Population based screening for chronic kidney disease: cost effectiveness study

Introduction

The cost effectiveness of general population screening for chronic
kidney disease (CKD) using estimating glomerular filtration rate (eGFR)
has recently been assessed and published 1, and discussed 2, in the BMJ.
The conclusion was that although population based screening is not cost
effective, targeted screening of people with diabetes 'seems attractive'.
Such screening requires an appropriate tool and recently concerns have
been raised, by the original authors, regarding the continued
appropriateness of the most commonly used MDRD equation for calculating
eGFR. They have suggested, and validated in a general population, a
modification of that equation.

Diabetic nephropathy is a major cause of kidney disease and CKD
staging is an important consideration in monitoring patients with
diabetes. This is usually done by estimating glomerular filtration rate by
calculation from serum creatinine measurements. Currently, the most widely
used formula for this is the 4-variable MDRD equation3, but a more recent
formula proposed by the same authors, the CKD-EPI equation4, is claimed to
give more accurate results. Although about 29% of subjects in the
populations used for the development of the equation were described as
having diabetes, it has not been extensively studied specifically in
patients with diabetes, a population with a high prevalence of CKD.

As we suspected that the MDRD formula might be under-estimating GFR
in some of our diabetic patients, thereby potentially precipitating
premature referral into renal services and promoting undue anxiety among
patients and their carers. We have compared the results of applying these
two equations for estimating GFR in a large, mixed population of patients
with diabetes attending a large district general hospital diabetes centre.

Patients and Methods

Approval for the study was given by the North Manchester Research
Ethics Committee.
We applied both the MDRD and CKD-EPI equations to a population of 1073
diabetic patients comprising 338 with type 1 and 733 with type 2 diabetes
mellitus. These patients were randomly selected from those attending a
specialist care general diabetes clinic. Their ages ranged from 18 to 95
years with a mean of 58 years and a median of 60 years. We classified
patients into CKD stages according to their eGFR results calculated by the
two equations using serum creatinine concentrations. Creatinine was
measured by the local laboratory on an Abbott Architect analyser using a
kinetic method based on a modified Jaffe reaction calibrated to values
assigned by isotope-dilution mass spectrometry. As we did not include
other markers of kidney disease, such as urinary albumin, in this study we
have not distinguished between Stage 0 (no evidence of kidney disease) and
Stage 11 patients.

Results

In this group of 1073 diabetic patients we found the median CKD-EPI
result was 71.4 mL min-1 1.73m-2 (95% CI 69.9 to 72.9) and was 5.6 mL min
-1 1.73m-2 greater than the median MDRD result of 65.8 mL min-1 1.73m-2
(95% CI 64.5 to 67.1). In terms of CKD staging this had a significant
effect, as can be seen from the Table. Using CKD-EPI, rather than the MDRD
equation, 135 more patients fell into the Stage 0/1 category, with 91
fewer in Stage 2 and 49 fewer in Stage 3 compared with MDRD staging.
Results were similar in those with type 1 and type 2 diabetes. The
predominant trend was to reclassify patients into less severe stages of
CKD. Very few moved into more advanced stages. Those who did were all
borderline cases and all but one in MDRD Stage 3 or higher.

Discussion and Conclusions

Levey and colleagues developed the CKD-EPI equation4 in response to
suggestions that their earlier MDRD equation3 was inaccurate in subjects
with normal or mildly impaired renal function. In their work-up of the
CKD-EPI equation they found that it appeared to be more accurate in terms
of agreement with measured GFR and classified fewer individuals than MDRD
into Stages 2 and 3, while the numbers in Stages 4 and 5 were less
affected. Our study on diabetic patients supports their findings with
respect to the staging. Whilst we cannot comment on relative accuracy, as
we did not measure GFR by a reference method, our data supports the
increasing belief that the MDRD equation underestimates eGFR in a
proportion of patients, including those with diabetes.

Use of the CKD-EPI equation for our patients resulted in 26.8% being
classified as Stage 0/1 compared with only 14.3% by the MDRD equation.
This was associated with a reduced prevalence from 47.0% to 38.5% in Stage
2 and a reduction from 38.7% to 34.6% in Stage 3 or higher, i.e. the group
requiring specialist referral and more rigorous monitoring. However, it is
still of concern that 33.8% of patients randomly selected from those
attending a specialist care general diabetes clinic were classified as
having some significant form of CKD (Stage 3 or 4) when compared to the
prevalence of 6.7% observed by Levey et al in NHANES, a survey of a cross-
section of the non-institutionalised US population4. The mean age of that
population was not given and may have differed from ours, but the NHANES
age bands clearly showed diminishing renal function with age, as indicated
by CKD-EPI. Between 40-50y of age 2.1% of subjects were in CKD Stages 3
or 4; age 60-69y 10.8% and age 70 and above 38%. Nevertheless, our figures
underline the impact of diabetes (both type 1 and type 2) on renal
function and the potential burden on renal services.

Our findings, from a mixed UK diabetes clinic population, add to the
body of evidence on the CKD-EPI equation from the original work of Levey
et al4. Their work provided evidence in the general population that the
CKD-EPI equation gives a more accurate estimate of GFR than the MDRD
version by formulating their new equation using data from iothalamate
clearance studies. Recently, Sabanayagam et al.5 applied the CKD-EPI
equation to a multi-ethnic Asian population and, like us, found it
resulted in the classification of a greater number as Stage 1 CKD with
reduced prevalence of Stage 2. Accurate evaluation of renal function is of
relevance and importance in many areas of medicine including when used as
a safety criterion for certain medications and interventions. The
accumulating data in the literature suggest that the CKD-EPI equation will
gain acceptance and supersede the MDRD equation, our findings support such
a move. There remains the task of convincing laboratories of the need to
modify the software in their computer systems to achieve uniformity
nationally and internationally wherever possible.

References

1)Manns B, Hemmelgran B, Tonelli M, Au F, Chiasson TC, Doug J,et al.
Population based screening for chronic kidney disease: cost effectiveness
study. BMJ 2010; 341:c5869.

2)Kiberd B. Screening for chronic kidney disease. BMJ 2010; 341:c5734.

3) Levey AS, Coresh J, Greene T, Stevens LA, Zhang YL, Hendriksen S, et
al. Using standardized serum creatinine values in renal disease study
equation for estimating glomerular filtration rate. Ann Intern Med 2006;
145: 247-54.

4) Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI,
et al. A new equation to estimate glomerular filtration rate. Ann Intern
Med 2009; 150: 604-12.

5) Sabanayagam C, Wong TY, Shyong E. The CKD-EPI equation and MDRD study
equation find similar prevalence of chronic kidney disease in Asian
populations. Ann Intern Med 2009; 151: 892-3.

Table: Comparison of numbers of patients attributed to different CKD Stages when eGFR results are calculated using the conventional 4-variable MDRD equation and the new CKD-EPI equation applied to a population of 1073 diabetic subjects.