0
Reviews |

Estimating Equations for Glomerular Filtration Rate in the Era of Creatinine Standardization: A Systematic Review FREE

Amy Earley, BS; Dana Miskulin, MD, MS; Edmund J. Lamb, PhD; Andrew S. Levey, MD; and Katrin Uhlig, MD, MS
[+] Article and Author Information

From Tufts Medical Center, Boston, Massachusetts, and Kent and Canterbury Hospital, Canterbury, Kent, United Kingdom.

Grant Support: By KDIGO.

Potential Conflicts of Interest: Ms. Earley and Drs. Miskulin and Uhlig report the following: Grant (money to institution): National Kidney Foundation; Support for travel to meetings for the study or other purposes: National Kidney Foundation. Ms. Earley and Dr. Uhlig further report: Fees for participation in review activities such as data monitoring boards, statistical analysis, end point committees, and the like (money to institution): National Kidney Foundation. Dr. Miskulin further reports: Grant: National Kidney Foundation; Employment: Dialysis Clinic. Dr. Lamb: Support for travel to meetings for the study or other purposes: Kidney Disease: Improving Global Outcomes; Grants/grants pending (money to institution): National Institutes of Health. Dr. Levey: Support for travel to meetings for the study or other purposes (money to institution): National Kidney Foundation; Board membership (money to institution): National Kidney Foundation; Grants/grants pending (money to institution): National Kidney Foundation. Disclosures can also be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M11-2267.

Requests for Single Reprints: Katrin Uhlig, MD, MS, Tufts Medical Center, 800 Washington Street, Box 391, Boston, MA 02111; e-mail, kuhlig@tuftsmedicalcenter.org.

Current Author Addresses: Ms. Earley and Drs. Miskulin, Levey, and Uhlig: Tufts Medical Center, 800 Washington Street, Box 391, Boston, MA 02111.

Dr. Lamb: East Kent Hospitals University National Health Service Trust, Kent and Canterbury Hospital, Ethelbert Road, Canterbury, Kent CT1 3NG, United Kingdom.

Author Contributions: Conception and design: A. Earley, E.J. Lamb, A.S. Levey, K. Uhlig.

Analysis and interpretation of the data: A. Earley, D. Miskulin, E.J. Lamb, A.S. Levey, K. Uhlig.

Drafting of the article: A. Earley, D. Miskulin, E.J. Lamb, A.S. Levey, K. Uhlig.

Critical revision of the article for important intellectual content: A. Earley, D. Miskulin, E.J. Lamb, A.S. Levey, K. Uhlig.

Final approval of the article: A. Earley, E.J. Lamb, A.S. Levey, K. Uhlig.

Statistical expertise: K. Uhlig.

Obtaining of funding: K. Uhlig.

Administrative, technical, or logistic support: A. Earley, E.J. Lamb, K. Uhlig.

Collection and assembly of data: A. Earley, D. Miskulin, E.J. Lamb, A.S. Levey, K. Uhlig.


Ann Intern Med. 2012;156(11):785-795. doi:10.7326/0003-4819-156-11-201203200-00391
Text Size: A A A

Background: Clinical laboratories are increasingly reporting estimated glomerular filtration rate (GFR) by using serum creatinine assays traceable to a standard reference material.

Purpose: To review the performance of GFR estimating equations to inform the selection of a single equation by laboratories and the interpretation of estimated GFR by clinicians.

Data Sources: A systematic search of MEDLINE, without language restriction, between 1999 and 21 October 2011.

Study Selection: Cross-sectional studies in adults that compared the performance of 2 or more creatinine-based GFR estimating equations with a reference GFR measurement. Eligible equations were derived or reexpressed and validated by using creatinine measurements traceable to the standard reference material.

Data Extraction: Reviewers extracted data on study population characteristics, measured GFR, creatinine assay, and equation performance.

Data Synthesis: Eligible studies compared the MDRD (Modification of Diet in Renal Disease) Study and CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equations or modifications thereof. In 12 studies in North America, Europe, and Australia, the CKD-EPI equation performed better at higher GFRs (approximately >60 mL/min per 1.73 m2) and the MDRD Study equation performed better at lower GFRs. In 5 of 8 studies in Asia and Africa, the equations were modified to improve their performance by adding a coefficient derived in the local population or removing a coefficient.

Limitation: Methods of GFR measurement and study populations were heterogeneous.

Conclusion: Neither the CKD-EPI nor the MDRD Study equation is optimal for all populations and GFR ranges. Using a single equation for reporting requires a tradeoff to optimize performance at either higher or lower GFR ranges. A general practice and public health perspective favors the CKD-EPI equation.

Primary Funding Source: Kidney Disease: Improving Global Outcomes.

Editor's Note
Context

  • Multiple methods are used to estimate glomerular filtration rate (GFR) from serum creatinine level.

Contribution

  • This review summarized data from cross-sectional studies that compared 2 or more creatinine-based GFR estimating equations to a reference GFR measurement. Studies from North America, Europe, and Australia showed that the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation performed better at higher GFRs and the Modification of Diet in Renal Disease (MDRD) Study equation performed better at lower GFRs. Neither equation performed as well in Asian or African populations as it did in North American or European populations.

Implication

  • The performance of the CKD-EPI and MDRD Study equations varies across populations and GFR ranges.

—The Editors


Estimates of glomerular filtration rate (GFR) from serum creatinine levels are now reported by more than 80% of clinical laboratories in the United States (1). Accurate estimation of GFR is important for detecting and staging chronic kidney disease (CKD), determining drug dosages, and stratifying risk (2). The equation from the MDRD (Modification of Diet in Renal Disease) Study is most frequently used but is known to be less accurate at higher GFRs and in racial and ethnic groups outside of North America, Europe, and Australia. Thus, researchers have developed and validated other GFR estimating equations to overcome these limitations.

The increasing use of GFR estimating equations has led to an appreciation of the effect of differences in creatinine assays on the accuracy of GFR estimates (34). In 2006, a serum matrix standard reference material (SRM) was prepared by the National Institute of Standards and Technology and submitted to the Joint Committee for Traceability in Laboratory Medicine (5). Use of this material, in combination with the isotope-dilution mass spectrometry reference method, was intended to assist reagent manufacturers in achieving better consensus among methods (6). By the end of 2009, the calibration of most clinical laboratory methods was traceable to the SRM and isotope-dilution mass spectrometry (1).

Our goal was to systematically review the performance of creatinine-based GFR estimating equations for use with standardized serum creatinine measurements to inform selection of a single equation that laboratories could use to estimate GFR and to help clinicians interpret estimated GFR.

Data Sources and Searches

We conducted a systematic search in MEDLINE from 1999 (the year before the 4-variable MDRD Study equation was published [7]) to 21 October 2011, with no language restrictions. Because the MDRD Study equation was the first equation to be reexpressed for use with standardized creatinine measurements (8), our search allowed us to identify other equations based on the MDRD Study equation that could also be reexpressed for use with standardized creatinine measurements. We also supplemented our search with references from 2 previous reviews (910). Appendix Table 1 lists our keywords and search strategy.

Table Jump PlaceholderAppendix Table 1.  

Search Strategy

Study Selection

We included cross-sectional studies in any adult clinical or research population comparing GFR estimates from at least 2 creatinine-based estimating equations with a reference method of GFR measurement. Acceptable reference methods for measuring GFR in the development and validation populations were the urinary or plasma clearance of an exogenous filtration marker. We included equations that were originally developed by using standardized serum creatinine measurements or those for which the coefficients were subsequently recalculated for use with standardized creatinine measurements, which we refer to as reexpressed. We also included 1 study for which a conversion factor could be applied post hoc (11). We required evaluation of the equation in a data set that was external to the one in which it was developed, with independent sampling of the development and validation populations. Creatinine assays had to be traceable to the SRM (5). Acceptable assay methods included those standardized against isotope-dilution mass spectrometry of the reference material of the National Institute of Standards of Technology. Studies were also acceptable if they used a conversion factor derived from calibration of samples across the range of creatinine concentrations represented in the study population, with methods traceable to isotope-dilution mass spectrometry. We excluded studies that used an assay that was not traceable to the SRM and those with unclear traceability. In some cases, these studies were developed with assays and analytic material that are no longer available; the performance of the equations described in such studies was considered to be irrelevant to current practice. The minimum size was arbitrarily set at 100. We also looked for results of performance in subgroups by GFR, age, and race that had at least 50 members.

Data Extraction and Quality Assessment

We extracted data on study population characteristics, reference standards, measured GFRs, methods of creatinine calibration, and equation performance (bias, precision, and accuracy). We noted whether participants had conditions that could affect serum creatinine level through non-GFR determinants, such as conditions that alter creatinine generation (chronic illness or steroids), or medications that inhibit creatinine secretion (trimethoprim). Data from each article were systematically extracted by 1 of the authors. Other authors reviewed the methods section of studies describing development and evaluation of an equation to identify the creatinine assays and to confirm the study design and results. Our stringent selection criteria ensured that all reviewed studies were of good methodological quality.

Statistical Analysis

Summary measures of the differences between measured and estimated GFR were compared for each equation. No single metric captures all of the important information for evaluating performance of GFR estimating equations (10). In general, measures on the raw scale tend to emphasize errors at higher GFRs, whereas measures on the percentage or log scale tend to emphasize errors at lower GFRs (10). Bias, an expression of systemic error in estimated GFR, is defined as the median or mean of the differences between estimated and measured GFR. We used bias on the raw scale because it is easier to interpret. Precision is an expression of the random variation or “spread” of estimated GFR values around the measured GFR. The root mean square error of the regression of estimated GFR versus measured GFR, or log estimated GFR versus log measured GFR, is considered to be a direct measure of precision. Indirect measures, such as interquartile range or SD for the differences between estimated GFR and measured GFR, were included if the root mean square error was not provided. Accuracy is affected by both bias and imprecision and was expressed as the percentage of estimated GFR values within 30% of measured GFR (P30). This measure of P30 was first reported as a measure of the accuracy of the MDRD Study equation (12) and was subsequently recommended by the National Kidney Foundation Disease Outcomes Quality Initiative in its CKD guidelines (13). An error of 30% is considered large; for example, a 30% error in a patient with measured GFR of 50 mL/min per 1.73 m2 could lead to an estimated GFR as low as 35 or as high as 65 mL/min per 1.73 m2. Some have suggested that 20% would be a more appropriate threshold for a large error. We have included alternative values for accuracy (such as P10, P15, or P20) if they were reported. For our conclusions, we focused primarily on P30 because it is a measure of large errors that would be important to clinicians, is less influenced by small bias due to differences in measurement methods or regression to the mean, and was most consistently reported across studies.

We tabulated results separately for adult populations of largely Northern European ancestry (North American, European, and Australian populations) and those from other locations and separated the results for subgroups by GFR range, race or ethnicity, and age. Within each table, we organized the studies first by similar reference measurement method and then by size.

Role of the Funding Source

The authors are members of the evidence review team and 2 workgroup experts of the ongoing Kidney Disease: Improving Global Outcomes (KDIGO) guideline on evaluation and management of CKD. The evidence review team was supported by KDIGO to conduct systematic reviews and provide methods support. The judgments and interpretations in the article are those of the authors. The funding source did not participate in the design, conduct, or reporting of the study.

Our search yielded 3250 abstracts; of these, 100 articles were reviewed in full text and 23 met our inclusion criteria (Figure 1). The main reason for exclusion was that the equations had not been developed or reexpressed for use with creatinine assays traceable to the SRM. Appendix Table 2 lists equations that met our criteria, and Appendix Table 3 lists the equations that were not traceable to the SRM.

Grahic Jump Location
Figure 1.

Summary of evidence search and selection.

SCr = serum creatinine; SRM = standard reference material.

Grahic Jump Location
Table Jump PlaceholderAppendix Table 2.  

Information on Development of Equations Based on Serum Creatinine Assays That Are Traceable to the Standard Reference Material

Table Jump PlaceholderAppendix Table 3.  

Overview Table of Equations Developed to Predict GFR Based on Serum Creatinine Assays Not Traceable to the Standard Reference Material

Estimating Equations Developed in Adult Populations in North America, Europe, or Australia

Twelve studies, comprising 12 898 patients, met our criteria (Table 1) (1426). Study populations were the general population in 3 studies (1417), kidney transplant recipients in 3 studies (1820), individuals before kidney donation in 1 study (21) and before and after kidney donation in another study (22), patients with cancer in 1 study (23), and a heterogeneous population in 3 studies (2426). All studies compared the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation with the MDRD Study equation. Methods of measuring GFR included clearance of iothalamate in 5 studies, technetium–diethylenetriamine pentaacetic acid (Tc-DTPA) in 3 studies, inulin in 1 study, iohexol in 1 study, chromium–ethylenediamine tetraacetic acid (Cr-EDTA) in 1 study, and various markers in 1 study.

Table Jump PlaceholderTable 1.  

Performance Comparison of Creatinine-Based GFR Estimating Equations in North America, Europe, and Australia

Across the 12 studies, P30 ranged from 59% to 95%. The CKD-EPI equation was more accurate than the MDRD Study equation in 10 studies and less accurate in 2 studies. Bias ranged from 14.6 to −22 mL/min per 1.73 m2. The CKD-EPI equation was less biased than the MDRD Study equation in 7 studies and more biased in 5 studies. Two studies reported the root mean square error as a measure of precision. In 6 of the 10 studies that reported a measure of precision, the CKD-EPI equation was more precise than the MDRD Study equation; precision for the MDRD Study equation was better or the same in the other 4 studies. In 5 studies (1417, 2526), performance measures were consistently better for the CKD-EPI equation than for the MDRD Study equation. In 2 studies (1819), both of which were conducted in kidney transplant recipients, performance measures were better for the MDRD Study equation. In 1 study (18), all patients received trimethoprim.

Figure 2 shows differences in accuracy and bias according to measured GFR and GFR measurement method. The CKD-EPI equation seems to be more accurate and less biased in studies with higher mean measured GFRs (approximately >60 mL/min per 1.73 m2), whereas the MDRD Study equation has greater accuracy and less bias at lower GFRs. No clear pattern by GFR measurement method was observed. Appendix Table 4 shows subgroups stratified by GFR or clinical characteristics in studies in which these subgroup data were reported (1920, 2526). Within each study, the differences in accuracy and bias were larger at higher GFRs and smaller at lower GFRs.

Grahic Jump Location
Figure 2.

Differences in accuracy and bias between estimated GFR by CKD-EPI and MDRD Study equations in North America, Europe, and Australia.

Difference in accuracy (top), as measured by P30 (P30 for CKD-EPI minus P30 for MDRD), is plotted against mean measured GFR in the study population. Difference in bias (bottom) (absolute value for bias for estimated GFR by MDRD Study equation minus absolute value for bias for estimated GFR by CKD-EPI equation) is plotted against mean measured GFR in the study population. CKD-EPI = Chronic Kidney Disease Epidemiology Collaboration; GFR = glomerular filtration rate; MDRD = Modification of Diet in Renal Disease; P30 = percentage of estimated GFR values within 30% of measured GFR.

* Denotes study in which all patients received trimethoprim.

† Could be reported in mL/min.

Grahic Jump Location
Table Jump PlaceholderAppendix Table 4.  

Performance of Creatinine-Based GFR Estimating Equations in North America, Europe, and Australia in Subgroups by GFR

One study (27) compared a modification of the CKD-EPI equation that used a 4-level race or ethnicity coefficient (black, Asian, and Native American or Hispanic vs. white and other) with a modification that used a 2-level coefficient (black vs. white and other) in racial or ethnic groups in North America and Europe (Appendix Table 5). The development data set had few Asian or Native American or Hispanic patients, and the improvements in bias and precision in these groups were not clinically meaningful. Only 2 studies (26, 28) provided results by age subgroups. In both studies, the differences in bias between equations were generally smaller in subgroups of older patients (data not shown).

Table Jump PlaceholderAppendix Table 5.  

Performance of Creatinine-Based GFR Estimating Equations in North America, Europe, and Australia in Subgroups by Race

Estimating Equations Developed in Adult Populations Outside North America, Europe, or Australia

Eight studies met our criteria (Table 2) (2934). The study populations were from Asia and Africa, including Japanese, Chinese, Thai, Korean, South African black, or multiethnic Singapore populations and included general population in 3 studies (27, 2930), patients with CKD in 3 studies (11, 3132), kidney transplant recipients in 1 study (33), and a heterogeneous population in 1 study (34). All studies examined the MDRD Study and CKD-EPI equations or modifications thereof. In 6 studies (11, 2831, 34), the MDRD Study or CKD-EPI equations were modified by adding or removing a coefficient to improve the performance of the equation in the development data set or a new equation was developed by using the same variables (Appendix Table 2). Methods for measuring GFR included Tc-DTPA in 3 studies, inulin in 2 studies, Cr-EDTA in 2 studies, and iothalamate in 1 study.

Table Jump PlaceholderTable 2.  

Performance Comparison of Creatinine-Based GFR Estimating Equations Outside of North America, Europe, and Australia

Comparing the performance of equations across studies is limited because locally derived equations were generally not tested in other studies or populations. In these studies, the unmodified MDRD Study and CKD-EPI equations were less accurate (P30 ranging from 29% to 94%) than in the studies of populations in North America, Europe, and Australia. In 5 studies, adding (11, 29, 31, 34) or removing (30) a coefficient improved the accuracy of GFR estimation (11, 2931, 34). However, in 1 study (27), modifying the CKD-EPI equation by substituting a 4-level race or ethnicity coefficient derived in a North American and European population for the 2-level coefficient did not significantly improve accuracy in Chinese, Japanese, or South African black populations. Coefficients developed in 1 study of an ethnic or racial population did not improve equation accuracy in a study of another ethnic or racial population (30, 33). The Japanese and Chinese coefficients also vary widely (Appendix Table 2) (29, 34). In 3 studies that compared the CKD-EPI and MDRD Study equations (29, 3132) (with or without modification), the CKD-EPI equation was more accurate.

Three studies examined the performance of estimating equations by GFR strata (Appendix Table 6) (2930, 32). In 2 studies, the differences in accuracy and bias were larger at higher GFRs and smaller at lower GFRs. In 1 study (29), the differences in bias between equations were smaller in older subgroups than in younger subgroups (data not shown).

Table Jump PlaceholderAppendix Table 6.  

Performance of Creatinine-Based GFR Estimating Equations Outside North America, Europe, and Australia in Subgroups by GFR

Only the MDRD Study and CKD-EPI equations and modifications thereof have been expressed for use with creatinine assays traceable to the SRM and concurrently compared with measured GFR in adults. In studies of North American, European, or Australian populations, the CKD-EPI equation was more accurate (had a higher P30) than the MDRD Study equation in 10 of 12 studies. The comparison of bias was more variable. For both accuracy and bias, the CKD-EPI equation performed better at higher GFRs (approximately >60 mL/min per 1.73 m2) and the MDRD Study equation performed better at lower GFRs. Within studies, the differences in bias were greater at higher GFRs than at lower GFRs.

Data on adults outside of North America, Europe, and Australia are more limited. Neither the MDRD Study nor the CKD-EPI equation performs as well in these locations as it does in North America and Europe. Equation performance can be improved by deriving local “race/ethnicity” coefficients; however, the modified or new equations generally do not exhibit the same level of accuracy as the CKD-EPI or MDRD Study equations do in North American, European, or Australian populations. The coefficients also do not seem to be generalizable beyond the local population, possibly because of differences in GFR measurement methods or differences in populations in addition to race or ethnicity (35).

Despite estimating GFR from the same variables (age, sex, race, and serum creatinine level), the CKD-EPI equation generally yields higher values for estimated GFR than does the MDRD Study equation. The difference in equation performance can be explained by differences in the development populations. The MDRD Study equation was developed in a study population with CKD and a mean GFR of 40 mL/min per 1.73 m2, whereas the CKD-EPI equation was developed in a more diverse study population, including participants with and without CKD, with a mean GFR of 68 mL/min per 1.73 m2. Our observation of differences in equation performance in various GFR ranges probably reflects differences in non-GFR determinants of creatinine (in particular, creatinine generation due to muscle mass and diet) and regression to the mean. Our observation of larger differences in estimated GFR between the equations at higher GFRs probably reflects equation development on the logarithmic scale.

Other differences in equation performance can be explained by differences between the development and the validation populations. The CKD-EPI and MDRD Study equations were developed in North American and European populations that mainly consisted of black and white persons. Our observation that both equations perform less well in other racial and ethnic groups is consistent with known racial and ethnic differences in muscle mass and diet (35). No participants were receiving trimethoprim in either development population. Our observation of marked underestimation of GFR with both equations in persons receiving trimethoprim is consistent with the known inhibition of creatinine secretion by this drug (36).

Both the MDRD Study and CKD-EPI equations were developed by using iothalamate clearance as the reference standard for GFR. The validation studies used various filtration markers, with small differences in clearance compared with iothalamate (37). Use of filtration markers other than iothalamate in validation studies of these equations would introduce a systematic bias. The observation of variation in differences in equation performance among studies with similar mean GFRs may be due in part to differences in GFR measurement method.

The observed and expected differences in performance by range of GFR suggest that we cannot optimize the performance of any equation for all clinical populations across a wide range of GFRs. Because the goal is to select a single estimating equation for routine use by clinical laboratories, the tradeoff of optimizing performance at either higher or lower GFR ranges must be accepted. Because the difference in bias (on the raw scale) between the equations is greater at higher GFRs, using the CKD-EPI equation would lead to smaller average bias in clinical populations with a wide range of GFRs.

As in diagnostic test evaluation, the magnitude of differences in performance between the CKD-EPI and MDRD Study equations are best appreciated by considering the implications for clinical practice. Reporting estimated GFR by using the MDRD Study equation is widespread, so a change in the estimating equation used by clinical laboratories would have profound implications. Because the differences between the equations are greater at higher GFRs, the implications of introducing the CKD-EPI equation would be larger for public health and general clinical practice than for nephrology practices. Applying the CKD-EPI rather than the MDRD Study equation to the U.S. adult population would lead to a higher average estimated GFR, less sensitivity but more specificity for detecting a GFR less than 60 mL/min per 1.73 m2, and a lower prevalence estimate but a higher risk profile for persons with an estimated GFR in this range (25, 3840). This would potentially enable more efficient use of resources in caring for patients with decreased estimated GFR. Another consequence of using the CKD-EPI equation would be to allow reporting of estimated GFR as a numerical value throughout the full range, rather than limit it to lower values (for example, <60 mL/min per 1.73 m2, as currently recommended for the MDRD Study equation in the United States) (1). However, using the CKD-EPI equation would slightly increase bias (overestimation) at lower GFRs. Nephrologists and others caring for patients with low GFR would need to be aware of this limitation.

Estimation methods need to be further improved. Even in North America, Europe, and Australia, the CKD-EPI equation does not meet the 2002 KDOQI benchmark of P30 greater than 90% (13). This level of performance may be beyond expectation for creatinine-based equations. Standardization has reduced but not eliminated bias due to differences in creatinine assays, and differences among GFR measurement methods remain another source of bias. Imprecision in estimating equations is quantitatively more important, probably because of variation in non-GFR determinants of serum creatinine level and imprecision in measured GFR. The former reflects the large number of factors affecting creatinine generation and reducing imprecision may require including additional variables, such as measures of muscle mass or diet; the latter reflects the fallibility of the reference standard rather than errors in the estimating equations and can be improved only by more precise measures of GFR. Although urinary clearance of inulin is the gold standard, it is difficult to use, and studies are needed to compare the bias and precision of alternative methods to enable calibration of GFR measurement methods. In addition, statistical adjustment for measurement error may allow more accurate evaluation of estimated GFR (41).

Outside North America, Europe, and Australia, bias remains an issue because of systematic variation in differences in creatinine generation. Incorporating locally derived coefficients can minimize this bias. However, the lack of generalizability of these coefficients probably reflects the contributions of multiple factors that differ across populations in various locations. Developing or reexpressing existing equations for local populations requires significant resources and may not be feasible in all locations.

The accuracy and worldwide generalizability of GFR estimating equations might be improved by using such alternative filtration markers as cystatin C, which is less dependent on muscle mass. However, all filtration markers have non-GFR determinants, so it is unlikely that imprecision will be eliminated by any single marker. Using multiple markers can improve precision by minimizing the contribution of any 1 non-GFR determinant (42). In June 2010, the Institute for Reference Materials and Measurements released a reference material for cystatin C measurement (4345). Reagent manufacturers are in the process of recalibrating their assays against this standard, and estimating equations for use with standardized cystatin C measurements are being evaluated (46).

A strength of our review is that the stringent selection criteria eliminated older and smaller studies. This generated an evidence base of good-quality studies and makes the conclusions directly relevant to the current era of creatinine assays. Our review also highlights the shortcomings of the existing literature. We propose criteria for studies that are developing and validating GFR estimating equations (Table 3) to enhance the quality and comparability of future studies.

Table Jump PlaceholderTable 3.  

Suggested Criteria for Developing and Validating GFR Estimating Equations

Our review has limitations. It was restricted to studies that compared at least 2 equations. Although this omits studies that developed and validated a single GFR estimating equation, it probably did not alter our conclusions because such equations are unlikely to have been developed in large populations and tested as widely as the equations that we included. We may also have inadvertently excluded studies that did use creatinine assays standardized against SRM. This emphasizes the importance of our recommendation that investigators provide full information about their methods.

In summary, neither the CKD-EPI nor the MDRD Study equation is optimal across all populations and GFR ranges. Using a single equation for reporting estimated GFR requires a tradeoff to optimize performance at either higher or lower GFR ranges. A general practice and public health perspective favors adopting the CKD-EPI equation in North America, Europe, and Australia and using it as a comparator for new equations in all locations. Whether the precision of creatinine-based equations can be substantially improved without adding other variables is uncertain. Equations using other filtration markers instead of or in addition to creatinine hold promise in this regard. Going forward, new equations should be rigorously developed, validated, and reported.

Miller WG.  Estimating glomerular filtration rate [Editorial]. Clin Chem Lab Med. 2009; 47:1017-9.
PubMed
CrossRef
 
Stevens LA, Coresh J, Greene T, Levey AS.  Assessing kidney function—measured and estimated glomerular filtration rate. N Engl J Med. 2006; 354:2473-83.
PubMed
CrossRef
 
Coresh J, Astor BC, McQuillan G, Kusek J, Greene T, Van Lente F, et al..  Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate. Am J Kidney Dis. 2002; 39:920-9.
PubMed
CrossRef
 
Vickery S, Stevens PE, Dalton RN, van Lente F, Lamb EJ.  Does the ID-MS traceable MDRD equation work and is it suitable for use with compensated Jaffe and enzymatic creatinine assays? Nephrol Dial Transplant. 2006; 21:2439-45.
PubMed
CrossRef
 
Myers GL, Miller WG, Coresh J, Fleming J, Greenberg N, Greene T, et al., National Kidney Disease Education Program Laboratory Working Group.  Recommendations for improving serum creatinine measurement: a report from the Laboratory Working Group of the National Kidney Disease Education Program. Clin Chem. 2006; 52:5-18.
PubMed
CrossRef
 
Panteghini M, Myers GL, Miller WG, Greenberg N, International Federation of Clinical Chemistry and Laboratory Medicine.  The importance of metrological traceability on the validity of creatinine measurement as an index of renal function. Clin Chem Lab Med. 2006; 44:1287-92.
PubMed
CrossRef
 
Levey AS, Greene T, Kusek J, Beck G.  A simplified equation to predict glomerular filtration rate from serum creatinine [Abstract]. J Am Soc Nephrol. 2000; 11:155A.
 
Levey AS, Coresh J, Greene T, Stevens LA, Zhang YL, Hendriksen S, et al., Chronic Kidney Disease Epidemiology Collaboration.  Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med. 2006; 145:247-54.
PubMed
 
Coresh J, Stevens LA.  Kidney function estimating equations: where do we stand? Curr Opin Nephrol Hypertens. 2006; 15:276-84.
PubMed
CrossRef
 
Stevens LA, Zhang Y, Schmid CH.  Evaluating the performance of equations for estimating glomerular filtration rate. J Nephrol. 2008; 21:797-807.
PubMed
 
Ma YC, Zuo L, Chen JH, Luo Q, Yu XQ, Li Y, et al..  Modified glomerular filtration rate estimating equation for Chinese patients with chronic kidney disease. J Am Soc Nephrol. 2006; 17:2937-44.
PubMed
CrossRef
 
Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D.  A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999; 130:461-70.
PubMed
 
National Kidney Foundation.  K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis. 2002; 39:S1-266.
PubMed
 
Cirillo M, Lombardi C, Luciano MG, Bilancio G, Anastasio P, De Santo NG.  Estimation of GFR: a comparison of new and established equations [Letter]. Am J Kidney Dis. 2010; 56:802-4.
PubMed
CrossRef
 
Jones GR, Imam SK.  Validation of the revised MDRD formula and the original Cockcroft and Gault formula for estimation of the glomerular filtration rate using Australian data. Pathology. 2009; 41:379-82.
PubMed
 
Jones GR.  Use of the CKD-EPI equation for estimation of GFR in an Australian cohort [Letter]. Pathology. 2010; 42:487-8.
PubMed
CrossRef
 
Michels WM, Grootendorst DC, Verduijn M, Elliott EG, Dekker FW, Krediet RT.  Performance of the Cockcroft–Gault, MDRD, and new CKD-EPI formulas in relation to GFR, age, and body size. Clin J Am Soc Nephrol. 2010; 5:1003-9.
PubMed
CrossRef
 
Kukla A, El-Shahawi Y, Leister E, Kasiske B, Mauer M, Matas A, et al..  GFR-estimating models in kidney transplant recipients on a steroid-free regimen. Nephrol Dial Transplant. 2010; 25:1653-61.
PubMed
CrossRef
 
Pöge U, Gerhardt T, Stoffel-Wagner B, Sauerbruch T, Woitas RP.  Validation of the CKD-EPI formula in patients after renal transplantation. Nephrol Dial Transplant. 2011; 26:4104-8.
PubMed
 
White CA, Akbari A, Doucette S, Fergusson D, Knoll GA.  Estimating glomerular filtration rate in kidney transplantation: is the new chronic kidney disease epidemiology collaboration equation any better? Clin Chem. 2010; 56:474-7.
PubMed
CrossRef
 
Lane BR, Demirjian S, Weight CJ, Larson BT, Poggio ED, Campbell SC.  Performance of the chronic kidney disease-epidemiology study equations for estimating glomerular filtration rate before and after nephrectomy. J Urol. 2010; 183:896-901.
PubMed
 
Tent H, Rook M, Stevens LA, van Son WJ, van Pelt LJ, Hofker HS, et al..  Renal function equations before and after living kidney donation: a within-individual comparison of performance at different levels of renal function. Clin J Am Soc Nephrol. 2010; 5:1960-8.
PubMed
CrossRef
 
Redal-Baigorri B, Stokholm KH, Rasmussen K, Jeppesen N.  Estimation of kidney function in cancer patients. Dan Med Bull. 2011; 58:A4236.
PubMed
 
Eriksen BO, Mathisen UD, Melsom T, Ingebretsen OC, Jenssen TG, Njølstad I, et al..  Cystatin C is not a better estimator of GFR than plasma creatinine in the general population. Kidney Int. 2010; 78:1305-11.
PubMed
CrossRef
 
Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, et al., CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration).  A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009; 150:604-12.
PubMed
 
Murata K, Baumann , Saenger AK, Larson TS, Rule AD, Lieske JC.  Relative performance of the MDRD and CKD-EPI equations for estimating glomerular filtration rate among patients with varied clinical presentations. Clin J Am Soc Nephrol. 2011; 6:1963-72.
PubMed
CrossRef
 
Stevens LA, Claybon MA, Schmid CH, Chen J, Horio M, Imai E, et al..  Evaluation of the Chronic Kidney Disease Epidemiology Collaboration equation for estimating the glomerular filtration rate in multiple ethnicities. Kidney Int. 2011; 79:555-62.
PubMed
CrossRef
 
Stevens LA, Schmid CH, Greene T, Zhang YL, Beck GJ, Froissart M, et al..  Comparative performance of the CKD Epidemiology Collaboration (CKD-EPI) and the Modification of Diet in Renal Disease (MDRD) Study equations for estimating GFR levels above 60 mL/min/1.73 m2. Am J Kidney Dis. 2010; 56:486-95.
PubMed
CrossRef
 
Horio M, Imai E, Yasuda Y, Watanabe T, Matsuo S.  Modification of the CKD epidemiology collaboration (CKD-EPI) equation for Japanese: accuracy and use for population estimates. Am J Kidney Dis. 2010; 56:32-8.
PubMed
CrossRef
 
van Deventer HE, George JA, Paiker JE, Becker PJ, Katz IJ.  Estimating glomerular filtration rate in black South Africans by use of the modification of diet in renal disease and Cockcroft–Gault equations. Clin Chem. 2008; 54:1197-202.
PubMed
CrossRef
 
Praditpornsilpa K, Townamchai N, Chaiwatanarat T, Tiranathanagul K, Katawatin P, Susantitaphong P, et al..  The need for robust validation for MDRD-based glomerular filtration rate estimation in various CKD populations. Nephrol Dial Transplant. 2011; 26:2780-5.
PubMed
CrossRef
 
Teo BW, Xu H, Wang D, Li J, Sinha AK, Shuter B, et al..  GFR estimating equations in a multiethnic Asian population. Am J Kidney Dis. 2011; 58:56-63.
PubMed
CrossRef
 
Yeo Y, Han DJ, Moon DH, Park JS, Yang WS, Chang JW, et al..  Suitability of the IDMS-traceable MDRD equation method to estimate GFR in early postoperative renal transplant recipients. Nephron Clin Pract. 2010; 114:c108-17.
PubMed
CrossRef
 
Matsuo S, Imai E, Horio M, Yasuda Y, Tomita K, Nitta K, et al., Collaborators developing the Japanese equation for estimated GFR.  Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis. 2009; 53:982-92.
PubMed
CrossRef
 
Rule AD, Teo BW.  GFR estimation in Japan and China: what accounts for the difference? [Editorial]. Am J Kidney Dis. 2009; 53:932-5.
PubMed
CrossRef
 
Berglund F, Killander J, Pompeius R.  Effect of trimethoprim-sulfamethoxazole on the renal excretion of creatinine in man. J Urol. 1975; 114:802-8.
PubMed
 
Stevens LA, Levey AS.  Measured GFR as a confirmatory test for estimated GFR. J Am Soc Nephrol. 2009; 20:2305-13.
PubMed
CrossRef
 
Matsushita K, Selvin E, Bash LD, Astor BC, Coresh J.  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; 55:648-59.
PubMed
CrossRef
 
Skali H, Uno H, Levey AS, Inker LA, Pfeffer MA, Solomon SD.  Prognostic assessment of estimated glomerular filtration rate by the new Chronic Kidney Disease Epidemiology Collaboration equation in comparison with the Modification of Diet in Renal Disease Study equation. Am Heart J. 2011; 162:548-54.
PubMed
CrossRef
 
White SL, Polkinghorne KR, Atkins RC, Chadban SJ.  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 AusDiab (Australian Diabetes, Obesity and Lifestyle) Study. Am J Kidney Dis. 2010; 55:660-70.
PubMed
CrossRef
 
Kwong YT, Stevens LA, Selvin E, Zhang YL, Greene T, Van Lente F, et al..  Imprecision of urinary iothalamate clearance as a gold-standard measure of GFR decreases the diagnostic accuracy of kidney function estimating equations. Am J Kidney Dis. 2010; 56:39-49.
PubMed
CrossRef
 
Stevens LA, Coresh J, Schmid CH, Feldman HI, Froissart M, Kusek J, et al..  Estimating GFR using serum cystatin C alone and in combination with serum creatinine: a pooled analysis of 3,418 individuals with CKD. Am J Kidney Dis. 2008; 51:395-406.
PubMed
 
Blirup-Jensen S, Grubb A, Lindstrom V, Schmidt C, Althaus H.  Standardization of Cystatin C: development of primary and secondary reference preparations. Scand J Clin Lab Invest Suppl. 2008; 241:67-70.
PubMed
 
Grubb A, Blirup-Jensen S, Lindström V, Schmidt C, Althaus H, Zegers I, IFCC Working Group on Standardisation of Cystatin C (WG-SCC).  First certified reference material for cystatin C in human serum ERM-DA471/IFCC. Clin Chem Lab Med. 2010; 48:1619-21.
PubMed
CrossRef
 
Zegers I, Auclair G, Schimmel H, Emons H, Blirup-Jensen S, Schmidt C, et al..  Certification of Cystatin C in the Human Serum. Reference Material ERM-DA471/IFCC. Luxembourg: Publications Office of the European Union; 2010.
 
Inker LA, Eckfeldt J, Levey AS, Leiendecker-Foster C, Rynders G, Manzi J, et al..  Expressing the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) cystatin C equations for estimating GFR with standardized serum cystatin C values [Letter]. Am J Kidney Dis. 2011; 58:682-4.
PubMed
CrossRef
 
Imai E, Horio M, Nitta K, Yamagata K, Iseki K, Hara S, et al..  Estimation of glomerular filtration rate by the MDRD study equation modified for Japanese patients with chronic kidney disease. Clin Exp Nephrol. 2007; 11:41-50.
PubMed
 
Cockcroft DW, Gault MH.  Prediction of creatinine clearance from serum creatinine. Nephron. 1976; 16:31-41.
PubMed
CrossRef
 
Rule AD, Larson TS, Bergstralh EJ, Slezak JM, Jacobsen SJ, Cosio FG.  Using serum creatinine to estimate glomerular filtration rate: accuracy in good health and in chronic kidney disease. Ann Intern Med. 2004; 141:929-37.
PubMed
 
Jelliffe RW.  Estimation of creatinine clearance when urine cannot be collected. Lancet. 1971; 1:975-6.
PubMed
 
Jelliffe RW.  Letter: Creatinine clearance: bedside estimate. Ann Intern Med. 1973; 79:604-5.
PubMed
 
Mawer GE, Lucas SB, Knowles BR, Stirland RM.  Computer-assisted prescribing of kanamycin for patients with renal insufficiency. Lancet. 1972; 1:12-5.
PubMed
 
Hull JH, Hak LJ, Koch GG, Wargin WA, Chi SL, Mattocks AM.  Influence of range of renal function and liver disease on predictability of creatinine clearance. Clin Pharmacol Ther. 1981; 29:516-21.
PubMed
CrossRef
 
Gates GF.  Creatinine clearance estimation from serum creatinine values: an analysis of three mathematical models of glomerular function. Am J Kidney Dis. 1985; 5:199-205.
PubMed
 
Bjornsson TD, Cocchetto DM, McGowan FX, Verghese CP, Sedor F.  Nomogram for estimating creatinine clearance. Clin Pharmacokinet. 1983; 8:365-9.
PubMed
CrossRef
 
Walser M, Drew HH, Guldan JL.  Prediction of glomerular filtration rate from serum creatinine concentration in advanced chronic renal failure. Kidney Int. 1993; 44:1145-8.
PubMed
 
Nankivell BJ, Gruenewald SM, Allen RD, Chapman JR.  Predicting glomerular filtration rate after kidney transplantation. Transplantation. 1995; 59:1683-9.
PubMed
CrossRef
 

Figures

Grahic Jump Location
Figure 1.

Summary of evidence search and selection.

SCr = serum creatinine; SRM = standard reference material.

Grahic Jump Location
Grahic Jump Location
Figure 2.

Differences in accuracy and bias between estimated GFR by CKD-EPI and MDRD Study equations in North America, Europe, and Australia.

Difference in accuracy (top), as measured by P30 (P30 for CKD-EPI minus P30 for MDRD), is plotted against mean measured GFR in the study population. Difference in bias (bottom) (absolute value for bias for estimated GFR by MDRD Study equation minus absolute value for bias for estimated GFR by CKD-EPI equation) is plotted against mean measured GFR in the study population. CKD-EPI = Chronic Kidney Disease Epidemiology Collaboration; GFR = glomerular filtration rate; MDRD = Modification of Diet in Renal Disease; P30 = percentage of estimated GFR values within 30% of measured GFR.

* Denotes study in which all patients received trimethoprim.

† Could be reported in mL/min.

Grahic Jump Location

Tables

Table Jump PlaceholderAppendix Table 1.  

Search Strategy

Table Jump PlaceholderAppendix Table 2.  

Information on Development of Equations Based on Serum Creatinine Assays That Are Traceable to the Standard Reference Material

Table Jump PlaceholderAppendix Table 3.  

Overview Table of Equations Developed to Predict GFR Based on Serum Creatinine Assays Not Traceable to the Standard Reference Material

Table Jump PlaceholderTable 1.  

Performance Comparison of Creatinine-Based GFR Estimating Equations in North America, Europe, and Australia

Table Jump PlaceholderAppendix Table 4.  

Performance of Creatinine-Based GFR Estimating Equations in North America, Europe, and Australia in Subgroups by GFR

Table Jump PlaceholderAppendix Table 5.  

Performance of Creatinine-Based GFR Estimating Equations in North America, Europe, and Australia in Subgroups by Race

Table Jump PlaceholderTable 2.  

Performance Comparison of Creatinine-Based GFR Estimating Equations Outside of North America, Europe, and Australia

Table Jump PlaceholderAppendix Table 6.  

Performance of Creatinine-Based GFR Estimating Equations Outside North America, Europe, and Australia in Subgroups by GFR

Table Jump PlaceholderTable 3.  

Suggested Criteria for Developing and Validating GFR Estimating Equations

References

Miller WG.  Estimating glomerular filtration rate [Editorial]. Clin Chem Lab Med. 2009; 47:1017-9.
PubMed
CrossRef
 
Stevens LA, Coresh J, Greene T, Levey AS.  Assessing kidney function—measured and estimated glomerular filtration rate. N Engl J Med. 2006; 354:2473-83.
PubMed
CrossRef
 
Coresh J, Astor BC, McQuillan G, Kusek J, Greene T, Van Lente F, et al..  Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate. Am J Kidney Dis. 2002; 39:920-9.
PubMed
CrossRef
 
Vickery S, Stevens PE, Dalton RN, van Lente F, Lamb EJ.  Does the ID-MS traceable MDRD equation work and is it suitable for use with compensated Jaffe and enzymatic creatinine assays? Nephrol Dial Transplant. 2006; 21:2439-45.
PubMed
CrossRef
 
Myers GL, Miller WG, Coresh J, Fleming J, Greenberg N, Greene T, et al., National Kidney Disease Education Program Laboratory Working Group.  Recommendations for improving serum creatinine measurement: a report from the Laboratory Working Group of the National Kidney Disease Education Program. Clin Chem. 2006; 52:5-18.
PubMed
CrossRef
 
Panteghini M, Myers GL, Miller WG, Greenberg N, International Federation of Clinical Chemistry and Laboratory Medicine.  The importance of metrological traceability on the validity of creatinine measurement as an index of renal function. Clin Chem Lab Med. 2006; 44:1287-92.
PubMed
CrossRef
 
Levey AS, Greene T, Kusek J, Beck G.  A simplified equation to predict glomerular filtration rate from serum creatinine [Abstract]. J Am Soc Nephrol. 2000; 11:155A.
 
Levey AS, Coresh J, Greene T, Stevens LA, Zhang YL, Hendriksen S, et al., Chronic Kidney Disease Epidemiology Collaboration.  Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med. 2006; 145:247-54.
PubMed
 
Coresh J, Stevens LA.  Kidney function estimating equations: where do we stand? Curr Opin Nephrol Hypertens. 2006; 15:276-84.
PubMed
CrossRef
 
Stevens LA, Zhang Y, Schmid CH.  Evaluating the performance of equations for estimating glomerular filtration rate. J Nephrol. 2008; 21:797-807.
PubMed
 
Ma YC, Zuo L, Chen JH, Luo Q, Yu XQ, Li Y, et al..  Modified glomerular filtration rate estimating equation for Chinese patients with chronic kidney disease. J Am Soc Nephrol. 2006; 17:2937-44.
PubMed
CrossRef
 
Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D.  A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999; 130:461-70.
PubMed
 
National Kidney Foundation.  K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis. 2002; 39:S1-266.
PubMed
 
Cirillo M, Lombardi C, Luciano MG, Bilancio G, Anastasio P, De Santo NG.  Estimation of GFR: a comparison of new and established equations [Letter]. Am J Kidney Dis. 2010; 56:802-4.
PubMed
CrossRef
 
Jones GR, Imam SK.  Validation of the revised MDRD formula and the original Cockcroft and Gault formula for estimation of the glomerular filtration rate using Australian data. Pathology. 2009; 41:379-82.
PubMed
 
Jones GR.  Use of the CKD-EPI equation for estimation of GFR in an Australian cohort [Letter]. Pathology. 2010; 42:487-8.
PubMed
CrossRef
 
Michels WM, Grootendorst DC, Verduijn M, Elliott EG, Dekker FW, Krediet RT.  Performance of the Cockcroft–Gault, MDRD, and new CKD-EPI formulas in relation to GFR, age, and body size. Clin J Am Soc Nephrol. 2010; 5:1003-9.
PubMed
CrossRef
 
Kukla A, El-Shahawi Y, Leister E, Kasiske B, Mauer M, Matas A, et al..  GFR-estimating models in kidney transplant recipients on a steroid-free regimen. Nephrol Dial Transplant. 2010; 25:1653-61.
PubMed
CrossRef
 
Pöge U, Gerhardt T, Stoffel-Wagner B, Sauerbruch T, Woitas RP.  Validation of the CKD-EPI formula in patients after renal transplantation. Nephrol Dial Transplant. 2011; 26:4104-8.
PubMed
 
White CA, Akbari A, Doucette S, Fergusson D, Knoll GA.  Estimating glomerular filtration rate in kidney transplantation: is the new chronic kidney disease epidemiology collaboration equation any better? Clin Chem. 2010; 56:474-7.
PubMed
CrossRef
 
Lane BR, Demirjian S, Weight CJ, Larson BT, Poggio ED, Campbell SC.  Performance of the chronic kidney disease-epidemiology study equations for estimating glomerular filtration rate before and after nephrectomy. J Urol. 2010; 183:896-901.
PubMed
 
Tent H, Rook M, Stevens LA, van Son WJ, van Pelt LJ, Hofker HS, et al..  Renal function equations before and after living kidney donation: a within-individual comparison of performance at different levels of renal function. Clin J Am Soc Nephrol. 2010; 5:1960-8.
PubMed
CrossRef
 
Redal-Baigorri B, Stokholm KH, Rasmussen K, Jeppesen N.  Estimation of kidney function in cancer patients. Dan Med Bull. 2011; 58:A4236.
PubMed
 
Eriksen BO, Mathisen UD, Melsom T, Ingebretsen OC, Jenssen TG, Njølstad I, et al..  Cystatin C is not a better estimator of GFR than plasma creatinine in the general population. Kidney Int. 2010; 78:1305-11.
PubMed
CrossRef
 
Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, et al., CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration).  A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009; 150:604-12.
PubMed
 
Murata K, Baumann , Saenger AK, Larson TS, Rule AD, Lieske JC.  Relative performance of the MDRD and CKD-EPI equations for estimating glomerular filtration rate among patients with varied clinical presentations. Clin J Am Soc Nephrol. 2011; 6:1963-72.
PubMed
CrossRef
 
Stevens LA, Claybon MA, Schmid CH, Chen J, Horio M, Imai E, et al..  Evaluation of the Chronic Kidney Disease Epidemiology Collaboration equation for estimating the glomerular filtration rate in multiple ethnicities. Kidney Int. 2011; 79:555-62.
PubMed
CrossRef
 
Stevens LA, Schmid CH, Greene T, Zhang YL, Beck GJ, Froissart M, et al..  Comparative performance of the CKD Epidemiology Collaboration (CKD-EPI) and the Modification of Diet in Renal Disease (MDRD) Study equations for estimating GFR levels above 60 mL/min/1.73 m2. Am J Kidney Dis. 2010; 56:486-95.
PubMed
CrossRef
 
Horio M, Imai E, Yasuda Y, Watanabe T, Matsuo S.  Modification of the CKD epidemiology collaboration (CKD-EPI) equation for Japanese: accuracy and use for population estimates. Am J Kidney Dis. 2010; 56:32-8.
PubMed
CrossRef
 
van Deventer HE, George JA, Paiker JE, Becker PJ, Katz IJ.  Estimating glomerular filtration rate in black South Africans by use of the modification of diet in renal disease and Cockcroft–Gault equations. Clin Chem. 2008; 54:1197-202.
PubMed
CrossRef
 
Praditpornsilpa K, Townamchai N, Chaiwatanarat T, Tiranathanagul K, Katawatin P, Susantitaphong P, et al..  The need for robust validation for MDRD-based glomerular filtration rate estimation in various CKD populations. Nephrol Dial Transplant. 2011; 26:2780-5.
PubMed
CrossRef
 
Teo BW, Xu H, Wang D, Li J, Sinha AK, Shuter B, et al..  GFR estimating equations in a multiethnic Asian population. Am J Kidney Dis. 2011; 58:56-63.
PubMed
CrossRef
 
Yeo Y, Han DJ, Moon DH, Park JS, Yang WS, Chang JW, et al..  Suitability of the IDMS-traceable MDRD equation method to estimate GFR in early postoperative renal transplant recipients. Nephron Clin Pract. 2010; 114:c108-17.
PubMed
CrossRef
 
Matsuo S, Imai E, Horio M, Yasuda Y, Tomita K, Nitta K, et al., Collaborators developing the Japanese equation for estimated GFR.  Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis. 2009; 53:982-92.
PubMed
CrossRef
 
Rule AD, Teo BW.  GFR estimation in Japan and China: what accounts for the difference? [Editorial]. Am J Kidney Dis. 2009; 53:932-5.
PubMed
CrossRef
 
Berglund F, Killander J, Pompeius R.  Effect of trimethoprim-sulfamethoxazole on the renal excretion of creatinine in man. J Urol. 1975; 114:802-8.
PubMed
 
Stevens LA, Levey AS.  Measured GFR as a confirmatory test for estimated GFR. J Am Soc Nephrol. 2009; 20:2305-13.
PubMed
CrossRef
 
Matsushita K, Selvin E, Bash LD, Astor BC, Coresh J.  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; 55:648-59.
PubMed
CrossRef
 
Skali H, Uno H, Levey AS, Inker LA, Pfeffer MA, Solomon SD.  Prognostic assessment of estimated glomerular filtration rate by the new Chronic Kidney Disease Epidemiology Collaboration equation in comparison with the Modification of Diet in Renal Disease Study equation. Am Heart J. 2011; 162:548-54.
PubMed
CrossRef
 
White SL, Polkinghorne KR, Atkins RC, Chadban SJ.  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 AusDiab (Australian Diabetes, Obesity and Lifestyle) Study. Am J Kidney Dis. 2010; 55:660-70.
PubMed
CrossRef
 
Kwong YT, Stevens LA, Selvin E, Zhang YL, Greene T, Van Lente F, et al..  Imprecision of urinary iothalamate clearance as a gold-standard measure of GFR decreases the diagnostic accuracy of kidney function estimating equations. Am J Kidney Dis. 2010; 56:39-49.
PubMed
CrossRef
 
Stevens LA, Coresh J, Schmid CH, Feldman HI, Froissart M, Kusek J, et al..  Estimating GFR using serum cystatin C alone and in combination with serum creatinine: a pooled analysis of 3,418 individuals with CKD. Am J Kidney Dis. 2008; 51:395-406.
PubMed
 
Blirup-Jensen S, Grubb A, Lindstrom V, Schmidt C, Althaus H.  Standardization of Cystatin C: development of primary and secondary reference preparations. Scand J Clin Lab Invest Suppl. 2008; 241:67-70.
PubMed
 
Grubb A, Blirup-Jensen S, Lindström V, Schmidt C, Althaus H, Zegers I, IFCC Working Group on Standardisation of Cystatin C (WG-SCC).  First certified reference material for cystatin C in human serum ERM-DA471/IFCC. Clin Chem Lab Med. 2010; 48:1619-21.
PubMed
CrossRef
 
Zegers I, Auclair G, Schimmel H, Emons H, Blirup-Jensen S, Schmidt C, et al..  Certification of Cystatin C in the Human Serum. Reference Material ERM-DA471/IFCC. Luxembourg: Publications Office of the European Union; 2010.
 
Inker LA, Eckfeldt J, Levey AS, Leiendecker-Foster C, Rynders G, Manzi J, et al..  Expressing the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) cystatin C equations for estimating GFR with standardized serum cystatin C values [Letter]. Am J Kidney Dis. 2011; 58:682-4.
PubMed
CrossRef
 
Imai E, Horio M, Nitta K, Yamagata K, Iseki K, Hara S, et al..  Estimation of glomerular filtration rate by the MDRD study equation modified for Japanese patients with chronic kidney disease. Clin Exp Nephrol. 2007; 11:41-50.
PubMed
 
Cockcroft DW, Gault MH.  Prediction of creatinine clearance from serum creatinine. Nephron. 1976; 16:31-41.
PubMed
CrossRef
 
Rule AD, Larson TS, Bergstralh EJ, Slezak JM, Jacobsen SJ, Cosio FG.  Using serum creatinine to estimate glomerular filtration rate: accuracy in good health and in chronic kidney disease. Ann Intern Med. 2004; 141:929-37.
PubMed
 
Jelliffe RW.  Estimation of creatinine clearance when urine cannot be collected. Lancet. 1971; 1:975-6.
PubMed
 
Jelliffe RW.  Letter: Creatinine clearance: bedside estimate. Ann Intern Med. 1973; 79:604-5.
PubMed
 
Mawer GE, Lucas SB, Knowles BR, Stirland RM.  Computer-assisted prescribing of kanamycin for patients with renal insufficiency. Lancet. 1972; 1:12-5.
PubMed
 
Hull JH, Hak LJ, Koch GG, Wargin WA, Chi SL, Mattocks AM.  Influence of range of renal function and liver disease on predictability of creatinine clearance. Clin Pharmacol Ther. 1981; 29:516-21.
PubMed
CrossRef
 
Gates GF.  Creatinine clearance estimation from serum creatinine values: an analysis of three mathematical models of glomerular function. Am J Kidney Dis. 1985; 5:199-205.
PubMed
 
Bjornsson TD, Cocchetto DM, McGowan FX, Verghese CP, Sedor F.  Nomogram for estimating creatinine clearance. Clin Pharmacokinet. 1983; 8:365-9.
PubMed
CrossRef
 
Walser M, Drew HH, Guldan JL.  Prediction of glomerular filtration rate from serum creatinine concentration in advanced chronic renal failure. Kidney Int. 1993; 44:1145-8.
PubMed
 
Nankivell BJ, Gruenewald SM, Allen RD, Chapman JR.  Predicting glomerular filtration rate after kidney transplantation. Transplantation. 1995; 59:1683-9.
PubMed
CrossRef
 

Letters

NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Comments

Submit a Comment
Estimating Equations for Glomerular Filtration Rate in the Old: As Age Advances, the Formulae Concur Even Less!
Posted on June 19, 2012
T.S. Dharmarajan, MD, FACP, AGSF, Professor of Medicine and Associate Dean, Edward P. Norkus PhD, FACN, Associate Professor, Preventive and Community Medicine
New York Medical College, Valhalla, NY, Montefiore Medical Center (North division), Bronx, NY
Conflict of Interest: None Declared

To the Editor: We appreciate the review, “Estimating equations for glomerular filtration rate in the era of creatinine standardization”, by Earley et al (1), comparing the glomerular filtration rates (GFR) derived using the Modification of Diet in Renal Disease (MDRD) and the Chronic Kidney Disease Epidemiology Collaboration Initiative (CKD-EPI) formulae from publications between 1999 and 2011. The authors noted the CKD-EPI formula performed better at higher GFRs (>60 mL/min/1.73 m2) than the MDRD equation, while the MDRD equation did better at lower GFRs. They concluded that neither the CKD-EPI nor the MDRD formulae was optimal for all populations and GFR ranges, and acknowledged the paucity of data comparing the formulae across different age subgroups.

In February 2012, we published data that compared GFR estimates derived from the MDRD, CKD-EPI and Cockcroft-Gault (CG) formulae in a cross-section of 1535 older Americans (59 - 104 years) (2). Our sample included 29% White, 36% African American and 35% Hispanic adults, with 55% residents in long-term care; 39% had medical records indicating renal insufficiency determined by their primary physician. We observed a significant disconnect in CKD staging, with the potential to influence recommendations for monitoring and management, based on the formula used. In our report, the C-G formula provided significantly lower GFR estimates than either the MDRD or CKD-EPI formulae in the entire sample and in the subset of individuals classified as having renal insufficiency.

We also compared GFR estimates after stratifying our sample into four age categories (59-69, 70-79, 80-89 and 90+ years). The three formulae produced nearly identical GFR estimates, across race, in individuals aged <70 years. However, in subjects 70 - 104 years, the GFR estimates significantly differed across formulae. Individuals >69 years were classified into lower CKD stages (levels 3-5) significantly more often using the C-G and CKD-EPI formulae compared to the MDRD formula (C-G>CKD-EPI>MDRD); conversely, classification into CKD stages 1 and 2 (better function) occurred significantly more often with the MDRD formula (Table 1).

Our study's aim was simply to determine if GFR estimates differed across the three equations. We did not determine the accuracy or bias in the derived GFR estimates based on standard reference methods (urinary or plasma clearance of an exogenous marker). Nevertheless, we believe our findings that the CKD-EPI (and C-G) formulae classifies older, White, African American and Hispanic Americans into lower CKD stages more often than the MDRD equation are relevant in the old for several reasons. Based on the National Health and Nutrition Examination Survey (NHANES) data, the prevalence of CKD stages 3-5 in adults above 60 years was 22% (3). The increasing decline in muscle mass with age (sarcopenia) renders serum creatinine values less reliable as a marker of renal function in the old, emphasizing the need for reliable means to estimate GFR. Decline in renal function may occur with age and commonly does from disease (3, 4). In practice, accurate staging of CKD is essential: to accurately determine dosing of medications handled by the kidney; determine risk for surgery; assess risk for contrast use in imaging studies; plan nutritional therapy in CKD; and recognize and address complications linked to stage of kidney disease (5).

Based on the formula used, disconnect in GFR value and consequent staging of CKD, approximately half of older adults with CKD may be misclassified and inappropriately monitored. These implications emphasize a need for the ideal formula to estimate GFR, especially for the geriatric patient.

References:

1. Earley, A, Miskulin D, Lamb EJ et al. Estimating equations for glomerular filtration rate in the era of creatinine standardization. Ann Intern Med., 2012; 156:785-795

2. Dharmarajan TS, Yoo J, Russell RO, Norkus EP. Chronic kidney disease staging in nursing home and community older adults: Does the choice of Cockcroft-Gault, Modification of Diet in Renal disease or the Chronic Kidney Disease Epidemiology Collaboration Initiative equations matter? JAMDA, 2012; 13:151-155.

3. Coresh J, Selvin E, Stevens LA et al. Prevalence of chronic kidney disease in the United States. JAMA. 2007;298:2038-47

4. Lindeman RD, Tobin J, Shock NW. Longitudinal studies on the rate of decline in renal function with age. J Am Geriatr Soc. 1985; 33: 278-8

5. Fink HA, Ishani A, Taylor BC et al. Screening for, monitoring and treatment of chronic kidney disease stages 1-3: A systematic review for the U.S. Preventive Serviuces Task Force and for an American College of Physicians Clinical Practice Guideline. Ann Intern Med.2012; 156:570-81

Performance of CKD-EPI equation in diabetic individuals
Posted on August 29, 2012
Sandra P Silveiro, Ariana A. Soares, Letícia S. Weinert, Eduardo G. Camargo
Endocrinology Division, Hospital de Clínicas de Porto Alegre, Brazil
Conflict of Interest: None Declared

TO THE EDITOR: Earley and colleagues have elegantly presented the results of a timely systematic review “Estimating equations for glomerular filtration rate in the era of creatinine standardization - a systematic review” (1), concluding that the performance of the CKD-EPI and MDRD study equations varies across populations and GFR ranges, taking into account the use of a standardized creatinine measurement. The authors required, as inclusion criteria, the minimum number of a 100 individuals, the use of a glomerular filtration rate (GFR) reference method and the use of an IDMS traceable creatinine method. As the review encompassed the period of 1999 up to October 2011, we would like to share our results published in the November 2011 issue of Diabetes Care (2), because not only our data fulfills the required inclusion criteria, but also describes the findings of a South-Brazilian population, not evaluated in the systematic review. In our cross-sectional study, we evaluated 105 patients with type 2 diabetes, with a mean age of 57±8 years; 53 (50%) men and 90 (86%) white. Forty-six (44%) patients had microalbuminuria, and 14 (13%) had macroalbuminuria - all patients had GFRs >60 mL/min/1.73 m2, as measured by 51Cr-EDTA single-injection method. Measured 51Cr-EDTA GFR was 103±23, CKD-EPI GFR was 83±15 (bias: 20), and MDRD GFR was 78±17 mL/min/1.73 m2 (bias: 24). Accuracy P30 (95%CI) was 67% (58–74) for CKD-EPI and 64% (56–75) for MDRD. We concluded that both equations pronouncedly underestimated GFR in type 2 diabetic patients with GFR above 60 mL/min/1.73 m2. These findings confirmed our previous report of a significantly worse performance of both equations in patients with diabetes when compared with healthy individuals, mainly due to the fact that diabetic patients had higher serum creatinine levels than the healthy group, even after matched by GFR (3). In contrast, when previously analyzing a group of 96 healthy volunteers with normal GFR, we observed the improved accuracy (P30) of CDK-EPI equation as compared to the MDRD equation (85% vs. 69%, respectively, P <0.001) (4). We congratulate the authors on the excellent job of reporting the situation of estimating equations on the present days. A more appropriated interpretation of the equations performance allows the eventual development of strategies to improve the identified misrepresentations

References

1) Earley A, Miskulin D, Lamb EJ, Levey AS, Uhlig K. Estimating equations for glomerular filtration rate in the era of creatinine standardization - A systematic review. Early release. Ann Intern Med 2012.

2) Silveiro SP, Araújo GN, Ferreira MN, Souza FD, Yamaguchi HM, Camargo EG. Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation pronouncedly underestimates glomerular filtration rate in type 2 diabetes. Diabetes Care. 2011;34(11):2353-5.

3) Camargo EG, Soares AA, Detanico AB, Weinert LS, Veronese FV, Gomes EC, Silveiro SP. The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation is less accurate in patients with Type 2 diabetes when compared with healthy individuals. Diabet Med. 2011;28(1):90-5.

4) Soares AA, Eyff TF, Campani RB, Ritter L, Weinert LS, Camargo JL, Silveiro SP. Performance of the CKD Epidemiology Collaboration (CKD-EPI) and the Modification of Diet in Renal Disease (MDRD) Study equations in healthy South Brazilians. Am J Kidney Dis. 2010;55(6):1162-3

Submit a Comment

Summary for Patients

Clinical Slide Sets

Terms of Use

The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.

Toolkit

Want to Subscribe?

Learn more about subscription options

Advertisement
Related Articles
Topic Collections
PubMed Articles
Forgot your password?
Enter your username and email address. We'll send you a reminder to the email address on record.
(Required)
(Required)