Serum Creatinine

Brenner & Rector’s The Kidney, 6th ed., Copyright ? 2000 W. B. Saunders Company

Creatinine is a metabolic product of creatine and phosphocreatine, which are both found almost exclusively in muscle. Thus, creatinine production is proportional to muscle mass and varies little from day to day. However, production can change over longer periods of time if there are changes in muscle mass. Although diet ordinarily accounts for only a relatively small proportion of overall creatinine excretion, it is another source of variability in serum creatinine levels. Creatine from ingested meat is converted to creatinine and can be the source of up to 30% of total creatinine excretion. Thus, variability in meat intake can also contribute to variability in serum creatinine levels. The conversion of creatine to creatinine can occur with cooking. Since creatinine is readily absorbed from the gastrointestinal tract, ingesting cooked meat can lead to a rapid increase in serum creatinine levels.

Creatinine is small (mw 113 daltons), does not bind to plasma proteins, and is freely filtered by the renal glomerulus. However, it has long been appreciated that creatinine is also secreted by the renal tubule. Secretion is a saturable process that probably occurs via the organic cation pathway and is blocked by some commonly used medications including cimetidine, trimethoprim, pyrimethamine, and dapsone. If tubular secretion of creatinine were constant, differences in serum creatinine and renal clearance could still reflect differences in glomerular filtration rate. However, evidence suggests that the secretion of creatinine varies substantially both in the same individuals over time and between different individuals. Particularly troublesome is the fact that the proportion of total renal creatinine excretion due to tubular secretion increases with decreasing renal function, which could have a dampening effect on serial measurements in individuals, because glomerular filtration rate could fall more rapidly than indicated by either serum creatinine or creatinine clearance.

While proportional tubular secretion of creatinine increases with decreasing glomerular filtration rate, total urine creatinine excretion actually declines Indeed, it has been shown that increased extrarenal creatinine degradation may be sufficient to entirely account for the decrease in urine creatinine excretion associated with declining glomerular filtration rate. The extrarenal degradation of creatinine has been attributed to its conversion to carbon dioxide and methylamine by bacteria in the intestine. The increase in extrarenal creatinine degradation with declining renal function can be expected to cause plasma creatinine to underestimate declines in glomerular filtration rate.

A number of methods are used to measure creatinine. The original Folin-Wu method used the Jaffe reaction, and the Jaffe reaction has been used with various modifications since. The method of Hare involved the isolation of creatinine by absorption on Lloyd’s reagent. More recently, the direct alkaline picrate method of Bronsnes and Taussky has been used. This method involves the complexing of creatinine with alkaline picrate and measurement using a colorimetric technique. The Jaffe reaction has also been adapted for use on autoanalyzers. Other methods currently in use employ O-nitrobenzaldehyde (Sakaguchi reaction) and imidohydrolase.

There is probably more variation in what laboratories report as the upper limit of normal for serum creatinine than for any other standard chemistry value. In the absence of procedures to remove noncreatinine chromogens, the upper limit of normal measured by the Jaffe reaction can be as high as 1.6 to 1.9 mg/dL for adults. The upper limit of normal for serum creatinine measured by autoanalyzer or the imidohydrolase methods is usually 1.2 to 1.4 mg/dL. Some laboratories report separate normal ranges for men and women and for adults and children.

A number of normal plasma constituents can interfere with creatinine measurement. Glucose, fructose, pyruvate, acetoacetate, uric acid, ascorbic acid, and plasma proteins can all cause the Jaffe colorimetric assay to yield falsely high creatinine values. The low levels of these substances generally do not interfere with the Jaffe assay of creatinine in urine. Normally, interfering chromogens increase the creatinine result by about 20%, but in some disease states the interference can be much greater. In diabetic ketoacidosis, for example, spurious elevations in serum creatinine can be significant. Cephalosporin antibiotics can also interfere with the Jaffe reaction. In marked renal insufficiency, serum creatinine increases and noncreatinine chromogens contribute proportionally less to the total reaction. In individuals with normal renal function, noncreatinine chromogens made up 14% (range 4.5% to 22.3%) of the total, while in individuals with serum creatinine levels ranging from 5.6 to 29.4 mg/dL, noncreatinine chromogens contributed only 5% (range 0% to 14.6%) to the total measured level. This same study found no effect of the noncreatinine chromogens on the variability of plasma values.

Several modifications in the classic Jaffe assay have been designed to remove interfering chromogens before analysis, including deproteinization with specific adsorption of creatinine using Fuller’s earth and ion-exchange resins, the measurement of Jaffe-positive chromogens before and after the destruction of creatinine with bacteria, and dialysis separation. These methods have largely been replaced by less costly and more convenient autoanalyzer techniques. Autoanalyzer methods utilize the Jaffe reaction, but separate creatinine from noncreatinine chromogens by the rate of color development, thus avoiding most of the interference seen with the standard Jaffe method. However, very high serum bilirubin levels can cause falsely low creatinine levels. Newer techniques measuring true serum creatinine give plasma levels that are slightly lower than those from the Jaffe assay method. The imidohydrolase method can be perturbed by extremely high glucose levels and by the antifungal agent 5-flucytosine.

Serum creatinine is probably the most widely used indirect measure of glomerular filtration rate, its popularity attributable to convenience and low cost. Unfortunately, serum creatinine is very insensitive to even substantial declines in glomerular filtration rate. Glomerular filtration rate measured by more accurate techniques (described later) may be reduced by up to 50% before serum creatinine becomes elevated. In addition, the correct interpretation of serum creatinine in the clinical setting is problematic. Failure to consider variation in creatinine production due to differences in muscle mass frequently leads to misinterpretation of serum creatinine levels. This confusion may be compounded by the use of standard normal ranges for creatinine levels that appear on routine laboratory reports.

For example, a serum creatinine that falls in the “normal” range may indicate a normal glomerular filtration rate in a young, healthy individual. However, the same serum creatinine in an elderly individual could indicate a twofold reduction in glomerular filtration rate due to a comparable reduction in muscle mass.

Muscle mass can also decline over a relatively short period of time. For example, significant declines in creatinine excretion were seen in renal transplant patients, especially those who had chronic declines in allograft function. The decline in creatinine excretion was likely due to decreases in muscle mass from multiple causes, including the effects of corticosteroids. As a result of this decline in muscle mass, changes in serum creatinine underestimated the amount of decline in renal function.

Failure to remember the potential effects of tubular secretion on serum creatinine, especially in patients with reduced renal function, may lead the clinician to believe that renal function is better than it actually is. Moreover, the potential for interference from plasma constituents and medications requires the clinician to know what assay is being used to measure serum creatinine. Based on whether the reported upper limit of normal for adults is high (1.4 to 1.9 mg/dL) or low (1.2 to 1.4 mg/dL), it may sometimes be possible to correctly surmise whether an unmodified alkaline picrate-Jaffe reaction (higher normal limits) or a newer method that removes interference with chromogens (lower normal limits) is being used. The clinician should also be aware of the precision of the assay. Precision is commonly measured by the coefficient of variation, which is the mean of replicate samples divided by the standard deviation.

Creatinine Clearance

Measuring creatinine clearance obviates some of the problems of using serum creatinine as a marker of glomerular filtration rate but creates others. Differences in steady-state creatinine production due to differences in muscle mass that affect serum creatinine should not affect creatinine clearance. Extrarenal elimination of creatinine should also have little influence on the ability of creatinine clearance to estimate glomerular filtration rate. However, the reliability of creatinine clearance is greatly diminished by variability in tubular secretion of creatinine and by the inability of most patients to accurately collect timed urine samples. Indeed, some have argued that creatinine clearance is a less reliable measure of glomerular filtration rate than serum creatinine and that the use of the creatinine clearance should be abandoned.

Tubular secretion of creatinine gives a creatinine clearance rate that overestimates the true glomerular filtration rate. The overestimation is reduced somewhat if serum and urine creatinine are both measured by the Jaffe method. As discussed, plasma constituents tend to falsely raise the serum creatinine measured by the Jaffe assay, while urine creatinine levels are largely unaffected. Thus, creatinine clearance determinations calculated from serum and urine creatinine levels measured with the Jaffe assay tend to be falsely low. In a given population of patients, this error will tend to cancel the error introduced by tubular creatinine excretion, and the creatinine clearance will more closely resemble the true glomerular filtration rate. However, the two errors are independent, and the occurrence of opposing errors of the same magnitude in the same patient is largely a result of chance. Thus, variability in the precision of creatinine clearance as an estimate of true glomerular filtration rate is not reduced and may be increased by this fortuitous combination of errors. Indeed, the creatinine clearance determined in 30 patients with a total chromogen method was only 9% higher than insulin clearance, whereas the true creatinine clearance was 31% higher. However, the correlation coefficient with inulin clearance compared to the true creatinine clearance was much better (0.96) than that of inulin clearance compared to the total chromogen creatinine clearance (0.86), suggesting that the latter technique was more accurate but less precise.

Prolonged storage of the urine can introduce error in the creatinine clearance determination by perturbing urine creatinine levels. High temperature and low urine pH enhance the conversion of creatine to creatinine in urine. Indeed, storing urine under adverse conditions for 24 hours was shown to cause a 20% increase in the amount of measured urine creatinine. This problem can be obviated by refrigerating urine samples and by measuring the urine creatinine level without undue delay.

Tubular secretion of creatinine would cause little difficulty if it were constant and a constant correction factor could be subtracted from creatinine clearance determinations to yield a more accurate estimate of glomerular filtration rate. Unfortunately, interpatient and intrapatient variability in tubular creatinine secretion makes this impossible. The tendency for tubular secretion to increase proportionally with declining levels of renal function, for example, decreases the usefulness of creatinine clearance determinations to accurately reflect glomerular filtration rate in patients with renal disease.

As discussed in the Urea Clearance section, all renal clearance techniques that rely on measuring a marker of glomerular filtration rate in the urine are subject to the vagaries of urine collection. Variability in the adequacy of timed urine samples can introduce substantial error in the clearance determination. Carrying out collections under direct supervision can enhance the accuracy of timed collections. However, decreasing the collection time may increase the contribution of errors due to incomplete bladder emptying, especially if urine volumes are not increased with water loading. In addition, short-interval urine collections negate the advantages of time-averaged glomerular filtration rate estimates made from 24-hour urine collection. The cost of the procedure can also be substantially higher if trained personnel are used to directly supervise urine collections in a clinic setting.

In principle, the renal clearance of creatinine is the urine creatinine excretion divided by the area under the plasma creatinine concentration-time curve over the period of time in which the urine was sampled. In practice, creatinine clearance is usually measured by determining the urine creatinine excretion and sampling a single plasma creatinine value. It is then assumed that the plasma creatinine was constant over the time of the urine collection. Plasma creatinine remains relatively constant over 24 hours if food intake and activity are also constant. However, in a 24-hour period there may be substantial variability in plasma creatinine levels, largely due to effects of diet. Thus, under usual clinical conditions, the assumption that plasma creatinine levels are constant during the period of urine collection may not be valid and may in fact be a source of error.

The day-to-day coefficient of variation for serum creatinine is approximately 8%. Since two creatinine determinations must be made to calculate a creatinine clearance, the coefficient of variation of the creatinine clearance should be higher than that of serum creatinine. Indeed, the coefficient of variation of creatinine clearance could be expected to be at least 11.3% (the square root of 2 times the square of 8%). This is in fact similar to the coefficient of variation for creatinine clearance reported in at least one investigation. Others have reported a day-to-day coefficient of variation for creatinine clearance, when carried out in the routine clinical setting, as high as 27%.


Since tubular secretion of creatinine is a major limitation on creatinine clearance, several investigators have tried to enhance the accuracy of creatinine clearance by blocking tubular creatinine secretion with the H2 receptor antagonist cimetidine. In these studies, cimetidine substantially improved the creatinine clearance estimate of glomerular filtration rate in patients with mild to moderate renal impairment. However, in many patients, tubular secretion of creatinine was not completely blocked, and the cimetidine-enhanced creatinine clearance still overestimated glomerular filtration rate in these individuals. Moreover, only a limited number of patients have been studied so far, and the optimal dosing schedule for cimetidine has not yet been determined.

A cimetidine-enhanced creatinine clearance measurement requires no more cooperation from the patient than a standard creatinine clearance. Cimetidine is very safe; indeed, it was recently reported that the incidence of adverse reactions during prolonged treatment of 622 patients with cimetidine (10.9%) was similar to that seen during treatment of 516 patients with placebo (10.1%). Moreover, the cimetidine-enhanced creatinine clearance can be measured in most clinical laboratories. Thus, the technique may be especially useful for patients who live in areas where more expensive glomerular filtration rate measurement techniques are not readily available. Although it will not replace more accurate methods for measuring glomerular filtration rate, the cimetidine-enhanced creatinine clearance could prove to be a cost effective alternative in many clinical situations.


The need to collect a urine sample remains a major limitation of the creatinine clearance technique, with or without cimetidine enhancement. Therefore, there have been many attempts to mathematically transform or correct serum creatinine so that it may more accurately reflect glomerular filtration rate (Table 26-3). Under ideal conditions, glomerular filtration rate, as measured by a marker such as creatinine, should be equal to the inverse of the creatinine multiplied by a constant rate of creatinine excretion. Thus, in an ideal situation changes in inverse creatinine should be directly proportional to changes in glomerular filtration rate. However, the ideal situation is just that–ideal. Changes in creatinine production, extrarenal elimination, and tubular secretion of creatinine can all create errors in the use of inverse creatinine to measure changes in glomerular filtration rate. Indeed, none of the shortcomings of using serum creatinine as a marker of glomerular filtration rate are avoided by using inverse creatinine.

One of the problems with using creatinine or its inverse as a measure of glomerular filtration rate is that interpatient and intrapatient differences in creatinine production often

TABLE 26-3 — Formulae for Estimating Glomerular Filtration Rate Using Serum Creatinine and Other Clinical Parameters




(100/Cr) – 12 if male (80/Cr) – 7 if female

mL/min/1.73 m2


(Wt ? (29.3 – 0.203 ? Age) ? (1.00 – 0.03 ? Cr))/(Cr ? 14.4), if male (Wt ? (25.3 – 0.175 ? Age) ? (1.00 – 0.03 ? Cr))/(Cr ? 14.4), if female



(98 – 16 ? (Age – 20)/20)/Cr, multiply by 0.90 if female

mL/min/1.73 m2


((140 – Age) ? (Wt))/(72 ? Cr), multiply by 0.85 if female


Cockcroft & Gault

((145 – Age)/Cr) – 3, multiply by 0.85 if female

mL/min/70 kg


(27 – (0.173 ? Age))/Cr, if male (25 – (0.175 ? Age))/Cr, if female



7.57/(Cr ? 0.0884) – 0.103 ? Age + 0.096 ? Wt – 6.66, if male

mL/min/1.73 m2


6.05/(Cr ? 0.0884) – 0.080 ? Age + 0.080 ? Wt – 4.81, if female

(height2 )

170 ? Cr-.999 ? Age-.176 ? (0.762 if female) ? (1.180 if black) ? SUN-.170 ? Alb.318

mL/min/1.73 m2


Cr = serum creatinine (mg/dL); Wt = body weight (kg); SUN = serum urea nitrogen (mg/dL); Alb = serum albumin (g/dL).

occur. Variations in creatinine production due to age- and sex-related differences in muscle mass have been measured and have been incorporated in formulae to improve the ability of serum creatinine to estimate glomerular filtration rate. The most widely used formula is that of Cockcroft and Gault, which reduces the variability of serum creatinine estimates of glomerular filtration measured in a population of men and women of different ages. However, the formula does not take into account differences in creatinine production between individuals of the same age and sex or even in the same individual over time. The formula systematically overestimates glomerular filtration rate in individuals who are obese or edematous. Moreover, it does not take into account extrarenal elimination, tubular handling, or inaccuracies in the laboratory measurement of creatinine that each contribute to error in the serum creatinine estimate of glomerular filtration rate.

Glomerular filtration rate has probably never been measured with more accuracy in a large population of patients than in the Modification of Diet in Renal Disease study. Levey and co-workers used the isotopically measured glomerular filtration rate determinations in this study to derive a formula for estimating glomerular filtration rate using only readily measurable clinical variables. Significantly, they derived the formula on a randomly selected subset of patients from the whole population and then tested the formula in the remainder of the population. A formula that used only plasma chemistries and patient characteristics was able to predict 90.3% of the variability in isotopically measured glomerular filtration rate.

It cannot be assumed that formulae to predict renal function derived from data from one patient population will be valid when applied to another population. For example, few diabetic individuals were included in some of the original studies that examined formulas for predicting glomerular filtration rate, and when these formulas were subsequently tested in diabetic patients, they were found by some investigators to be inaccurate. Thus, it is important to test the accuracy of a formula in each patient population that may reasonably be expected to differ from the population that was used to derive the formula. In particular, investigators should test the accuracy of the formula in populations with characteristics that can be expected to affect the variables in the formula that differ from the characteristics of the population used to derive the formula. At a minimum, a candidate formula should probably be tested in adults, children, diabetics, and nondiabetics and among different races. If the formula is to be used in renal transplant recipients, it should probably be tested in that population as well.

Print This Post Print This Post