NDT Advance Access originally published online on June 16, 2007
Nephrology Dialysis Transplantation 2007 22(10):2894-2899; doi:10.1093/ndt/gfm289
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Use of GFR equations to adjust drug doses in an elderly multi-ethnic group—a cautionary tale
Division of Nephrology, UBC, Centre for Health Evaluation and Outcome Sciences, and Department of Pharmacy, St Paul's Hospital, Vancouver BC
Correspondence and offprint requests to: Dr Adeera Levin, 1081 Burrard Street, Providence Wing Rm 6010A, Vancouver, BC, V6Z 1Y8. Email: alevin{at}providencehealth.bc.ca
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Background. Glomerular filtration rate (GFR) is the best index of kidney function. Mathematical estimations of GFR, based on serum creatinine (SCr), are a clinically useful method to follow renal function, but have certain limitations which need to be considered. Convention supports the use of Cockcroft–Gault (CG) for the purposes of drug dosing. The impact of using the modification of diet in renal disease (MDRD) formula has not been formally evaluated with respect to drug dosing; especially in an elderly multi-ethnic population. A cross-sectional study of long-term care facility patients was conducted to demonstrate the impact of the use of different formulae in the elderly for the purposes of medication dosing.
Methods. Patients with ESRD were excluded. GFR was calculated for all subjects using the four-variable modified MDRD equation (re-expressed using isotope-dilution mass spectrometry-based creatinine values) and the CG equation (corrected for body surface area). Discordance was defined as a reclassification of one stage of chronic kidney disease (CKD) by using a different formula. Calculated GFR from each formula was used to calculate the doses of two drugs: amantadine and digoxin, to demonstrate the potential impact of the use of different formulae on the risk of drug toxicity.
Results. A total of 180 patients were identified with a mean age of 85 years, of which 30% were Asian. Mean MDRD-GFR and CG-GFR in the same group were different (72.9 ml/min/1.73 m2 vs 52.1 ml/min/1.73 m2). Only 37.2% of the patients were categorized in the same stage of CKD by both methods. When MDRD was used in place of CG to determine drug dose adjustments, we found that 20% fewer patients would have qualified for a dose reduction of amantadine, which would have translated to a higher total cumulative dose delivered.
Conclusions. The use of CG and MDRD provided discordant estimations in over 60% of the elderly patients. While the importance of these equations cannot be questioned, caution should be exercised in situations where they have not been prospectively validated. Therefore, their interchangeable use cannot be advocated in the dosing of medications until further prospective validations are performed.
Keywords: eGFR equations; elderly; drug dosing; kidney function; validation
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Glomerular filtration rate (GFR) is widely accepted as the preferred index of kidney function and recognized as defining chronic kidney disease (CKD) [1,2]. Mathematical estimations of GFR, based on serum creatinine concentration, have emerged as a clinically useful method to follow kidney function [2–4].
These equations are routinely used in many centres to guide dosing of medications based on kidney function. With the evolution and availability of multiple formulae, there is a potential for confusion in clinical practice. The Cockcroft–Gault (CG) equation has been the traditional method of estimating GFR for drug dosing, reflected by instructions in drug package inserts and pharmaceutical practice.
Since October 2003, all laboratories in British Columbia report an estimated GFR with every serum creatinine assessment, using the modification of diet in renal disease (MDRD) formula. The MDRD study equation, which uses age, sex and race (but not weight), along with serum creatinine (SCr), has been more widely utilized in those areas where computerized laboratory estimation of GFR has been advocated, partly due to the fact that weight is not required in its calculation. When corrected for creatinine calibration and body surface area in CKD patients, the correlation of CG and MDRD is very high (r2 = 0.92) [5]. Accordingly, a re-expressed four variable MDRD formula, calibrated to isotopic dilution techniques for creatinine has been endorsed for mathematical estimations for GFR [6]. However, neither MDRD formula, nor the CG has been formally validated for use in calculating drug dosages for patients with varying degrees of renal dysfunction. Despite this, CG appears to be used by most pharmacists.
Since implementation of automatic reporting of MDRD calculated GFR in British Columbia, pharmacists have noted large discrepancies in drug doses calculated using MDRD and CG equations in elderly nursing home patients. With increasing awareness of the need for improved assessment of kidney function with regards to drug dosing, there is understandable concern about whether to change the conventional use of CG in drug dosage adjustment to the use of the readily available MDRD.
The objective of this study was to describe the degree and direction of discordance between the CG and MDRD calculations of GFR in an elderly, multi-ethnic population and its impact on medication dosing with the broader aim that some direction for future research and understanding could be developed.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
A cross-sectional study was conducted. Institutional Ethical Committee approval was obtained prior to initiation of the study.
Study population
Patients residing at two Providence Health Care Long Term Care Facilities (LTCF) between 1 June, 2003 and 30 June, 2004 were included in the study. Patients with end-stage renal disease (ESRD) requiring renal replacement therapy were excluded. Annual SCr measurements are routinely performed at a central lab for all Providence Health Care LTCF, as is ongoing in tracking of medications and weights. As of October 2003, SCr measurements have been standardized across all laboratories in British Columbia. For all subjects in this study, lab testing was done at a single laboratory where the use of calibrated creatinines was implemented prior to the institution of the provincial standardization programme.
Study design
Baseline demographic information including age, gender, ethnicity, SCr, weight and height was abstracted from charts. Ethnicity was self-reported by patients or family members and categorized as Caucasian, Asian and other. Asian patients included people of Chinese and Japanese descent. The ‘other’ ethnic group included patients of Indian and African descent.
Methods of serum creatinine determination
The method for creatinine determination used at the laboratory is an enzymatic colorimetric assay produced by Roche Diagnostics and was one of the first creatinine methods to be standardized against isotope-dilution mass spectrometry (IDMS). The concentration of creatinine present is measured photometrically on the Bayer Advia 1650 automated analyser. This method is quite specific and does not have non-creatinine chromogens, unlike the Jaffe chemical method. All laboratories in British Columbia subscribe to the creatinine standardization programme provided by Healthmetrx [7]. As part of the programme, we analysed three Healthmetrx creatinine specimens with reference values determined by IDMS on three runs to determine if any further correction factor was required with this method. On all three specimens the lab's results were within 2 µmol/l of reference value, therefore, no correction factor was required for the SCr. Note that criteria considered acceptable for the BC Healthmetrx program is reference value eGFR ± 20%.
Methods of GFR estimation
The following formulae were used for the purposes of this article:
Re-expressed four variable MDRD equation using IDMS-based creatinine values (MDRD-GFR): (ml/min/1.73 m2)
175 x standardized SCr–1.154 x age–0.203 x 1.212 [if black] x 0.742 [if female]
Cockcroft–Gault corrected for body surface area (BSA) (CG-GFR): (ml/min/1.73 m2)
[140 – age] x weight(kg)/ SCr (µmol/l) x 1.21 x 0.85 [if female] x BSA/1.73 m2
Cockcroft Gault equation (CG-CrCl): (ml/min)
[140 – age] x weight(kg)/ SCr (µmol/l) x 1.21 x 0.85 [if female]
MDRD unadjusted for BSA (MDRD-CrCl): (ml/min)
175 x SCr–1.154 x age–0.203 x 1.212 [if black] x 0.742 [if female] x 1.73 m2/BSA
Definitions of discordance
Mean MDRD-GFR and CG-GFR was compared in the same group of patients. Specifically, re-expressed four variable MDRD equation using IDMS-based creatinine values and the CG formula for creatinine clearance corrected for BSA were used for direct comparison in the analysis. To further evaluate the discordance between the two formulae in this population, we classified patients by K/DOQI-recommended CKD stages using each formula. Discordance was defined, for the purposes of this analysis, as a reclassification of the stage of CKD of at least 1 based on the estimation formula used. Lastly, to demonstrate the impact of the discordance on drug dosing, the dosages of two commonly utilized drugs, amantadine and digoxin, were calculated according to pharmacy recommendations using each estimation formula and were compared [8,9]. As dosing recommendations for these medications are based on creatinine clearance (ml/min), both the MDRD and CG formulas were unadjusted for BSA.
Statistical analysis
Means (±SD) and frequencies were used to describe data, as appropriate. The t-tests, chi-squared tests and analysis of variance (ANOVA) with trend analyses were used to compare continuous and categorical variables, as appropriate. All statistical analyses were performed using Stata v9.2.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Patient baseline characteristics
A total of 180 patients were identified from two LTCF, representing the entire population (no patients were receiving dialysis). The mean age was 85 years and 78.2% were female. Fifty-four patients (30%) were Asian, 122 (67.7%) were Caucasian and 4 (2.2%) were classified as other. Table 1 describes the key characteristics of the cohort with respect to weight, height, BMI, BSA, SCr and calculated GFR by the four different methods. Notably, the patients were of small physical size (with a mean BMI of 22.2).
|
Since a significant portion of the population was Asian, baseline characteristics were compared between the Asian and non-Asian groups. Asian patients were lighter, shorter and had lower BSA compared with the non-Asian patients. However, there were no significant differences in age, BMI or mean SCr. Mean CG-CrCl tended to be lower in the Asian group, however this did not achieve statistical significance. Furthermore, there was no difference in the CG-estimated GFR, once it was adjusted for BSA.
The distribution by stage of CKD using the MDRD-GFR and CG-GFR is shown in Figure 1.
|
Discordance between CG and MDRD
The mean MDRD-GFR and CG-GFR differed substantially in the same group of patients, with MDRD estimating a higher level of kidney function compared with CG (72.9 ml/min/1.73 m2 vs 52.1 ml/min/1.73 m2).
In total, only 67 patients (37.2%) were classified in the same stage of CKD regardless of the formula used. MDRD calculated a higher GFR when compared with the CG calculation, resulting in a reclassification in the stage of CKD by one stage in 106 people (58.9%) and by two stages in seven people (3.9%). Figure 2 outlines these findings in detail.
|
Ethnicity of discordant and concordant groups
Patients were then stratified by ethnicity. Asian patients trended towards having a greater frequency of discordance as compared to non-Asian patients (70.4 vs 59.2% discordance, P = 0.16), but this was not statistically significant. Note that the Asian patients had lower weights as compared with the non-Asian population.
Impact on drug dosing—amantadine
Current dosing guidelines for amantadine require adjustments based on estimations of kidney function (Table 3). Convention supports the use of CG to calculate the drug doses. If MDRD had been used (even when unadjusted for BSA) to calculate the creatinine clearance in place of CG, 126 (70%) patients would have required a downward adjustment in the dose, compared with 164 (91.2%) if CG were used. Also, the total cumulative dose of amantadine used by all patients would have been significantly greater using the MDRD formula (94.6 g) than if the CG formula were used (70.1 g) (P < 0.01).
|
Impact of drug dosing—digoxin
Maintenance dosing guidelines for digoxin require adjustment for renal impairment. Those with CrCl <50 ml/min require a reduction in dosage. Again, current practice over the last three decades has been to use the CG formula. Using this formula the majority of patients in our study population (104/179 patients) would have undergone downward dose adjustment for this drug with a narrow therapeutic window. However, use of the MDRD formula would have led to adjustments in only 46 patients. Thus, 58 patients (32.2%) in this population would not have had their dose adjusted downward and may have received a higher dose of digoxin if MDRD had been used in place of CG for the purposes of drug dosing.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
In this analysis of 180 elderly multi-ethnic patients, we describe the potential for complications with the interchangeable use of two independently validated equations for the estimation of renal function. The two formulae used in the same group of patients classified <40% of patients in the same stage of CKD, with a significant difference in mean MDRD-GFR and CG-GFR. Furthermore, using MDRD in place of CG for the purposes of drug dosing, fewer patients would have required downward dosage adjustments. In the case of amantadine, this would potentially result in patients receiving a significantly larger cumulative drug dose. Without actual measures of kidney function, we cannot say which equation approximates true kidney function, and thus which equation and dosing regimen would have resulted in higher or lower drug toxicity effects. Alternatively, the different estimates, and thus dosing, may result in reduced doses such that therapeutic efficacy may be compromised.
In this small-sized elderly population mean MDRD-GFR was substantially higher than the CG-GFR. This may reflect an underestimation of GFR by the CG formula due to the small physical size of the cohort and the importance of weight in the CG calculation. Again, as validation of the equations was not the purpose of this study, the issue of which is more accurate cannot be answered.
What we have demonstrated is that in a very elderly multi-ethnic population these formulae are not interchangeable, and thus we suggest that caution be used when determining doses until proper studies have been conducted. Previous studies have examined the use of these equations in the elderly, with variable results. Lamb et al. [10] published an observational study comparing the performance of MDRD and CG with a gold standard measurement in a largely Caucasian elderly population. The study suggested that MDRD and CG are both valid predictors of GFR in an elderly population. In a later publication, the same group compared MDRD and CG using different SCr measurements in elderly patients (mean age 80 years) and reported that regardless of the serum creatinine method used, CG tended to underestimate GFR, while MDRD overestimated GFR compared with EDTA GFR estimation [11]. Our study population was even older than the population in the Lamb study.
This finding was also reported by Foissart et al. [12], who compared the use of MDRD and CG to Cr-EDTA GFR in a cohort of over 2000 European subjects. In all age groups, including a subgroup of subjects aged 65 years or older, MDRD underestimated GFR relative to Cr-EDTA GFR. However, to our knowledge, the use of these equations in a very elderly multi-ethnic cohort has not been assessed and needs to be studied in a prospective manner with the use of gold standard measurement of GFR.
It is undisputed that the formulae for estimating kidney function are important tools for the purpose of increasing awareness of early kidney disease, helping clinicians to interpret serum creatinine and identifying earlier moderate CKD. However, there are important caveats to the use of these equations. Neither equation has been validated in all ethnic groups nor in the extremes of age, nor in those with non-Western dietary habits. Furthermore, we must consider the utility of these equations in the context of emerging methods of assessment of kidney dysfunction in the elderly which have important implications on the use of these equations. Potentially, lab information system programmes could be developed whereby estimates of GFR were not reported in those over 80 years of age, and cystatin C measurements were substituted instead. The logistics of this should be explored or better methods of estimating should be developed.
We demonstrated a thought experiment using a clinical scenario of drug dosing of commonly used drugs in the elderly to highlight these caveats. However, we were not able to quantify the absolute magnitude of the impact of drug toxicities. The tables presented aim to demonstrate the potential impact of the use of these different formulae on cumulative doses. As stated earlier, it is not known which formula is closer to the truth, therefore, it is up to the clinicians to determine the relative risk of overdosing or underdosing any particular drug in their different patient groups.
As reporting of GFR estimations become routine in clinical practice, it is important for physicians and allied health professionals to understand the limitations and nuances in the application of these formulae. As our understanding of renal functional assessment evolves and newer methods of assessment become available, we must consider which equation should be used in which cohort of patients and should aim to employ a more targeted approach in our use of these equations. To that end, prospective studies evaluating the predictive ability of MDRD and CG in elderly, multi-ethnic populations against a measured GFR as control are essential. Furthermore, the impact on dosing based on either formula should be examined using clinical outcomes of adverse events and serum drug levels in an attempt to identify the most accurate formula in this patient population.
Several limitations of this study have been noted. Given the retrospective study design, we were not able to assess potential co-factors that may have influenced SCr levels or kidney function, such as diet, use of nephrotoxic medications, and co-existing disease. Furthermore, the absence of gold standard measurements of GFR does not allow us to conclude which equation is a better estimate of kidney function: we can simply conclude that they are discordant in this population.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
The use of the CG and MDRD formulae to estimate kidney function provided discordant estimations in over 60% of elderly multi-ethnic patients in Vancouver long-term care facilities. The interchangeable use of these equations could result in significant differences in the cumulative doses provided to the population depending on the equation used, and is therefore not recommended. Currently, given the absence of a gold standard validated equation in this elderly, multi-ethnic population we feel that caution is warranted when using these formulae for the purpose of drug dosing.
Conflict of interest statement. None declared.
|
65 years of agea
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
- Smith H. Comparative physiology of the kidney. In:. In: Kidney: Structure and Function in Health and Disease (1951) New York: Oxford University Press. 5250.
- K/DOQI. Clinical Practice Guidelines for CKD: Evaluation, classification, and stratification. Kidney Disease Outcome Quality Initiative. In: Am J Kidney Dis (2002) 39:S1.[CrossRef][Web of Science][Medline]
- Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron (1976) 16:31.[Web of Science][Medline]
- 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.
[Abstract/Free Full Text] - Coresh J, Astor BC, McQuillan G, 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.[CrossRef][Web of Science][Medline]
- Levey AS, Coresh J, Greene T, et al. 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.
[Abstract/Free Full Text] - http://www.ceqal.com/services.php.
- Compendium of Pharmaceuticals and Specialties. (2006).
- McGeer ASD, Tamblyn SE, Kolbe F, Orr P, Aoki FY. Use of antiviral prophylaxis in influenza outbreaks in long-term care facilities. Can J Infect Dis (2000) 11:187–192.
- Lamb EJ, Webb MC, Simpson DE, Coakley AJ, Newman DJ, O'Riordan SE. Estimation of glomerular filtration rate in older patients with chronic renal insufficiency: is the modification of diet in renal disease formula an improvement? J Am Geriatr Soc (2003) 51:1012.[CrossRef][Web of Science][Medline]
- Lamb EJ, Wood J, Stowe HJ, O'Riordan SE, Webb MC, Dalton RN. Susceptibility of glomerular filtration rate estimations to variations in creatinine methodology: a study in older patients. Ann Clin Biochem (2005) 42(Pt 1):11.[CrossRef][Web of Science][Medline]
- Froissart M, Rossert J, Jacquot C, Paillard M, Houillier P. Predictive performance of the modification of diet in renal disease and Cockcroft-Gault equations for estimating renal function. J Am Soc Nephrol (2005) 16:763.
[Abstract/Free Full Text]
Accepted in revised form: 17. 4.07
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. A Chapman, A. J Kaufenberg, P. Anderson, A. Khokhar, and C. G Schrock Safety and Effectiveness of a Modification of Diet in Renal Disease Equation-Based Potassium Replacement Protocol Ann. Pharmacother., March 1, 2009; 43(3): 436 - 443. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. P. Moranville and H. R. Jennings Implications of using modification of diet in renal disease versus Cockcroft-Gault equations for renal dosing adjustments Am. J. Health Syst. Pharm., January 15, 2009; 66(2): 154 - 161. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P B. Bookstaver, J. W Johnson, T. P McCoy, D. Stewart, and J. C Williamson Modification of Diet in Renal Disease and Modified Cockcroft-Gault Formulas in Predicting Aminoglycoside Elimination Ann. Pharmacother., December 1, 2008; 42(12): 1758 - 1765. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. J. Glassock and C. Winearls Screening for CKD with eGFR: Doubts and Dangers Clin. J. Am. Soc. Nephrol., September 1, 2008; 3(5): 1563 - 1568. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||