% in patients with higher GFR). There was a higher proportion of females in the lower GFR group (45.2%) compared with the higher GFR group (33.5%), and the lower GFR group was slightly younger (mean age=53.6 years) compared with the higher GFR group (mean age=55.4 years). Most risk factors assessed varied by GFR, with comparatively lower hematocrit, albumin, LDL cholesterol, 24-hour urine sodium excretion, glucose, and urine urea nitrogen and comparatively higher serum phosphorus, UP/Cr, and high uric acid in the lower GFR group. GFR groups were similar in triglyceride levels, body mass index, smoking history, and prevalence of heart disease. There was no difference between the two groups in the proportion randomized to low versus high BP target. The mean duration of follow-up was 47.4 months (SD=17.3), whereas the median was 48.6 months (Table 1). AKaplan Meier survival plot of time to clinical composite outcome stratified by degree of proteinuria and high versus low GFR group shows a trend to decreased unadjusted survival (i.e., worse patient outcomes) caused by high proteinuria in patients with lower GFR compared with patients with higher GFR. The P value testing the significance of this trend (using Cox regression to test for an interaction between strata) was 0.16 (Figure 1). Table 1. Baseline characteristics by GFR comparison group High GFR$45 ml/min per 1.73 m 2 (n=603) Low GFR,45 ml/min per 1.73 m 2 (n=491) P Value a Demographics Sex (% female) ,0.001 Mean age (yr; SD) 55.4 (10.3) 53.6 (11.1) Comorbid conditions Heart disease (%) b BMI (.30 kg/m 2 ; %) Uric acid (.8.0 mg/dl; %) ,0.001 Smoking history (%) Serum albumin (,4.3 g/dl; %) ,0.001 Mean baseline laboratory values (SD) Hematocrit (mean %) 40.6 (4.61) 38.0 (4.98),0.001 Phosphorus (mg/dl) 3.40 (0.73) 3.67 (0.63),0.001 Glucose (mg/dl) 96.1 (19.9) 93.7 (16.4) 0.03 Mean arterial BP (mmhg) 113 (15.3) 115 (16.9) 0.16 Urine urea nitrogen (g/d) 8.90 (3.97) 7.53 (3.40),0.001 Urine sodium (g/d) 3.81 (2.03) 3.53 (1.91) 0.02 Triglycerides (mg/dl) c 137 (77.3) 144 (85.0) 0.21 LDL cholesterol (mg/dl) 139 (39.6) 133 (42.5) 0.03 Urine protein/creatinine ratio 0.19 (0.38) 0.50 (0.61),0.001 GFR (ml/min per 1.73 m 2 ) 55.8 (6.03) 33.1 (7.17),0.001 AASK study-related characteristics Randomized to low BP target (%) d Composite outcome (%) e ,0.001 Renal outcome (%) f ,0.001 BMI, body mass index; AASK, African American Study of Kidney Disease. a P value testing for difference between low and high GFR groups (t test for continuous variables and chi-squared test for categorical variables). b Includes left ventricular hypertrophy and history of heart disease; 16% of records are missing. c For triglycerides, 23% of records missing. d Low BP target was goal mean arterial pressure of 92 mmhg or less. e Composite clinical outcome includes death, development of ESRD, or reduction in GFR. f Composite renal outcome includes development of ESRD or reduction in GFR.

4 78 Clinical Journal of the American Society of Nephrology Figure 1. Kaplan Meier survival curve stratified by GFR group and level of proteinuria. Proteinuria stratified by median level of protein/ creatinine ratio (0.081). A trend to decreased survival (i.e., worse patient outcomes) caused by proteinuria in patients with lower GFR compared with patients with higher GFR is illustrated by a wider gap between solid lines (low GFR) compared with a smaller gap between dashed lines (high GFR). In an unadjusted Cox regression, the interaction between lower GFR and higher proteinuria strata was not statistically significant (P=0.16). Multivariate Analysis Stratified by CKD Severity In the multivariate analysis testing for interaction between the six selected risk factors and GFR while controlling for other baseline risk factors, baseline GFR significantly modified the association between proteinuria and clinical composite outcome. Specifically, higher UP/Cr values were more closely associated with clinical composite outcome at lower levels of GFR compared with higher levels of GFR (P=0.002) after controlling for other potential risk factors (Table 2). Based on the estimated coefficients, the increased hazard of the clinical composite outcome associated with a doubling of proteinuria ranged from 30% (95% CI=21% 39%) with a GFR of 50 to 55% (95% CI=40% 72%) with a GFR of 25 (Figure 2). The level of baseline GFR did not modify the association between any of the other potential risk factors tested and the clinical composite outcome. In a test for the interaction between the six selected risk factors and GFR in their association with the secondary renal composite outcome (ESRD or decline in GFR), proteinuria continued to interact with GFR (P,0.001). Specifically, higher UP/Cr values were more closely associated with ESRD or reduction in GFR at lower levels of GFR compared with higher levels of GFR. The interaction between heart disease and GFR level was of borderline statistical significance (P=0.04), suggesting a stronger association between heart disease and GFR decline or ESRD at higher levels of GFR. However, this interaction was not statistically significant after accounting for multiple comparisons. Baseline GFR did not modify the association between any of the other tested risk factors and the renal composite outcome (Supplemental Table 1). Discussion This study shows that, of six previously identified risk factors for CKD progression or death in the AASK cohort, after controlling for the effect of other baseline risk factors in a multivariate analysis, proteinuria was more closely associated with CKD progression or death in patients with more advanced CKD compared with patients with less advanced CKD. In a separate analysis censoring for death, proteinuria continued to be more closely associated with ESRD or GFR decline at lower levels of baseline GFR. The finding that the association between proteinuria and CKD progression, ESRD, or death is stronger at lower GFR levels adds an additional dimension to the importance of proteinuria in African Americans. It has previously been shown that lower GFR and higher proteinuria alone can explain much of the observed differences in development of ESRD between African Americans and other races (13). Our findings suggest that, at least in African Americans with hypertensive nephrosclerosis, the interaction between proteinuria and GFR may also contribute to the observed racial disparities.

5 Clin J Am Soc Nephrol 8: 75 81, January, 2013 GFR and Risk Factors for Clinical Outcomes, Erickson et al. 79 Table 2. Results from multivariate model testing for interaction between risk factors, GFR, and composite clinical outcome HR Lower 95% CI Upper Categorical variables Uric acid.8.0 mg/dl Low BP randomization group BMI.30 kg/m Serum albumin,4.3 g/dl Continuous variables Age (10 yr) Glucose (10 mg/dl) Urine urea nitrogen (g/d) Mean arterial pressure (10 mmhg) Urine sodium (g/d) LDL cholesterol (10 mg/dl) Triglycerides (10 mg/dl) Covariates tested for interaction (stratified by GFR) Male sex GFR= GFR= Heart disease GFR= GFR= Smoking history GFR= GFR= Hematocrit (5%) GFR= GFR= Phosphorus (0.1 mg/dl) GFR= GFR= Doubling of urine protein/creatinine GFR= a GFR= a Supplemental Table 1 has detailed regression results, including estimated coefficient for GFR level. HR, hazard ratio; CI, confidence interval; BMI, body mass index. a Statistically significant interaction term (P=0.002). There is reason to suspect that the observed interaction between proteinuria and GFR may be unique to African Americans, because several studies involving other populations have failed to identify this interaction. For instance, a study of men at high risk for cardiovascular disease based on the Multiple Risk Factor Intervention Trial found that, although dipstick proteinuria and low estimated GFR were independently associated with development of ESRD, there was no interaction (14). Although a study of a community-based cohort in Alberta, Canada found significant interactions between proteinuria and GFR in determining death and CKD progression, the interactions operated in the opposite direction of what we observed in that study, lower GFR reduced the effect of proteinuria on death or CKD progression (15). This finding is important from both pathophysiologic and clinical perspectives. Pathophysiologically, it is consistent with the hypothesis that proteinuria perpetuates glomerular injury, leading to a progressive cycle of renal dysfunction through mechanisms such as cytokine release, promotion of interstitial damage, upregulation of the renin angiotensin aldosterone system, and altered glomerular hemodynamics (16 21). Clinically, it reinforces that African Americans with advanced kidney disease and proteinuria could benefit substantially from close monitoring and aggressive efforts to reduce proteinuria. This finding is particularly important considering the finding that the magnitude of reduction in proteinuria was associated with the risk of ESRD in patients enrolled in the AASK Trial (22). There is an increasing interest in developing tools for predicting the risk of CKD progression using a so-called renal risk score, which is similar to the Framingham Risk Score for cardiovascular disease (23,24). A validation study testing several predictive models found that they were accurate but noted that minority groups, including African Americans, were underrepresented in their data (25). This study suggests that, to accurately develop prognostic models in African Americans, it may be necessary to consider the stronger association between proteinuria and adverse outcomes in patients with more advanced CKD. There are several limitations with this study. First, by using a clinical composite outcome as the primary endpoint, we were unable to differentiate the effect of the various risk factors on each specific outcome (such as death, ESRD, or CKD progression). However, this composite was necessary to avoid bias because of informative censoring that could have occurred if the analysis was restricted to more specific competing outcomes. Using the composite outcome also increased the power of the study to detect interaction effects. Second, large numbers of missing data associated with risk factors, such as triglycerides and left ventricular hypertrophy, necessitated the use of multiple imputation. Third, because this study is a secondary analysis, these results will need to be replicated in additional populations to improve generalizability. Previous studies have highlighted the importance of proteinuria in explaining the increased risk of CKD progression observed in the African-American population. By showing that the severity of CKD modifies the relationship between proteinuria and adverse clinical outcomes in a cohort of African Americans with measured GFR, our study provides an additional pillar to our understanding of why renal outcomes are worse in African Americans. Acknowledgments The authors thank Wolfgang C. Winkelmayer, ScD, for assistance with development of the statistical model and Maria Montez Rath for assistance in statistical methods. K.E. was supported in part by Agency for Healthcare Research and Quality Grant F32 HS Disclosures None.

6 80 Clinical Journal of the American Society of Nephrology Figure 2. Increased hazard of composite outcome associated with doubling of proteinuria stratified by GFR. Hazard ratios were obtained from estimated coefficients for proteinuria, GFR, and the interaction between proteinuria and GFR. References 1. National Kidney Foundation; : K/DOQI clinical practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Am J Kidney Dis 39[2 Suppl 1]: S1 S266, Coresh J, Astor BC, Greene T, Eknoyan G, Levey AS: Prevalence of chronic kidney disease and decreased kidney function in the adult US population: Third National Health and Nutrition Examination Survey. Am J Kidney Dis 41: 1 12, US Renal Data System; : USRDS 2009 annual data report: Atlas of end-stage renal disease in the United States. In: National Institutes of Health, Bethesda, MD, US Renal Data System, Lewis J, Greene T, Appel L, Contreras G, Douglas J, Lash J, Toto R, Van Lente F, Wang X, Wright JT Jr; AASK Study Group; : A comparison of iothalamate-gfr and serum creatinine-based outcomes: Acceleration in the rate of GFR decline in the African American Study of Kidney Disease and Hypertension. JAmSoc Nephrol 15: , Hunsicker LG, Adler S, Caggiula A, England BK, Greene T, Kusek JW, Rogers NL, Teschan PE: Predictors of the progression of renal disease in the Modification of Diet in Renal Disease Study. Kidney Int 51: , 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 351: , Klag MJ, Whelton PK, Randall BL, Neaton JD, Brancati FL, Stamler J: End-stage renal disease in African-American and white men. 16-year MRFIT findings. JAMA 277: , Norris KC, Greene T, Kopple J, Lea J, Lewis J, Lipkowitz M, Miller P, Richardson A, Rostand S, Wang X, Appel LJ: Baseline predictors of renal disease progression in the African American Study of Hypertension and Kidney Disease. JAmSocNephrol17: , Lea J, Cheek D, Thornley-Brown D, Appel L, Agodoa L, Contreras G, Gassman J, Lash J, Miller ER 3rd, Randall O, Wang X, McClellan W; AASK Study Investigators; : Metabolic syndrome, proteinuria, and the risk of progressive CKD in hypertensive African Americans. Am J Kidney Dis 51: , CollettD:Modeling Survival Data in Medical Research, 2nd Ed., Boca Raton, FL, Chapman & Hall/CRC, Rubin D: Multiple Imputation for Nonresponse in Surveys, New York, John Willey & Sons, Allison PD: Missing Data, Thousand Oaks, CA, Sage Publications, Inc., McClellan WM, Warnock DG, Judd S, Muntner P, Kewalramani R, Cushman M, McClure LA, Newsome BB, Howard G: Albuminuria and racial disparities in the risk for ESRD. JAmSoc Nephrol 22: , Ishani A, Grandits GA, Grimm RH, Svendsen KH, Collins AJ, Prineas RJ, Neaton JD: Association of single measurements of dipstick proteinuria, estimated glomerular filtration rate, and hematocrit with 25-year incidence of end-stage renal disease in the multiple risk factor intervention trial. J Am Soc Nephrol 17: , Hemmelgarn BR, Manns BJ, Lloyd A, James MT, Klarenbach S, Quinn RR, Wiebe N, Tonelli M; Alberta Kidney Disease Network; : Relation between kidney function, proteinuria, and adverse outcomes. JAMA 303: , Largo R, Gómez-Garre D, Soto K, Marrón B, Blanco J, Gazapo RM, Plaza JJ, Egido J: Angiotensin-converting enzyme is upregulated in the proximal tubules of rats with intense proteinuria. Hypertension 33: , Wang SN, LaPage J, Hirschberg R: Role of glomerular ultrafiltration of growth factors in progressive interstitial fibrosis in diabetic nephropathy. Kidney Int 57: , Yu HT: Progression of chronic renal failure. Arch Intern Med 163: , Hayslett JP: Functional adaptation to reduction in renal mass. Physiol Rev 59: , Johnson HA, Vera Roman JM: Compensatory renal enlargement. Hypertrophy versus hyperplasia. Am J Pathol 49: 1 13, Remuzzi G, Bertani T: Pathophysiology of progressive nephropathies. N Engl J Med 339: , Lea J, Greene T, Hebert L, Lipkowitz M, Massry S, Middleton J, Rostand SG, Miller E, Smith W, Bakris GL: The relationship

7 Clin J Am Soc Nephrol 8: 75 81, January, 2013 GFR and Risk Factors for Clinical Outcomes, Erickson et al. 81 between magnitude of proteinuria reduction and risk of endstage renal disease: Results of the African American study of kidney disease and hypertension. Arch Intern Med 165: , Wilson PW, D Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB: Prediction of coronary heart disease using risk factor categories. Circulation 97: , Taal MW, Brenner BM: Predicting initiation and progression of chronic kidney disease: Developing renal risk scores. Kidney Int 70: , Tangri N, Stevens LA, Griffith J, Tighiouart H, Djurdjev O, Naimark D, Levin A, Levey AS: A predictive model for progression of chronic kidney disease to kidney failure. JAMA 305: , 2011 Received: April 2, 2012 Accepted: September 11, 2012 Published online ahead of print. Publication date available at www. cjasn.org. This article contains supplemental material online at asnjournals.org/lookup/suppl/doi: /cjn /-/ DCSupplemental.