Kidney Markers

Urea

Blood urea nitrogen

The waste product of protein breakdown, made in the liver and cleared by the kidneys.

It is a rough indicator of kidney function but a poor one on its own, because it moves with hydration, protein intake, bleeding into the gut and liver function.

A meal changes it, and not in the direction usually claimed

The common statement is that urea falls after eating, and that this is why a kidney panel is better taken fasted. Both halves need correcting.

After a standard light mixed meal, urea does fall slightly at 1 and 2 hours. Measured alongside creatinine and uric acid, only uric acid moved by an amount that would change how a result is read. The urea fall was real and trivial.

After a protein-heavy meal urea production rises instead. That is the same mechanism as bleeding into the gut, since blood in the gut is a protein load.

So the reason to fast before a kidney panel is not urea at all. It is creatinine, which a meal containing cooked meat raises substantially, and which in turn drags the eGFR down. The figures are in the creatinine entry below. NICE puts it as a direct instruction. Do not eat any meat in the 12 hours before a blood test for eGFR.

Where the numbers come from

There is no published healthy threshold for urea. The population reference range is assay-dependent. A typical lab interval runs about 2.5 to 7.8 millimoles per litre, and that is a reference interval and not a guideline figure.

Urea is read alongside creatinine more often than on its own. A urea raised out of proportion to creatinine is associated with dehydration, with a very high protein intake, with catabolic states, and with bleeding into the gut.

One mechanism point gets stated wrongly almost everywhere. When dehydration raises urea, filtration is genuinely reduced as well. That is what the word prerenal means. Urea rises disproportionately because the kidney reabsorbs more urea when flow through the tubules is slow, and not because filtration has been left untouched.

The ratio, and how well it actually works

The urea to creatinine ratio is described with far more confidence than the data support, so here is what is behind it.

First, the units trap. The UK ratio divides urea in millimoles per litre by creatinine in micromoles per litre and multiplies by a thousand, which gives figures of the order of 40 to 100. The American version divides blood urea nitrogen by creatinine with both in milligrams per decilitre, which gives figures of the order of 8 to 15, with a commonly quoted cut-off of 20. Almost every published threshold you will find is the American one. The 2 numbers are not interchangeable and a threshold taken from one and applied to the other is meaningless.

Second, the accuracy. The use it is best tested for is separating upper from lower gastrointestinal bleeding. At the widely quoted American threshold above 30, the ratio picks up fewer than 4 in 10 upper bleeds, and it still flags about 1 in 10 people who do not have one. Lower the threshold to 22 and it catches about two-thirds, at the cost of flagging far more people wrongly. In UK units the equivalent threshold is 100, and it identifies around 70% of upper bleeds in people already presenting with suggestive symptoms.

So the thresholds are evidence-based and modest, and they were derived in people presenting acutely and not in people taking a routine blood test. Current lab-medicine guidance goes further and advises against relying on the ratio on its own to separate reduced kidney perfusion from other causes of acute kidney injury. Separately, observational work in the seriously ill has found a high ratio there tracking with a worse outcome rather than with reduced perfusion.

High protein intake, and how little it actually moves

A high-protein diet raises urea, and the size of that effect is smaller than the internet suggests.

In resistance-trained men eating 3.32 against 2.51 grams of protein per kilogram of body weight per day, over a full year, there was no difference in urea, creatinine, eGFR or the urea to creatinine ratio. At 3.4 against 2.3 grams per kilogram over 8 weeks, nothing on a basic metabolic panel moved either.

That is 2 trials at intakes far above anything a guideline recommends, over 8 weeks and over a year, finding no measurable movement. There is no published figure for how far a given protein intake shifts urea, and no professional body publishes a urea target to judge it against.

Creatinine

A waste product of normal muscle metabolism, produced at a fairly steady rate and cleared by the kidneys, mostly by filtration and partly by active secretion in the tubules. It is the standard basis for estimating kidney function. How filtration works, and what the tubules do after it, is in Kidneys, in the Body Guide.

Because it comes from muscle, it is affected by how much muscle you carry. A heavily muscled person can have a creatinine that looks high while their kidneys are entirely normal, and a frail person can have a normal-looking creatinine while function is reduced.

That is not an inference. NICE says it directly, and names the groups. Interpret a creatinine-based eGFR with caution in adults at extremes of muscle mass, for example bodybuilders, people who have had an amputation, or people with muscle wasting disorders. Reduced muscle mass leads to overestimation of filtration, and increased muscle mass to underestimation.

What NICE does not attach to that caution is a figure, a correction factor, or an alternative. The caution is official. The quantification does not exist.

A meal containing meat, with the numbers

It rises after a meal, particularly cooked meat, because cooking converts creatine in the meat to creatinine, which is then absorbed ready-made.

The effect is large enough to matter. In UK testing, median creatinine rose from 80.5 to 101.0 micromoles per litre 1 to 2 hours after a cooked-meat meal, and the eGFR calculated from it fell from 84.0 to 59.5 millilitres per minute per 1.73 square metres. Using a standardised meal containing about 54 grams of protein, the same effect appeared at every stage of kidney disease tested, it moved more than a third of those already at stage 3a down into stage 3b, and it had gone after 12 hours of fasting.

A meal without meat has little effect. This is the basis of the NICE instruction quoted in the urea entry above, and the scale of it is the thing to hold onto. A single meal moved the eGFR further than the distance between 2 filtration categories.

Where the numbers come from

No professional body publishes a healthy threshold for creatinine itself. The population reference range is assay-dependent, and the numbers that matter are attached to eGFR, which is calculated from it.

Creatine supplementation, and what the trials actually show

Creatine raises creatinine too, by a different route from cooked meat. Meat delivers creatinine ready-made. Supplementing enlarges the body's creatine pool, and a small fraction of that pool, around 1.7% a day, converts to creatinine on its own. Same endpoint, different mechanism, different time course.

This needs saying plainly because it is common in exactly the people most likely to be looking hard at a creatinine near the top of the range. The size of the effect was established by work published in 2025 and 2026.

Creatine raises serum creatinine by about 0.14 milligrams per decilitre, roughly 12 micromoles per litre, and lowers creatinine-based estimates of filtration by around 10.8 millilitres per minute. A second look at it put the creatinine rise at 0.13, with no change in urea and no difference between periods under and over a month.

The part that settles it. Where filtration was measured directly with an injected marker instead of calculated from creatinine, it did not change at all, and what movement there was ran the other way. Albumin and protein in the urine, and urinary creatinine, were all unchanged. The effect disappears the moment filtration is measured instead of estimated, which makes it a change in creatinine metabolism and not kidney injury.

A single trial makes the same point concretely. Taking 3 or 5 grams a day for 35 days, creatinine rose and the calculated eGFR fell, every value stayed inside the reference range, and 2 markers of actual kidney injury did not move.

So on current evidence, the rise in creatinine on creatine is a change in the marker and not a change in filtration.

eGFR

Estimated glomerular filtration rate

An estimate of how much plasma the kidneys filter each minute, standardised to a body surface area of 1.73 square metres, and the standard way kidney function is described and staged.

It is calculated, not measured. The equation uses creatinine together with age and sex, which is why both are needed before it can be reported.

It inherits creatinine's dependence on muscle mass in full, including the NICE caution quoted in the creatinine entry above. Cystatin C is the second marker used when creatinine alone is thought to be misleading, and the cystatin C entry below sets out what it does and does not fix, because the usual summary of that overstates it.

Which equation, and why that is not a detail

There is more than one equation, they disagree with each other, and the disagreement is currently unresolved between the UK, Europe and the US. A reader who does not know which one produced their number cannot compare it with anything.

In the UK the equation in current use is CKD-EPI 2009, without any ethnicity adjustment. NICE's 2021 guideline specifies the CKD-EPI creatinine equation, and the UK Kidney Association resolves which version that means. Current NICE guidance is the 2009 equation, used without an ethnicity adjustment factor, and the newer CKD-EPI 2021 equation should not be used because it has not been validated in UK populations. A UK Kidney Association patient safety alert in 2023 said the same thing, that all UK labs should use CKD-EPI 2009 without an ethnicity coefficient and that CKD-EPI 2021 should not be introduced until it has been validated in a UK cohort. European lab medicine reached the same conclusion in 2022.

The ethnicity adjustment, which the UK removed in 2021

Until 2021, UK guidance told labs to adjust the eGFR upward for people of Black ethnicity. NICE removed that recommendation in its 2021 guideline. Its reasoning was that muscle mass varies within ethnic groups and not only between them, so an adjustment applied to a whole group is inaccurate for many individuals in it. The UK Kidney Association's position is that the adjustment may not be valid or accurate, that it has been shown to overestimate filtration in people identifying as Black in the UK, and that it should be removed from UK practice. A joint statement from UK kidney professional, patient and research organisations in August 2021 named the removal as the major change in the new guideline.

The US removed race from its equations at the same time and reached a different answer about what to replace it with. The joint task force of the National Kidney Foundation and the American Society of Nephrology recommended in September 2021 that the US adopt the new CKD-EPI 2021 equation, which estimates filtration without a race variable, and that cystatin C be used more often alongside creatinine.

So the UK and the US agree that race should not be in the equation, and disagree about the replacement. The UK and Europe kept the 2009 equation and dropped the coefficient. The US moved to the 2021 equation. European lab medicine has since pointed to a third equation called EKFC, which its federation argues suits European populations better than CKD-EPI 2021 does. That is a commentary on the KDIGO guideline rather than a formal position statement. Three positions, all current, all defensible on the same imperfect evidence.

UK labs are not all doing the same thing

This is the part that matters most practically and is almost never said. A national UK quality assessment scheme with over 400 participating labs found that in February 2022, at most 44% of them were correctly reporting CKD-EPI 2009. The same audit found around 32% still using the older MDRD equation, 31% still using an older creatinine method, and 15% still applying an ethnicity coefficient that had been withdrawn the year before, figures the UK Kidney Association repeated in its 2023 patient safety alert.

The practical consequence is that an eGFR depends partly on which lab ran the sample, and that 2 eGFRs from different labs are not necessarily comparable even with identical creatinine.

What chronic kidney disease actually requires

Chronic kidney disease is defined as an abnormality of kidney structure or function, present for at least 3 months, with implications for health.

A reduced filtration estimate is one route into that definition and not the only one. It is possible to meet it with a completely normal eGFR and an abnormal urine result. It is equally possible to have a filtration estimate in the 60 to 89 band and not meet it at all. The UK Kidney Association puts the second half explicitly. An eGFR between 60 and 90 is mildly reduced kidney function, and it requires other evidence of kidney disease before it counts as chronic kidney disease.

That distinction carries a lot of weight, because the category name sounds like a diagnosis and is not one.

NICE confirms a new eGFR below 60 by repeating the test within 2 weeks, and the diagnosis needs the abnormality to persist across at least 2 occasions at least 90 days apart.

Where the numbers come from

KDIGO's 2024 guideline sets the categories used to describe kidney filtration, all in millilitres per minute per 1.73 square metres. G1 is 90 or above and is labelled normal or high. G2 is 60 to 89 and is labelled mildly decreased. G3a is 45 to 59, G3b is 30 to 44, G4 is 15 to 29, and G5 is below 15 and is labelled kidney failure. An eGFR below 60 is flagged as low.

Two things about the categories are easy to miss. The descriptive labels matter, because normal or high and mildly decreased are not the same kind of statement as kidney failure. And the categories themselves are not new. They date from KDIGO's 2012 guideline and were carried forward unchanged in 2024.

KDIGO says explicitly not to assume chronicity from a single abnormal result, because it could reflect a recent acute event. That is the same point as the paragraph above, and it is why a single low filtration estimate from a one-off test cannot establish the diagnosis by definition. It is a reason to repeat and to see a doctor, and not a diagnosis in itself.

And full staging combines eGFR with a urine albumin-to-creatinine ratio. That is a urine test, and it is not part of any blood panel, ours or anyone else's. So a blood result alone gives you one half of the staging picture.

How much spread sits around the number

The boundaries are sharper than the measurement behind them, and the best evidence on that is recent and British.

In healthy people being assessed as prospective kidney donors in the UK, whose filtration was measured directly with an injected marker and averaged 86.5, most equations placed more than 80% of estimates within 30% of the true value. Within 10% of the true value, they managed between 29% and 57%. In other words, in exactly the kind of healthy population that buys a blood test, the best available equation landed within a tenth of the right answer about half the time. Measured filtration is still the recommendation when assessing prospective donors.

Measured against filtration determined directly, in the largest UK comparison to date and published in 2026, every creatinine-based equation overestimates on average. The best is out by 1.4 millilitres per minute. CKD-EPI 2021, the newest, is out by 13.3. The proportion of estimates landing within 30% of the true value runs from 86.0% for the best equation down to 70.1% for CKD-EPI 2021, against an accepted standard of 80%. Getting the KDIGO category right runs from 55.4% to 65.0%, so between a third and nearly half of people are put in the wrong band.

Accuracy was worst at the young end. In adults aged 18 to 25, the MDRD equation and both CKD-EPI equations overestimated most and performed least well. Peak accuracy was at ages 40 to 45. A body mass index below 20 and a creatinine below 50 micromoles per litre were both associated with substantial overestimation. And in South Asian participants, CKD-EPI 2021 overestimated measured filtration by 17.7 millilitres per minute, with only 58.3% of estimates within 30%.

Norwegian work published in 2026 found the same thing in both adults and children. Both CKD-EPI equations came out least accurate of those tested, and both overestimated in people under 25 and over 65.

The error does not always run the same way. The systematic overestimation in those studies is driven by low muscle mass. In high muscle mass NICE says the bias runs the other direction, towards underestimation. Both are true, and neither is a correction factor you can apply.

Filtration falls with age in healthy kidneys

In carefully screened healthy kidney donors, filtration declines by about 6.3 millilitres per minute per 1.73 square metres per decade. The categories above use the same fixed boundaries at every age, so they sit differently on a 25-year-old than on a 75-year-old.

The published critique of that runs in both directions, which is not what most people expect. A fixed threshold of 60 tends to over-diagnose in older people, whose filtration has fallen for reasons that carry much less risk at that age. And because the level at which risk starts to rise is higher in the young, the same fixed threshold tends to under-diagnose in young adults. One reworking of pooled cohort data put the level at which risk starts to rise at below 75 for people aged 18 to 54, below 60 for those aged 55 to 64, and below 45 for those 65 and over.

Whether the boundaries should be age-adapted is an active argument and not a settled position. A 2026 analysis in older adults showed what is at stake. Switching from CKD-EPI 2021 to either of 2 age-adapted equations gave filtration estimates 12 to 15 millilitres per minute lower, which raised the apparent prevalence of chronic kidney disease in that population from 21% to between 37% and 46%, without identifying any additional people at higher risk. A separate 2026 analysis comparing EKFC against CKD-EPI found that women were consistently reclassified into milder categories more often than men, and were 4-5 times more likely to be reclassified out of chronic kidney disease altogether, with no change in how well kidney failure was predicted.

None of that means the number is useless. It means the number is an estimate with a real spread around it, produced by an equation that is currently contested, and that a single point value carries less precision than its 2 significant figures imply.

The specific question about heavy training

The reasoning that a muscular person's creatinine-based eGFR reads falsely low is sound physiology and it is NICE's own stated caution. The direct evidence in that population is almost nonexistent, and the one study that exists does not say what the reasoning predicts.

It compared bodybuilders using anabolic steroids alongside supplements against bodybuilders using supplements alone. The creatinine-based eGFR was around 120 millilitres per minute in both groups, which is high and not low. Cystatin C-based estimates were slightly lower in the steroid group, and urine albumin and protein measures were significantly higher in that group.

So a single small study, and what it hints at is early kidney damage in steroid users, and not a falsely low estimate in trained people generally. That is the honest position. The muscle-mass caution rests on physiology and on NICE, and not on a study in that population.

The related question about protein has been tested properly, by feeding people both diets in turn. Raising protein from 15% to 25% of energy raised the cystatin C-based eGFR by about 4 millilitres per minute. A high protein intake raises filtration instead of lowering it.

How old the guidance actually is

NICE's kidney guideline was published on 25 August 2021. It has been amended 4 times since, in November 2021, March 2022, November 2022 and November 2023. None of those amendments touched the eGFR equation, the ethnicity adjustment, or cystatin C.

Against that, the largest UK validation of eGFR equations was published in August 2026, the European federation's paper on a different equation was 2025, the UK work that most bears on cystatin C reported in 2024 and 2026, and the creatine evidence is 2025 and 2026. So UK guidance on this marker predates essentially all of the evidence above, which is a statement of where things stand and not a criticism of the guidance.

Sodium

The main electrolyte in the fluid outside cells, central to fluid balance, nerve conduction and blood pressure regulation.

It is regulated within a narrow range, so genuine changes are meaningful. Low sodium, hyponatraemia, is the most common electrolyte abnormality in medicine. High sodium usually reflects dehydration.

What actually causes a low sodium in someone otherwise well

The likeliest explanation by a wide margin is a medication, and the likeliest medication is a thiazide diuretic. The risk is highest in older people, in women, and at lower body weight. The timing catches people out. It often appears soon after starting the drug, and it can equally emerge after months or years of uneventful use. Several other common drugs do it or compound it, including SSRIs, tricyclic antidepressants, carbamazepine and anti-inflammatories.

The other main route is a condition in which the body holds on to water inappropriately, which has the unwieldy name of syndrome of inappropriate antidiuretic hormone secretion and is the most common cause in people who are not in hospital. Several of the drugs above act through it.

Drinking very large volumes of water quickly will also do it, and there are 2 specific versions of that, both of which affect people who are otherwise healthy.

Endurance exercise. Exercise-associated hyponatraemia is defined as a sodium below 135 after exercise whether or not there are symptoms, and it is more common than most people assume. Across published series, around 8% of marathon runners were below 135 after the race, along with 11% of Ironman triathletes, 33% of rugby players after an 80-minute match, and 67% of a small group of ultramarathon runners during the race. The mechanism is drinking hypotonic fluid beyond thirst combined with a hormonal response that holds water back, and not losing salt in sweat.

Very high alcohol intake with almost no food. There is not enough solute in the diet to carry the water out. It has a name in the literature and a substantial case history behind it.

An underactive thyroid appears on most lists and deserves demoting. Overt hypothyroidism is present in 1.2% of people with a sodium of 136 or above, 2.4% at 130 to 135 and 3.5% at 129 or below, which is a real gradient. Read the other way, 96.5% of people with a sodium at or below 129 do not have overt hypothyroidism. It has been argued that the association is coincidence and not cause. Severe untreated hypothyroidism can do it. Ordinary treated hypothyroidism essentially does not. The thyroid's own markers are in Thyroid, in the Body Guide.

Where the numbers come from

The population reference range is assay-dependent. It is one of the narrowest ranges on any report, because the body defends sodium concentration tightly. Its within-person variation is among the lowest of any routine marker, at around 0.5 to 1%.

Worth being clear about what this measures, because the name misleads. Sodium concentration is a measure of water balance, not of salt intake and not of kidney filtration. The body holds the concentration steady by moving water instead of by moving sodium, so a low result usually means too much water relative to sodium and not too little salt. Eating more or less salt over a few days does not move it.

The one study that argues the top of the range is not inert

No professional body publishes an optimal sodium. There is one large observational study behind it, and it is unusual enough to need reporting carefully.

It followed American adults, first measured between the ages of 45 and 66, for 25 years, using serum sodium as a proxy for how well hydrated they habitually were. Those whose middle-age sodium sat above 142 millimoles per litre, which is inside the normal range on almost any report, had a 39% higher rate of developing chronic disease over follow-up. Above 144, the rate of dying earlier than expected was 21% higher. Above 142 also carried up to 50% higher odds of scoring as biologically older than their chronological age on a panel of biomarkers. The lowest-risk band was reported as 138 to 142.

The framing has to stay attached to it. This is an association and nothing more, sodium is standing in for habitual hydration rather than being something to manipulate, and no trial has tested whether changing the number changes the outcome. What it does establish is that variation inside the reference range is not as meaningless as a reference interval implies.

One measurement note, from a 2026 study of how markers hold up under posted transport conditions. Sodium is stable across the transport conditions that matter, with one exception. It drifts if a sample gets genuinely hot in transit, at around 37 degrees. That is one reason samples are best posted promptly and not left in a hot car.

Potassium

Not offered, and here is why

A major electrolyte, central to nerve conduction, muscle contraction and heart rhythm. It is one of the most clinically important results in medicine, which is why its absence from a kidney profile needs explaining.

Red blood cells are full of potassium, at around a hundred times the concentration of the fluid around them. Two separate things move it from one to the other, and only one of them is visible to the lab.

The one the lab can see. Haemolysis, meaning cells rupturing between the needle and the analyser and spilling their contents into the serum. Every gram per litre of haemoglobin freed that way adds roughly 0.27 to 0.33 millimoles per litre of potassium. A lab can measure how haemolysed a sample is and flag it.

The one it cannot. Cells hold potassium in by actively pumping it back, and that pump needs both warmth and fuel. As a sample sits, the glucose in it runs down and the pump slows. If the sample is kept cold, the pump slows further, because the cold is itself what stops it. Potassium then drifts out of entirely intact cells. Nothing ruptures, no haemoglobin is released, and there is no haemolysis for anyone to detect. The result is simply wrong, and it looks fine.

The scale of it

A 2024 study took routine outpatient samples and handled them the way real posted samples are handled, shaking them to simulate transport and holding them refrigerated. Serum potassium rose by an average of 0.48 millimoles per litre by 4 hours, 0.64 by 6 hours and 0.94 by 8 hours. At the start, fewer than 1% read above the reference range. At 8 hours, 38% did. In lithium heparin plasma it was worse still, reaching 82% of samples above range at 8 hours.

Haemolysis was negligible and explained none of it.

Eight hours is not a long time in the post.

Two things about this run counter to intuition

Refrigerating the sample makes it worse. The rise is smaller at 25 degrees than at 4, because at room temperature the pump is still working. Cold-chain shipping, which sounds like the careful option, is actively counterproductive for this one marker. UK primary care data show the same thing from the other direction. Community potassium results run higher in cold weather than in warm.

Sustained heat pushes it the other way. At around body temperature the pump overshoots and the result comes back spuriously low instead. So there is no storage temperature that makes a posted potassium trustworthy, only different directions of error.

How the sample is taken matters as much as how it travels

This is the part people can actually control, and the numbers are large.

Clenching the fist during a draw raises potassium by up to 1.0 millimole per litre, and repeated handgrip exercise by up to 1.4. A 2026 case report notes the effect is worst in the first tube drawn. A tourniquet left on for a minute or more produces a clinically significant rise. A very high platelet count adds around 0.05 per hundred units above the normal range, and a very high white cell count around 0.6.

What that adds up to in real UK community samples

Across 2002 to 2005, before phlebotomy staff were retrained to stop patients clenching, 9% of UK primary care potassium results were at or above 5.2 millimoles per litre and 0.9% crossed the threshold that triggers a phone call from the lab. After retraining, those figures fell to 6% and 0.5%. So roughly a third to a half of the above-range potassium results in UK community samples were artefact from technique alone, before a single sample had spent any time in the post.

A falsely raised potassium is not a harmless error. It is the kind of result that sends someone to A&E for a problem they do not have. And because the drift out of intact cells leaves no visible trace, a posted sample cannot reliably distinguish a genuinely high potassium from a drifted one. That is why it is not reported on posted blood.

There is a genuine evidence gap here. Work published in 2026 on venous samples held at the temperatures and delays a posted sample meets found potassium, phosphate and iron consistently outside acceptable limits, while sodium, calcium, ferritin and AST only went out at 37 degrees. What nobody has published is potassium in finger-prick capillary samples posted to a lab. It has not been done.

If potassium specifically matters to you, it needs a sample analysed close to where it was taken, which in practice means asking your GP.

Sodium is a different matter, and we do offer it. Sodium sits mostly outside cells instead of inside them, at around 140 millimoles per litre outside against 10 to 20 inside, so neither mechanism above moves it much. It is not completely immune, as the note at the end of the sodium entry explains, but the difference between the 2 markers here is not marginal.

Cystatin C

A small protein produced at a fairly constant rate by nearly every cell in the body and cleared by the kidneys, and it is handled differently from creatinine once it gets there, which is set out in Kidneys, in the Body Guide. Like creatinine it can be used to estimate filtration rate, and it is the standard second marker when creatinine alone is thought to be misleading.

Its advantage is that it is not made in muscle, so it does not carry creatinine's dependence on how much muscle you have, what you ate, or when you last trained. That makes the comparison between the 2 markers informative in a way neither is on its own.

What it is not is a cleaner marker

This is where the usual summary overstates the case.

Measured against filtration determined directly, cystatin C on its own performs about the same as creatinine on its own. In the original external validation, the proportion of estimates landing within 30% of measured filtration was 87.2% for creatinine, 85.9% for cystatin C and 91.5% for the 2 combined. So estimates using cystatin C as the sole marker are not more accurate than creatinine-based ones, which suggests the unknown non-filtration influences on cystatin C are similar in size to those on creatinine.

The UK evidence agrees. Against filtration measured directly in English adults, 90.2% of creatinine-based estimates landed within 30% of the true value, 89.5% of cystatin C-based estimates, and 94.9% for the 2 combined. Cystatin C alone was marginally worse than creatinine alone. And in healthy people being assessed as kidney donors, the creatinine-based equations agreed with measured filtration better than the cystatin C-based ones did.

The gain comes from combining 2 imperfect markers, and not from one of them being clean. That is a different claim, and it is the one the evidence supports.

Muscle mass explains less of the gap than people assume

The difference between a creatinine-based and a cystatin C-based estimate in the same person is a real signal, and it is not simply a muscle measurement. Everything usually blamed for that difference, taken together, explains only 36% of its variation. That includes muscle mass, protein intake, clearance of mid-sized molecules, obesity and inflammation. Allowing for all of them does not weaken the link between that difference and death or heart failure.

Where muscle has been measured the link is real but weak. The difference between the 2 estimates correlates only weakly with markers of muscle mass.

What does affect cystatin C

It has its own influences. They are simply different ones, and for anyone reading this page they deserve more attention than the muscle question.

Thyroid status moves it more than anything else does. Treating an underactive thyroid raises cystatin C from about 0.84 to 1.10 milligrams per litre, a rise of roughly 31%, while creatinine falls from 86 to 76 micromoles per litre over the same period. Treating an overactive thyroid did the reverse, lowering cystatin C from 1.32 to 0.95 while creatinine rose. The 2 markers move in opposite directions on thyroid treatment, and the size of the cystatin C swing is larger than the muscle-mass problem it is being used to solve.

Diabetes raises it by about 8.5% while lowering creatinine by about 3.9%.

Inflammation raises it, tracking with a higher CRP, a higher white cell count and a lower albumin, and again lowers creatinine at the same time.

Smoking, and body fat, both raise it independently of filtration. In a general population sample, greater weight and greater height each predicted a higher cystatin C independently of filtration.

Sex and age both shift it, by about 9.2% lower in women and about 4.3% lower per 20 years younger, which is one reason the equations need both.

Glucocorticoids are widely reported to raise it. There is no clean human effect size to quote.

That list matters particularly here, because a good share of the people most interested in cystatin C are on levothyroxine, carrying a raised body mass index, or training hard enough to keep CRP up.

Where the numbers come from

There is no separate threshold for cystatin C itself, and no reason to expect one. It is an input to a filtration estimate, and it is the estimate that carries the categories set out under eGFR. The population reference range is assay-dependent, and cystatin C is more assay-dependent than most. The 2026 English study found that test calibration critically affected the measurement, and its accuracy figures above are the ones obtained after recalibration. Standardisation of the assay across platforms is still an open problem.

International and UK guidance currently disagree about this marker

KDIGO's 2024 guideline recommends that where cystatin C is available, the filtration category should be estimated from creatinine and cystatin C combined, and that the combined estimate should be used in situations where the creatinine-based one is less accurate and filtration affects a decision. It also says explicitly that the difference between the 2 estimates may itself be informative, in both direction and size.

NICE does not recommend cystatin C at all. Its 2014 guideline did, in a narrow situation, and its 2021 guideline removed that recommendation. The stated reasoning was that the committee could not be confident in the accuracy of cystatin C-based estimates, that there were serious limitations in the quality of the evidence, and that while cystatin C-based equations may reduce false-positive results they are likely to increase false-negative ones. The committee also noted that a 30% deviation from measured filtration, which is the standard the accuracy figures above are judged against, is still an unacceptably large difference. The UK Kidney Association's 2023 alert on eGFR does not mention cystatin C either. A large UK trial reporting in 2024 found that its data did not support using cystatin C for monitoring filtration in stage 3 chronic kidney disease, and no evidence that adding it to existing monitoring would be cost-effective.

Both positions are defensible readings of the same imperfect evidence, and the UK one is the one your NHS results will reflect.

The practical point for anyone who trains heavily or supplements creatine

A creatinine-based estimate can read low while the kidneys are entirely normal, and a second marker that is not made in muscle helps separate those. What the evidence supports is that the disagreement between the 2 estimates carries the information, in both direction and size, and that the combined estimate is more accurate than either figure alone.

What it does not support is treating cystatin C as the arbiter. Nobody has tested cystatin C against directly measured filtration in resistance-trained people or in people supplementing creatine. That work does not exist, so cystatin C is the better bet here and not a proven one.

Uric acid

Urate

The end product of purine breakdown, cleared mainly by the kidneys. When concentrations are high enough it can crystallise in joints, which is what gout is.

It is also associated with kidney stones, and observationally with metabolic and cardiovascular risk. Whether it causes that risk has been tested directly and repeatedly, and the answer is set out below.

It rises after a purine-rich meal, with alcohol, with dehydration and during rapid weight loss.

Where the numbers come from

NICE guideline NG219 on gout carries the committed figures.

In someone with the symptoms and signs of gout, a serum urate of 360 micromoles per litre, or 6 milligrams per decilitre, or more confirms the clinical diagnosis. If it is below 360 during a flare and gout is still strongly suspected, NICE repeats the measurement at least 2 weeks after the flare has settled, because urate can read misleadingly low during an attack.

For treatment, NICE aims for a serum urate below 360, and considers a lower target below 300 micromoles per litre, or 5 milligrams per decilitre, for people with tophi or chronic gouty arthritis, or who keep having frequent flares despite being below 360.

The 360 figure is not the crystallisation point

It is usually described that way and it is not correct.

NICE's own evidence review puts the physiological saturation threshold at around 380 micromoles per litre. The older lab figure for monosodium urate crystallising out of serum at body temperature is about 404, which is also the figure the American College of Rheumatology uses to define hyperuricaemia. NICE's stated reasons for choosing 360 were that it is more attainable and requires lower drug doses which may improve how well people stick to treatment, and that it reflected what primary care was already doing.

So there are 3 numbers in circulation and they mean different things. Around 360 is a treatment target chosen partly for practicality. Around 380 is NICE's figure for where urate starts coming out of solution. Around 404 is the older lab saturation figure and the American definition of a raised urate.

Two caveats from NICE's own work

It found no evidence comparing different target levels against each other, so the targets above are a consensus position and not a measured cliff edge.

And it states plainly that there is not enough evidence to show any specific diet prevents flares or lowers serum urate, advising a healthy balanced diet instead. That is worth publishing given the volume of dietary advice attached to this marker, and it sits alongside rather than against the note above that a purine-rich meal raises the number on the day.

What the diagnostic threshold requires

Symptoms and signs. A raised urate in someone with no symptoms is not gout, and NICE's figures are not written for that situation. Diuretics raise urate, thiazides particularly, as do excess alcohol and excess body weight.

That point can be put more strongly than it usually is, because the major bodies have all published on it. The American College of Rheumatology's 2020 guideline conditionally recommends against starting urate-lowering treatment in people with a raised urate, no previous flares and no tophi, on evidence it describes as of moderate or high certainty. Its stated reason is the arithmetic. On the available trial data, 24 people would need treating for 3 years to prevent one gout flare. KDIGO's 2024 kidney guideline likewise suggests not using urate-lowering agents in people with chronic kidney disease and a raised urate but no symptoms, for the purpose of slowing kidney disease.

The contrast with actual gout is the whole story. For someone who has gout and is on treatment, the same American guideline strongly recommends continuing to a target below 6 milligrams per decilitre, on moderate and high-quality evidence. Same marker, same number, opposite recommendations, and the thing that separates them is whether there are symptoms.

The causal question has been tested directly, and the answer was no

This is the part that matters, because urate is widely treated as something to drive down on its own account.

In 2020, 2 trials of allopurinol for kidney protection reported together.

The first took people with stage 3 or 4 chronic kidney disease, at high risk of getting worse, and no gout. Kidney function declined at 3.33 millilitres per minute per 1.73 square metres a year on allopurinol and 3.23 on placebo. Serious adverse events ran at 46% against 44%.

The second took people with type 1 diabetes and early to moderate kidney disease, ran for 3 years, and measured filtration directly with an injected marker instead of estimating it. Urate fell from 6.1 to 3.9 milligrams per decilitre on treatment and did not move on placebo. The difference in measured filtration between the 2 groups was 0.001 millilitres per minute per 1.73 square metres. Albumin loss into the urine afterwards was about 40% higher on allopurinol than on placebo.

So a trial that cut urate by a third, measured filtration directly instead of estimating it, and ran for 3 years, found an effect of one thousandth of a millilitre per minute, and a hint of harm on albumin loss. On the evidence from those 2 trials, lowering urate does not improve filtration measured either way, and it does not justify urate-lowering treatment for kidney protection on its own.

The genetic evidence points the same way

Inherited variants that raise urate for life make it possible to ask whether urate itself does the damage. That test has been run several times, and each time gout was used as a check that the method could find an effect when one was there.

On kidney outcomes it found nothing, for either filtration or chronic kidney disease, in a test large enough to have found an effect the size of the one seen in ordinary observational work. That observed association is about 2 millilitres per minute lower filtration, and roughly 50% higher odds of chronic kidney disease, for every milligram per decilitre of urate. The same variants raised the odds of gout between 3 and 6-fold, so the method was working. Reducing urate is unlikely to reduce the risk of developing chronic kidney disease.

For cardiovascular outcomes the pattern repeats. The association disappears once the analysis allows for the variants that affect things other than urate. Using only variants that affect urate and nothing else, there is no association with type 2 diabetes, coronary disease, ischaemic stroke or heart failure, while those same variants raise the odds of gout nearly 6-fold. A third test, large enough to find meaningful effects on cognition, Alzheimer's disease, coronary disease, heart attack, blood pressure and stroke, found no consistent evidence of any.

So urate causes gout, and for everything else it has been tested repeatedly and has not held up. On current evidence it behaves like a marker that travels alongside metabolic risk and not a cause of it.

Units

The UK reports urate in micromoles per litre and the US in milligrams per decilitre. The 2 NICE figures above are 6 and 5 milligrams per decilitre.