Can the Albumin Creatinine Ratio Be Improved?

The albumin-creatinine ratio can be improved, and researchers have been working on the problem from several angles at once. The test itself is already more informative than the older urine dipstick, but it suffers from wide day-to-day swings, built-in biases related to body composition, and a lack of standardization across laboratories. Fixes range from something as simple as changing when you collect the urine sample to more ambitious efforts like replacing the standard ratio with a mathematically adjusted estimate or layering in entirely new biomarkers alongside it.

Why the Standard ACR Falls Short

The urine albumin-to-creatinine ratio works by dividing the amount of albumin in a urine sample by the amount of creatinine. The idea is that creatinine, which your kidneys filter at a fairly steady rate, acts as a stand-in for how concentrated or dilute the sample happens to be. A higher ratio suggests the kidneys are leaking more albumin than they should. It is the standard screening tool for early kidney damage, and combining it with estimated kidney filtration rate and other markers substantially improves the ability to predict who will progress to kidney failure or die from kidney-related causes.

The trouble is that creatinine is not the neutral yardstick it is assumed to be. Creatinine production depends heavily on muscle mass, which means two people with the same actual albumin leak can get very different ACR results. Someone with low muscle mass produces less creatinine, which inflates the ratio and can flag a problem that does not exist. Research has shown that a high ACR in people with low muscle mass more often reflects low urinary creatinine than actual kidney damage or cardiovascular disease.¹ Conversely, a muscular person may get a falsely reassuring result because their high creatinine output drags the ratio down.

On top of that body-composition bias, the ACR bounces around dramatically from one day to the next. For people whose true albumin levels are in the normal range, a single spot ACR measurement would need to change by more than 400% before you could be confident the shift was real rather than just noise.² Even in people with heavy protein loss in the urine, a change of roughly 50% could still be within normal fluctuation. That variability comes from shifts in blood pressure, kidney filtration pressure, hydration, physical activity, and other short-term physiological factors.³

And the laboratory side is no better. Commercially available albumin assays can disagree by more than 40% when measuring the same sample, depending on which method a lab uses.⁴ When the test itself, the biology behind it, and the lab running it all introduce their own errors, you end up with a tool that is clearly better than a dipstick but leaves a lot of room for improvement.

Collecting the Sample at the Right Time

One of the simplest improvements costs nothing and requires no new technology: use the first urine of the morning instead of a random spot sample collected whenever the patient happens to visit the clinic. A study comparing different collection strategies found that the first morning void had a day-to-day variability of about 19%, which was similar to a full 24-hour urine collection and significantly lower than a random spot sample.⁵ The 24-hour collection has long been treated as the gold standard, but it is inconvenient and prone to its own errors when patients forget to collect every void. A first morning sample captures a reasonably concentrated overnight collection without the hassle.

How you handle the sample after collection also matters. Albumin degrades in stored urine, and samples kept at standard freezer temperatures of around minus 20°C lost more than 30% of their albumin after 12 months. Storage at minus 80°C on the day of collection preserved the albumin much more reliably, and pre-treating the sample before freezing at the warmer temperature did not help.⁶ For clinical decisions based on a single stored sample, this kind of degradation could easily push a result across a diagnostic threshold in the wrong direction.

Adjusting the Cutoffs

The standard ACR threshold for diagnosing moderately increased albuminuria is 30 mg/g, the same number regardless of whether the patient is a 25-year-old male athlete or a 70-year-old woman with low muscle mass. Given how much creatinine production varies between people, a single cutoff misclassifies a meaningful number of patients. One straightforward fix is to use sex-specific thresholds. Because men generally excrete more creatinine than women, a lower ACR cutoff for men and a higher one for women better aligns the test with actual albumin excretion. Research suggests cutoffs of about 17 mg/g for men and 25 mg/g for women improve the test’s accuracy.⁷ In people with obesity, the problem gets more complicated because fat mass and muscle mass interact differently with creatinine excretion, and optimal cutoffs would ideally account for age and adiposity too.

There is also growing evidence that the standard 30 mg/g threshold may be too high for catching early kidney trouble in certain populations. In people with type 2 diabetes, ACR values well below the traditional cutoff have been associated with progression to later-stage kidney disease. One study found that ACR values as low as about 8 to 11 mg/g were linked to increased risk of kidney disease progression, with slightly different thresholds for men and women.⁸ Lowering the diagnostic threshold in high-risk groups could catch disease earlier, though it also raises the question of how many healthy people would be flagged unnecessarily.

Replacing the Ratio With a Calculated Estimate

Rather than just tweaking the cutoff, some researchers have proposed replacing the raw ACR with a formula-derived value called the estimated albumin excretion rate, or eAER. The idea is to take the ACR and mathematically adjust it using factors that predict creatinine excretion, like age, sex, and body size. This converts the ratio back into something closer to an actual excretion rate without requiring the patient to collect urine for 24 hours.

The results have been promising. In one validation study, the eAER fell within 30% of the actual measured excretion rate for about 60% of participants, compared to only 39% when using the raw ACR. The ACR also had a consistent negative bias, underestimating the true excretion rate by around 20%, while the eAER had a much smaller error.⁹ A separate study from a large Dutch population cohort confirmed that the eAER was both more accurate and less biased than the ACR for estimating true albumin loss, and performed more consistently across age and sex groups.¹⁰

The eAER approach is appealing because it does not require any new lab equipment or sample types. It uses the same spot urine ACR that clinics already collect, just plugs it into a correction formula. The main barrier to wider adoption is inertia: clinical guidelines, electronic health records, and years of accumulated research are built around the raw ACR. Transitioning to a calculated measure requires updating workflows, decision thresholds, and the evidence base that supports them.

Fixing the Lab Side

Even if you collect the sample perfectly and apply the right formula, the result is only as good as the measurement. The 40%-plus disagreement between commercial albumin assays is a serious problem, and it means a patient tested at one lab could be classified differently than if the same sample went to a lab down the street. Efforts to develop a reference measurement procedure for urine albumin are underway, and standardizing the assays would eliminate one of the biggest sources of noise in ACR results.⁴

One candidate for a more precise reference method is liquid chromatography combined with mass spectrometry. This approach breaks down the albumin protein into specific fragments and measures them directly, rather than relying on antibody-based reactions that can vary between manufacturers. In testing, this method showed a strong correlation with the standard immunoturbidimetric approach and provided consistent results at low albumin concentrations where antibody-based tests tend to struggle.¹¹ It is unlikely to replace routine clinical assays directly, given the cost and complexity of the equipment, but it could serve as the benchmark that all other assays are calibrated against.

Point-of-Care Devices

At the other end of the technology spectrum, simpler and cheaper devices are being developed to bring ACR testing to primary care clinics, pharmacies, and even patients’ homes. An earlier generation of point-of-care devices showed reasonable but imperfect performance, with a sensitivity of about 83% and specificity of about 80% for detecting albuminuria in primary care settings.¹² Those numbers were good enough for screening but left some room for false positives and missed cases.

Newer devices appear to be closing the gap. A recently validated point-of-care device called MyACR uses colorimetric spectroscopy and showed perfect sensitivity and specificity for detecting severe nephropathy when compared with a standard lab method in a set of over 100 high-risk patient samples.¹³ The device also transmits results digitally, which matters for remote monitoring. Meanwhile, paper-based microfluidic sensors using quantum dot technology have shown high sensitivity at the very low albumin concentrations needed to catch early-stage kidney damage, and they are cheap to produce.¹⁴ These are still in development, but the direction is clear: making ACR testing faster, cheaper, and more accessible while maintaining accuracy at the thresholds that matter most clinically.

Adding New Biomarkers Alongside ACR

Some of the most ambitious efforts to improve ACR do not try to fix the ratio itself but instead combine it with other measurements that capture different aspects of kidney health. Albumin leaks through the kidney’s filtering units, the glomeruli, so ACR primarily reflects glomerular damage. But kidney disease can also involve damage to the tubules, the tiny tubes that reabsorb useful substances and secrete waste after filtration. Tubular damage often occurs alongside or even before glomerular damage, and ACR alone misses it.

Several tubular biomarkers have shown promise. NGAL and KIM-1, both of which rise when tubular cells are injured, are substantially elevated in people with chronic kidney disease compared to healthy controls.¹⁵ These markers correlate with declining kidney function and could help identify people whose kidneys are deteriorating even when the ACR looks reassuring. On the other side of the equation, markers of healthy tubular function like epidermal growth factor (EGF) and uromodulin (UMOD) tend to be lower in people whose kidney disease is progressing. Higher baseline levels of both were independently associated with a lower risk of diabetic kidney disease getting worse.¹⁶

Urinary podocyte markers represent yet another angle. Podocytes are specialized cells that form part of the kidney’s filtering barrier, and when they are shed into the urine, it signals structural damage at the glomerular level that may precede the kind of albumin leak the ACR picks up.¹⁷ None of these biomarkers are ready to replace ACR in routine clinical use — cost, assay availability, and the lack of established clinical thresholds are all barriers — but the evidence is building that a panel approach, combining ACR with one or more of these markers, would give a much more complete picture of kidney health than any single test.

Combining ACR With Better Filtration Estimates

You do not necessarily need novel biomarkers to get more out of the data already being collected. Combining ACR with cystatin C-based estimates of kidney filtration rate, rather than the standard creatinine-based estimate, significantly improves risk prediction. A study in patients with an original diagnosis other than kidney disease — specifically, people who had experienced strokes — found that both reduced cystatin C-based filtration rate and elevated ACR independently predicted death and disability. Combining the two markers improved risk stratification even in patients who did not meet the formal criteria for chronic kidney disease.¹⁸

This finding dovetails with earlier research showing that defining chronic kidney disease using creatinine, cystatin C, and ACR together dramatically outperforms any single marker. People classified as having kidney disease by all three markers had roughly a fivefold higher risk of death compared to those identified by creatinine alone.¹⁹ The clinical implication is that ACR becomes a much more powerful tool when it is not asked to do all the heavy lifting on its own.

Machine Learning Models That Build on ACR

Machine learning offers another path to squeezing more predictive value out of ACR without changing the test itself. A random forest model trained on routine clinical data — lab values, vital signs, demographics — to predict kidney disease progression at two and five years achieved strong accuracy, with an area under the curve of about 0.88 at two years. ACR was the single most influential variable in the model, but the algorithm also pulled in kidney filtration rate, blood urea, hemoglobin, age, and blood albumin levels. The top 30% of patients flagged as high or intermediate risk by the model accounted for 87% of the people who actually progressed within two years.²⁰

What is interesting about these models is not that they replace ACR but that they reveal how much additional information is already sitting in the patient’s chart, waiting to be combined intelligently. A single ACR value in isolation is noisy and biased. That same value fed into an algorithm alongside other routine measurements becomes part of a much sharper prediction. Community screening programs are also catching on to this idea. Adding quantitative ACR to standard screening with kidney filtration rate and dipstick testing identified substantial additional albuminuria and changed risk classifications for a meaningful number of people in cardiovascular high-risk cohorts.²¹

ACR in Settings Beyond Chronic Kidney Disease Screening

Most of the conversation about improving ACR centers on its traditional role: screening for and monitoring chronic kidney disease. But the test is increasingly being explored in acute settings as well. In intensive care units, where acute kidney injury develops rapidly and existing detection methods rely on a rise in serum creatinine that can lag by hours or days, ACR has drawn interest as a cheap, readily available early-warning signal. The test’s low cost makes it attractive compared with novel biomarkers that can run hundreds of dollars per assay, even if its diagnostic role in acute kidney injury is still being defined.²² Whether ACR can fill that niche as effectively as it fills the chronic screening role remains an open question, but the economic argument for trying is hard to ignore when budgets are tight and kidneys are failing fast.