Professional Documents
Culture Documents
Age and Ageing, Volume 40, Issue 5, 1 September 2011, Pages 583589,
https://doi.org/10.1093/ageing/afr011
Published: 10 March 2011
Abstract
Background: detecting chronic kidney disease (CKD) may have important implications for the
management of older and frail people. We aimed at investigating whether clinical setting
(nursing home: NH versus hospital: H) affects the agreement between glomerular filtration rate
(GFR) values estimated by Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI),
Cockcroft-Gault (CG) and Modification of Diet in Renal Disease (MDRD) equations.
Design: observational study.
Setting: comparison between NH residents and H patients.
Subjects: we used data from 177 NH residents, and 439 H patients.
Methods: the agreement between estimating equations and the odds of a discrepancy >25%
between formulas in relation to setting (NH versus H) were investigated.
Results: the agreement between MDRD and CKD-EPI formulas was good either in NH (k = 0.82)
or H (k = 0.87) patients, while corresponding figures for CG indicate only a fair agreement with
CKD-EPI (k = 0.50 for both populations). Setting (NH versus H) was associated with
discordance between MDRD and CKD-EPI (OR = 3.97; 95% CI = 1.759.01), but not between CG
and EPI (OR = 1.25; 95% CI = 0.871.81).
Conclusions: in NH residents, MDRD and CKD-EPI formulas yield highly concordant GFR
values, but CG behaves differently in up to one-third of patients. Such findings have important
implications in dosing drugs cleared by the kidney. Setting should be taken into consideration
in studies for validation of GFR equations.
Keywords: glomerular filtration rate, nursing home, hospital, drugs, elderly
Issue Section: Research Papers
Introduction
Chronic kidney disease (CKD) is very common in nursing home (NH) residents [1] and
represents a risk factor for drug overdosage and adverse drug reactions (ADRs) [2].
Unfortunately, sarcopenia, whose prevalence in NH residents ranges between 40 and 85%
[3], reduces creatinine production making serum creatinine (Scr) a poorly reliable marker
of CKD, whereas logistic difficulties prevent the measurement of the glomerular filtration
rate (GFR) by radionuclide or iothalamate-based methods in NH residents. These
methodological problems impact guidelines for drug dosing in these patients. Indeed, most
of the regulatory authorities mandate the use of the Cockcroft-Gault (CG) formula, which
estimates creatinine clearance (CCr) and antedates the Modification of Diet in Renal
Disease (MDRD) by 23 years, to adjust dosing, but recently the MDRD has been occasionally
suggested as the reference equation [4]. More important, following recommendation by the
National Kidney Foundation [5], the MDRD-derived GFR is more and more frequently part
of computer generated laboratory reports. This might improperly lead to adjust drug
dosing according to MDRD equation that actually estimates something different from what
CG measures (i.e. GFR versus CCr). Finally, a third equation, the Chronic Kidney Disease
Epidemiology Collaboration (CKD-EPI) [6], has recently been proposed as more accurate
than the MDRD in the upper range of GFR values and is gaining growing popularity, but it
has been developed and tested in a youngadult population.
On average, MDRD seems more accurate than CG in the CKD population, while CG formula
should be preferred to the MDRD for estimating renal function in at-risk subjects with
normal Scr, e.g. diabetics and stage 1 and 2 CKD patients, and in elderly patients [7].
However, it is unknown whether in NH residents CG and MDRD are plagued with a risk of
misclassification comparable with that observed in hospitalised and severely diseased
patients [8]. Therefore, the present study was aimed at comparing results provided by CG,
MDRD and CKD-EPI formulas in NH residents and to verify whether discrepancy mirrors
that characterising elderly people admitted to acute care medical wards (H). The final
objective of our analysis was to assess to which extent the discrepancy in estimated CCr or
GFR impacts adjustment of drug dosing.
Methods
NH patients
Data were collected in a 80-beds skilled NH facility in Taranto, Italy, as previously
described [9]. Reasons for admission to the NH were most frequently acute disabling
conditions (e.g. hip fracture or stroke) in patients with several comorbid conditions and
little or no rehabilitation potential. Over a 8-month period, 210 patients were initially
screened. Of them, 30 patients were excluded because we could not reliably measure
weight and/or height, and 3 patients were excluded because of missing values for any of
the variables used to calculate estimated GFR, leaving 177 patients for the analysis.
H patients
We used data from a collaborative observational study group, the PharmacosurVeillance in
the elderly CarePVC, based in community and university hospitals located throughout
Italy, aimed at surveying drug consumption, occurrence of ADRs and quality of hospital
care [10]. Overall, 690 patients were enrolled in the study. Twenty-five patients who died
during hospital stay were excluded from the analysis, as were patients enrolled in long-
term care/rehabilitation units (n = 159), and patients having missing values for any of the
variables used to calculate estimated GFR (n = 67), leaving a final sample of 439 patients
for the analysis. All of them were successfully tracked during the follow-up period.
Renal function estimates
Scr was measured by standardised Jaffe method in all laboratories of participating centres
while patients clinical status was stable, and in the absence of dehydration or
hyperhydration. The cut-off used for Scr was 1.26 mg/dl in males and 1.04 mg/dl in
females [11]. CCr was estimated by CG [12], and GFR through MDRD [13] and CKD-EPI [6]
formulas.
Since the MDRD and CKD-EPI formulas are corrected for body surface area (BSA), while CG
formula is not, we resolved to adjust CG-estimated values for BSA in order to minimise
discrepancies between the three different methods.
Analytic approach
Outcomes of this study were: the agreement between CCr/GFR formulas, the impact of
setting on discrepancies between formulas and the implication of discrepancy on drug
dosing.
In both study populations, demographics, socioeconomic and clinical data were collected
together with detailed information on pharmacological therapy, cognitive status (mini-
mental state examination, MMSE), mood (Geriatric Depression Scale, GDS) and disability
(basic activities of daily living, ADL). The number of diagnoses was used as an index of
overall comorbidity. Selected diagnoses known to be associated with CKD, such as
hypertension, congestive heart failure, diabetes, COPD, coronary heart disease and
cerebrovascular disease, were also separately considered in the analysis. Body mass index
and serum albumin were also included in the analysis as indices of nutritional status.
Second, in order to evaluate whether the risk of overestimating renal function was setting-
dependent, we estimated the odds of an absolute discrepancy (>25%) between MDRD and
CKD-EPI in the NH compared with the H population. The latter analysis was repeated using
the CG-BSA formula instead of the MDRD.
Finally, we provided some examples of how differences in the estimated CCr or GFR as a
function of the formula used translates in drug dosage adjustment [16, 17]. To this aim, we
selected out renally cleared cardiovascular drugs and antibiotics which are commonly
prescribed in elderly and frail population of nursing home residents [18].
All analyses were performed using SPSS (version 10.0; SPSS, Inc., Chicago, IL, USA).
Results
NH patients H patients P-
(n = 177) (n = 439) value
Dependent in at least 1
BADL 52.5 30.3 0.001
The average agreements of the MDRD and CG-BSA formulas with the CKD-EPI formula are
shown in Table 2, upper panel: the MDRD equation on average yielded higher and CG-BSA
lower values than CKD-EPI, both in the NH and H populations.
Table 2.
Average agreement (upper panel) and agreement between MDRD-, CG- and CKD-
EPI-based classification of renal function (lower panel)
Upper panel
CG-
BSA 9.5 26.97.9
MDRD-
GFR 4.3 6.114.7
CG-
BSA 8.8 22.54.9
Average 95% agreement
agreement intervals
CKD-EPI
<30
(n = 15) 3059 60+ (n = 102)
% (n = 60) % % Weighted
Lower panel
MDRD
<30 73.3 0 0
CG-
BSA
CKD-EPI
<30
(n = 43) 3059 60+ (n = 209)
% (n = 187) % % Weighted
MDRD
<30 90.7 0 0
CG-
BSA
The agreement between MDRD and CKD-EPI formulas in identifying patients with normal
or mildly reduced renal function, moderate and severe CKD was good either in NH or H
patients, while corresponding figures for CG-BSA indicate only a fair agreement with CKD-
EPI in both study populations (Table 2, lower panel).
All patients carrying discordance between MDRD and CKD-EPI had normal Scr values, the
corresponding proportion among patients with concordant values was 61.9%. Mean
estimated GFR was higher in patients with discordant values (CKD-EPI: 90.7 7.2 versus
57.6 20.6 ml/min/1.73 m2, P 0.001; MDRD: 123.6 17.2 versus
61.8 22.9 ml/min/1.72 m2, P < 0.001). Correlates of discordance greater than 25% of
CKD-EPI value are shown in Table 3. Dependence and setting were significantly associated
with discordance between MDRD and CKD-EPI, while female gender was a negative
correlate of discordance (Table 3, upper panel).
Table 3.
Correlates of the observed discordance between GFR formulas
Crude
Discordance 25% Discordance >25% OR
of CKD-EPI of CKD-EPI (95%
value n = 591 value n = 25 CI)
Upper panel
1.01
(0.95
Age, years 80.3 6.5 80.8 8.1 1.07)
0.34
(0.14
Gender, female 58.2 32.0 0.79)
MDRD versus CKD-EPI
Crude
Discordance 25% Discordance >25% OR
of CKD-EPI of CKD-EPI (95%
value n = 591 value n = 25 CI)
0.96
(0.88
BMI, kg/m2 25.4 4.5 24.7 2.9 1.05)
0.76
Serum albumin, (0.40
g/dl 3.6 0.6 3.5 0.7 1.41)
3.22
Dependent in at (1.40
least 1 BADL 35.5 64.0 7.42)
2.21
Cognitive (0.91
impairment 53.8 72.0 5.36)
1.08
(0.93
No. of diagnoses 5.3 2.4 5.8 2.9 1.26)
0.61
(0.27
Hypertension 67.7 56.0 1.36)
0.40
Congestive heart (0.10
failure 17.8 8.0 1.73)
MDRD versus CKD-EPI
Crude
Discordance 25% Discordance >25% OR
of CKD-EPI of CKD-EPI (95%
value n = 591 value n = 25 CI)
0.84
(0.33
Diabetes 27.4 24.0 2.13)
0.76
(0.31
COPD 33.8 28.0 1.85)
0.43
Coronary heart (0.13
disease 23.9 12.0 1.48)
1.02
Cerebrovascular (0.38
disease 19.6 20.0 2.78)
2.54
(0.91
Fracture 9.0 20.0 7.04)
0.96
(0.22
Dementia 8.3 8.0 4.20)
Study group
Crude
Discordance 25% Discordance >25% OR
of CKD-EPI of CKD-EPI (95%
value n = 591 value n = 25 CI)
3.97
(1.75
NH 27.4 60.0 9.01)
Lower panel
1.19
(1.15
Age, years 78.2 5.8 84.6 6.2 1.23)
0.59
(0.42
Gender, female 61.4 48.5 0.83)
0.85
(0.83
BMI, kg/m2 27.1 4.0 21.8 3.0 0.87)
0.61
Serum albumin, (0.45
g/dl 3.7 0.6 3.5 0.7 0.82)
MDRD versus CKD-EPI
Crude
Discordance 25% Discordance >25% OR
of CKD-EPI of CKD-EPI (95%
value n = 591 value n = 25 CI)
1.81
Dependent in at (1.28
least 1 BADL 32.0 46.1 2.56)
1.42
Cognitive (1.01
impairment 51.7 60.3 1.99)
0.90
(0.86
No. of diagnoses 5.4 2.4 5.1 2.6 1.03)
0.77
(0.54
Hypertension 69.2 63.2 1.09)
0.46
Congestive heart (0.28
failure 20.6 10.8 0.77)
0.54
(0.36
Diabetes 31.1 19.6 0.81)
0.73
(0.50
COPD 35.9 28.9 1.04)
MDRD versus CKD-EPI
Crude
Discordance 25% Discordance >25% OR
of CKD-EPI of CKD-EPI (95%
value n = 591 value n = 25 CI)
0.82
Coronary heart (0.55
disease 24.5 21.1 1.23)
0.58
Cerebrovascular (0.36
disease 22.3 14.2 0.91)
3.23
(1.86
Fracture 5.8 16.7 5.62)
1.11
(0.61
Dementia 8.0 8.8 2.03)
Study group
1.25
(0.87
NH 27.2 31.9 1.81)
The prevalence of normal Scr level was 69.1% in discordant patients and 60.7% in
concordant patients (P = 0.04) when considering CG versus CKD-EPI. Although differences
were not statistically significant, mean GFR or CCr values were higher in patients with
discordant values also in this analysis (CKD-EPI: 60.6 17.6 versus
58.1 22.8 ml/min/1.73 m2, P = 0.176; CG-BSA: 66.8 20.6 versus
63.1 27.9 ml/min/1.73 m2, P = 0.100). Setting did not affect discordance between CG-BSA
and CKD-EPI. Age, dependence and cognitive impairment were significant direct correlates
of discordance, while female gender, BMI and serum albumin, as well as selected diagnoses
such as congestive heart failure, diabetes and cerebrovascular disease qualified as negative
correlates of the outcome (Table 3, lower panel).
In the Appendix 2, Supplementary data available in Age and Ageing online, we present
dosage adjustment needed in relation to GFR or CCr values for selected drugs frequently
used in NH residents. To provide an example of how adjustment depends upon the
equation used to estimate renal function, one could consider a 85-year-old male patient
having 1.3 mg/dl Scr and a body weight of 70 kg. Such patient will have a CG-BSA-
estimated CCr of 39.0 ml/min/1.73 m2, a MDRD-estimated GFR of 52 ml/min/1.73 m2 and a
CKD-EPI-estimated GFR of 50 ml/min/1.73 m2. Such different estimates will lead to
extremely different dosage adjustment for selected drugs frequently used in elderly and
frail patients (see Appendix 1, Supplementary data available in Age and Ageing online, for
examples).
Discussion
Limitations of this study deserve consideration. First, the quality of data collection in a
multicentre population is unlikely to reach the degree of accuracy achieved in a single
centre highly skilled nursing home. Second, patients for whom no estimate of height and/or
weight was available, i.e. the most disabled and at greatest risk of renal failure and ADRs,
could not contribute to our analysis. Third, we take care of measuring Scr and weight when
the patient was normally hydrated, but we relied on clinical judgment and laboratory
analyses, not on bio-impedenziometry. Finally, our study did not include a direct
measurement of GFR. Thus, we cannot draw any definitive conclusion about the
relationship between setting and accuracy of the equations studied.
Conclusions
This study shows that in a frail NH population MDRD and CKD-EPI formulas yield highly
concordant GFR values, but CG-BSA behaves differently in up to one-third of patients. As a
general rule, it seems reasonable to adjust renally cleared drugs dosing according to the
recommendation provided by the manufacturer and, if no equation is recommended, refer
to the one proved more reliable in the reference population. Research on the validity of
available equations in NH patients, having a gold standard GFR measurement, is highly
desirable and could improve the management of a dramatically rising proportion of elderly
people.
Key points
- MDRD and CKD-EPI formulas yield highly concordant GFR values also in a frail NH
population.
- The agreement between CG-BSA and CKD-EPI is only fair in both populations.
- These findings have relevant implications, especially in dosing drugs. It seems reasonable
to adjust renally cleared drugs dosing according to the recommendation provided by the
manufacturer and, if no equation is recommended, refer to the one proved more reliable in
the reference population.
- Further research on the validity of available equations in NH patients, having a gold
standard GFR measurement, is highly desirable and could improve the management of a
rapidly expanding population.
Conflicts of interest
None declared.
Funding
The PVC study was partially supported by a grant from the Italian Ministry of Health (RF-
INR-2005-127640). A complete list of participating centres has been previously published
[10].
References
1 McClellan WM, Resnick B, Lei L, et al. . Prevalence and severity of chronic kidney disease and
anemia in the nursing home population, J Am Med Dir Assoc , 2010, vol. 11 (pg. 33-41)
https://doi.org/10.1016/j.jamda.2009.07.003
3 Bauer JM, Kaiser MJ, Sieber CC. Sarcopenia in nursing home residents, J Am Med Dir Assoc ,
2008, vol. 9 (pg. 545-51) https://doi.org/10.1016/j.jamda.2008.04.010
4 de Lemos ML, Hsieh T, Hamata L, et al. . Evaluation of predictive formulae for glomerular
filtration rate for carboplatin dosing in gynecological malignancies, Gynecol Oncol , 2006, vol. 103 (pg.
1063-9) https://doi.org/10.1016/j.ygyno.2006.06.024
5 Brosius FCIII, Hostetter TH, Kelepouris E, et al. . Detection of chronic kidney disease in patients
with or at increased risk of cardiovascular disease: a science advisory from the American Heart
Association Kidney and Cardiovascular Disease Council; the councils on high blood pressure research,
cardiovascular disease in the young, and epidemiology and prevention; and the quality of care and
outcomes research interdisciplinary working group: developed in collaboration with the National Kidney
Foundation, Circulation , 2006, vol. 114 (pg. 1083-7)
https://doi.org/10.1161/CIRCULATIONAHA.106.177321
6 Levey AS, Stevens LA, Schmid CH, et al. . A new equation to estimate glomerular filtration rate,
Ann Intern Med , 2009, vol. 150 (pg. 604-12)
7 Helou R. Should we continue to use the Cockcroft-Gault formula?, Nephron Clin Pract , 2010,
vol. 116 (pg. c172-86) https://doi.org/10.1159/000317197
8 Smilde TD, van Veldhuisen DJ, Navis G, Voors AA, Hillege HL. Drawbacks and prognostic value
of formulas estimating renal function in patients with chronic heart failure and systolic dysfunction,
Circulation , 2006, vol. 114 (pg. 1572-80) https://doi.org/10.1161/CIRCULATIONAHA.105.610642
11 Jones CA, McQuillan GM, Kusek JW, et al. . Serum creatinine levels in the US population: third
national health and nutrition examination survey, Am J Kidney Dis , 1998, vol. 32 (pg. 992-9)
https://doi.org/10.1016/S0272-6386(98)70074-5
12 Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine, Nephron ,
1976, vol. 16 (pg. 31-41) https://doi.org/10.1159/000180580
13 Levey AS, Greene T, Kusek JW, Beck GJ. MDRD Study GroupA simplified equation to predict
glomerular filtration rate from serum creatinine, J Am Soc Nephrol , 2000, vol. 11 pg. A0828
14 Altman DG. , Some Common Problems in Medical Research. Practical Statistics for Medical
Research , 1997LondonChapman & Hall
15 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, vol. 5 (pg. 1003-9) https://doi.org/10.2215/CJN.06870909
16 Munar MY, Singh H. Drug dosing adjustments in patients with chronic kidney disease, Am Fam
Physician , 2007, vol. 75 (pg. 1487-96)
17 Schmid P, Fischer AG, Wuillemin WA. Low-molecular-weight heparin in patients with renal
insufficiency, Swiss Med Wkly , 2009, vol. 139 (pg. 438-52)
18 Blass DM, Black BS, Phillips H, et al. . Medication use in nursing home residents with advanced
dementia, Int J Geriatr Psychiatry , 2008, vol. 23 (pg. 490-6) https://doi.org/10.1002/gps.1921
19 Perrone RD, Madias NE, Levey AS. Serum creatinine as an index of renal function: new insights
into old concepts, Clin Chem , 1992, vol. 38 (pg. 1933-53)
21 Gill J, Malyuk R, Djurdjev O, Levin A. Use of GFR equations to adjust drug doses in an elderly
multi-ethnic groupa cautionary tale, Nephrol Dial Transplant , 2007, vol. 22 (pg. 2894-9)
https://doi.org/10.1093/ndt/gfm289
22 Melloni C, Peterson ED, Chen AY, et al. . Cockcroft-Gault versus modification of diet in renal
disease: importance of glomerular filtration rate formula for classification of chronic kidney disease in
patients with non-ST-segment elevation acute coronary syndromes, J Am Coll Cardiol , 2008, vol. 51 (pg.
991-6) https://doi.org/10.1016/j.jacc.2007.11.045
24 Goldberg TH, Finkelstein MS. Difficulties in estimating glomerular filtration rate in the elderly,
Arch Intern Med , 1987, vol. 147 (pg. 1430-3) https://doi.org/10.1001/archinte.147.8.1430
25 Pequignot R, Belmin J, Chauvelier S, et al. . Renal function in older hospital patients is more
accurately estimated using the Cockcroft-Gault formula than the modification diet in renal disease
formula, J Am Geriatr Soc , 2009, vol. 57 (pg. 1638-43) https://doi.org/10.1111/j.1532-
5415.2009.02385.x
26 Stevens LA, Coresh J, Feldman HI, et al. . Evaluation of the modification of diet in renal disease
study equation in a large diverse population, J Am Soc Nephrol , 2007, vol. 18 (pg. 2749-57)
https://doi.org/10.1681/ASN.2007020199
27 Macdonald J, Marcora S, Jibani M, et al. . GFR estimation using cystatin C is not independent of
body composition, Am J Kidney Dis , 2006, vol. 48 (pg. 712-9) https://doi.org/10.1053/j.ajkd.2006.07.001
The Author 2011. Published by Oxford University Press on behalf of the British Geriatrics Society.
All rights reserved. For Permissions, please email: journals.permissions@oup.com