Glomerular filtration rate (GFR) is the best single measure of kidney function. A subnormal GFR is an important early indicator of possible chronic kidney disease. However, it also does not identify the cause. It is also not a ‘catch all’, for example, in early diabetes mellitus, GFR will typically be normal or even increased, due to ‘hyperfiltration’.
Optimal detection and risk stratification of early kidney disease also requires measurement of urine albumin, rather than just eGFR.
Creatinine clearance (qv) has been used to estimate GFR, but requires an accurate 24-hour urine collection, making it an unsuitable screening test.
Serum creatinine (qv) alone has often been used as a simple screen but is strongly influenced by muscle mass, diet and other factors, which are hard to quantify. A ‘normal’ result may be falsely reassuring in patients with low muscle mass, especially older women. Therefore other ‘simple’ ways to indirectly estimate GFR have been widely sought.
The CKD-EPI (standing for ‘Chronic Kidney Disease Epidemiology Collaboration’) formula is now widely used throughout Australia, as recommended by a recent MJA position statement.
This is an improvement from the previous MDRD (‘modification of diet in renal disease’) formula and has been validated in an Australian context (White, 2010). Specifically, it shows better correlation with true GFR, especially in the 60-90mL/min/1.73m2 range.
This results in more appropriate stratification of individuals for long-term clinical risk of CVD, all-cause mortality, and end stage kidney disease. eGFR values using this formula are slightly higher in patients with normal or near-normal kidney function, especially in younger patients (<60 years), resulting in less misclassification of patients as having CKD (fewer with calculated eGFR <60mL/min/1.73m2). Conversely, older patients (men >70 years and women >75 years) have slightly lower eGFR results.
Calculation of GFR using the CKD-EPI formula is available through the Kidney Health Australia website, accessible at https://kidney.org.au/health-professionals/detect/calculator-and-tools#section-gfr-calculator.
Most laboratories now automatically provide a calculated eGFR using this formula when serum creatinine is requested, however, it may still be unreliable in some circumstances.
Specifically, the formula is unreliable and should not be used in the following situations:
|Groups where the CKD-EPI formula is unreliable and should not be used or interpreted with caution:|
|Children under age 18||formula not validated in this group eGFR generally not reported;( see Schwartz formula, above)|
|Rapidly changing kidney function, e.g. acute kidney injury||if eGFR is <60mL/min/1.73m2, consider situations where it may be unreliable and repeat in 14 days|
|Patients on dialysis
|creatinine varies widely pre- and post-dialysis. (eGFR usually not reported if laboratory is aware)|
|formula not validated (eGFR not reported if laboratory is aware). Use serum creatinine as standard test.|
|Extremes of body size||assumed relationship of muscle mass to weight and circulating plasma volume is perturbed|
|Very high or very low muscle mass:||falsely high eGFR: skeletal muscle diseases with significant loss of muscle bulk, as well as those with paraplegia and amputees
falsely low eGFR: very high muscle mass (e.g. body builders)
|Unusual diets||falsely high eGFR, e.g. vegetarians (low serum creatinine)
falsely low eGFR, e.g. protein or creatine supplements, recent cooked meat meal (high serum creatinine; see text)
|Severe liver disease||associated with reduced liver formation of creatinine from creatine, falsely high eGFR|
|Drug effects||e.g. trimethoprim, cimetidine, probenicid (reduced tubular creatinine secretion); fibrates (mechanism unclear). For more discussion on drug effects, see Creatinine, serum|
|Ref: Johnson, et al (2012)|
Sources of inaccuracy
The formula is dependent on accurate serum creatinine measurement. While assays are now internationally standardised, an assay biased slightly high also gives a falsely low eGFR; one biased low gives a falsely high eGFR (there is a worldwide move to improve creatinine assays).
The CKD-EPI formula only gives an estimate. Typical measurement uncertainty for serum creatinine is 3.5%, which is acceptable, since patients may vary from day to day (biological) by 7%. However, when these variations are combined with the formula error of up to 15%, the total error of eGFR may be up to ±20% of true GFR.
The CKD-EPI formula provides better accuracy than the MDRD equation and can be used up to 90mL/min/1.73m2 and most labs now report a calculated eGFR up to this level. However, the formula is less reliable over 90mL/min/1.73m2 and a numerical result may not be provided by the laboratory.
The CKD-EPI formula is less well validated in non-European groups, but appears reliable with minimal bias in South East Asians, Africans, Indian and Chinese patients living in Western countries (Stevens, 2011). It also appears useful in indigenous Australians, whose ‘linear’ body build is associated with less muscle and more fat for a given weight than African Americans. The 2012 Position Statement recommended it be used ‘prudently’ in non-European groups.
Use for drug dosing
Unlike the Cockcroft-Gault formula, no extra information, such as patient height or weight, is needed for the CKD-EPI calculation and the result is normalised to a patient of average body surface area (1.73m2). This underestimates the ‘real’ (unadjusted) GFR in large patients and overestimates it in small patients. To find an ‘uncorrected’ result, for example, for drug dosing, a correction factor must be applied and height and weight then do become necessary.
Precise dose-adjustments for renally excreted drugs are seldom needed (usually only in a hospital setting) in which case published formulae or creatinine clearance measurements should be used. In general practice, drug dose adjustments, when needed, are more approximate and based on, say, a halving of the dose for GFR below a certain threshold. The eGFR result is usually sufficient in such situations.
Effect of ACE inhibitors (ACEi) and angiotensin receptor blockers (ARBs)
These will be expected to lower eGFR due to their effects on renal microvascular circulation with a reduction in glomerular filtration pressure. If the eGFR falls more than 25%, then Kidney Health Australia recommends that the drug should be stopped and nephrologist consultation considered, to exclude bilateral renal artery stenosis.
Note, however, that other causes of such a fall in eGFR to consider also include aggressive diuretic therapy, worsening systolic heart failure and introduction of some other drug affecting creatinine concentration. For a list of such drugs, see Creatinine, serum; Kidney disease, acute and chronic.
By age 65, about 25% of patients have an eGFR below 60mL/min/1.73m2, rising to 50% over age 80. However, age itself should not be considered a physiological or age-appropriate explanation for an eGFR <60mL/min/1.73m2, since the risk of adverse outcomes (for example, cardiovascular and all-cause mortality, renal disease progression) is also increased in patients over age 65 with reduced eGFR.
Fasting status and recent meat intake
The formula was derived in patients who were not required to be fasting and the possible influence of recent meat intake is included within the eGFR calculation. However, a recent cooked meat meal can raise serum creatinine by 20umol/L after 1-2 hours.
Kidney Health Australia recommends (2012) that if the initial eGFR is <60mL/min/1.73m2, follow-up measurements of serum creatinine (and eGFR calculation) should be performed when the patient has fasted and, most importantly, has not eaten a cooked meat meal for at least four hours prior.
The approach to a mildly reduced eGFR result
No further investigation is needed in the 60-90mL/min/1.73m2 range, which does not by itself indicate kidney disease. However, other findings, such as albuminuria, active sediment, clinical history and examination, may justify further evaluation and nephrologist consultation.
If the reported eGFR is <60mL/min/1.73m2, then situations causing it to be unreliable or misleading should be considered (see table above). The patient should be retested, preferably on a fasting sample, after 14 days. Note that age alone should not be assumed as an explanation for a reduced eGFR.
Possible nephrotoxic drugs, over the counter medications and herbal agents that the patient may be taking should also be considered. For a list of common ones, see Kidney disease, acute and chronic; Herbs.
If the follow-up eGFR after 14 days is stable (within ±20%), then the eGFR should be repeated again within three months. If reduced eGFR is present on at least three occasions for a minimum of ≥3 months and is stable, then the patient has chronic kidney disease.
If the initial reduced eGFR is confirmed on repeat, then additional tests to consider include (Kidney Health Australia, Management of Chronic Kidney Disease in General Practice, 2012):
Referral to a nephrologist should be considered.
Resource websites (Accessed March 2018)
Kidney Health Australia www.kidney.org.au
U.S. National Kidney Foundation www.kidney.org
U.S. National Kidney Disease Education Program www.nkdep.nih.gov
General Practice Pathology is a regular column each authored by an Australian expert pathologist on a topic of particular relevance and interest to practising GPs. The authors provide this editorial free of charge as part of an educational initiative developed and coordinated by Sonic Pathology.