Why is albuminuria described as more informative than eGFR at a normal eGFR?

My eGFR is 92 and has been for years, which I had taken to mean my kidneys were not a topic. A urine albumin:creatinine ratio was added to a recent panel and came back at 56 mg/g, which the report categorised as A2. I now gather that this is the finding and that the normal eGFR was never much reassurance on its own.

I would like to understand why. Specifically:

• Why does albuminuria carry information that eGFR does not, when both are described as kidney tests?
• How reproducible is a UACR? I have read that it varies enormously between samples, which would make a single value hard to act on.
• The trial evidence in this drug class seems to be built around eGFR thresholds and hard endpoints. Where does albuminuria fit into that, and did it move in the trials?

A worked interpretation would help, including the units, because my report gives mg/g and everything I read locally uses mg/mmol and I keep having to guess which threshold applies.

Because they measure different things: eGFR estimates how much the kidney is filtering, albuminuria measures whether the filter is leaking. Damage precedes loss of function by years, so at a normal eGFR the albumin measurement carries almost all the information and the eGFR carries almost none.

The mechanism, briefly

The glomerular filtration barrier — endothelial glycocalyx, basement membrane, podocyte slit diaphragms — normally excludes albumin almost entirely, and the small amount that crosses is retrieved by the proximal tubule. Albumin appearing in urine means either the barrier has become permeable or tubular retrieval has failed, and in the diabetic and hypertensive kidney it means both, driven by the same raised intraglomerular pressure that produces hyperfiltration.

The important structural point: total GFR is defended by recruitment and hypertrophy of remaining nephrons. You can lose a substantial fraction of your nephrons and maintain a normal total eGFR, because the survivors hyperfilter to compensate. So a normal eGFR is compatible with considerable established damage, and eGFR only starts falling once compensation is exhausted. Albuminuria appears while the compensation is still working.

That is why the staging system is two-dimensional: a G stage from eGFR and an A stage from albuminuria, with risk determined by the combination. Your G1 A2 is not "normal kidneys"; it is a defined risk category with a substantially higher event rate than G1 A1, and — this is the part worth internalising — a higher risk than some people with a mildly reduced eGFR and no albuminuria.

Worked interpretation and the units

UACR is the urine albumin concentration divided by the urine creatinine concentration in the same sample. Dividing by creatinine corrects for how dilute the sample was, which is why a spot sample works at all and why a plain urine albumin concentration is nearly useless.

Worked, from a lab report giving albumin 45 mg/L and creatinine 0.80 g/L:

• UACR = 45 ÷ 0.80 = 56 mg/g.
• Convert to SI: mg/mmol = (mg/g) ÷ 8.84. So 56 ÷ 8.84 = 6.3 mg/mmol.

The category boundaries, which are the same boundaries expressed in two systems:

• A1: under 30 mg/g, equivalently under 3 mg/mmol. Normal to mildly increased.
• A2: 30 to 300 mg/g, equivalently 3 to 30 mg/mmol. Moderately increased. Your 56 mg/g sits in the lower part of this band.
• A3: above 300 mg/g, above 30 mg/mmol. Severely increased.

The conversion factor of 8.84 is creatinine's molar mass relationship, and forgetting it produces an eight-fold error, which is the most common unit mistake in this area. If someone quotes "a UACR of 30" you cannot interpret it until you know the units, because 30 mg/g is the bottom of A2 and 30 mg/mmol is the bottom of A3.

Note also that risk rises continuously below the A1 boundary. There is no albumin excretion rate that is protective; 20 mg/g carries more risk than 5 mg/g, and the boundary at 30 is an administrative convenience.

Reproducibility, and why one value is not enough

You are right to be suspicious. Within-person variability of UACR is large — reported coefficients of variation of roughly 30 to 40% are typical, considerably worse than any blood analyte you are tracking. Run that through the reference change value:

RCV = 2.77 × sqrt(CVa² + CVi²) = 2.77 × sqrt(25 + 1225) = 2.77 × 35.4 = 98%

So detecting a real change from two samples requires roughly a halving or a doubling. This is why the convention is to confirm an abnormal result with two of three samples over three to six months, and why a 30 to 40% reduction is the figure usually quoted as a meaningful therapeutic response — it is barely above the noise floor for a single pair and only becomes convincing across repeated measurements.

Things that transiently raise UACR and should be excluded before repeating:

• Vigorous exercise in the preceding 24 hours.
• Urinary tract infection.
• Fever or any acute illness.
• Menstruation, or any contamination with blood.
• Marked hyperglycaemia in the hours before the sample.
• A very high protein meal.
• Standing upright all day — orthostatic proteinuria is a real phenomenon, and it is the reason first-morning samples are preferred.

Practically: use first-morning samples, avoid hard training the day before, and do not act on a single value in either direction.

edited 16 Jul 2026 by lyoph_cake — removed a claim I could not source

One addition on volume status, because it is the confounder that ruins albuminuria series in this specific population and it works in the opposite direction to the one people expect.

UACR is a ratio, and dehydration moves both terms. Urine albumin concentration rises with concentrated urine, but so does urine creatinine, and because the ratio was designed to cancel exactly that, mild dehydration affects UACR less than it affects either component. That is the good news.

The bad news is that it does not cancel perfectly, and the residual bias depends on which way the water went:

• Very dilute urine, from a large water load, drives urine creatinine low. Because creatinine excretion is fairly constant per unit time while albumin excretion is more variable, very dilute samples tend to produce higher and noisier UACR values. A urine creatinine below roughly 0.3 g/L (2.5 mmol/L) is a reason to distrust the ratio.
• Very concentrated urine, typically a first-morning sample after a long fast, is the preferred condition but at the extreme can push in the other direction.
• Reduced muscle mass lowers urine creatinine excretion for the same reason it lowers plasma creatinine. Over months of substantial lean-mass loss, the denominator of your UACR shrinks, and UACR drifts upwards at constant albumin excretion. This is the mirror image of the plasma creatinine artefact, and it is much less widely discussed. Someone losing 4 kg of lean mass over a year may see UACR rise by 8 to 12% for that reason alone.

That last point is worth flagging because it cuts against the drug in the same way the plasma creatinine artefact flatters it. If you are tracking both plasma eGFR and UACR through a large weight loss, one measure is biased optimistic and the other pessimistic, for the same underlying reason. The defence is the same in both cases: know that it is happening, prefer the trend over the level, and get a timed or 24-hour collection if a decision genuinely hinges on the number.

A 24-hour urine albumin excretion avoids the creatinine denominator entirely and is the reference measurement. It is also unpleasant, frequently collected incorrectly, and rarely necessary. Reserve it for the case where the spot ratios are equivocal and the decision matters.

Finally, on the acute case: during a week of vomiting and poor intake, do not measure UACR at all. You will get a value that reflects volume status and acute illness, and you will spend three months trying to explain it.