Journal of the American Society of Nephrology
2007 JASN IMPACT FACTOR 7.111 HOME   AUTHOR INFO   EDITORIAL BOARD   SUBSCRIBE   FEEDBACK   ALERTS   HELP 
    advanced
CURRENT ISSUE ARCHIVES JASN Express ONLINE SUBMISSION


Published ahead of print on December 21, 2005
J Am Soc Nephrol 17: 487-496, 2006
© 2006 American Society of Nephrology
doi: 10.1681/ASN.2005010034
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
ASN.2005010034v1
17/2/487    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Mattman, A.
Right arrow Articles by White, C. T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Mattman, A.
Right arrow Articles by White, C. T.

Clinical Nephrology

Estimating Pediatric Glomerular Filtration Rates in the Era of Chronic Kidney Disease Staging

Andre Mattman*, Shaun Eintracht*, Thomas Mock*, Geraldine Schick, David W. Seccombe{dagger},{ddagger}, Robert Morrison Hurley§,|| and Colin Thomas White§,||

* Laboratory Medicine and § Division of Nephrology, Department of Pediatrics, British Columbia’s Children’s and Women’s Hospital, {dagger} Canadian External Quality Assessment Laboratory, {ddagger} Department of Pathology and Laboratory Medicine, and || Faculty of Medicine, Department of Pediatrics, University of British Columbia, Vancouver, British Columbia, Canada

Address correspondence to: Dr. Colin White, British Columbia’s Children’s Hospital, Division of Nephrology, 4480 Oak Street, ACB K4-151, Vancouver, British Columbia, Canada, V6H 3V4. Phone: 604-875-2272; Fax: 604-875-3649; E-mail: cwhite{at}cw.bc.ca

Received for publication January 10, 2005. Accepted for publication November 1, 2005.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 Conclusion
 References
 
With the use of information from a database of pediatric patients with concomitant nuclear GFR and serum creatinine (Cr), estimated GFR equations were derived on the basis of local laboratory methods and population. These formulas then were compared with those recommended by the National Kidney Foundation for estimating GFR in children. For this, their ability to estimate accurately an individual’s true GFR and chronic kidney disease stage, identify patients whose true GFR was <60 ml/min per 1.73 m2, and to identify correctly deterioration in an individual’s GFR over time was compared. Next, two methods to estimate GFR in children without the use of height or weight were developed. The first was a height- and weight-independent formula; the second was a novel approach using the Schwartz formula and calculating a Cr cutoff based on age-based estimates of height and GFR level of interest, i.e., <60 ml/min per 1.73 m2. Our results suggest that if local laboratory constants are derived and a height is known, then the Schwartz formula offers the most accuracy with least mathematical complexity to perform in the clinical setting. If height is not available but the local laboratory constants have been derived, then the British Columbia’s Children’s Hospital 2 formula is of value; however, in the setting of estimating pediatric renal function in the outpatient laboratory, where neither of these factors is commonly known, an approach whereby a Cr cutoff for a GFR of interest is developed is suggested. Provided are Cr levels that are based on a reference method of Cr measurement to facilitate this approach for the clinician.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 Conclusion
 References
 
The prevalence of ESRD is rapidly increasing in the adult population, with >1 million people affected worldwide by the end of 2001 (1) and at a cost to the North American ESRD programs of $25.2 billion in 2002 (2). The National Kidney Foundation (NKF) has recognized that to "further improve dialysis outcomes, it is necessary to improve the health status of those who reach ESRD... from earliest kidney damage through the various stages of progression to kidney failure" (3). As a first step toward achieving this goal, the NKF developed a chronic kidney disease (CKD) classification system, the NKF Disease Outcomes Quality Initiative (NKF-DOQI) classification, to "improve/standardize" care of the patient at all stages of CKD (3).

In the adult nephrology literature, a move toward classification and management of CKD on the basis of GFR has led to the proposal for widespread implementation of equations to calculated estimated GFR (eGFR) (4). These equations are often derived from studies in patients with particular illnesses, e.g., cancer (5,6), renal dysfunction (7,8), and organ transplantation (9–11). They use some combination of easily obtained demographics (e.g., height, weight, gender, race) and serum measurements (e.g., creatinine [Cr], urea, albumin) (12) to calculate the individual’s GFR or Cr clearance.

Similarly, the pediatric members of the NKF CKD working group clearly stated that "major emphasis is placed on the identification of children and adolescents with CKD by measuring the protein-to-Cr ratio in spot urine specimens and by estimating the GFR from serum Cr using prediction equations" (13).

Recognizing that eGFR would play a vital role in the new CKD classification, the NKF reviewed the published literature on formulas to determine which were most accurate and clinically applicable. In adults, the formulas suggested were the Modification of Diet in Renal Disease (MDRD) formulas by Levey et al. (12) or the Cockcroft-Gault formula (CG) (14). In pediatrics, they recommended using either the Schwartz (15), or the Counahan-Barratt (CB) formulas (16) for estimating GFR (13).

Although recommended by the NKF, both the Schwartz and the CB formulas do have clinically important limitations. For example, a recent study that evaluated an optimized form of the Schwartz equation against a concurrent iothalamate GFR showed that the SD of the difference of these two methods was approximately 10 ml/min per 1.73 m2; in other words, the relative error of the GFR (as defined as ±2 SD) estimated by the Schwartz formula approximated ±20 ml/min per 1.73 m2 (8). For a GFR of 75 ml min per 1.73 m2, this degree of uncertainty encompasses a wide range of GFR values from <60 to >90 ml/min per 1.73 m2, or CKD stages 1 to 3. These relative errors will be substantially higher when Cr calibration differences from laboratory to laboratory are not corrected (12).

A second limitation of these eGFR formulas in pediatrics is the requirement for both a height and a serum Cr measurement. Height is not readily available or verifiable in the clinical laboratories that obtain the blood sample and report the Cr results, thus discouraging the routine application of GFR estimation equations to the pediatric population.

With these limitations of pediatric eGFR formulas in mind, the goals of this study were to (1) derive eGFR equations based on our local laboratory methods and patient population; (2) compare these formulas with those recommended by the NKF for use in estimating GFR in children as related to each formula’s ability to (a) accurately estimate an individuals true GFR, (b) correctly identify the patient whose true GFR was <60 ml/min per 1.73 m2, and (c) assess the ability of each formula to identify correctly a deterioration of an individuals GFR over time; and finally (3) develop a height- and weight-independent method for estimating GFR in children and identifying those who are at risk for having significant renal dysfunction.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 Conclusion
 References
 
We undertook a retrospective review of data on all patients who were referred to the British Columbia’s Children’s Hospital (BCCH) nephrology service over the 5-yr period from 1998 to 2003 and had both a two-point single injection of technetium-99m-diethylenetriaminepentaacetic acid (99mTc-DTPA) nuclear GFR measurement (nGFR) and a serum Cr measurement (enzymatic Cr method; Vitros 950/250) within 1 d of each other. During this period, there was no change in either the laboratory or nuclear medicine equipment or protocols used. All studies were performed during the regular care or investigation of the patient, and the vast majority was done on an outpatient basis. The nGFR was considered to be the patient’s true GFR. This study received approval from the University of British Columbia ethics review board.

Patients
The data set consisted of 267 unique patients with a total of 473 observations. Of these, 116 patients had multiple database entries, accounting for 206 of the 473 observations. Each observation included the patient’s height, weight, diagnosis, Cr, age, gender, and nGFR. One infant who was younger than 1 yr was excluded from the study, leaving a total of 266 patients (472 observations). The patient demographics are summarized in Table 1, and diagnoses are summarized in Table 2.


View this table:
[in this window]
[in a new window]

  		Table 1. Patient demographicsa

 

View this table:
[in this window]
[in a new window]

  		Table 2. Patient diagnoses for model and validation groupsa

 
Derivation of Local Formulas
To derive our local eGFR formulas, we used the first available set of paired nGFR and Cr measurements for each patient (n = 266) and randomly assigned them between a modeling (n = 180) and a validation group (n = 86). There were no statistically significant differences between the modeling and validation groups with respect to mean age, nGFR, serum Cr, height, or weight.

The derivation of all formulas and constants was performed using only the modeling group’s data, whereas the validation group data were used to compare each formula in terms of accuracy, ability to identify CKD staging, and ability to identify patients with a GFR <60 ml/min per 1.73 m2. The only exception to this is in the data on the serial eGFR, for which the various formulas were applied to all patients, irrespective of original group, with two or more nGFR.

Two novel eGFR formulas were subsequently developed using data from the modeling group. The first (BCCH1) was developed using a linear regression of the patient data (aside from diagnosis) onto nGFR (ml/min per 1.73 m2). Forward stepwise regression was used to narrow down a list of candidate variables. For the BCCH1 formula, these variables included weight, height, age, presence of nephrotic syndrome, gender, inverse of Cr and inverse of Cr squared, age x height, age/Cr, height/Cr, gender x age, and gender/Cr. The second formula (BCCH2) was derived using only easily available demographic variables: the inverse of Cr, age, and gender. Model selection was based on the Akaike Information Criterion (S-plus Statistical Software, Seattle, WA) (17). Akaike Information Criterion selection serves to optimize the accuracy of the modeled equation while minimizing the number of included covariates.

Comparison with Published Formulas
Including our two locally derived equations, BCCH 1 and 2, we compared a total of seven previously published formulas with respect to their performance in estimating the true or measured GFR. These formulas included those suggested by the NKF, e.g., Schwartz, Counahan-Barratt, CG, as well as one of the abbreviated forms of the MDRD formulas (aMDRD) (12) and the more recent pediatric-derived formula proposed by Leger et al. (18) (Table 3). Of note, the adult-derived aMDRD equation has never been recommended for use in the population under 18 yr of age. We included this formula in our comparison studies only for the sake of completeness.


View this table:
[in this window]
[in a new window]

  		Table 3. eGFR formulas: Accuracy versus nuclear GFRa

 
To compare fairly the accuracy of these formulas, we empirically developed constants to correct for our local laboratory Cr method. These constants were developed by iteratively changing the constant associated with the Cr variable to minimize the mean relative error between the given eGFR formula and the nGFR measurement within the model group. The constants and the optimized forms of the established eGFR formulas are listed in Table 3; all references to any eGFR formulas from this point on can be assumed to mean the locally optimized form of that equation and may be designated by a preceding "o"; for example, oSch would be read as "the optimized form of the Schwartz equation."

The statistical comparison of the eGFR formulas consisted of evaluation of Pearson correlation coefficients, determination of the mean and SD of the relative error [100 x (eGFR – nGFR)/average (eGFR, nGFR)] and the mean absolute value of the relative error [100 x |(eGFR – nGFR)|/average (eGFR, nGFR)]. We also compared the respective eGFR formulas with respect to the proportion of eGFR results that agreed within 30% of the matched nGFR result.

Overall accuracy of the formulas in terms of CKD staging was based on achieving concordance with the "true" CKD staging as determined by nGFR. Sensitivity and specificity of each formula in terms of correctly identifying a patient with a true GFR <60 ml/min per 1.73 m2 also were calculated. As NKF-DOQI CKD staging in pediatrics has been recommended only for the population older than 2 yr, children who were between 1 and 2 yr of age were dropped from the validation group, leaving n = 83 for these latter comparisons (13).

Serial GFR
We subsequently analyzed data on 116 patients with two or more paired measurements of Cr and nGFR (206 repeat measurements). We determined the change in the nGFR between the initial and the final measurements in the database and evaluated the sensitivity and the specificity of the eGFR formulas in correctly identifying a 30% decrease in nGFR. The 30% decrease in nGFR was chosen arbitrarily in recognition that accuracy within this percentage was deemed "acceptable" by the NKF-DOQI CKD workgroup when they evaluated and recommended the various eGFR formulas (19). The time interval between the first and final measurements averaged 2.2 yr (range 0.3 to 4.1 yr).

Height- and Weight-Independent GFR Estimation
We evaluated two separate height- and weight-independent GFR estimation methods. The first was the BCCH 2 formula as described above. The second and, we believe, novel approach was that of defining GFR-specific Cr cutoffs and was designed for use in pediatric laboratory reporting of eGFR. For example, to identify patients with a GFR <60 ml/min per 1.73 m2, we determined the Cr value, based on our locally optimized Schwartz formula, above which a child of that age and gender would have an eGFR <60 ml/min per 1.73 m2. In other words, we took the Schwartz formula (15) eGFR = k x Ht/serum Cr, which can be rearranged as serum Cr = k x Ht/eGFR, derived the k value locally, used an age- and gender-based estimate for the height, and then set the eGFR to the value of interest, e.g., <60 ml/min per 1.73 m2.

This approach requires an age- and gender-specific estimation of height; we chose to use the third percentile value for age/gender as defined in the current Centre for Disease Control published growth curves (using the height of a child at the 97th percentile means the Cr cutoffs as generated would represent an eGFR between 72 and 74 ml/min per 1.73 m2 depending on the age and the gender of the child). We compared the results achieved by using these newly derived cutoff values, in terms of their ability to identify correctly patients whose nGFR was <60 ml/min per 1.73 m2, with those achieved when using the optimized Schwartz equation (15).

To enable other laboratories to make use of our Cr cutoff values, we converted these Vitros Cr values into equivalent values on the basis of a reference Cr method that measures Cr via isotope dilution gas chromatography/mass spectrometry (20,21). A conversion equation was developed from data obtained through a provincial external quality assurance program for Cr clinical laboratory Cr measurements. This program has been operating for the past 3 yr and includes 38 participating laboratories with either a Vitros 250 (n = 28) or a Vitros 950 (n = 10) Cr method. During a 6-mo period, each laboratory measured Cr levels in six different samples ranging in concentration from 37 to 132 µmol/L. Reference values for these samples were obtained by quadruplicate measurements of the same samples with the reference method (20,21). This comparison study yielded a conversion equation of the form Vitros 250/950 (Cr) = (1.0342 x reference value Cr) + 9.588. All comparative statistics were performed with the software SPSS for Windows release 12.0.0 (SPSS, Chicago, IL).


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 Conclusion
 References
 
Derivation of Local Formulas
Two novel pediatric eGFR equations were developed and are described below. The first equation, BCCH1, describes a quadratic relationship between the Ln (eGFR) and the inverse of the Cr measurement. Height is a prominent factor in the equation, whereas weight has a more minor but still significant effect. Age, gender, and the presence or absence of the nephrotic syndrome did not have an impact on the final equation. The second equation, BCCH2, makes use of Cr, age, and gender variables only.

BCCH1: Ln(eGFR) (ml/min per 1.73 m2) = 1.18 + (0.0016 x Wt [kg]) + (0.01 x Ht [cm]) + [(149.5/Cr) – (2141/Cr2)]

BCCH2: eGFR (ml/min per 1.73 m2) = –61.56 + [5886 x (1/Cr)] + (4.83 x age [years]) + 10.02 (if male)

Comparison with Published Formulas
Using data from the validation group, all eGFR formulas were compared as outlined below. However, other than in the cohort of patients who were older than 15 yr (n = 22), the adult-derived aMDRD (12) formula performed significantly worse than any other formula evaluated.

All eGFR formulas, except the aMDRD, produced values that were significantly correlated with the nGFR values (P < 0.005, one tailed significance; Table 3). The Sch (15) and BCCH1 formulas had the strongest correlations (r = 0.83) followed by the CB (16) (r = 0.79), Leger (18) (r = 0.74), CG (14) (r = 0.73), BCCH2 (r = 0.72), and aMDRD (r = 0.31).

The Schwartz formula, after optimization, was the most accurate formula evaluated, with an absolute percentage error of 18.0 ± 12.9%, followed by the BCCH1 at 18.1 ± 12.0% and the CB at 19.5 ± 14.1%. The BCCH2 formula, with a percentage error of 22.7 ± 16.8% and two outliers, CG formula with 22.1 ± 15.9% and one outlier, and the Leger formulas with an absolute percentage error of 24.3 ± 15.8 and one outlier all yielded much poorer results. The aMDRD formula was by far the least accurate, with a percentage error of 42.7 ± 27% (Table 3, Figure 1).


Figure 1
View larger version (13K):
[in this window]
[in a new window]
[as a PowerPoint slide]
  		Figure 1. Absolute value of the relative errors for estimated GFR (eGFR) formulas. The y axis represents the percentage absolute value of the relative error of the eGFR formulas (%). The thick central line indicates the mean absolute value of the relative error, the solid box indicates the 25th to 75th percentiles, and the error bars indicate the 2.5th to 97.5th percentiles, respectively. The outliers ({circ}) represent values that are >1.5 times the interquartile range above the 75th percentile.

 
As a result of the optimization process that was applied to each equation, all eGFR formulas had a mean relative error that was close to and evenly distributed around zero; however, the confidence limits for the relative error for all formulas tested were broad (Table 3), with the minimal 90% confidence limits for any formula tested being –30 to 39% for the BCCH1 eGFR formula (Table 3). Figure 2 demonstrates distribution of the relative error according to the GFR for the BCCH1, BCCH2, optimized Schwartz, and CB formulas. In terms of the percentage of optimized eGFR results within 30% of the nGFR, this value ranged from as low as 35% for the aMDRD formula to 83% for the BCCH1 formulas (Figure 3).


Figure 2
View larger version (34K):
[in this window]
[in a new window]
[as a PowerPoint slide]
  		Figure 2. Bland-Altmann plots for nuclear medicine GFR versus British Columbia’s Children’s Hospital 1 (BCCH1) (a), BCCH2 (b), oSch (c), and oCB (d) eGFR formulas, respectively. The solid line indicates the mean and the dashed lines indicate ±2 SD for the relative difference between nuclear GFR (nGFR) and the respective eGFR formulas.

 

Figure 3
View larger version (17K):
[in this window]
[in a new window]
[as a PowerPoint slide]
  		Figure 3. Agreement between eGFR values and the nGFR values for patients in the validation group.

 
Concordance with CKD Stage
Table 4 outlines the concordance rates, both exact stage and within one stage, for the eGFR formulas. The BCCH 1 and 2, oSch, and oCB formulas all performed similarly with a same stage concordance ranging from 67 to 71% and within one stage concordance of 100%.


View this table:
[in this window]
[in a new window]

  		Table 4. Accuracy of eGFR formulas in classifying CKD stages

 
Identification of a Patient with GFR <60 ml/min per 1.73 m2
The relative sensitivity and specificity of the eGFR formulas for identifying an nGFR <60 ml/min per 1.73 m2 are summarized in Table 4. Aside from the BCCH2 and the aMDRD, all other formulas achieved a sensitivity of 86%. All formulas, aside from that of Leger (88%) and the aMDRD (75%), achieved a specificity of ≥96%.

Serial Changes in GFR
Altogether, 206 follow-up measurements were performed on the 116 patients who were older than 5 yr. Figure 4 illustrates the relative change in the oSch GFR (Schfinal – Schinitial)/Schinitial versus the relative change in the nGFR (nGFRfinal – nGFRinitial)/nGFRinitial. A decrease of 30% or more in the oSch eGFR had a sensitivity of 63%, specificity of 96%, a positive predictive value (PPV) of 56%, and a negative predictive value (NPV) of 97% in detecting a clinically significant 30% deterioration in the nGFR. These results are compared with the performance of the other eGFR formulas in Table 5.


Figure 4
View larger version (17K):
[in this window]
[in a new window]
[as a PowerPoint slide]
  		Figure 4. Relative change in values of the optimized Schwartz equation as a function of relative change in nGFR. The dotted lines represent the 30% thresholds for designating a clinically significant decrease in GFR over time. These reference lines designate four quadrants that correspond to true-positive (TP), false-negative (FN), true-negative (TN), and false-positive (FP) results.

 

View this table:
[in this window]
[in a new window]

  		Table 5. Ability of serial eGFR measurements to accurately identify a minimum 30% relative decrease in nGFRa

 
Height- and Weight-Independent GFR Estimation
As current pediatric eGFR formulas all require height or weight data, they are not adapted easily for use in an outpatient laboratory setting, where obtaining these values would be almost impossible. Therefore, we derived a novel formula that was dependent only on age, gender, and Cr level (BCCH2) and fared well in comparison with the other formulas as tested (Table 3, Figure 1).

Our second method for determining a height- and weight-independent GFR estimate was to use the optimized Schwartz equation to identify age- and gender-specific Cr cutoffs that would identify children with nGFR values <60 ml/min per 1.73 m2. The resulting Cr cutoff values are listed in Table 6 in two forms: the Vitros values obtained in this study and their equivalent values in terms of a reference method (21,22). Using this approach, we were able to identify correctly 87% of patients in the validation group with an eGFR <60 ml/min per 1.73 m2 while maintaining a specificity of 93% (Table 4), very similar to the best results obtained from the use of optimized eGFR equations when height is available.


View this table:
[in this window]
[in a new window]

  		Table 6. Gender- and age-dependent creatinine cutoff valuesa

 

   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
Conclusion
 References
 
GFR, a summation of the filtration rate of each functional nephron, provides an estimate of the functional renal mass (22). Accurate knowledge of GFR is often vital for provision of appropriate medical care, including the appropriate dosing of many drugs. As well, diagnosis of a reduced GFR, i.e., CKD, often heralds other comorbidities (23), and early identification may lead to earlier interventions and minimization of morbidity and mortality. The routine chemistry measurement of Cr is a well-established marker of GFR. This has led to efforts to encourage laboratories reflexively to report eGFR values whenever a Cr test is ordered. These eGFR values are derived from formulas that combine serum Cr with demographic variables that are related to Cr production rates.

However, a number of pitfalls are inherent in the development and the use of such a formula, the first being the intrinsic biologic variability in the GFR itself, which has substantial within-day (24) and between-day variability (25). The physiologic variability in the production and in the extraglomerular elimination of Cr will also contribute to variability in Cr-based eGFR measurements; if one then factors in interlaboratory variation in terms of measurement of Cr (26), then it is clear that even a perfect eGFR estimate may differ substantially from a nuclear GFR estimate. This variability is reflected in the guidelines proposed by the NKF-DOQI CKD workgroup, who chose to define the accuracy of the various GFR estimation equations as clinically acceptable if the estimated GFR value is within 30% of the true measured GFR (23).

As well, because most of these formulas were derived from patients in an outpatient setting, there is a presumption that the patient will be in a steady state before and during any measurements or testing. Many patients in whom GFR estimations would be of greatest use, e.g., patients with cancer or rapidly deteriorating renal function or those who are receiving multiple nephrotoxins, will not meet that presumption.

In our study we set out to derive our own local eGFR formulas and compare them against the currently recommended NKF-DOQI eGFR formulas as well as that of Leger et al. (18) and one of the aMDRD formulas (12). To do so fairly, we derived laboratory-specific constants for the Cr term in each of the previously published equations.

By all measures of accuracy, the aMDRD formula as tested fared much worse than any of the other formulas in our pediatric population. This finding could be predicted as the aMDRD formula was derived from an adult cohort in whom the relationship between GFR and age is much different than in the pediatric population. Similarly, the other adult formula suggested by the NKF-DOQI group, CG, has an age term that adversely affects its accuracy when used in a pediatric population. This can be explained in part by realizing that in adults, muscle bulk and, therefore, Cr production rates decline with age; i.e., the youngest adults have the highest serum Cr values for a given body surface area–normalized GFR. In contrast, the situation is reversed in pediatrics, in which the youngest children are expected to have the lowest serum Cr for a given body surface area–normalized GFR.

Although our locally derived formula, BCCH1, performed as well as any formula tested in estimating an individual GFR, its mathematical complexity could not be justified for use in a day-to-day clinic setting by any significant improvement in accuracy over either the optimized Schwartz (15) or CB formula (16). When optimized for our local laboratory method, the Schwartz formula fared much better than the nonoptimized form, with an improvement in accuracy of between 20 and 25% (Figure 3), comparable to other published studies (8,27–29). The Schwartz formula and the CB formula have simpler mathematical forms that facilitate day-to-day use in a clinic or office setting.

Hogg et al. (13) clearly outlined the importance of early identification of a child with a persistent moderate to severe decrease in GFR, i.e., <60 ml/min per 1.73 m2 (NKF-DOQI stage 3) for >3 mo, as at this level of dysfunction, children are more likely to have associated complications and hence benefit from early identification (30).

In our validation group, the optimized Schwartz formula (15) had a PPV of 86% and a NPV of 97% in identifying individuals with an nGFR value of 59 ml/min per 1.73 m2 or less. We believe that these figures are robust enough for clinicians to use this optimized equation to justify further, more definitive investigations, such as a nuclear GFR, when a child’s eGFR result suggests that he or she has significant renal impairment.

In terms of the utility of serial eGFR measurements for monitoring patients, the results achieved with the optimized Schwartz formula are equally encouraging. Overall, 108 (93%) of the 116 paired nGFR measurements did not show a decrease of 30% over time. The optimized Schwartz eGFR formula correctly classified 96% of those 108 patients. Of the small number of patients who did demonstrate progressive renal insufficiency, five (63%) of eight were identified correctly for a PPV of 56% and a NPV of 97%. We believe that these results are sufficiently high to guide initial clinical decision making and justify further definitive testing as indicated.

Although the pediatric eGFR formulas have merit for clinicians who need to identify and monitor patients with CKD, pediatric eGFR formulas present unique challenges for the clinical laboratories. The first problem is that all published pediatric eGFR formulas use patient height. As laboratories generally do not measure patient height at the time of blood collection, pediatric eGFR results cannot be reported easily. The second problem is that pediatric eGFR formulas make use of constants that are specific to a particular method for Cr measurement. As illustrated in Figure 3, a clinical laboratory must derive a laboratory-specific constant to optimize the accuracy of the eGFR equations. These two problems essentially have ruled out the use of a pediatric eGFR formula in an outpatient laboratory setting up until now. To address the problems that are inherent with reporting eGFR in children from the laboratory viewpoint, we took two approaches: (1) Derivation of a non–height-, non–weight-based eGFR formula and (2) derivation of optimized Schwartz formula–based Cr values that would identify children who have a GFR <60 ml/min per 1.73 m2 (NKF-DOQI stage 3).

Our height- and weight-independent formula, BCCH2, performed comparably to other eGFR equations; its performance was only slightly below that of the optimized Schwartz on most measures (Figures 1 through 3, Tables 3 through 5). Hence, this formula does allow a reasonable estimate of GFR even in the absence of a known height or weight. There are two significant drawbacks to its routine use on a clinical basis: (1) The need for each center to derive local laboratory constants for the formula and (2) that similar to other non–height-containing formulas, e.g., CG and aMDRD, occasional estimates of GFR are grossly inaccurate (Figures 1 and 2).

Our second approach to identifying patients with a GFR <60 ml/min per 1.73 m2 was to use the optimized Schwartz formula to derive Cr cutoffs that are specific to a child’s age and gender. These values were obtained by making the assumption that a child of a given age was at the third percentile for age- and gender-appropriate height. The third percentile was chosen to maximize the sensitivity of the Cr cutoffs. This produced a level of accuracy near that obtained by using our best eGFR formulas (Table 4).

Aside from eliminating the need for clinical laboratories to obtain height information on pediatric patients, this Cr cutoff approach also frees clinical laboratories from the obligation to obtain a laboratory-specific constant for the Schwartz or an alternate pediatric eGFR formula. We have presented Cr cutoffs (Table 6) in terms of the Vitros Cr method that is used in our laboratory and also in terms of a reference Cr method (by definition, a nonbiased method). In British Columbia, all clinical laboratories subscribe to an external quality control service for Cr measurement that tracks the bias between the clinical laboratory result and a reference method. Thus, all laboratories in British Columbia immediately can adjust and adopt our Cr cutoff values to aid in the identification of children who are at high risk for CKD. It is likewise a simple matter for other clinical laboratories to determine the bias of their Cr methods as compared with a reference method and similarly use our values as presented.

There are a number of limitations to our study and potentially the broad applicability of its results. First is that all estimates of accuracy, etc., for the formulas and results are based on a pediatric nephrology referral population with a high likelihood of renal dysfunction. The PPV and NPV would differ substantially in a healthy cohort with a lower prevalence of disease. In addition, one would expect that the specificity of identifying a GFR <60 ml/min per 1.73 m2 would be lower in a healthy cohort in which the relationship between serum Cr and GFR may be substantially different (31).

A second limitation is the issue of basing our true GFR values on a plasma disappearance method on the basis of 99Tc-DTPA sampling. Although inulin clearance is classically held to be the gold standard for measuring GFR (32), it is cumbersome to perform and is not available in North America (33). 99Tc-DTPA is cleared primarily by glomerular filtration (34) and has a clearance that approximates that of inulin (35); the main concern with its use is that it can dissociate from DTPA during a clearance study, thereby adversely affecting the results (36). However, the two-sample 99Tc-DTPA method used in our center has been proposed as a valid method for measuring GFR in children by the Pediatric Committee of the European Association of Nuclear Medicine for routine use in children when measuring GFR on a clinical basis, so we believe that it is a valid clinical standard with which to compare our eGFR results (37).

One additional limitation of our study is that the Cr cutoff values presented in Table 6 cannot be considered definitive for several reasons. First, the cutoffs were developed as part of a retrospective review. Second, the Cr measurements were performed on a Vitros analyzer, which differs significantly from other Cr methods. Although we have presented Cr cutoff values in terms of the Vitros assay and in terms of reference method for Cr, these latter values were obtained indirectly. Thus, these Cr cutoffs need to be validated in a prospective study using a reference method.


   Conclusion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Conclusion
References
 
Although there are some advantages in terms of accuracy to be gained by use of our locally derived eGFR formula, BCCH1, its mathematical complexity leads us to concur with the NKF-DOQI workgroup and suggests that for ease of calculation and to maintain the highest accuracy across the pediatric age group in estimating GFR, clinicians should use either the Schwartz (15) or the CB (16) formula for children who are younger than 18 yr, presuming that an appropriate constant for the local laboratory method has been validated. The optimized Schwartz formula in particular is very robust and may also be used to identify and follow GFR in pediatric patients with CKD.

In situations in which height is not available but local laboratory-specific constants are developed for our non–height- or weight-containing formula, BCCH2, we believe that it provides a useful tool for the clinician who wishes to estimate a child’s renal function with the view to proceeding to further testing in the case in which they are estimated to have a GFR <60 ml/min per 1.73 m2.

In situations in which neither height nor local constants for the various eGFR formula are available, we suggest laboratory conversion of current NKF-DOQI CKD stages to new cutoff values of Cr as outlined above, specifically targeting the stage 2/3 transition at the level of 60 ml/min per 1.73 m2. This method is simpler to apply to an outpatient setting than the derivation of local constants or eGFR formula and can identify reliably children who are at risk for a low GFR and then may go on to benefit from timely investigations, referral, and, potentially, therapy.

Finally, we remind the reader that presently the best accuracy of any of the published eGFR formulas in adults or pediatrics rarely reaches the level of 80% of values closer than ±30% to the true GFR. Therefore, caution should be exercised in simply applying any formula to an individual patient for whom the need for accuracy in estimating the GFR is vital to produce a "high stakes" decision, e.g., decisions regarding dosing of nephrotoxic or chemotherapeutic drugs (5,6). In these scenarios, obtaining a more invasive but accurate measurement of the GFR can be justified on the basis of its importance in directing potentially harmful therapy for the individual.


   Acknowledgments
 
Preliminary results of this article were presented at the 48th Annual Conference of the Canadian Society of Clinical Chemists and Canadian Association of Medical Biochemists (London, Ontario, Canada) and subsequently published as an abstract in Clinical Biochemistry (Vol. 37, p 737) in 2004.

We thank Drs. David Lirenman and James Carter for contribution to this study by their exemplary care of these patients over many years.


   Footnotes
 
A.M. and S.E. contributed equally to this work.

Published online ahead of print. Publication date available at www.jasn.org.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Conclusion
 References
 

  1. Moeller S, Gioberge S, Brown G: ESRD patients in 2001: Global overview of patients, treatment modalities and development trends. Nephrol Dial Transplant 17 : 2071 –2076, 2002[Free Full Text]
  2. USRDS: The United States Renal Data System. Am J Kidney Dis 42[Suppl 5] : S1 , 2003
  3. National Kidney Foundation: K/DOQI clinical practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Am J Kidney Dis 39[Suppl 1] : S14 –S266, 2002[CrossRef]
  4. McCullough PA: Cardiorenal risk: An important clinical intersection. Rev Cardiovasc Med 3 : 71 –76, 2002[CrossRef][Medline]
  5. Cole M, Price L, Parry A, Keir MJ, Pearson AD, Boddy AV, Veal GJ: Estimation of glomerular filtration rate in paediatric cancer patients using 51CR-EDTA population pharmacokinetics. Br J Cancer 90 : 60 –64, 2004[CrossRef][Medline]
  6. Hjorth L, Wiebe T, Karpman D: Correct evaluation of renal glomerular filtration rate requires clearance assays. Pediatr Nephrol 17 : 847 –851, 2002[CrossRef][Medline]
  7. Filler G, Lepage N: Should the Schwartz formula for estimation of GFR be replaced by cystatin C formula? Pediatr Nephrol 18 : 981 –985, 2003[CrossRef][Medline]
  8. Hellerstein S, Berenbom M, DiMaggio S, Erwin P, Simon SD, Wilson N: Comparison of two formulae for estimation of glomerular filtration rate in children. Pediatr Nephrol 19 : 780 –784, 2004[CrossRef][Medline]
  9. Fulladosa X, Moreso F, Narvaez JA, Grinyo JM, Seron D: Estimation of total glomerular number in stable renal transplants. J Am Soc Nephrol 14 : 2662 –2668, 2003[Abstract/Free Full Text]
  10. Gonwa TA, Jennings L, Mai ML, Stark PC, Levey AS, Klintmalm GB: Estimation of glomerular filtration rates before and after orthotopic liver transplantation: Evaluation of current equations. Liver Transpl 10 : 301 –309, 2004[CrossRef][Medline]
  11. Pierrat A, Gravier E, Saunders C, Caira MV, Ait-Djafer Z, Legras B, Mallie JP: Predicting GFR in children and adults: a comparison of the Cockcroft-Gault, Schwartz, and modification of diet in renal disease formulas. Kidney Int 64 : 1425 –1436, 2003[CrossRef][Medline]
  12. 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 130 : 461 –470, 1999[Abstract/Free Full Text]
  13. Hogg RJ, Furth S, Lemley KV, Portman R, Schwartz GJ, Coresh J, Balk E, Lau J, Levin A, Kausz AT, Eknoyan G, Levey AS: National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative clinical practice guidelines for chronic kidney disease in children and adolescents: Evaluation, classification, and stratification. Pediatrics 111 : 1416 –1421, 2003[Abstract/Free Full Text]
  14. Cockcroft DW, Gault MH: Prediction of creatinine clearance from serum creatinine. Nephron 16 : 31 –41, 1976[Medline]
  15. Schwartz GJ, Haycock GB, Edelmann CM Jr, Spitzer A: A simple estimate of glomerular filtration rate in children derived from body length and plasma creatinine. Pediatrics 58 : 259 –263, 1976[Abstract/Free Full Text]
  16. Counahan R, Chantler C, Ghazali S, Kirkwood B, Rose F, Barratt TM: Estimation of glomerular filtration rate from plasma creatinine concentration in children. Arch Dis Child 51 : 875 –878, 1976[Abstract]
  17. Burnham K, Anderson D: Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach, New York, Springer-Verlag, 2002
  18. Leger F, Bouissou F, Coulais Y, Tafani M, Chatelut E: Estimation of glomerular filtration rate in children. Pediatr Nephrol 17 : 903 –907, 2002[CrossRef][Medline]
  19. Clinical Practice Guidelines for Chronic Kidney Disease: Evaluation, classification, and stratification. Guideline 4. Estimation of GFR. Am J Kidney Dis 39 : S76 –S92, 2002[CrossRef]
  20. Thienpont LM, De Leenheer AP, Stockl D, Reinauer H: Candidate reference methods for determining target values for cholesterol, creatinine, uric-acid, and glucose in external quality assessment and internal accuracy control. II. Method transfer. Clin Chem 39 : 1001 –1006, 1993[Abstract/Free Full Text]
  21. Thienpont L, Franzini C, Kratochvila J, Middle J, Ricos C, Siekmann L, Stockl D: Analytical quality specifications for reference methods and operating specifications for networks of reference laboratories. Discussion paper from the members of the external quality assessment (EQA) Working Group B1 on target values in EQAS. Eur J Clin Chem Clin Biochem 33 : 949 –957, 1995[Medline]
  22. Rose BD: Clinical Physiology of Acid-Base and Electrolyte Disorders. Renal Circulation and Glomerular Filtration Rate, New York, McGraw Hill, 1984
  23. Clinical Practice Guidelines for Chronic Kidney Disease: Evaluation, classification, and stratification. Guideline 5. Assessment of proteinuria. Am J Kidney Dis 39 : S93 –S102, 2002
  24. Rehling M, Moller ML, Thamdrup B, Lund JO, Trap-Jensen J: Reliability of a 99mTc-DTPA gamma camera technique for determination of single kidney glomerular filtration rate. A comparison to plasma clearance of 51Cr-EDTA in one-kidney patients, using the renal clearance of inulin as a reference. Scand J Urol Nephrol 20 : 57 –62, 1986[Medline]
  25. van Acker BA, Koomen GC, Arisz L: Drawbacks of the constant-infusion technique for measurement of renal function. Am J Physiol 268 : F543 –F552, 1995[Medline]
  26. 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]
  27. Bokenkamp A, Domanetzki M, Zinck R, Schumann G, Byrd D, Brodehl J: Cystatin C—A new marker of glomerular filtration rate in children independent of age and height. Pediatrics 101 : 875 –881, 1998[Abstract/Free Full Text]
  28. Morris MC, Allanby CW, Toseland P, Haycock GB, Chantler C: Evaluation of a height/plasma creatinine formula in the measurement of glomerular filtration rate. Arch Dis Child 57 : 611 –615, 1982[Abstract]
  29. Springate JE, Christensen SL, Feld LG: Serum creatinine level and renal function in children. Am J Dis Child 146 : 1232 –1235, 1992[Abstract]
  30. Clinical practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Guideline 1. Definition and staging of chronic kidney disease. Am J Kidney Dis 39 : S46 –S64, 2002[CrossRef]
  31. 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 141 : 929 –937, 2004[Abstract/Free Full Text]
  32. Smith H: Measurement of the filtration rate. In: The Kidney. Structure and Function in Health and Disease, New York, Oxford University Press, 1951 , p 59
  33. Filler G, Foster J, Acker A, Lepage N, Akbari A, Ehrich JH: The Cockcroft-Gault formula should not be used in children. Kidney Int 67 : 2321 –2324, 2005[CrossRef][Medline]
  34. Piepsz A, Denis R, Ham HR, Dobbeleir A, Schulman C, Erbsmann F: A simple method for measuring separate glomerular filtration rate using a single injection of 99mTc-DTPA and the scintillation camera. J Pediatr 93 : 769 –774, 1978[Medline]
  35. LaFrance ND, Drew HH, Walser M: Radioisotopic measurement of glomerular filtration rate in severe chronic renal failure. J Nucl Med 29 : 1927 –1930, 1988[Abstract/Free Full Text]
  36. Carlsen JE, Moller ML, Lund JO, Trap-Jensen J: Comparison of four commercial Tc-99m(Sn)DTPA preparations used for the measurement of glomerular filtration rate: Concise communication. J Nucl Med 21 : 126 –129, 1980[Abstract/Free Full Text]
  37. Piepsz A, Colarinha P, Gordon I, Hahn K, Olivier P, Sixt R, van Velzen J: Guidelines for glomerular filtration rate determination in children. Eur J Nucl Med 28 : BP31 –BP36, 2001[Medline]



This article has been cited by other articles:


Home page
 Nephrol Dial TransplantHome page
C. M. Soares, J. S. S. Diniz, E. M. Lima, G. R. Oliveira, M. R. Canhestro, E. A. Colosimo, A. C. Simoes e Silva, and E. A. Oliveira
Predictive factors of progression to chronic kidney disease stage 5 in a predialysis interdisciplinary programme
Nephrol. Dial. Transplant., October 13, 2008; (2008) gfn547v2.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
ASN.2005010034v1
17/2/487    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Mattman, A.
Right arrow Articles by White, C. T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Mattman, A.
Right arrow Articles by White, C. T.

HOME CURRENT ISSUE ARCHIVES JASN Express ONLINE SUBMISSION AUTHOR INFO
EDITORIAL BOARD SUBSCRIBE FEEDBACK ALERTS HELP