Renal function (GFR or creatinine clearance) as estimated by serum creatinine and 24-hour urine collection should in theory closely correlate.

However, I have run into a case where they differ greatly. The serum creatinine was measured twice at two separate labs and were in normal range at 0.89 and 1.0 mg/dL, yielding a calculated CrCl of 115 and approx 90. However, the 24-hour urine creatinine clearance results yielded 38 ml/min and 5 ml/min, which would be either acute kidney injury / acute renal failure, or chronic kidney disease / failure (CKD stage 3b and CKD stage 5 respectively). The test was again repeated a third time and again resulted in 1.0mg/dl but this time a CrCl 70 mg/dl.

This is a huge difference with large ramifications in terms of diagnosis and treatment. The question is:

What could be the reason for this, and which measurement is the more reliable one?

I've read several studies that concluded that the serum creatinine alone was actually a better predictor of GFR than urine creatine clearance over 24, 4, and 2h. Every cohort study I found online shows an upper margin of error of around 30% where predicted values from serum creatinine fall and measured clearance levels.

There are various factors at play here:

  1. The GFR to serum creatinine is a hyperbolic relation which is a bad function for estimating values greater than 60 ml/min and even as low as 45 ml/min: hyper]([![enter image description here]1 As you can see the entire range of values for GFR is estimated over a small fraction of .5 mg/dl difference whereas values lower than 30 range over 20 mg/dl serum creatinine.

  2. The jaffe assay which was used reacts creatinine with picric acid in alkaline solution and spectrophotometrically compared to a baseline creatinine sample with are both measured against a blank. This particular reaction is not as precise as the less common enzymatic assay which does not react with with nonspecific targets

"The nonspecificity of Jaffe's reaction causes falsely elevated creatinine results in the presence of protein, glucose, acetoacetate, ascorbic acid, guanidine, acetone, cephalosporins, aminoglycosides (mainly streptomycin), ketone bodies, α-keto acids, and other organic compounds." -wiki

  1. The standardization of measurements within labs and between labs was found to introduce about 5% error(and in 1 sample 25%) when 5 samples over the entire GFR range were sent to 26 different instruments: enter image description here

After reading many studies online it appears the estimated GFR calculated by formulas are rather inaccurate predictors of GFR, and the creatine clearance is a rather imprecise measurement as well. Here's one study showing standard errors that span 100% when comparing calculated vs measured gfr although it suffers from having only 100 subjects: enter image description here One large cohort study of 45000 has good news for young healthy individuals with low creatinine clearance values: The lower the value the slower the decline with age so that even at the lowest 2.5% of 70Ml your average projection is not kidney failure 40 years later: enter image description here

  • 1
    This could be a good question about the correlation between serum creatinine and urine creatinine in accurately determining renal function, but unfortunately sounds like a personal advice request - especially with the "is the test valid" portion. Please present this question more in line with how a professional/student would, including relevant background and reference, or it may end up being closed as an advice request.
    – DoctorWhom
    Commented Aug 19, 2019 at 6:12
  • @DoctorWhom ok, i'll try
    – none
    Commented Aug 19, 2019 at 10:38
  • @DoctorWhom did I make the question more pedagogical or worse?
    – none
    Commented Aug 19, 2019 at 10:47
  • This is improved. It would be best if you included the link to the studies you reference, but this is a more academic level question. Given your efforts, I will further edit to aid you - please just add the reference.
    – DoctorWhom
    Commented Aug 19, 2019 at 22:12
  • Thinking again, the overreaction of jaffe doesn't apply here as that would increase creatinine results and the issue is a lower amount of creatinine detected.
    – none
    Commented Aug 21, 2019 at 5:25

1 Answer 1


A lot of creatinine clearance calculations exist, and a lot depends on the patient (ethnic origin, sex, age, height, weight), so any discrepancies can be considered normal.

The point of multiple calculations led me to the Wikipedia article on creatinine, and to the section on serum creatinine, which admits that the two tests produce incompatible results, saying:

A rise in blood creatinine concentration is a late marker, observed only with marked damage to functioning nephrons. Therefore, this test is unsuitable for detecting early-stage kidney disease.

going on to suggest a better test, eGFR:

A better estimation of kidney function is given by calculating the estimated glomerular filtration rate (eGFR)

On the other hand, urine creatinine tests are used in doping tests:

Creatinine concentration is checked during standard urine drug tests.

so perhaps your patient cheated the testing process somehow to avoid being discovered.

  • 1
    Although this is to a degree true, the results usually are not so dramatically different. There is a way to diagnose kidney disease, stage the degree of it, and determine subsequent treatment. For this reason, he/she is asking which is the more reliable method of determining renal function, and why would they differ so greatly.
    – DoctorWhom
    Commented Aug 19, 2019 at 22:25
  • I've seen your posts only in the mathematics stack exchange so maybe I would have gotten a better answer if I included some of the actual estimation equations, pearson's coefficients instead of the biological aspects. The calulated value is 4 sd's below the average 107 which is midway between the two estimated values. Unfortunately cohort studies are hard to find. Of the first four studies from google/scholar, 3 had less than 100 patients and the pearson coefficient ranged from .25 to .79, making it impossible to tell how accurate one can expect the formula estimates to be. The largest study.
    – none
    Commented Aug 21, 2019 at 9:47
  • I found so far has only 293 patients and states that P30 (percentage of estimates within 30%) was 86% nature.com/articles/s41598-019-40416-w. Thus it's plausible that the 70 ml test made in the 3rd lab could could plausibly fall in the 14% larger than 30% error.
    – none
    Commented Aug 21, 2019 at 9:51
  • These are comparing the mdrd and Cockkroft gault formula values to creatinine clearance. A more appropriate study to look for would compare creatinine clearance to inuline or other precise marker to see how accurate it is which might explain the large range of values for creatinine clearance, but I haven't put enough search time in yet to find one.
    – none
    Commented Aug 21, 2019 at 10:01
  • Just found one (with 146 patients) and r^2=.44 making CC also a poor estimator in that study. sjkdt.org/…
    – none
    Commented Aug 21, 2019 at 10:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.