NHANES III: Influence of Race on GFR Thresholds and Detection of Metabolic Abnormalities
Robert N. Foley,
Changchun Wang,
Areef Ishani and
Allan J. Collins
United States Renal Data System Coordinating Center, Minneapolis, Minnesota
Correspondence: Dr. Robert N. Foley, United States Renal Data System, 914 South 8th Street, Suite S-253, Minneapolis, MN 55404. Phone: 612-347-5979; Fax: 612-347-5878; E-mail: rfoley{at}usrds.org
Received for publication December 28, 2006.
Accepted for publication May 24, 2007.
Whether the creatinine-based glomerular filtration rate (GFR)thresholds used to define chronic kidney disease (CKD) identifymetabolic abnormalities similarly in minority and nonminoritypopulations is unknown. We addressed this question among adultparticipants in the Third National Health and Nutrition ExaminationSurvey (NHANES III) (n = 15,837). GFR was estimated from serumcreatinine values and metabolic abnormalities were defined by5th or 95th percentile values. After adjustment for age, demographiccharacteristics, and GFR, black participants were significantlymore likely than white participants to have abnormal levelsof systolic and diastolic blood pressure, hemoglobin, phosphorus,and uric acid. Hispanic subjects were significantly more likelyto have abnormal levels of systolic blood pressure, hemoglobin,bicarbonate, and phosphorus. Among participants with GFR <60 mL/min per 1.73 m2, black participants were significantlymore likely to have abnormal levels of systolic and diastolicblood pressure, hemoglobin, and uric acid; Hispanic subjectswere significantly more likely to have abnormal systolic bloodpressure levels. Metabolic abnormalities were more common inminority populations, and low GFR appeared to have a multiplicativeeffect. Defining CKD using a single GFR threshold may be disadvantageousfor minority populations because metabolic abnormalities arepresent at higher levels of GFR.
Estimates indicate that approximately 19 million adults in theUnited States have chronic kidney disease (CKD),1 and recentpublic health initiatives have focused on harmonizing case definitionsand early identification in the general population.2–4Several community-based studies have shown a graded associationbetween GFR and the risk for cardiovascular disease and death,providing further support for the hypothesis that earlier detectionof CKD leads to public health improvement.1,5–9 Risk factorsfor CKD include older age, hypertension, diabetes, cardiovasculardisease, and family history of CKD.
Several national guidelines have recommended the use of serumcreatinine levels to measure GFR, with 60 ml/min per 1.73 m2considered a watershed value, in part because treatable renalabnormalities become increasingly prevalent as GFR falls belowthis level.2,10–12 At the level of public health policy,demonstrating a GFR value below this threshold before embarkingon an exhaustive search for renal complications would seem tobe a rational use of these guidelines. Surprisingly, it is unknownwhether such a two-stage strategy performs similarly in minorityand nonminority populations. In particular, if GFR thresholdsare to become the gatekeeper to more intensive investigationand intervention, then it would seem important to know whetherrenal abnormalities develop at similar GFR values among individualsof different races and ethnicities. Hence, the objectives ofthis national study were to test the following hypotheses amongthe adult population of the United States:
Overall prevalenceof metabolic abnormalities varies by race/ethnicity.
Prevalenceof metabolic abnormalities varies by race/ethnicity,independentof GFR level.
Prevalence of renal abnormalities varies byrace/ethnicity amongindividuals with GFR <60 ml per minper 1.73 m2.
Of the 15,837 study participants, 76.9% were white, 10.4% wereblack, 5.1% were Hispanic, and 7.6% were of other race or ethnicity(Table 1). The corresponding mean GFR values were 90.8 ml/minper 1.73 m2 for white individuals, 104.7 for black individuals,108.5 for Hispanic individuals, and 99.8 for others (P <0.0001).
Table 1. Population characteristics compared by race and ethnicity (N= 15,837)a
Table 2 shows comparisons of mean values of metabolic and BPvariables, by race and ethnicity, in the overall populationand in the subgroups defined by GFR category. In the overallpopulation, participants with GFR <60 ml/min per 1.73 m2had higher systolic BP (SBP), potassium, phosphorus, and uricacid levels and lower hemoglobin levels than participants withGFR 60 ml/min per 1.73 m2. Compared with white individuals,black individuals had higher SBP, diastolic BP (DBP), phosphorus,and uric acid levels and lower potassium and hemoglobin levels;Hispanic individuals had higher phosphorus and lower SBP, DBP,and potassium levels. Formal testing showed statistically significantinteractions between race and ethnicity and GFR for SBP, DBP,and uric acid. Among individuals with GFR <60 ml/min per1.73 m2, black individuals had higher SBP, DBP, and uric acidlevels and lower hemoglobin levels; Hispanic individuals hadhigher SBP and phosphorus levels and lower bicarbonate and calciumlevels.
Table 2. BP and laboratory variables compared by categories of GFR and race and ethnicitya
Table 3 is similar to Table 2, with adjustment for the characteristicsshown in Table 1. In the overall population, individuals withGFR <60 ml/min per 1.73 m2 had higher SBP, DBP, potassium,calcium, phosphorus, and uric acid levels and lower hemoglobinand bicarbonate levels than individuals with GFR 60 ml/min per1.73 m2. Compared with white individuals, black individualshad higher SBP, DBP, and phosphorus levels and lower potassiumand hemoglobin levels; Hispanic individuals had higher SBP andphosphorus levels and lower potassium and hemoglobin levels.Formal testing showed a statistically significant interactionbetween race and ethnicity and GFR for DBP, bicarbonate, anduric acid. Among individuals with GFR <60 ml/min per 1.73m2, black individuals had higher SBP, DBP, and uric acid levelsand lower hemoglobin levels; Hispanic individuals had higherSBP levels and lower hemoglobin levels.
Table 3. Multivariate analysis of BP and laboratory variables with linear regressiona
Table 4 shows unadjusted odds risk for laboratory and BP variables5th or 95th percentiles. In the overall population, individualswith GFR <60 ml/min per 1.73 m2 were more likely to haveabnormal levels of each variable studied except calcium. Blackindividuals were more likely to have abnormal levels of eachvariable except high potassium and low calcium, and Hispanicindividuals were more likely to have abnormal levels of bicarbonate,phosphorus, and uric acid. Formal testing showed a statisticallysignificant interaction between race and ethnicity and GFR forabnormal levels of SBP, DBP, potassium, and uric acid. Amongindividuals with GFR <60 ml/min per 1.73 m2, black individualswere more likely to have abnormal SBP, DBP, hemoglobin, anduric acid levels; Hispanic individuals were more likely to haveabnormal SBP, DBP, bicarbonate, calcium, and phosphorus levels.
Table 4. Unadjusted OR of BP and laboratory variables 5th or 95th percentiles, compared by categories of GFR and race/ethnicitya
With covariate adjustment (Table 5), individuals with GFR <60ml/min per 1.73 m2 were more likely to have abnormal levelsof each variable studied except SBP, DBP, and calcium. Blackindividuals were more likely to have abnormal SBP, DBP, hemoglobin,and phosphorus levels, and Hispanic individuals were more likelyto have abnormal SBP, hemoglobin, bicarbonate, and phosphoruslevels. Formal testing showed statistically significant interactionsbetween race and ethnicity and GFR for abnormal levels of SBP,DBP, potassium, and uric acid. Among individuals with GFR <60ml/min per 1.73 m2, black individuals were more likely to haveabnormal SBP, DBP, hemoglobin, and uric acid levels; Hispanicindividuals were more likely to have abnormal SBP levels.
Table 5. Adjusted OR of BP and laboratory variables 5th or 95th percentiles, compared by categories of GFR and race/ethnicitya
Among individuals with GFR <60 ml/min per 1.73 m2, adjustedodds ratios (OR) for abnormal levels of several variables werehigher among participants from minority populations. Figure 1shows adjusted OR of detecting these abnormalities when differentGFR thresholds (in 5-ml/min per 1.73 m2 increments) were usedfor case definition among black, Hispanic, and other-race participants,using a fixed threshold value of 60 ml/min per 1.73 m2 for whiteparticipants. With this approach, none of the GFR thresholdsled to statistical neutrality for abnormal SBP, DBP, or hemoglobinlevels. Statistical neutrality was achieved with GFR thresholdsof 65, 80, and 65 ml/min per 1.73 m2, respectively, for phosphorus,uric acid, and the presence of one or more abnormalities amongblack, Hispanic, and other-race participants.
Figure 1. OR, from logistic regression models, of BP and laboratory variables 5th or 95th percentiles. In each model, white individuals with GFR <60 ml/min per 1.73 m2 are compared successively with black, Hispanic, and other-race individuals with GFR thresholds varying in 5-ml/min per 1.73 m2 increments from 60 to 90. Adjustment was made for age, gender, body mass index, born outside United States, self-reported diabetes, self-reported hypertension, angiotensin-converting enzyme inhibitor therapy, diuretic therapy, serum ferritin, and red blood cell folate. (Top) P < 0.05 for all odds ratios. (Bottom) P < 0.05 for phosphorus at GFR 60 ml/min per 1.73 m2; uric acid at GFR 60, 65, 70, and 75 ml/min per 1.73 m2; and one or more abnormalities at GFR 60 ml/min per 1.73 m2.
We found that metabolic abnormalities were more common in blackand Hispanic adults than in white adults, an association thatwas evident regardless of whether CKD was present. As expected,the presence of CKD seemed to multiply prevalence estimates,irrespective of race. In aggregate, these observations seemto suggest that CKD management strategies based on single GFRthresholds may be disadvantageous to populations in which thesecomplications are more prevalent, namely ethnic and racial minorities.In particular, the findings suggest that GFR thresholds >60ml/min per 1.73 m2 may be appropriate for detecting severalmetabolic abnormalities in minority populations.
It was been known for several years that the burden of ESRD(requiring renal replacement therapy) differs substantiallyin different racial and ethnic groups in the United States,with much higher event rates among black individuals.13–17Progress has been made in the arena of CKD, especially withregard to interventions that slow the progression of importantcauses, such as diabetic nephropathy and hypertensive nephrosclerosis.18,19
We found that black individuals have a lower prevalence of GFR<60 ml/min per 1.73 m2, mirroring recent findings of theReasons for Geographic and Racial Differences in Stroke (REGARDS)study.20 People of Hispanic or Latino ethnicity now form thelargest single minority population in the United States,21 andthere is little reason to suspect that they should have intrinsicallylower risk for kidney disease; unfortunately, very few studieshave attempted to quantify the burden of CKD in this population.
We found that, unlike several other variables, mean potassiumlevels were lower in black individuals, when adjustment wasmade for GFR. A substantial body of evidence suggests that blackindividuals may be relatively potassium deficient compared withwhite individuals. For example, urinary potassium excretionseems to be lower on random diets22–25 and on diets withfixed potassium contents.26,27 Given the reciprocal relationshipbetween potassium deficiency and sodium retention, intrinsicdifferences in potassium handling may contribute to higher-than-expectedprevalence of hypertension among black individuals.28
Formal, gold-standard measurement of GFR was not a design featureof the Third National Health and Nutrition Examination Survey(NHANES III). Therefore, it was not possible to determine whetherassociations between serum creatinine and true GFR values differin community-dwelling adults of different races and ethnicitiesand similar age and gender distribution. Similarly, it was notpossible to determine whether complications associated withdeclining GFR develop at different GFR levels or whether thehigher prevalence of GFR-associated complications in minoritygroups with GFR <60 ml/min per 1.73 m2 reflects a greaterprevalence of these complications in general, irrespective ofGFR level. The data presented here tend to support the latterhypothesis. For example, in this study, compared with whiteparticipants, black participants had higher adjusted OR forhigh BP and low hemoglobin levels, regardless of whether GFRlevels were <60 ml/min per 1.73 m2. For Hispanic participants,similar between-GFR category parallels were seen for high BP,low hemoglobin, and high phosphorus.
The limitations of our study should be pointed out. It was cross-sectional,and longitudinal measures were not available. Mirroring clinicalreality, gold-standard measures of GFR, such as those basedon inulin or isotope clearance methods, were not used. Limitationsnotwithstanding, we believe that this study has useful features.The study design facilitates quantification of the burden andthe complications of CKD in a nationally representative sample.Overall, this study suggests that strategies in which detectionof treatable renal abnormalities are predicated on a singlethreshold of estimated GFR might be disadvantageous to racialminorities. Research focusing on efficient, equitable identificationof covert kidney disease is urgently needed.
Design
NHANES III, conducted between 1988 and 1994, used stratified,multistage, probability sampling methods to assemble a nationwideprobability sample of the noninstitutionalized population ofthe United States.29 Calibration factors can have an impacton creatinine-based estimates of glomerular filtration, andNHANES III data have been directly calibrated with referencestandards. All NHANES III participants aged 20 yr or older wereeligible for determination of hematologic and biochemical profilesat the mobile examination center.
Measurements and Definitions GFR.
Serum creatinine, measured at White Sands Research Center (Alamogordo,NM) with the modified kinetic Jaffe reaction and a Hitachi 737analyzer (Boehringer Mannheim, Indianapolis, IN), was recalibratedto results obtained at the Cleveland Clinic (Cleveland, OH),using the method of Coresh et al.30 Estimated GFR levels werederived from the re-expressed Modification of Diet in RenalDisease (MDRD) Study formula, namely, 175 x (serum creatininevalue)–1.154 x age–0.203 x (0.742 for women) x (1.21if black).12
Metabolic Abnormalities.
These were defined by the fifth or 95th percentiles of theirrespective distributions in the overall population. The specificthreshold values were as follows: SBP 156.0 or DBP 90.7 mmHg,potassium 4.6 mmol/L, hemoglobin 118 g/L, bicarbonate 22.2 mmol/L,calcium 2.13 mmol/L, phosphorus 1.36 mmol/L, and uric acid 0.47mmol/L.
Other Variables.
Self-reported diabetes was defined as an affirmative answerto the question, "Have you ever been told by a doctor that youhave diabetes or sugar diabetes?" Self-reported hypertensionwas defined as an affirmative answer to the question, "Haveyou ever been told by a doctor or other health professionalthat you have hypertension, also called high BP?"
Statistical Analyses 2 analysis and ANOVA were used for unadjusted comparisons ofbaseline variables between racial/ethnic groups. Consideredas continuous parameters, unadjusted and adjusted associationsof metabolic and BP variables were explored with ANOVA and multiplelinear regression, respectively. Unadjusted and adjusted logisticregression analyses were used to explore the corresponding associationswhen abnormal values of metabolic and BP variables were consideredas binary (yes/no) variables. National estimates of each parameterwere adjusted for the sampling weights implicit in complex surveydesigns, using SUDAAN software (Research Triangle Institute,Research Triangle Park, NC) for complex sample surveys. SASVersion 8.2 (SAS Institute, Cary, NC) was used for data assembly.
The data reported here were analyzed by the US Renal Data Systemusing public-use NHANES files. This study was performed as adeliverable under contract N01-DK-9-2343 (National Instituteof Diabetes and Digestive and Kidney Diseases, National Institutesof Health, Bethesda, MD).
We thank US Renal Data System colleagues Beth Forrest for manuscriptpreparation and regulatory assistance and Nan Booth, MSW, MPH,for manuscript editing.
Footnotes
Published online ahead of print. Publication date availableat www.jasn.org.
Coresh J, Byrd-Holt D, Astor BC, Briggs JP, Eggers PW, Lacher DA, Hostetter TH: Chronic kidney disease awareness, prevalence, and trends among US adults, 1999 to 2000.
J Am Soc Nephrol 16
: 180
–188, 2005[Abstract/Free Full Text]
National Kidney Foundation: K/DOQI clinical practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Kidney Disease Outcomes Quality Initiative.
Am J Kidney Dis 39
: S1
–S246, 2002[CrossRef][Medline]
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, 2003[Medline]
Levey AS, Coresh J, Balk E, Kausz AT, Levin A, Steffes MW, Hogg RJ, Perrone RD, Lau J, Eknoyan G: National Kidney Foundation practice guidelines for chronic kidney disease: Evaluation, classification, and stratification.
Ann Intern Med 139
: 137
–147, 2003[Abstract/Free Full Text]
Fried LF, Shlipak MG, Crump C, Bleyer AJ, Gottdiener JS, Kronmal RA, Kuller LH, Newman AB: Renal insufficiency as a predictor of cardiovascular outcomes and mortality in elderly individuals.
J Am Coll Cardiol 41
: 1364
–1372, 2003[Abstract/Free Full Text]
Henry RM, Kostense PJ, Bos G, Dekker JM, Nijpels G, Heine RJ, Bouter LM, Stehouwer CD: Mild renal insufficiency is associated with increased cardiovascular mortality: The Hoorn Study.
Kidney Int 62
: 1402
–1407, 2002[CrossRef][Medline]
Muntner P, He J, Hamm L, Loria C, Whelton PK: Renal insufficiency and subsequent death resulting from cardiovascular disease in the United States.
J Am Soc Nephrol 13
: 745
–753, 2002[Abstract/Free Full Text]
Anavekar NS, McMurray JJ, Velazquez EJ, Solomon SD, Kober L, Rouleau JL, White HD, Nordlander R, Maggioni A, Dickstein K, Zelenkofske S, Leimberger JD, Califf RM, Pfeffer MA: Relation between renal dysfunction and cardiovascular outcomes after myocardial infarction.
N Engl J Med 351
: 1285
–1295, 2004[Abstract/Free Full Text]
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
: 1296
–1305, 2004[Abstract/Free Full Text]
Joint Specialty Committee on Renal Medicine of the Royal College of Physicians and the Renal Association at RCoGP:
Chronic Kidney Disease in Adults: UK Guidelines for Identification, Management and Referral, London, 2006
Johnson D: The CARI guidelines. Evaluation of renal function.
Nephrology (Carlton) 10[Suppl 4]
: S133
–S176, 2005
Stevens LA, Coresh J, Greene T, Levey AS: Assessing kidney function: Measured and estimated glomerular filtration rate.
N Engl J Med 354
: 2473
–2483, 2006[Free Full Text]
Li S, McAlpine DD, Liu J, Li S, Collins AJ: Differences between blacks and whites in the incidence of end-stage renal disease and associated risk factors.
Adv Ren Replace Ther 11
: 5
–13, 2004[CrossRef][Medline]
Easterling RE: Racial factors in the incidence and causation of end-stage renal disease (ESRD).
Trans Am Soc Artif Intern Organs 23
: 28
–33, 1977[Medline]
Whittle JC, Whelton PK, Seidler AJ, Klag MJ: Does racial variation in risk factors explain black-white differences in the incidence of hypertensive end-stage renal disease?
Arch Intern Med 151
: 1359
–1364, 1991[Abstract]
Feldman HI, Klag MJ, Chiapella AP, Whelton PK: End-stage renal disease in US minority groups.
Am J Kidney Dis 19
: 397
–410, 1992[Medline]
Tell GS, Hylander B, Craven TE, Burkart J: Racial differences in the incidence of end-stage renal disease.
Ethn Health 1
: 21
–31, 1996[Medline]
Hovind P, Tarnow L, Parving HH: Remission and regression of diabetic nephropathy.
Curr Hypertens Rep 6
: 377
–382, 2004[Medline]
Ruggenenti P, Schieppati A, Remuzzi G: Progression, remission, regression of chronic renal diseases.
Lancet 357
: 1601
–1608, 2001[CrossRef][Medline]
McClellan W, Warnock DG, McClure L, Campbell RC, Newsome BB, Howard V, Cushman M, Howard G: Racial differences in the prevalence of chronic kidney disease among participants in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Cohort Study.
J Am Soc Nephrol 17
: 1710
–1715, 2006[Abstract/Free Full Text]
Berenson GS, Voors AW, Dalferes ER Jr, Webber LS, Shuler SE: Creatinine clearance, electrolytes, and plasma renin activity related to the blood pressure of white and black children: The Bogalusa Heart Study.
J Lab Clin Med 93
: 535
–548, 1979[Medline]
Voors AW, Dalferes ER Jr, Frank GC, Aristimuno GG, Berenson GS: Relation between ingested potassium and sodium balance in young blacks and whites.
Am J Clin Nutr 37
: 583
–594, 1983[Abstract/Free Full Text]
Sorof JM, Forman A, Cole N, Jemerin JM, Morris RC: Potassium intake and cardiovascular reactivity in children with risk factors for essential hypertension.
J Pediatr 131
: 87
–94, 1997[CrossRef][Medline]
Lasker N, Hopp L, Grossman S, Bamforth R, Aviv A: Race and sex differences in erythrocyte Na+, K+, and Na+-K+-adenosine triphosphatase.
J Clin Invest 75
: 1813
–1820, 1985[Medline]
Luft FC, Rankin LI, Bloch R, Weyman AE, Willis LR, Murray RH, Grim CE, Weinberger MH: Cardiovascular and humoral responses to extremes of sodium intake in normal black and white men.
Circulation 60
: 697
–706, 1979[Free Full Text]
Gallen IW, Rosa RM, Esparaz DY, Young JB, Robertson GL, Batlle D, Epstein FH, Landsberg L: On the mechanism of the effects of potassium restriction on blood pressure and renal sodium retention.
Am J Kidney Dis 31
: 19
–27, 1998[Medline]
Suh A, DeJesus E, Rosner K, Lerma E, Yu W, Young JB, Rosa RM: Racial differences in potassium disposal.
Kidney Int 66
: 1076
–1081, 2004[CrossRef][Medline]
National Center for Health Statistics: National Health and Nutrition Examination Survey. Available at: http://www.cdc.gov/nchs/nhanes.htm. Accessed April 18, 2007
Coresh J, Astor BC, McQuillan G, Kusek J, Greene T, Van Lente F, Levey AS: Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate.
Am J Kidney Dis 39
: 920
–929, 2002[CrossRef][Medline]
This article has been cited by other articles:
H. N. Ibrahim, C. Wang, A. Ishani, A. J. Collins, and R. N. Foley Screening for Chronic Kidney Disease Complications in US Adults: Racial Implications of a Single GFR Threshold
Clin. J. Am. Soc. Nephrol.,
November 1, 2008;
3(6):
1792 - 1799.
[Abstract][Full Text][PDF]
O. M. Gutierrez, M. Mannstadt, T. Isakova, J. A. Rauh-Hain, H. Tamez, A. Shah, K. Smith, H. Lee, R. Thadhani, H. Juppner, et al. Fibroblast Growth Factor 23 and Mortality among Patients Undergoing Hemodialysis
N. Engl. J. Med.,
August 7, 2008;
359(6):
584 - 592.
[Abstract][Full Text][PDF]
Top
Abstract
Introduction
60 ml/min per 1.73 m2. Compared with white individuals, black individuals had higher SBP, diastolic BP (DBP), phosphorus, and uric acid levels and lower potassium and hemoglobin levels; Hispanic individuals had higher phosphorus and lower SBP, DBP, and potassium levels. Formal testing showed statistically significant interactions between race and ethnicity and GFR for SBP, DBP, and uric acid. Among individuals with GFR <60 ml/min per 1.73 m2, black individuals had higher SBP, DBP, and uric acid levels and lower hemoglobin levels; Hispanic individuals had higher SBP and phosphorus levels and lower bicarbonate and calcium levels. 
5th or 
2 analysis and ANOVA were used for unadjusted comparisons of baseline variables between racial/ethnic groups. Considered as continuous parameters, unadjusted and adjusted associations of metabolic and BP variables were explored with ANOVA and multiple linear regression, respectively. Unadjusted and adjusted logistic regression analyses were used to explore the corresponding associations when abnormal values of metabolic and BP variables were considered as binary (yes/no) variables. National estimates of each parameter were adjusted for the sampling weights implicit in complex survey designs, using SUDAAN software (Research Triangle Institute, Research Triangle Park, NC) for complex sample surveys. SAS Version 8.2 (SAS Institute, Cary, NC) was used for data assembly. 
