Predictive and prognostic performance of urinary albumin-to-creatinine ratio for acute kidney injury: a systematic review and meta-analysis

Abstract

The urine albumin-to-creatinine ratio (UACR) is a well-established marker for chronic kidney disease, but its utility in predicting acute kidney injury remains uncertain. This systematic review and meta-analysis aimed to evaluate predictive performance for AKI development and prognostic performance for AKI progression in hospitalized adults. A comprehensive search of Ovid MEDLINE, Embase, and CENTRAL databases identified 13 studies (n = 10,438) on AKI incidence and three studies (n = 1596) on AKI progression. Elevated UACR was associated with an increased risk of AKI (pooled OR 1.39; 95% CI 1.08–1.79) and AKI progression (pooled OR 3.76; 95% CI 2.59–5.45). The pooled sensitivity and specificity for AKI prediction were 0.71 (95% CI 0.59–0.80) and 0.67 (95% CI 0.56–0.76), respectively, with an area under the curve (AUC) of 0.74. However, there was high heterogeneity across studies, and UACR thresholds for AKI prediction varied widely. Despite these limitations, UACR appears to be a promising, low-cost biomarker for predicting AKI, particularly in high-risk settings such as cardiac surgery. Standardization of thresholds and further validation are needed to support its clinical implementation.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-37717-2.

Introduction

Acute kidney injury (AKI), an abrupt decline in kidney function, is independently associated with increased morbidity, prolonged hospitalization, and higher healthcare costs^1–4. Traditional markers such as serum creatinine have delayed kinetics and fail to promptly reflect acute changes in kidney function⁵, hindering early diagnosis. Thus, the timely detection of AKI remains a clinical challenge. Although several novel biomarkers have been investigated, many are costly, not routinely available in real-time, and of uncertain clinical utility due to inconsistent performance across populations, timing, and assay platforms^6–8. Their roles in AKI classification and prognostication remain controversial.

The urine albumin-to-creatinine ratio (UACR) is a well-established marker of kidney damage in chronic kidney disease (CKD)⁹. It quantifies the amount of albumin excreted in urine relative to creatinine. Under normal physiological conditions, urinary albumin excretion is minimal due to efficient renal filtration and reabsorption. Typically, UACR values remain below 30 mg/g (3.39 mg/mmol), while higher levels suggest kidney damage or CKD^10–12.

Although proteinuria and albuminuria have been associated with increased AKI risk, particularly in patients with pre-existing CKD or undergoing high-risk procedures^13–15. UACR is not currently recognized as a predictive or prognostic for AKI in international consensus guidelines^16,17. Its potential value lies in its broad availability, standardization, and low cost. However, current literature primarily reports associations between elevated UACR and AKI, without consensus on diagnostic thresholds, timing, or prognostic significance¹⁸. Accordingly, we conducted a systematic review and meta-analysis to evaluate the role of UACR as a predictive biomarker for AKI development and a prognostic biomarker for AKI progression.

Methods

This systematic review adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Detail in Supplementary Material S1)¹⁹ and was registered in PROSPERO (protocol number: CRD420250648041).

Type of studies

We included studies that evaluated the urinary albumin-to-creatinine ratio (UACR) as a predictive tool for AKI or AKI progression. Eligible studies enrolled hospitalized adult patients aged 18 years or older, assessed UACR at least at one time point, and reported AKI as an outcome. We included all study designs except case reports, letters, conference abstracts, and preprints. Studies were excluded if they lacked relevant predictive data, specifically, if they did not report sufficient information to construct a 2 × 2 contingency table (i.e., true positives [TP], false positives [FP], true negatives [TN], and false negatives [FN]) for calculating sensitivity, specificity, or odds ratios, were not published in English or were not full original research articles. We also excluded studies involving pregnant patients, animal models, urological surgery, or substance-induced AKI. (e.g., contrast media or cisplatin nephrotoxicity) due to their distinct pathophysiological mechanisms and atypical UACR dynamics.

Search strategy and study selection

A comprehensive literature search was conducted using Ovid MEDLINE, EMBASE, and the Cochrane Library to identify relevant studies. The search was limited to English-language publications, and gray literature databases or preprint servers were not systematically searched. The search strategy incorporated terms related to “urine albumin-to-creatinine ratio,” “acute kidney injury,” “biomarkers,” “predictive value,” and “diagnostic accuracy,” along with relevant synonyms and subject headings. Reference lists of included studies and relevant review articles were also manually screened to identify additional eligible studies. The full search strategy is detailed in Supplementary Material S2, and the final search was conducted on February 5, 2025.

Duplicate records were removed using Covidence (https://www.covidence.org). Two independent reviewers (NK and JI) then screened titles and abstracts, followed by full-text assessment based on predefined inclusion and exclusion criteria. Discrepancies were resolved through discussion, with a third reviewer (NR) consulted when consensus could not be reached.

Data extraction

Data were independently extracted by two authors (NK and JI). The following information was collected from each study: year of publication, country, sample size, population type, mean or median age, incidence of AKI or AKI progression, the criteria used to define AKI or AKI progression (e.g., KDIGO, AKIN, RIFLE, or study-specific definitions), the timing of UACR measurement relative to AKI onset/diagnosis, and the reported UACR cutoff value. Studies were classified as evaluating predictive performance when UACR was measured before (or at baseline prior to) AKI onset to predict incident AKI, and as evaluating prognostic performance when UACR was measured after AKI diagnosis to predict AKI progression. If UACR was reported in mg/g, it was converted to mg/mmol by dividing the value by 8.84, consistent with established conversion standards. For studies that did not directly report sensitivity, specificity, or odds ratios (OR), we reconstructed these values using the reported sensitivity, specificity, and the absolute numbers of patients with and without AKI (prevalence). We calculated TP as (Sensitivity × number of AKI patients) and TN as (Specificity × number of non-AKI patients), with FP and FN derived by subtraction. We did not use statistical imputation for missing diagnostic accuracy data. Two studies‒G. Coca²⁰ and O. Molnar²¹‒ were conducted in similar settings and timeframes among cardiac surgery patients in North America, raising the possibility of overlapping patient cohorts. Although this could not be explicitly confirmed, both studies were retained in the analysis, with appropriate caution applied during interpretation.

We also extracted data on comparisons between UACR and other biomarkers (e.g., NGAL, Cystatin C), and on reported associations between UACR and clinical outcomes such as ICU or hospital length of stay and mortality, when available. These were narratively synthesized due to heterogeneity and limited reporting.

Quality assessment

Two authors (NK and JI) independently assessed the risk of bias and applicability concerns for each included study using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool²². The assessment was conducted in duplicate, and any discrepancies were resolved through discussed. When consensus could not be reached, a third author (NR) was consulted to adjudicate.

Statistical analysis

We conducted a meta-analysis of the included studies using R software version 4.3.3 (R Foundation for Statistical Computing, Vienna, Austria). Studies lacking sufficient predictive data were excluded from the meta-analysis. We generated forest plots to present the pooled sensitivity and specificity of UACR in predicting AKI. Additionally, we included studies that allowed for the calculation of ORs and performed a pooled diagnostic OR (DOR) analysis with corresponding 95% confidence intervals (CIs) to evaluate whether elevated UACR levels could predict AKI or its progression. Summary receiver operating characteristic (sROC) curves with 95% CIs were also created to visualize overall diagnostic performance.

Heterogeneity among studies was assessed using the Q statistic, its corresponding p-value, and the I² statistic. A p-value < 0.10 was considered indicative of statistically significant heterogeneity. The I² statistic quantifies the proportion of total variation attributable to between-study heterogeneity, with values interpreted as follows: 0–40%, 30–60% (moderate), 50–90% (substantial), and 75–100% (considerable) heterogeneity.

Subgroup analyses were performed based on patient population. To explore potential sources of heterogeneity, univariate meta-regression analyses were conducted using study-level covariates, including UACR cutoff value (mg/mmol), mean patient age (years), and baseline serum creatinine (µmol/L). Sensitivity analyses were also conducted to assess the robustness of the pooled estimates by sequentially excluding individual studies and evaluating the impact on overall results.

Publication bias was assessed using a funnel plot of log-transformed diagnostic odds ratios plotted against their standard errors, and asymmetry was formally evaluated using Egger’s regression test. Given the substantial clinical and methodological heterogeneity across included studies, a random-effects model²³ was used for all meta-analyses to account for between-study variability. All statistical tests were two-tailed, and a p-value < 0.05 was considered statistically significant.

Results

We identified a total of 2892 studies—2890 from databases (Embase = 2112; MEDLINE = 626; CENTRAL = 152) and 2 from citation searching. After removing 300 duplicates (1 manually and 299 by Covidence), 2592 articles remained for screening. Following title and abstract screening, 2457 articles were excluded. Full-text review was conducted for 135 articles, of which 119 were excluded for reasons outlined in Fig. 1. Ultimately, 16 studies were included in the final review.

Fig. 1.

PRISMA flow diagram of study selection. This diagram illustrates the systematic review process, detailing the number of records identified, screened, eligible, and included in the final meta-analysis. Reasons for exclusion at the full-text screening stage are specified. PRISMA Preferred reporting items for systematic reviews and meta-analyses.

Characteristics of included studies

A total of 16 studies^{13,15,20,21,24–35} were included in this review, and 13 studies^{13,15,20,21,24–32} assessed the utility of UACR for predicting AKI in hospitalized patients. Most studies excluded individuals with known severe CKD, except for Molnar et al.²¹ and Coca et al.²⁰, which included patients with serum creatinine > 2 mg/dL. The study by Tobe et al.²⁴ excluded only those receiving dialysis. The majority of studies (10 out of 13) employed a prospective cohort design; only three were retrospective^13,24,27.

Geographically, five of the 13 studies were conducted in Asia (China^25,28,31, Korea²⁹, and Japan²⁴), while the others originated from North America^13,20,21, Africa³² and Europe^15,26,27,30. Cardiac surgery patients represented the most common population, as seen in the studies by Coca et al.²⁰, Molnar et al.²¹, and George et al.¹³. Other populations included patients with sepsis²⁸, COVID-19^26,27,30, acute myocardial infarction¹⁵, cirrhosis with hepatic encephalopathy³², major burns²⁹, neurosurgical conditions²⁵, and those undergoing transcatheter aortic valve replacement (TAVR)²⁴.

Detailed study characteristics, including AKI definitions, UACR measurement timing, and comparisons with other biomarkers, are summarized in Table 1. Supplementary Table S1 summarizes three studies^33–35 that investigated UACR for predicting AKI progression, defined as worsening of AKI stage in all studies (e.g., from stage 1 to stage 2 or 3, or from stage 2 to stage 3). Two studies^33,34 involved cardiac surgery patients, while the third focused on individuals with acute decompensated heart failure³⁵. All three studies enrolled patients with established AKI and excluded those with advanced renal dysfunction.

Table 1.

Characteristics of studies evaluating urinary albumin-to-creatinine ratio (UACR) for predicting acute kidney injury (AKI).

Author, Year, Country	Design	N	Population	Age, year	AKI (%)	AKI criteria	Timing of UACR		Comparing to other biomarkers
G. Coca, 2012, USA20	Prospective cohort	1,159	Cardiac surgery	71.6 ± 10.1	409/1,159 (35.29%)	AKIN	Preoperative		No comparing with other biomarkers
Z. Zhang, 2012, China28	Prospective cohort	84	Septic patients	60.5 ± 13.3	36/84 (42.86%)	50% or greater increase in sCr from baseline, or an absolute increase of 0.3 mg/dL from baseline	The 2nd of ICU admission or after diagnosis of sepsis		No comparing with other biomarkers
O.Molnar, 2012, Canada21	Prospective cohort	1,198	Cardiac surgery (High risk for AKI)	71.5 ± 10.1	56/1,198 (4.71%)	The receipt of acute dialysis or a doubling in serum creatinine from the baseline preoperative value	0–6 h after surgery		No comparing with other biomarkers
K. George, 2015, USA13	Retrospective cohort	5,359	CABG patients	65.6 ± 7.6	1,436/5,359 (26.8%)	AKIN	Preoperative		No comparing with other biomarkers
Tziakas, 2015, Greece15	Prospective cohort	805	Acute MI	62.0 ± 13.0	118/805 (14.7%)	AKIN, RIFLE, KDIGO	Admission		UACR was better compared to urine NGAL, urine and plasma Cystatin-C
Yang, 2015, China31	Prospective cohort	317	ADHF	63.8 ± 15.8	104/317 (32.8%)	KDIGO	Admission		UAGT was better than urine NGAL, UACR, and clinical models
Yim, 2015, Korea29	Prospective cohort	97	Major burn	47.0 ± 14.9	40/97 (41.24%)	AKIN	at PBD 1, 3, 7, 14, 21 and 28		Serum cystatin C was better than UACR
Deng, 2020, China25	Prospective cohort	605	Post-op neurosurgery	52.0 [39.0–60.0]	67/605 (11.1%)	KDIGO	After surgery (ICU admission)		Serum cystatin C + uNAG outperformed individual biomarkers and other panels (uNAG + UACR or sCysC + UACR)
Luther, 2020, Sweden26	Prospective cohort	52	ICU COVID-19	59.3 ± 14.2	33/52 (63.5%)	KDIGO Creatinine criteria	ICU admission		No comparing with other biomarkers
Yildirim, 2021, Turkey27	Retrospective cohort	348	Hospitalized COVID-19	38.7 [30.8–46.6]	17/348 (4.9%)	KDIGO	Admission		Serum cystatin C was slightly better than UACR and D-dimer
Shahbah, 2022, Egypt32	Prospective case–control	67	Cirrhosis with HE	57.9 ± 12.8	16/67 (23.9%) in HE group	> 0.3 mg/dl from baseline in 48 h	Admission		No comparing with other biomarkers
Akihiro Tobe, 2022, Japan24	Retrospective cohort	206	TAVR patients	84.0 [80.5–86.5]	10/206 (9.5%)	An increase in serum creatinine ≥ 0.3 mg/dL or ≥ 1.5-fold from baseline within 48 h after TAVR	Pre-TAVR		No comparing with other biomarkers
Schnabel, 2023, Hungary30	Prospective cohort	141	Non-ventilated COVID-19	67.5 [55.8–78.2]	28/141 (19.7%)	KDIGO	twice weekly		No comparing with other biomarkers

Data presented as mean ± SD or median [interquartile range], AKI = Acute Kidney Injury, AKIN = Acute Kidney Injury Network, RIFLE = Risk, Injury, Failure, Loss, End-Stage, KDIGO = Kidney Disease: Improving Global Outcomes, UACR = Urine (or Urinary) Albumin-to-Creatinine Ratio, CABG = Coronary Artery Bypass Graft, MI = Myocardial Infarction, ADHF = Acute Decompensated Heart Failure, CRS = Cardiorenal Syndrome, TAVR = Transcatheter Aortic Valve Replacement, WRF = Worsening Renal Function, HE = Hepatic Encephalopathy, ICU = Intensive Care Unit, sCysC = Serum Cystatin C, uNAG = Urinary N-Acetyl-β-D-Glucosaminidase, NGAL = Neutrophil Gelatinase-Associated Lipocalin, UAGT = Urinary Angiotensinogen, PBD = Post-Burn Day(s), AUC = Area Under the Receiver Operating Characteristic Curve, OR = Odds Ratio, RR = Relative Risk, CI = Confidence Interval. To convert urine albumin-to-creatinine ratio (UACR) from SI units (mg/mmol) to conventional units (mg/g), multiply by 8.84. For example, 3.4 mg/mmol ≈ 30 mg/g.

Quality assessment revealed variability in the risk of bias across the included studies. Most demonstrated a low risk of bias in the index test and reference standard domains. However, concerns were noted in the patient selection and flow and timing domains, with several studies rated as having unclear or high risk of bias (Detailed in Supplementary Table S2).

Meta-analysis

UACR for prediction of AKI development

Eight studies^{13,15,20,21,24,26,28,30} comprising 9004 patients provided sufficient data to retrieve or calculate ORs for the association between elevated UACR and AKI development. The pooled OR, calculated using a random-effects model, was 1.39 (95% CI 1.08–1.79), suggesting a significant association between higher UACR levels and increased odds of AKI. However, substantial heterogeneity was observed across studies (I² = 89.3%, τ² = 0.066, Q(7) = 65.28, p < 0.001) (Shown in Table 2 and Fig. 2).

Table 2.

Diagnostic test performance of urinary albumin-to-creatinine ratio (UACR) in predicting acute kidney injury (AKI) and its progression across the included studies.

No	Author	UACR cutoff (mg/mmol)	Patient	Se	Sp	TP	FP	FN	TN	AUC	OR (95% CI)
AKI
1	G. Coca, 201220	≥ 3.39	Cardiac surgery	0.44†	0.71†	179	220	230	530	NR	1.88 (1.41–2.50)†
2	Z. Zhang, 201228	> 16.16	Septic	0.92	0.79	33	10	3	38	0.86 (0.77–0.94)	1.02 (1.01–1.03)
3	O. Molnar, 201221	≥ 6.10	Cardiac surgery	0.66†	0.40†	37	682	19	460	0.59	1.31 (0.72–2.40)†
4	K. George, 201513	≥ 3.39	Cardiac surgery	0.42†	0.67†	607	1289	829	2634	0.43 (0.19–0.68)	1.50 (1.32–1.70)†
5	Tziakas, 201515	> 7.54	Acute MI	0.68	0.76	80	165	38	522	NR	1.21 (1.03–1.48)
6	Yang, 201531	> 10.17	ADHF	0.68	0.66	71	72	33	141	0.71 (0.71–0.83)	NR
7	Yim, 201529	> 4.16 (PBD7)	Burn	0.58	0.56	23	25	17	32	PBD7: 0.61 (p = 0.069) PBD14: 0.70 (p = 0.001)	NR
8	Deng, 202025	> 5.00	Neurosurgery	0.45	0.82	29	97	36	443	0.65 (0.57–0.72)	NR
9	Luther, 202026	> 3.00	COVID	0.76†	0.32†	25	13	8	6	NR	1.44 (0.41–5.05)†
10	Yildirim, 202127	> 3.39	COVID	0.90	0.88	15	40	2	291	0.95 (0.91–0.98)	NR
11	Shahbah, 202232	> 10.34	HE/Cirrhosis	0.81	0.82	13	9	3	42	0.85	NR
12	Akihiro Tobe, 202224	≥ 3.39	TAVR	0.90†	0.51†	9	97	1	99	NR	9.19 (1.16–72.51)†
13	Schnabel, 202330	NR	COVID19	NR	NR	NR	NR	NR	NR	NR	1.48 (1.04–2.11)
AKI progression
14	Koyner, 201233	≥ 3.96	Cardiac surgery	0.84†	0.43†	37	185	7	141	NR	4.03 (1.74–9.33)†
15	Chen, 201635	≥ 11.84	ADHF	0.82†	0.38†	41	101	9	62	NR	2.8 (1.27–6.17)†
16	Su, 202434	> 7.51	Cardiac surgery	0.57	0.75	87	213	67	646	0.69	3.94 (2.77–5.61)†

^† Data calculated from 2 × 2 table, NR = Not report or cannot retrieve data, AKI = Acute kidney injury, AKIpro = Progression of acute kidney injury, MI = Myocardial infarction, ADHF = Acute decompensated heart failure, COVID-19 = Coronavirus disease 2019, HE = Hepatic encephalopathy, Cirrhosis = Liver cirrhosis, TAVR = Transcatheter aortic valve replacement, PBD7 = Post Burn Day 7, PBD14 = Post Burn Day 14, Se = Sensitivity, Sp = Specificity, TP = True positive, FP = False positive, FN = False negative, TN = True negative, AUC = Area under the receiver operating characteristic curve, OR = Odds ratio, CI = Confidence interval. To convert urine albumin-to-creatinine ratio (UACR) from SI units (mg/mmol) to conventional units (mg/g), multiply by 8.84. For example, 3.4 mg/mmol ≈ 30 mg/g.

Fig. 2.

Forest plot of pooled odds ratios for urinary albumin-to-creatinine ratio (UACR) in predicting acute kidney injury (AKI). The forest plot displays the individual and pooled odds ratios (ORs) for the risk of acute kidney injury (AKI) associated with elevated urinary albumin-to-creatinine ratio (UACR). Squares represent the OR of each included study, with the size of the square proportional to the study’s weight in the meta-analysis. Horizontal lines indicate 95% confidence intervals (CIs). The diamond represents the pooled OR (1.39) calculated using a random-effects model. AKI acute kidney injury, CI Confidence interval, OR Odds ratio, UACR Urinary albumin-to-creatinine ratio.

Pooled predictive performance

Twelve studies^{13,15,20,21,24–29,31,32} reported sensitivity and specificity estimates for UACR in predicting AKI. Based on a random-effects model, the pooled sensitivity was 0.71 (95% CI 0.59–0.80), and the pooled specificity was 0.67 (95% CI 0.56–0.76). The pooled DOR was 3.99 (95% CI 2.49–6.39), indicating moderate predictive performance. However, substantial heterogeneity was observed across studies (I² = 89.7% for sensitivity, I² = 97.8% for specificity, and I² = 90.5% for DOR; all p < 0.001), reflecting variability in predictive performance across populations and clinical settings (shown in Fig. 3).

Fig. 3.

Pooled sensitivity and specificity estimates of urinary albumin-to-creatinine ratio (UACR) for acute kidney injury (AKI) diagnosis. The left panel displays the pooled sensitivity (0.71), and the right panel displays the pooled specificity (0.67) of UACR for predicting AKI. Squares represent point estimates from individual studies, and horizontal lines indicate 95% confidence intervals (CIs). The diamonds at the bottom of each panel represent the pooled estimates derived from a random-effects model. AKI acute kidney injury, CI Confidence interval, UACR Urinary albumin-to-creatinine ratio.

Summary receiver operating characteristic (sROC) curve analysis

To evaluate the overall predictive performance of UACR in predicting AKI, we generated a sROC curve incorporating data from 12 studies^{13,15,20,21,24–29,31,32}. The pooled area under the curve (AUC) was 0.74 (95% CI 0.64 –0.84), indicating moderate predictive accuracy of UACR for AKI development and variability across studies (Shown in Fig. 4).

Fig. 4.

Summary receiver operating characteristic (sROC) curve for urinary albumin-to-creatinine ratio (UACR) predicting acute kidney injury (AKI). The plot illustrates the overall predictive performance of UACR. Each numbered circle represents the sensitivity and specificity of an individual study included in the meta-analysis (corresponding to the study numbers in Table 2). The red diamond indicates the summary operating point (pooled sensitivity and specificity). The solid black curve represents the fitted sROC curve. The shaded blue ellipse indicates the 95% confidence region, and the dashed blue contour outlines the 95% prediction region. The area under the curve (AUC) is 0.74, indicating moderate diagnostic accuracy. AKI Acute kidney injury, AUC Area under the curve, sROC Summary receiver operating characteristic, UACR Urinary albumin-to-creatinine ratio.

Subgroup analysis: cardiac surgery patients

In the subgroup of patients undergoing cardiac surgery, the pooled OR for subsequent AKI was 1.56 (95% CI 1.14–2.15), indicating a significant association between elevated UACR and increased risk of AKI. This analysis included three studies^13,20,21, with a combined total of 7325 patients. Heterogeneity across studies was low (I² = 15.4%, τ² = 0.0038, χ²₂ = 2.36, p = 0.307), suggesting more consistent predictive performance in this population (Shown in Supplementary Fig. S1).

UACR for prediction of AKI progression

Three studies^33–35, comprising a total of 1,596 patients, evaluated the association between elevated UACR levels and AKI progression. The pooled OR, calculated using a random-effects model, was 3.76 (95% CI 2.59–5.45), indicating a significant association between higher UACR levels and increased odds of AKI progression. Heterogeneity across studies was minimal (I² = 0.0%, τ² = 0, Q(2) = 0.63, p = 0.730). (Shown in Supplementary Fig. S2).

Pooled prognostic performance

Based on data from three studies^33–35, the pooled sensitivity and specificity estimates for UACR in predicting AKI progression were 0.75 (95% CI 0.34–0.94) and 0.53 (95% CI 0.16–0.87), respectively, based on a random-effects model. The pooled DOR was 3.76 (95% CI 2.78–5.08), with no heterogeneity observed (I² = 0%, p = 0.728). However, both pooled estimates exhibited substantial heterogeneity, with I² = 88.4% for sensitivity and I² = 98.6% for specificity, indicating variability in prognostic performance across different populations and clinical settings (Shown in Supplementary Fig. S3).

UACR comparing with other biomarkers

Out of the 13 included studies^{13,15,20,21,24–32}, five^{15,25,27,29,31} directly compared the predictive performance of the UACR with other biomarkers for AKI. Due to substantial heterogeneity in patient populations, clinical settings, biomarker types, and AKI definitions, a meta-analysis was not feasible. Only one study¹⁵ reported that UACR outperformed other biomarkers, specifically urinary Neutrophil Gelatinase-Associated Lipocalin (NGAL) and both urine and plasma Cystatin C, in predicting AKI among patients with acute myocardial infarction. In contrast, the remaining four studies^25,27,29,31 favored serum or urinary Cystatin C, NGAL, or urinary angiotensinogen (UAGT) over UACR. These findings suggest uncertainty regarding the comparative predictive performance of UACR (Shown in Table 1).

For the prediction of AKI progression, three studies^33–35 compared UACR with other biomarkers. All consistently found that plasma/urinary NGAL, UAGT, and urinary NAG (uNAG) provided superior predictive performance for identifying worsening of AKI stages (Supplementary Table S1).

Association between UACR and clinical outcomes

Only four studies examined the association between UACR and clinical outcomes. Coca et al.²⁰ and Zhang et al.²⁸ reported that higher UACR was associated with longer ICU and hospital length of stay. Molnar et al.²¹ found that elevated UACR was linked to increased risk of in-hospital death or dialysis, although the association with ICU stay was not significant after adjustment. George et al.¹³ demonstrated that UACR ≥ 300 mg/g was significantly associated with increased 90-, 180-, and 365-day mortality, longer hospital stays, and a higher incidence of AKI. Due to limited data and variability in outcome reporting, a meta-analysis was not feasible.

Sensitivity analysis

To assess the robustness of our findings, we conducted a sensitivity analysis by excluding the study Coca et al.²⁰ from the pooled analysis due to suspected overlapping data with Molnar et al. The association between higher UACR levels and AKI remained statistically significant, with a pooled OR of 1.37 (95% CI 1.01–1.85).

Meta-regression analysis

Meta-regression analyses were performed to explore potential sources of heterogeneity. Higher UACR cutoff values were associated with lower odds ratios (β = − 0.0318, 95% CI − 0.0407 to − 0.0229; p < 0.0001), accounting for 99.2% of the observed heterogeneity (Supplementary Fig. S4). Baseline serum creatinine also showed a positive association with the effect estimate (β = 0.0554, 95% CI 0.0001 to 0.1108; p = 0.0497); however, this analysis was limited to six studies (Supplementary Fig. S5). Increasing mean age was positively associated with the pooled odds ratio (β = 0.0594, 95% CI 0.0403 to 0.0784; p < 0.0001), explaining 97.2% of the between-study heterogeneity (Supplementary Fig. S6). Detailed meta-regression model statistics, including τ², I², and moderator test results, are provided in Supplementary Table S3.

Publication bias assessment

Publication bias was assessed using a funnel plot of diagnostic odds ratios and Egger’s regression test. Visual inspection of the funnel plot revealed asymmetry, with smaller studies tending to report larger effect sizes (Supplementary Fig. S7). Egger’s test confirmed statistically significant asymmetry (t = 3.30, df = 10, p = 0.008). However, this finding should be interpreted with caution given the substantial heterogeneity across studies, which can also contribute to funnel plot asymmetry.

Discussion

This systematic review and meta-analysis demonstrate that elevated UACR is associated with both the development and progression of AKI, particularly among cardiac surgery patients. Cutoff values used to define elevated UACR for predicting AKI incidence varied considerably, ranging from 3 to 16.16 mg/mmol. Meta-regression further demonstrated that higher UACR cutoff values were associated with lower odds ratios for AKI development.

Our findings align with prior research linking albuminuria to AKI risk. In cardiac surgery patients, Coca et al.²⁰ demonstrated that preoperative UACR, measured before AKI onset, independently predicted postoperative AKI, improving risk stratification. Molnar et al.²¹, along with a related study by Coca et al.²⁰, also found that elevated postoperative UACR was associated with subsequent AKI, suggesting that both pre- and early postoperative measurements may have clinical utility.

In patients with acute myocardial infarction, Tziakas et al.¹⁵ reported that admission UACR outperformed NGAL and cystatin C in predicting AKI, illustrating context-specific variability in biomarker performance. Nevertheless, across broader clinical settings, our review found that in other studies UACR was often outperformed by biomarkers such as NGAL and cystatin C for predicting both AKI incidence and progression^25,27,29,31.

The predictive value of UACR in detecting AKI likely stems from its role as an early marker of glomerular and systemic endothelial injury^36,37. Under normal conditions, the integrity of the glomerular filtration barrier prevents albumin from leaking into the urine³⁸. In patients at risk of AKI, systemic inflammation, oxidative stress, and hemodynamic instability can impair endothelial function and disrupt the glomerular barrier. Loss of the endothelial glycocalyx—an important regulator of vascular permeability—has been linked to both albuminuria and widespread endothelial dysfunction^39,40, resulting in increased glomerular permeability and urinary albumin excretion. Consequently, UACR may rise before any detectable change in serum creatinine. In addition, filtered albumin can overwhelm tubular reabsorptive capacity and induce tubular stress by activating proinflammatory and profibrotic pathways, thereby contributing to tubular injury and AKI progression⁴¹. However, not all AKI phenotypes present with albuminuria. Functional or predominantly hemodynamic AKI is often characterized by bland urine sediment with preserved glomerular barrier integrity, in which UACR may remain low^42,43. This distinction suggests that UACR may preferentially identify inflammatory or structural kidney injury rather than isolated functional changes, with potential implications for clinical risk stratification¹².

From a clinical perspective, elevated UACR may help identify patients at increased risk of AKI, particularly in the cardiac surgery setting where heterogeneity was low. In contrast, given the limited number of studies within other individual clinical contexts, our findings should not be extrapolated uniformly to all hospitalized populations. If validated, incorporation of UACR measurement into clinical assessment could prompt closer renal function monitoring and avoidance of nephrotoxic exposures in high-risk individuals. However, given the substantial heterogeneity across studies, the lack of standardized cutoff thresholds, and moderate predictive performance observed, UACR should not be used as a standalone criterion but rather interpreted alongside clinical context and established risk factors. Prospective studies are needed to define optimal thresholds and to evaluate whether UACR-guided management can meaningfully improve patient outcomes.

This systematic review has several strengths. First, to our knowledge, it is the first to assess the predictive utility of UACR for both AKI incidence and progression across a broad spectrum of hospitalized patients. Second, the review adheres rigorously to PRISMA guidelines¹⁹, and employs a comprehensive, systematic search strategy. Third, the inclusion of a large cumulative sample of 10,438 plus 1,596 patients enhances the statistical power and robustness of the findings. Fourth, the majority of included studies were prospective in design, with only three being retrospective, thereby strengthening the overall quality of evidence.

This study has several limitations:

1. High heterogeneity was observed across studies (I² > 85% for most pooled estimates), which substantially limits the generalizability of the pooled results. This heterogeneity likely reflects differences in AKI definitions (KDIGO, AKIN, RIFLE, or study-specific creatinine thresholds), patient populations and clinical settings, and the timing of UACR measurement. Meta-regression identified UACR cutoff values, mean age, and baseline serum creatinine as significant effect modifiers, partially explaining this variability.

2. UACR’s predictive utility may be limited in AKI phenotypes that do not manifest significant albuminuria, such as isolated prerenal (functional) AKI, where the glomerular barrier remains intact and urine sediment is typically bland. While this may reduce sensitivity for detecting all-cause AKI, it suggests a potential role for UACR in identifying inflammatory or structural kidney injury.

3. UACR cutoff thresholds varied widely across studies (ranging from 3 to 16.16 mg/mmol), precluding the recommendation of a single optimal threshold.

4. Potential cohort overlap between the cardiac surgery studies by Coca et al.²⁰ and Molnar et al.²¹ may have influenced subgroup findings, although sensitivity analyses confirmed the robustness of associations.

5. Limited availability of studies directly comparing UACR with other biomarkers constrained the scope of comparative analyses.

6. The majority of included studies were observational, limiting causal inference.

7. Egger’s test indicated statistically significant funnel plot asymmetry (p = 0.008), suggesting potential publication bias or small-study effects. However, given the substantial heterogeneity observed, this asymmetry may also reflect clinical and methodological diversity rather than true publication bias.

Despite its limitations, UACR remains a promising biomarker due to its broad availability, low cost, standardization, real-time accessibility, and non-invasive nature. Given the marked clinical heterogeneity across settings, pooled estimates may obscure pathophysiologic differences and should not be interpreted as a uniform effect across all hospitalized cohorts. When interpreted within appropriate clinical contexts—most notably cardiac surgery, where heterogeneity was low—it offers more consistent predictive value. Further research is needed to evaluate the prognostic utility of serial UACR measurements, define more precise cutoff thresholds, and determine associations with outcomes such as mortality and hospital length of stay. Additionally, direct comparisons with other biomarkers are warranted to inform more comprehensive and evidence-based risk stratification strategies.

Conclusions

This systematic review and meta-analysis demonstrate that elevated UACR is significantly associated with both the development and progression of acute kidney injury in hospitalized adults, particularly among patients undergoing cardiac surgery. However, high heterogeneity across studies (I² > 85%) limits the generalizability of these findings, and results should be interpreted with caution. UACR is a low-cost, non-invasive, and widely accessible biomarker that may support early risk stratification for AKI. Further research is needed to validate standardized diagnostic thresholds, evaluate the utility of serial UACR measurements, and determine its prognostic relevance for outcomes such as mortality and hospital length of stay. Comparative studies with other biomarkers are also essential to define UACR’s role in AKI prediction and risk stratification across diverse clinical settings.

Author contributions

Conceptualization and study design: NK, NR, ASN, RB. Data acquisition (screening, eligibility assessment, data extraction, and risk of bias assessment): NK, JI, NR. Formal analysis and statistical analysis: NK, NR, ASN. Investigation and validation (critical review of extracted data and analyses): NK, NR, YH, AC, JN, GE, ASN, RB. Writing—original draft: NK. Writing—review and editing: NK, NR, YH, AC, JN, GE, ASN, RB. Supervision/mentorship: ASN, RB. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.

Data availability

The data underlying this systematic review and meta-analysis are available from the corresponding author upon reasonable request. Individual study data were extracted from publicly available published articles. No unpublished primary data were used.

Declarations

Competing interests

The authors declare no competing interests.

Ethics declarations

Ethical approval was not required for this study, as it is a systematic review and meta-analysis of previously published data (Siriraj Institutional Review Board SIRB Protocol No. 687/2568 [Exempt]). The review protocol was registered in PROSPERO (registration number: CRD420250648041).