To assess someone’s kidney function, one would really like to measure their glomerular filtration rate (GFR). This can be done using a radioisotope tracer that is cleared from the blood solely by glomerular filtration, such as Cr-51 EDTA or Tc-99m DTPA. This technique is useful for research and when precise measurements of GFR are required. But it is impractical for routine repeated measurements.

Instead of measuring the clearance of an artificial tracer, routine assessment of kidney function uses the concentration of a substance produced by the body, creatinine, urea or cystatin C. Their concentration is determined by the balance between the rate of production and the rate of excretion.

When production and excretion have been stable for more than 24 h, an equilibrium concentration is reached. In equilibrium, the concentration is not affected by the volume of body water in which the substance is dissolved. Patients who are chronically fluid overloaded do not have a lower concentration due to dilution. The higher concentration found with dehydration is due to reduced kidney function, not haemoconcentration.

Creatinine would be the ideal choice for measuring kidney function if it was produced at a constant rate, freely filtered by the glomeruli, and neither reabsorbed nor secreted by the tubules. Unfortunately, it does not pass all these tests.

Creatinine is released into the bloodstream at a fairly constant rate from the breakdown of creatine in healthy skeletal muscles and is freely filtered by the glomeruli. In addition, 10–20 % of normal total creatinine excretion is by secretion into the tubules. As GFR declines that proportion increases and so serum creatinine does not increase as much as predicted from filtration. This is shown by comparing simultaneous measurements of serum creatinine and GFR using the gold standard inulin clearance technique (see Fig. 2.1).

The curved relationship between creatinine and GFR makes changes in serum creatinine hard to interpret correctly. Figure 2.2 shows a graph of serum creatinine against time in a man with declining kidney function. The shape of the line suggests that the rate of loss of kidney function accelerates over the years.

If we now add in GFR values, it is clear that kidney function has actually declined at a constant rate over the whole period (see Fig. 2.3).

Because of the inverse curved relationship between GFR and creatinine, the drop in GFR from 116 to 60 during the first 5 years causes a smaller increase in serum creatinine than the decline from 60 to 30 over the next 3 years.

To make serum creatinine results easy to interpret we need to convert them into estimates of the GFR (eGFR ). Equations to do the conversion have been derived from databases containing simultaneous measurements of GFR, by a radioisotope clearance technique, and serum creatinine concentration. The four-variable MDRD equation was derived from measurements performed on adult patients with chronic kidney disease in the Modification of Diet in Renal Disease (MDRD) study [2].

eGFR values are standardised to a body surface area of 1.73 m² and so their calculation does not require the patient’s body weight. This makes it possible for the laboratory to provide eGFR values automatically from the patient’s age and sex. As the laboratory does not usually have reliable information about ethnicity, the adjustment for Afro-Caribbean race is done manually.

For children, different equations are available [3]. The commonest used is the Schwarz equation [4].

The 90 % confidence intervals of the estimates are wide. Using the MDRD equation, 90 % of estimates are within 30 % of the directly measured GFR and 98 % within 50 %. In children using the Schwarz equation, 80 % of estimates are within 30 % of the measured GFR. The estimates are less accurate in people with very abnormal body composition, such as amputees and malnourished patients.

The MDRD equation was derived from patients known to have kidney disease and a raised serum creatinine. It is inaccurate for estimating GFR from serum creatinine values in the normal range, i.e. outside the dataset used to derive the formula, for example in pregnancy [5]. The eGFR is not appropriate for situations in which a more accurate measurement of GFR is required, such as in the assessment of a potential transplant donor.

Because the MDRD equation underestimates the true GFR when serum creatinine values are in the normal range, people are at risk of being labelled as having kidney disease based upon these estimates. To reduce this, guidelines recommend that laboratories use the more recently developed CKD-EPI equations [6]. CKD-EPI gives estimates that are closer to the measured GFR at lower levels of serum creatinine, applying a different equation when creatinine is in the normal range [7, 8].