Introduction of the CKD-EPI equation to estimate glomerular filtration rate in a Caucasian population

3176 Nephrol Dial Transplant (2011) 26: 3176 3181 doi: 10.1093/ndt/gfr003 Advance Access publication 16 February 2011 Introduction of the CKD-EPI equation to estimate glomerular filtration rate in a Caucasian population Jan A.J.G. van den Brand 1, Gerben A.J. van Boekel 1, Hans L. Willems 2, Lambertus A.L.M. Kiemeney 3, Martin den Heijer 3,4 and Jack F.M. Wetzels 1 1 Department of Nephrology, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands, 2 Department of Laboratory Medicine, Radboud University Medical Centre, Nijmegen, The Netherlands, 3 Department of Epidemiology, Biostatistics and Health Technology Assessment, Radboud University Medical Centre, Nijmegen, The Netherlands and 4 Department of Endocrinology, Radboud University Medical Centre, Nijmegen, The Netherlands Correspondence and offprint requests to: Jan A.J.G. van den Brand; E-mail: a.vandenbrand@nier.umcn.nl Abstract Background. Chronic kidney disease (CKD) is defined as the presence of kidney damage, albuminuria or a reduction in glomerular filtration rate (GFR). A GFR <60 ml/min/ 1.73m 2 alone is sufficient to diagnose CKD Stages III V. Recently, the new chronic kidney disease epidemiology collaboration (CKD-EPI) equation was introduced. It has been suggested to result in higher estimated glomerular filtration rates (egfrs) than the Modification of Diet in Renal Disease (MDRD 4 ) formula. Here, we assess consequences of introducing the CKD-EPI equation in a West European Caucasian population. Methods. Data were obtained from 6097 Caucasian participants of the Nijmegen Biomedical Study (2823 males and 3274 females). Serum creatinine values were determined using the Jaffe method, calibrated against mass spectrometry and Ó The Author 2011. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com

CKD-EPI and its effect on CKD prevalence 3177 were used to calculate egfr MDRD4 and egfr CKD-EPI.Demographic data, health status and information on medication use for all participants was obtained with a postal questionnaire. Results. The introduction of the CKD-EPI equation changed the curve of egfr by age, with higher values in the younger age groups and a steeper decline of egfr with ageing. As a consequence, younger people were more often classified to a higher GFR stage and older people, especially males, to a lower GFR stage. Conclusions. In comparison with the MDRD 4 formula, the CKD-EPI equation leads to higher estimates of GFR in young people and lower estimates in the elderly. On a population level, this may lead to higher estimates of kidney function. However, in routine clinical practice where the population is predominantly elderly, the opposite may be true. The introduction of egfr CKD-EPI necessitates reconsidering the definition of CKD. We suggest introducing age-dependent threshold values and/or the use of urinary albumin excretion to improve risk stratification. Keywords: chronic kidney disease; epidemiology; glomerular filtration rate Introduction According to the Kidney Disease Outcomes Quality Initiative, chronic kidney disease (CKD) is defined by the presence of kidney damage, albuminuria or a reduction of glomerular filtration rate (GFR) [1]. A reduced GFR, <60 ml/min/1.73m 2, is sufficient to diagnose CKD Stages III V. Thus, accurate estimation of GFR has become a heavily debated topic. The four variable equation developed in the Modification of Diet in Renal Disease (MDRD 4 ) Study is widely used in research and clinical practice. Its pros and cons and usefulness to define disease have been debated [2, 3]. The MDRD 4 formula may underestimate GFR in fairly young people and in women, thus overestimating the number of patients with CKD. GFR decreases with age and the risk associated with estimated glomerular filtration rate (egfr) values <60 ml/min/1.73m 2 is less evident in older people [4, 5]. This has led to a debate on the validity of using a cutoff value of 60 ml/min/1.73m 2 in elderly when defining CKD [2, 6]. Recently, the new CKD-EPI equation was developed [7]. Its mathematical properties allow for better estimation of measured GFR throughout its whole range, not just for values <60 ml/min/1.73m 2. However, this new equation was derived from a population consisting predominantly of young or middle-aged people with an average GFR of 70 ml/min/1.73m 2. It has been shown to result in higher estimates of GFR than the MDRD 4 formula and to provide a more appropriate risk classification in middle-aged individuals [8]. Little is known about the consequences of the introduction of the CKD-EPI equation for older people. The Nijmegen Biomedical Study is a population-based study with a sample stratified by age and gender. This allows for comparison of the MDRD 4 and CKD-EPI formulas across a wide range of ages. We assessed the consequences of introducing the CKD-EPI equation in our Caucasian population. Materials and methods Population Detailed methods of the Nijmegen Biomedical Study have been described elsewhere [5, 9]. In brief, the Nijmegen Biomedical Study is a populationbased cross-sectional survey conducted by the Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands. Age and sex stratified randomly selected adult inhabitants of the municipality of Nijmegen (Nijmegen, Lent and Oosterhout) were invited to fill out a postal questionnaire on lifestyle and medical history (n ¼ 22 452). Overall, 9371 people (41.7%) filled out and returned the questionnaire. In total, 68.9% of the respondents donated two 8.5 ml tubes of blood (n ¼ 6455). Respondents were asked whether they had been diagnosed with myocardial infarction, stroke, diabetes, hypertension or any kidney disease. Information about drug therapy during the last 6 months was gathered. Respondents who were not diagnosed with nor used medication associated with aforementioned diseases were defined as healthy. Analysis was limited to 6097 Caucasian respondents with valid data of whom 2823 were males and 3274 females. Of this population, 2356 respondents reported an underlying condition or associated drug therapy. Estimation of GFR Serum creatinine was measured using a Jaffe alkaline picrate assay (Abbott Aeroset analyser) [5]. The Jaffe assay was calibrated to isotope diluted mass spectrometry (IDMS)-traceable creatinine values using the Roche enzymatic creatinine assay. The version of the MDRD 4 equation reexpressed for IDMS-traceable creatinine was used to calculate egfr MDRD4 [10]. The egfr CKD-EPI was calculated as described by Levey et al. [7]. egfr was expressed in millilitre per minute per 1.73 m 2. Statistical analysis Analyses were performed using Stata 10 software. We calculated means, SD range and 5th, 25th, 50th, 75th and 95th percentiles per 5-year age group in healthy participants. LMS Chart Maker version 2.4 (Medical Research Council, UK) was used to obtain smoothed egfr values for males and females. M (the smoothed median of egfr versus age), S (smoothed coefficient of variation of egfr versus age) and L (smoothed Box-Cox power transformation against age) curves were allowed additional degrees of freedom until no substantial change in chi-square statistic (<4.0) occurred [11]. We created crosstabs to evaluate classification of patients per GFR group according to both the CKD-EPI and MDRD 4 equations. Results We calculated egfr using CKD-EPI for 6097 Caucasian participants of the Nijmegen Biomedical Study. Study population characteristics are presented in Table 1. Figures 1 and 2 show that the curve of egfr versus age changes with the introduction of the egfr CKD-EPI. For young people, the median egfr value is higher. In women, >75 and men >70 years of age, median egfr values are lower when the CKD-EPI is used. Furthermore, the variance of egfr within our population appears to decrease when the CKD-EPI is used. Supplementary tables with egfr values by 5-year age group for the total and healthy population are available at the website (etables 1 and 2). The effect of the introduction of the CKD-EPI equation on egfr stratification is illustrated for men and women by age in Tables 2 and 3. To provide more detailed insight, we have tabulated egfr in classes of 15 ml/min/1.73m 2. The tables clearly illustrate that egfr is higher using the CKD- EPI formula. This effect is most prominent at higher levels

3178 J.A.J.G. van den Brand et al. Table 1. Population characteristics of the Nijmegen Biomedical Study a Healthy population Males Females Total population n (% males) 1663 2073 6097 45% Age 53 (18 92) 48 (18 98) 58 (18 98) Medical history Myocardial infarction 0 0 448 8% Stroke 0 0 195 4% Hypertension 0 0 1487 27% Kidney disease 0 0 196 4% Diabetes mellitus 0 0 356 7% Medication Drug therapy for CVD 0 0 788 14% Diuretics 0 0 609 11% Anti-hypertensives 0 0 1060 19% Cholestrol-lowering drugs 34 2% 45 2% 621 12% Total number of drugs b 0 (0 13) 1 (0 20) 1 (0 50) Lifestyle Current smoker 418 25% 494 24% 1335 22% Never smoked 463 28% 904 44% 2042 34% Current alcohol use 1525 92% 1738 84% 5110 84% Sedentary lifestyle 473 28% 488 24% 1898 31% Laboratory Serum creatinine 79 (29 208) 65 (28 134) 73 (29 837) Serum urea 6.1 (2.3 19.6) 5.1 (2 19.3) 5.8 (2 39.2) Serum albumin 46 (28 59) 45 (13 57) 45 (13 59) a Data are presented as n and percentage or median and (range), respectively. CVD, cardiovascular disease. b Including vitamin supplements. of egfr and in women. Importantly, and not unexpectedly, the effect on reclassification is dependent on age. When compared to the MDRD 4 formula, CKD-EPI gave higher estimates of GFR in subjects under the age of 70 but lower estimates in people >70 years. Discussion We compared values of egfr calculated with the recently developed CKD-EPI equation and the MDRD 4 formula using the data of the Nijmegen Biomedical Study, a sex- and age-stratified population-based survey. Our data demonstrate that the CKD-EPI formula overall provides higher estimates of GFR than the MDRD 4 formula. However, the differences are dependent on age and gender. Whereas in younger age groups, the CKD-EPI formula provides higher estimates of GFR, the opposite was true in older people, especially males. Our data indicate that the prevalence of young persons with egfr < 60 ml/min/1.73m 2 will decrease. In contrast, the number of elderly persons with egfr < 60 ml/min/1.73m 2 will increase. Development of a new formula to estimate GFR was stimulated by some shortcomings of the MDRD 4 equation, Fig. 1. Smoothed egfr by age for Caucasian males (total population) in the Nijmegen Biomedical Study. The interrupted lines represent the egfr CKD-EPI and the connected lines the egfr MDRD4. Data are presented as the p95, median and p5. Fig. 2. Smoothed egfr by age for Caucasian females (total population) in the Nijmegen Biomedical Study. The interrupted lines represent the egfr CKD-EPI and the connected lines the egfr MDRD4. Data are presented as the p95, median and p5.

CKD-EPI and its effect on CKD prevalence 3179 Table 2. Estimated GFR stage for Caucasian Dutch males over and under 70 years of age in the Nijmegen Biomedical Study population using the CKD-EPI or MDRD4 equation CKD-EPI MDRD egfr >90 75 89 60 74 45 59 33 44 15 29 <15 Total <70 years >90 913 3 916 75 89 284 343 627 60 74 142 171 313 45 59 20 48 68 30 44 6 6 15 29 2 2 <15 2 2 Total 1197 488 191 48 6 2 2 1934 >70 years >90 46 108 154 75 89 241 9 250 60 74 9 247 20 276 45 59 2 122 20 144 30 44 50 3 53 15 29 9 2 11 <15 1 1 Total 46 358 258 142 70 12 3 889 Table 3. egfr stage for Caucasian Dutch females over and under 70 years of age in the Nijmegen Biomedical Study population using the CKD-EPI or MDRD4 equation a CKD-EPI MDRD egfr >90 75 89 60 74 45 59 33 44 15 29 <15 Total >70 years >90 1048 1048 75 89 570 290 860 60 74 321 189 510 45 59 54 56 110 30 44 6 4 10 15 29 1 2 3 <15 2 2 Total 1618 611 243 62 5 2 2 2543 >70 years >90 35 54 89 75 89 166 5 171 60 74 43 199 7 249 45 59 16 138 2 156 30 44 52 5 57 15 29 9 9 <15 0 0 Total 35 263 220 145 54 14 0 731 a Blank cells have no observations. i.e. underestimation of egfr in women and healthy young white males and its lower accuracy at egfr > 60 ml/min/ 1.73m 2 [12]. Levey et al. [7] reported that the use of the CKD-EPI equation indeed resulted in higher egfrs especially in individuals with egfr > 60 ml/min/1.73m 2. At first glance, our data provide support for the introduction of the CKD-EPI equation in routine clinical practice. The CKD-EPI equation can be used in patients with egfr > 60 ml/min/1.73m 2. Moreover, the use of the CKD-EPI formula would decrease the estimated prevalence of CKD in the general population when a fixed threshold of 60 ml/min/1.73m 2 is used since the lower prevalence of egfr < 60 ml/min/1.73m 2 in young people will outweigh the higher prevalence in the elderly. A similar conclusion was drawn by White et al. [13]. In a population-based survey (AusDiab), the authors found that the estimated CKD prevalence dropped from 13.1 to 11.5% when the CKD-EPI was used. They reported that the CKD prevalence decreased particularly in women but remained high in the elderly. However, our conclusions, as those in the AusDiab study, hold for data derived from a study of the general population and may not apply to clinical practice. In routine clinical practice, measurements of serum creatinine are performed in elderly more often than in young people [14] Thus, it remains unknown if the number of people with egfr < 60 ml/min/1.73m 2 that come to the attention of healthcare officers really decreases. Moreover, there is

3180 J.A.J.G. van den Brand et al. already debate on the role of age in the use of egfr as a predictor of cardiovascular outcome. The decline of egfr with age is considered by some to be a part of normal ageing [2, 4]. In elderly persons, the association between egfr and adverse outcomes is attenuated, and an egfr value between 45 and 60 ml/min/1.73m 2 is not an independent risk factor for cardiovascular events in the elderly [14, 15]. Conversely, in persons under the age of 60 years egfr values >60 ml/min/1.73m 2 are independently predictive of outcome [16]. Inconsiderate introduction of the CKD-EPI equation might have unwanted effects and may not help to settle the debate. Our data indicate that the number of elderly persons with a diagnosis of CKD will increase, whereas a relevant reduction of egfr in young persons may go unnoticed. We suggest that with introduction of the CKD-EPI formula, the classification of CKD should be redefined. A solution may be to assess the decline in GFR not just its absolute level. A rapid decline (>3 ml/min/1.73m 2 / year) in egfr is associated with significantly increased risk of cardiovascular and all-cause mortality in elderly [17]. Obviously, the use of this parameter will make classification more cumbersome. Another option may be to include urinary protein excretion. Hemmelgarn et al. [18] showed that patients with severe proteinuria but without overtly abnormal egfr had worse clinical outcomes than those with moderately reduced egfr and no proteinuria. The authors suggest that risk stratification based on egfr alone is fairly insensitive to clinically relevant gradients in risk. Urinary albumin excretion is considered another powerful marker for risk of both cardiovascular and allcause mortality. A recent meta-analysis showed that a urinary albumin/creatinine ratio >10 mg/g is associated with increased all-cause and cardiovascular mortality [19]. These findings suggest that urinary albumin or protein excretion could be used to improve the classification of CKD. Finally, one might consider to use age-dependent egfr cut-off values. Our study has some weak points. To find out if the CKD- EPI equation is representative for the elderly population, one requires measured GFR in a sufficient number of people aged >70. Our population contains a fairly large number of subjects >70 but only estimated and not measured GFR. We used a questionnaire to ascertain the health status of participants. So subjects unaware of an underlying illness may incorrectly be classified as healthy. This is particularly relevant for people with asymptomatic chronic diseases such as diabetes mellitus. Finally, our population consisted of West European Caucasian subjects, so inferences to other populations should be made with caution. Conclusion In comparison with the MDRD 4 formula, the CKD-EPI equation leads to higher estimates of GFR in young people and lower estimates in the elderly. On a population level, this will lead to higher estimates of kidney function. In routine clinical practice, this effect may be less apparent. The introduction of egfr CKD-EPI necessitates the fine-tuning of the definition of CKD. We suggest either introducing age-dependent threshold values or the use of urinary protein or albumin excretion to improve risk stratification. Supplementary data Supplementary data is available online at http://ndt. oxfordjournals.org. Acknowledgements. J.F.M.W. and J.A.J.G.B. are supported by a grant of the Dutch kidney Foundation (NSN: OW08). We thank Dick Glassock for reading the manuscript and helpful comments. Conflict of interest statement. There were no financial or other interests. References 1. National Kidney Foundation. Kidney Disease Outcomes Quality Initiative. Clinical practice guidelines for chronic kidney disease: evaluation classification, and stratification. Am J Kidney Dis 2002; 39: S1 S266 2. Glassock RJ, Winearls C. An epidemic of chronic kidney disease: fact or fiction? Nephrol Dial Transplant 2008; 23: 1117 1121 3. Eknoyan G. Chronic kidney disease definition and classification: the quest for refinements. Kidney Int 2007; 72: 1183 1185 4. O Hare AM, Choi AI, Bertenthal D et al. Age affects outcomes in chronic kidney disease. J Am Soc Nephrol 2007; 18: 2758 2765 5. Wetzels JF, Kiemeney LA, Swinkels DW et al. Age- and genderspecific reference values of estimated GFR in Caucasians: the Nijmegen Biomedical Study. Kidney Int 2007; 72: 632 637 6. O Hare AM, Bertenthal D, Covinsky KE et al. Mortality risk stratification in chronic kidney disease: one size for all ages? J Am Soc Nephrol 2006; 17: 846 853 7. Levey AS, Stevens LA, Schmid CH et al. A new equation to estimate glomerular filtration rate. Ann Intern Med 2009; 150: 604 612 8. Matsushita K, Selvin E, Bash LD et al. Risk implications of the new CKD Epidemiology Collaboration (CKD-EPI) equation compared with the MDRD Study equation for estimated GFR: the Atherosclerosis Risk in Communities (ARIC) Study. Am J Kidney Dis 2010; 55: 648 659 9. Hoogendoorn EH, Hermus AR, de Vegt F et al. Thyroid function and prevalence of anti-thyroperoxidase antibodies in a population with borderline sufficient iodine intake: influences of age and sex. Clin Chem 2006; 52: 104 111 10. Wetzels JF, Willems HL, den Heijer M. Age- and gender-specific reference values of estimated glomerular filtration rate in a Caucasian population: Results of the Nijmegen Biomedical Study. Kidney Int 2008; 73: 657 658 11. Cole TJ, Green PJ. Smoothing reference centile curves: the LMS method and penalized likelihood. Stat Med 1992; 11: 1305 1319 12. Stevens LA, Coresh J, Feldman HI et al. Evaluation of the modification of diet in renal disease study equation in a large diverse population. J Am Soc Nephrol 2007; 18: 2749 2757 13. White SL, Polkinghorne KR, Atkins RC et al. Comparison of the prevalence and mortality risk of CKD in Australia using the CKD Epidemiology Collaboration (CKD-EPI) and Modification of Diet in Renal Disease (MDRD) Study GFR estimating equations: the Aus- Diab (Australian Diabetes, Obesity and Lifestyle) Study. Am J Kidney Dis 2010; 55: 660 670 14. Raymond NT, Zehnder D, Smith SC et al. Elevated relative mortality risk with mild-to-moderate chronic kidney disease decreases with age. Nephrol Dial Transplant 2007; 22: 3214 3220 15. Roderick PJ, Atkins RJ, Smeeth L et al. CKD and mortality risk in older people: a community-based population study in the United Kingdom. Am J Kidney Dis 2009; 53: 950 960 16. van der Velde M, Bakker SJ, de Jong PE et al. Influence of age and measure of egfr on the association between renal function and cardiovascular events. Clin J Am Soc Nephrol 2010; 5: 2053 2059

CKD-EPI and its effect on CKD prevalence 3181 17. Shlipak MG, Katz R, Kestenbaum B et al. Rapid decline of kidney function increases cardiovascular risk in the elderly. J Am Soc Nephrol 2009; 20: 2625 2630 18. Hemmelgarn BR, Manns BJ, Lloyd A et al. Relation between kidney function, proteinuria, and adverse outcomes. JAMA 2010; 303: 423 429 19. Chronic Kidney Disease Prognosis Consortium, Matsushita K, van der Velde M et al. Association of estimated glomerular filtration rate and albuminuria with all-cause and cardiovascular mortality in general population cohorts: a collaborative meta-analysis. Lancet 2010; 375: 2073 2081 Received for publication: 23.8.10; Accepted in revised form: 3.1.11