Comparison of Two Different Semiquantitative Urinary Dipstick Tests with Albumin-to-Creatinine Ratio for Screening and Classification of Albuminuria According to KDIGO. A Diagnostic Test Study

Background: Semiquantitative dipstick tests are utilized for albuminuria screening. Methods: In a prospective cross-sectional survey, we analyzed the diagnostic test validity of the semiquantitative colorimetric indicator-dye-based Combur9-Test® and the albumin-specific immunochromatographic assay Micral-Test® for the detection of albuminuria, the distribution of the semiquantitative measurements within the albuminuria stages according to KDIGO, and the utility for albuminuria screening compared with an albumin-to-creatinine ratio (ACR) in a walk-in population. Results: In 970 subjects, albuminuria (≥30 mg/g) was detected in 12.7% (95% CI 85.6–96.3%) with the ACR. Sensitivity was 82.9% (95% CI 75.1–89.1%) and 91.9% (95% CI 88.7–96.9%) and specificity 71.5% (95% CI 68.4–74.6%) and 17.5% (95% CI 15.0–20.2%) for the Combur9-Test® and Micral-Test®, respectively. Correct classification to KDIGO albuminuria stages A2/A3 with the Combur9-Test® was 15.4%, 51.4%, and 87.9% at cut-offs of 30, 100, and ≥300 mg/dL, and with the Micral-Test® it was 1.8%, 10.5%, and 53.6% at cut-offs of 2, 5, and 10 mg/dL, respectively. Overall, disagreement to KDIGO albuminuria was seen in 27% and 73% with the Combur9-Test® and Micral-Test®, respectively. From the total population, 62.5% and 15.3% were correctly ruled out and 2.2% and 1% were missed as false-negatives by the Combur9-Test® and Micral-Test®, respectively. Conclusion: Compared to the Combur9-Test®, the utility of the Micral-Test® is limited, because the fraction of correctly ruled out patients is small and a large proportion with a positive Micral-Test® require a subsequent ACR conformation test.


Introduction
The burden of chronic kidney disease (CKD) is increasingly recognized as a global public health problem, with major negative implications on quality of life, premature death, and enormous costs on healthcare systems [1,2]. It is feared that with the ongoing increase in patients affected by the most common CKD risk factors-diabetes [3], hypertension [4], and obesity [5]-CKD prevalence will continue to rise globally, in both developed and developing countries [1,6,7]. Although only a small fraction of patients with CKD are progressing to end-stage renal disease (ESRD), the vast majority are affected by cardiovascular mortality [8,9]. Because early stages of CKD often remain undiagnosed, strategies are required not only for treatment regimens, but also for early detection [9,10]. Albuminuria is a sensitive surrogate marker for kidney impairment and is independent of a reduced estimated glomerular filtration rate (eGFR) associated with increased risk for chronic kidney disease [7,11,12]. According to the 2012 Kidney Disease, Improving Global Outcomes (KDIGO) guidelines, quantitative albumin-to-creatinine ratio (ACR) is the "first priority assessment test" for the evaluation of albuminuria [13]. However, for population screenings, research, and routine clinical check-ups, point-of-care (POC) urine dipstick devices are often utilized, especially in ambulatory or low-middle income settings, where rapid low-cost diagnostics are essential [9,[14][15][16]. In clinical practice and according to guidelines for the classification and diagnosis of albuminuria, a finding of semiquantitatively detected albuminuria requires a subsequent quantitative analysis for confirmation and classification to albuminuria stages [13]. Several studies have explored sensitivity and specificity measures of urine dipstick tests and their predictive values for the detection of albuminuria [10,[17][18][19][20][21]. However, relatively little is known about the usefulness of semiquantitative urine dipstick tests for albuminuria diagnosis and classification according to KDIGO guidelines [13].
The aim of this study was to assess the validity of a semiquantitative standard colorimetric indicator-dye-based multi dipstick (Combur9-Test ® ) and an albumin-specific immunochromatographic assay (Micral-Test ® ) compared with the quantitative ACR reference diagnostic test [13]. A further aim was to classify the dipstick results according to KDIGO albuminuria stages, and to assess the utility of the semiquantitative dipstick tests for an albuminuria screening in an outpatient clinic (OPC) walk-in population.

Study Population
This is a prospective cross-sectional study performed in a walk-in population of the outpatient clinic of the Bagamoyo District Hospital (BDH) as part of the RenalOne study as previously described [7]. Briefly, the BDH is located in Bagamoyo township and provides care for a semirural population. Bagamoyo district had approximately 300.000 inhabitants in the 2012 census [22]. The OPC was visited on average by 120 (range 41-164) patients daily. For the current study, one consultation hour of the general outpatient ward was designated to ensure a highly standardized procedure. From the newly registered patients of the OPC, 15 to 20 patients per day were consecutively seen. The call up of the patients from the OPC ward was done through the medical staff, without any involvement of the investigators. Pregnant women, patients younger than 18 years, and patients neither able nor willing to provide informed consent were excluded.

Reporting
We followed the STARD 2015 guidelines for reporting diagnostic accuracy studies, as described by Cohen and Korevaar et al. [23].

Measurements and Procedures
All data were collected in a case report form (CRF), translated from English to Swahili. In all participants, past medical histories and smoking status were assessed through interviews, and body weight and height, and blood pressure (BP) were measured and recorded. Glycated hemoglobin A1c (HbA1c) was measured from capillary blood by using a bedside DCA 2000+ Analyzer (Siemens Healthcare Diagnostics, 8047 Zürich, Switzerland). A random clean urine specimen was collected in all patients. Within 30 min after returning the urine specimen to the study team, two different types of semiquantitative dipstick tests and an ACR test were carried out, in order to detect and to determine the degree of albuminuria.
As index test, two semiquantitative urine dipstick tests were performed and recorded during the time of the ACR measurement: the colorimetric dipstick Combour9-Test ® (Bayer Diagnostics, 51368 Leverkusen, Germany) and the immunochromatographic dipstick Micral-Test ® (Roche Diagnostics, 6343 Rotkreuz, Switzerland). Both urine dipstick tests were carried out manually according to the manufacturer's instructions, read, interpreted, and recorded by the onsite investigator and full-time technician (N.H.), and supervised by a senior nephrologist (M.M.) [7]. The semiquantitative total urine protein detection with the Combour9-Test ® is based on the ability of the protein to change the dye color of an acid-base indicator, forming the anion of 3 ,3",5 ,5"-tetrachlorphenol-3,4,5,6tetrabromsulfophthalein in presence of protein in the urine, resulting in a gradational change of the indicator dye [24]. Although the Combur9-Test ® aims for total protein in the urine, the sensitivity to albumin is higher than to other proteins found in the urine; this is explained by the increased affinity of albumin to protons [24][25][26]. The rational in our study to consider a Combur9-Test ® value of ≥15-30 mg/dL (1+) as positive is based on the statement in the KDIGO 2012 guidelines that a reagent strip with (1+) protein positivity can substitute a quantitative albuminuria measurement where it not available [13]. Further, it is pointed out that a reagent strip (1+) protein and ACR ≥ 30 mg/g (≥3 mg/mmol) were associated with subsequent risk for CKD progression, acute kidney injury (AKI) [11], cardiovascular mortality, and all-cause mortality in the general population and in populations at risk [13,27,28]. The semiquantitative urinary dipstick Micral-Test ® is a chromatographic immunological procedure based on a gold-labelled monoclonal antibody with high specificity to human albumin. The correct handling is of significant importance for the test accuracy. With the correct amounts of fluid (urine) absorbed, the antibody-albumin conjugate is transported to a reactive detection pad. For the Micral-Test ® , the detection limit is 20 mg/L (2 mg/dL) [24]; cut-off levels are less clear and estimations for cardiovascular mortality are controversial, however several studies have shown that the test validity measures when compared to quantitative methods were strong at ≥20 mg/L (≥2 mg/dL) [29,30]. Further, at this cut-off, a sensitivity of 90% and a specificity of 96% were shown in a laboratorybased screening against the turbidimetric immunoassay [31]. Two index tests with different detection limits were chosen to analyze the influence of detection limits on models of albuminuria screening algorithms.
The reference test was a POC quantitative ACR measurement in a spot urine sample. According to the KDIGO guidelines, the urine ACR measurement is the "first priority test" for the evaluation of albuminuria for a CKD classification [13]. The ACR is a linear continuous value with a cut-off for a moderately increased albuminuria and a relevant risk for CKD at ACR ≥ 30 mg/g (≥3 mg/mmol) [13]. We performed the albumin/creatinine assay on the DCA 2000+ Analyzer (Siemens Healthcare Diagnostics, 8047 Zürich, Switzerland).) in the ambulatory setting of the OPC laboratory. The test cartridges were stored at 4 • C, and the calibration of the DCA 2000+ analyzer was regularly performed and recorded with the calibration card according to the manufacturer's guide. The albumin/creatinine assay is based on an immunoassay for the creatinine detection and an alkaline colorimetric assay for the creatinine test [32]. All reagents are contained in the test cartridge, all reaction steps, measurements, and calculations are performed automatically, and results are displayed on the instrument screen [32].
In the "Assessment of Performance" section of the United States (U.S.) Food and Drug Administration (FDA) "510(k) Safety and Effectiveness Summary (K963142)", it is declared that the DCA 2000+ Analyzer and the albumin/creatinine assay were studied in clinical settings and the results were substantially equivalent to results from methods used in clinical laboratory practice [32]. The linear detection range of the albumin/creatinine assay is an ACR of 1 to 2000 mg/g [32].

Outcomes/Aims
The primary outcome of this study was the assessment of the diagnostic test validity measures of an indicator-dye-based reagent strip (Combur9-Test ® ) and an albuminsensitive immunological assay (Micral-Test ® ), for the detection of albuminuria, and the distribution of the semiquantitative urine protein evaluation within the albuminuria stages according to KDIGO guidelines [13]. The secondary outcome was a utility analysis of the semiquantitative dipsticks for albuminuria screening.

Statistical Analyses
Statistical analyses were performed using STATA version 14 (StataCorp., College Station, TX, USA). Discrete variables were expressed as counts (percentage), and comparison between groups was done with Pearson's chi-square test or Fisher's exact test. Continuous variables were expressed as mean ± standard deviation (SD) if normally distributed or as median and range if not normally distributed, and t-test or Mann-Whitney test was used for comparison between groups.

Results
Overall, 1006 patients were recruited ( Figure 1). Five patients aged less than 18 years, 19 pregnant women, and eight patients with a missing ACR test were excluded, leaving 974 patients for the final analysis. In an additional four patients, Combur9-Test ® and/or Micral-Test ® results were missing, and therefore were not included in the comparison between the semiquantitative dipstick tests and the ACR test, leaving n = 970 for the assessment of the test validity, the distribution within albuminuria stages, and the semiquantitative dipstick test utility analysis for albuminuria screening algorithms.

Sensitivity and Specificity Analysis of Semiquantitative Urine Dipstick Tests Compared with Albumin-to-Creatinine Ratio Test
In Figure 2, the correlation of the semiquantitative dipstick test results with the ACR is summarized. For the Combur9-Test (Figure 2A and embedded table), area (a) shows n = 102 true positives, with a positive Combur9-Test ® and a positive ACR reference test, (b) shows n = 241 false positives, with a positive Combur9-Test ® and a negative ACR reference test, (c) shows n = 21 false negatives, with a negative Combur9-Test ® and a positive ACR reference test, and (d) shows n = 606 true negatives, with a negative Combur9-Test ® and a negative ACR reference test. For the Micral-Test ® (Figure 2B and embedded table), area (a) shows n = 113 true positives, with a positive Micral-Test ® and a positive ACR reference test, (b) shows n = 699 false positives, with a positive Micral-Test ® and a negative ACR reference test, (c) shows n = 10 false negatives, with a negative Micral-Test ® and a positive ACR reference test, and (d) shows n = 148 true negatives, with a negative Micral-Test ® and a negative ACR reference test.

Significance of Possible Models for Albuminuria Screening Algorithms Based on Combur9-Test ® or Micral-Test ® in Clinical Practice
In scenario A, the cut-off for Combur9-Test ® positivity was set at (1+) ≥30 mg/dL. A negative test had a high NPV of 96% and included 64.6% (n = 627/970) of the screened population ( Figure 5A). Without further confirmation of a negative test result, 2.2% (n = 21/970) and 17% (n = 21/123) cases with a relevant albuminuria of the whole population and of all albuminuria-positives would be missed, respectively. Correctly ruled out would be 62.5% (n = 606/970) and 71.5% (n= 606/847) of the whole population and all albuminuria-negatives, respectively. Of the 343 patients with a positive dipstick Combur9-Test ® and a subsequent ACR conformation test, 10.5% (n = 102/970) and 83% (n = 102/123) of the whole population and of all albuminuria-positives would be confirmed, and 24.8% (n = 241/970) and 70.2% (n = 241/343) of the whole population and of all Combur9-Test ® positives would be over tested, respectively. In summary, the proposed algorithm would mean that 17% of the true positive results would be missed (i.e., 2.2% of the whole population) and that an unnecessary additional test would have to be carried out in 70.2% of patients with a positive screening dipstick test (i.e., in 24.8% of the whole population).
In scenario A, for Micral-Test ® positivity, the cut-off was set at ≥2 mg/dL. A negative test had an NPV of 94% and included 16.3% (n = 158/970) of the screened population.
Without further confirmation of a negative test result, 1% (n = 10/970) and 8% (10/123) of cases with a relevant albuminuria of the whole population and of all albuminuria-positives would be missed, respectively. Correctly ruled-out would be 15.3% (n = 148/970) of the population and 17.5% (n = 148/847) of all albuminuria-negatives, respectively. Of the 83% (n = 812/970) with a positive dipstick Micral-Test ® and a subsequent ACR conformation test, 11.6% (113/970) of the population and 92% (n = 113/123) of all albuminuria-positives would be confirmed, and 72.1% (n = 699/970) of the population and 86% (n = 699/812) of all Micral-Test ® positives would be over tested, respectively. In summary, the proposed algorithm would mean that 8% of the true positive results would be missed (i.e., 1% of the whole population) and that an unnecessary additional test would have to be carried out in 86% of the patients with a positive screening dipstick test (i.e., in 72.1% of the whole population) ( Figure 5A).
In scenario B, the cut-off for Combur9-Test ® positivity was set at (≥2+) ≥100 mg/dL. A negative test had an NPV of 85.4% and included 89.4% (n = 867/970) of the screened population ( Figure 5B). Without further confirmation of a negative test result, 6.0% (n = 58/970) and 47% (n = 48/123) of cases with a relevant albuminuria of the whole population and of all albuminuria-positives would be missed, respectively. Correctly ruled out would be 83.4% (n = 809/970) and 95.5% (n= 809/847) of the population and all albuminurianegatives, respectively. Of the n = 103 patients with a positive dipstick Combur9-Test ® and a subsequent ACR conformation test, 6.7% (n = 65/970) and 53% (n = 65/123) of the whole population and of all albuminuria-positives would be confirmed, and 3.9% (n = 38/970) and 37% (n = 38/103) of the whole population and of all Combur9-Test ® positives would be over tested, respectively. In summary, the proposed algorithm would mean that 47% of the true positive results would be missed (i.e., 6% of the whole population) and that an unnecessary additional test would have to be carried out in 37% of patients with a positive screening dipstick test (i.e., in 3.9% of the whole population).
In scenario B, for Micral-Test ® positivity, the cut-off was set at ≥5 mg/dL. A negative test had an NPV of 98% and included 62.4% (n = 606/970) of the screened population. Without further confirmation of a negative test result, 1.9% (n = 18/970) and 15% (18/123) of cases with a relevant albuminuria of the whole population and of all albuminuria-positives would be missed, respectively. Correctly ruled out would be 60.6% (n = 588/970) of the population and 69.4% (n = 588/847) of all albuminuria-negatives, respectively. Of the 37.5% (n = 364/970) with a positive dipstick Micral-Test ® and a subsequent ACR conformation test, 10.8% (105/970) of the population and 85% (n = 105/123) of all albuminuria-positives would be confirmed, and 26.7% (n = 259/970) of the population and 71% (n = 259/812) of all Micral-Test ® positives would be over tested, respectively. In summary, the proposed algorithm would mean that 15% of the true positive results would be missed (i.e., 1.9% of the whole population) and that an unnecessary additional test would have to be carried out in 71% of the patients with a positive screening dipstick test (i.e., in 26.7% of the whole population) ( Figure 5A).

Discussion
We compared the performance of two different urine dipstick tests in nearly 1000 patients. Numerically, both tests had a comparable test validity and performance for test sensitivity and its NPV, but relevant differences in the specificity and the corresponding PPV. This finding is also reflected in the classification to KDIGO albuminuria stages A1-A3 ( Figure 3A,B). The Micral-Test ® showed weak allocation to all albuminuria stages when positive and was too sensitive to classify the large fraction of subjects without albuminuria to stage A1 (normal to mildly increased). The Combur9-Test ® has its limitations but was accurately classifying patients with strong positive (≥3+) results to the corresponding albuminuria stages A2/A3 (moderately/severely increased), and a majority of those subjects without albuminuria correctly to stage A1. Our data illustrate a dose-dependency between the two urine dipstick tests, which leads to a clear concordance with increasing albumin amounts in urine (Figure 4). This consistency makes it clear that both index tests detect the same individuals when albuminuria is present, whereas the immunochromatographic albumin-specific Micral-Test ® already detects albuminuria in concentrations in which the standard colorimetric indicator-dye-based multi dipstick Combur9-Test ® is still negative.
In the models for albuminuria screening algorithms, we clearly illustrate that both tests have a high accuracy in correctly ruling out patients with no further need for ACR conformation ( Figure 5A,B). However, the difference in absolute numbers was significantly different for the two surveyed index tests. The immunochromatographic albumin-specific Micral-Test ® had a specificity of 17.5% (n = 148/847 of all ACR negatives) and therefore was correctly ruling out only 15% (n = 148/970) of the population. While the colorimetric indicator-dye reagent strip Combur9-Test ® with a specificity of 71.5% (n = 606/847 of all ACR negatives) was four times more (62% of the population (n = 606/970)) efficient in correctly ruling out albuminuria-negatives from the population screened. The consequence of the high proportion already ruled out with the reagent strip Combur9-Test ® was resulting in a relevant smaller fraction (35% (n = 343/970)) of subsequent ACR testing compared with the Micral-Test ® , where 83% (n = 812/970) of the entire population would require an ACR conformation test.
Although the specificity improved with increasing cut-off level (i.e., ≥5 mg/dL or ≥10 mg/dL), the Micral-Test ® did not exceed a PPV of over 54%. Therefore, the chance for a correct detection of a clinically relevant albuminuria according to the KDIGO guidelines with the albumin-specific dipstick test was nearly fifty-fifty, even at its highest test positivity level (≥10 mg/dL) [13]. The vast majority (72.1%; n = 699/970) of Micral-Test ® positive patients were not confirmed to have albuminuria in a ACR test, and therefore were over tested. Similarly, for the Combur9-Test ® , more than two-thirds (70.3%, n = 241/343) of the patients with a positive dipstick test result were not confirmed to have albuminuria in an ACR conformation test. However, the decisive difference is that the overall number for a required ACR conformation test was significantly smaller with the Combur9-Test ® (35%; n = 343/970) compared with the Micral-Test ® (72.1%; n = 699/970). The indicatordye-based Combur9-Test ® strip is of good value to correctly rule out patients with no albuminuria, but the semiquantitative detection device has its limitations, with weak test validity measures for ≥1+ (30 mg/dL) and ≥2+ (100 mg/dL) cut-offs, and therefore these quantifications are of little value for albuminuria classification according to KDIGO [13].
If the cut-off level was increased by one level for each index test, the performance of the Combur9-Test ® ≥2+ (100 mg/dL), in regard to correctly ruling out patients, further increased to 83.4%, but at the cost of a relevant number of patients that were missed and judged falsely as negatives (47% of all true positives). If the cut-off level for the Micral-Test ® was increased to ≥5 mg/dL, the overall distribution of correctly detected, ruled out, missed, and over tested patients was nearly equal to the Combur9-Test ® with its regular cut-off levels at ≥1+ (30 mg/dL) Figure 5C(a,d).
With these results from the colorimetric indicator-dye-based urine test strip, we confirm the findings of large population-based studies from across the globe, and several smaller studies in at-risk cohorts, including some from SSA [17,21,[36][37][38][39][40]. However, contrary to the opinion of other authors [41], the utility value of the Micral-Test ® to correctly rule out patients is only partially correct, because the fraction of correctly ruled out healthy patients was comparatively small, and a large proportion with a positive Micral-Test ® would require a subsequent ACR conformation test according to guideline-orientated clinical practice [13].
In our study, we face some limitations. A change in prevalence may lead to different test validity measures and must be considered also for event (albuminuria) probabilities [42]. There are several reasons why albuminuria prevalence might be overestimated with a single-point measurement of ACR. First, transient albuminuria (due to the dayto-day variability) could not be excluded due to the cross-sectional study design with a missing second urine sample for confirmation, but this is the common practice in prevalence studies and therefore the results should be comparable and to a certain degree generalizable [6,14,43]. The second reason why our prevalence could be overestimated is the study setting within an outpatient clinic, enrolling a clientele which could have a higher estimated albuminuria or CKD prevalence, compared with a community-based setting [14,43,44]. Finally, due to the variability of albuminuria during the day, the random collection of urine samples could skew the prevalence rate [45], but this should not influence the finding of the concordance and agreement between index and reference tests. Both semiquantitative urine dipstick tests are dependent on urine concentration and the urine specific gravity [46], which was not assessed in our laboratory. Further, the albuminuria screening algorithm models were built under the assumption that we detected patients with persistent and not transient albuminuria. However, a strength of our study remains that the index dipstick tests and the ACR reference test were carried out simultaneously from the same urine specimen, supporting the concordance and agreement of the tests.

Conclusions
The two semiquantitative index tests differ strongly in specificity, especially at their lowest positivity levels. Compared to the Combur9-Test ® , the more expensive and more specific Micral-Test ® does not add any significant benefit to clinical practice and would require a much higher rate of unnecessary subsequent ACR clarification testing, in order to diagnose albuminuria according to guideline recommendations [13]. The Combur9-Test ® with the ability to rule out albuminuria accurately, without a relevant loss of missed albuminuria cases, is therefore a valid initial diagnostic tool to screen for albuminuria and CKD in a walk-in population with an assumed albuminuria prevalence of 10%. These findings are of importance considering the distribution of resources in clinical practice and research settings, where upon a positive dipstick test result a subsequent ACR conformation test should be executed.
Author Contributions: All authors made contributions to the study. N.C.H. carried out the fieldwork in the laboratory and clinic, collected, entered, and analyzed the data, conceptualized, wrote and illustrated the manuscript. M.M. designed the study, wrote and revised the manuscript, cleaned data, and supervised the fieldwork. A.H. coordinated the fieldwork and the approval of ethical clearance, informed the Bagamoyo community authorities about the study, translated protocols from English to Swahili, and revised the manuscript. K.R. coordinated this study onsite and provided important intellectual content input for study design. I.M.K. carried out clinical fieldwork and supported the translation of the protocols. A.S. revised statistical analysis. S.A. coordinated the fieldwork and communicated with local authorities. C.F.R.H. coordinated the fieldwork and provided important intellectual content input for the study design. All authors have read and agreed to the published version of the manuscript.

Funding:
The study was supported by a project fund from the University Hospital Basel (VFWAWFPool -section medicine) and the "Freiwillige Akademische Gesellschaft Basel (FAG, No. 02/12/11)". The sponsors did not influence study design, or collection, analysis and interpretation of data, writing of the report, or the decision to submit the report for publication. Informed Consent Statement: All included patients signed an informed consent form in Swahili. For illiterate patients, the informed consent was read, and a fingerprint of the index finger was used instead of a signature.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.