Skip to content
BY 4.0 license Open Access Published by De Gruyter March 15, 2021

Evaluating chronic kidney disease in rural South Africa: comparing estimated glomerular filtration rate using point-of-care creatinine to iohexol measured GFR

  • Sean Currin ORCID logo EMAIL logo , Mwawi Gondwe , Nokthula Mayindi , Shingirai Chipungu , Bongekile Khoza , Lungile Khambule , Tracy Snyman , Stephen Tollman , June Fabian , Jaya George and on behalf of the ARK Consortium

Abstract

Objectives

The prevalence of chronic kidney disease is rising rapidly in low- and middle-income countries. Serum creatinine and estimation of glomerular filtration rate (GFR) are critical diagnostic tools, yet access to centralised laboratory services remains limited in primary care resource-limited settings. The aim of this study was to evaluate point-of-care (POC) technologies for serum creatinine measurement and to compare their performance to a gold standard measurement using iohexol measured GFR (mGFR).

Methods

POC creatinine was measured using iSTAT® and StatSensor® devices in capillary and venous whole blood, and laboratory creatinine was measured using the compensated kinetic Jaffe method in 670 participants from a rural area in South Africa. GFR estimating equations Chronic Kidney Disease Epidemiology Collaboration and Modification of Diet in Renal Disease (CKD-EPI and MDRD) for POC and laboratory creatinine were compared to iohexol mGFR.

Results

Calculated GFR for laboratory and POC creatinine measurements overestimated GFR (positive bias of 1.9–34.1 mL/min/1.73 m2). However, all POC devices had less positive bias than the laboratory Jaffe method (1.9–14.7 vs. 34.1 for MDRD, and 8.4–19.9 vs. 28.6 for CKD-EPI). Accuracy within 30% of mGFR ranged from 0.56 to 0.72 for POC devices and from 0.36 to 0.43 for the laboratory Jaffe method. POC devices showed wider imprecision with coefficients of variation ranging from 4.6 to 10.2% compared to 3.5% for the laboratory Jaffe method.

Conclusions

POC estimated GFR (eGFR) showed improved performance over laboratory Jaffe eGFR, however POC devices suffered from imprecision and large bias. The laboratory Jaffe method performed poorly, highlighting the need for laboratories to move to enzymatic methods to measure creatinine.

Introduction

Chronic kidney disease (CKD) has a prevalence of 10.7% in sub-Saharan Africa, and is associated with significant morbidity and mortality [1]. Low- and middle-income countries (LMIC) suffer a disproportionate mortality associated with CKD when compared to high-income countries [2]. Therefore, screening strategies that aim to identify patients with early stages of CKD in primary care settings and ideally, prevent progression of CKD, are of profound importance.

Glomerular filtration rate (GFR) is considered the best overall index of kidney function and is an important criterion in the diagnosis and staging of CKD [3]. GFR cannot be quantitated directly. However, it can be assessed by the measurement of exogenous (inulin, Cr-EDTA, iohexol, iothalamate) or endogenous markers [4]. Exogenous markers are impractical for routine clinical use due to their limited access, high cost and duration of time required for testing. Estimated GFR (eGFR) equations based on endogenous markers and additional factors, such as age and sex, are recommended for the routine assessment of kidney function [3]. The creatinine based Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) and Modification of Diet in Renal Disease (MDRD) equations are the most commonly used [5]. The 2009 CKD-EPI creatinine equation is recommended for use in adults [3].

In resource-limited settings a lack of access to central laboratory services hinders the ability to implement effective CKD screening programs. Even when central laboratory services are available, logistical factors in primary health care clinics can impact the accuracy of creatinine measurements, such as sub-optimal storage of collected samples and delays in transport to the central laboratory [6]. Point-of-care (POC) testing offers an important solution to this problem by offering access to actionable results in real-time for clinical decision making [7]. While the 2012 KDIGO guidelines recognise that limited laboratory access is a problem, they fall short of recommending POC testing in such settings [8].

The aim of our study was to evaluate the performance of POC creatinine by comparing eGFR equations to iohexol measured GFR (mGFR), and to compare this performance with that of laboratory eGFR. Additionally, we looked at the performance of the MDRD and CKD-EPI equations for these different creatinine measurements.

Materials and methods

Our study formed part of the African Research on Kidney (ARK) study, which aimed to determine the population prevalence of CKD and associated risk factors in three sub-Saharan African countries, Malawi, South Africa, and Uganda. Detailed methods for the ARK study have been published [9]. The South African component was conducted in the Agincourt Health and Demographic Surveillance Site in Bushbuckridge, a rural sub-district of the Mpumalanga province in South Africa.

Participants from the community were invited to participate in this second phase of the study, to measure GFR using intravenous iohexol. The initial study comprised a population-based sample of 2021 adults aged 20–80 years to determine CKD prevalence and investigate associated risk factors. For the iohexol study, we stratified a subsample of the main study by sex and eGFR to ensure adequate sampling across all stages of CKD.

Measured GFR was determined by plasma excretion of iohexol. A 5 mL bolus of intravenous iohexol (Omnipaque™, GE Healthcare, Illinois) was administered in the antecubital fossa of one arm, followed by repeat plasma sampling from the antecubital fossa of the contralateral arm after 5, 120, 180 and 240 min. Prior to iohexol administration a baseline venous blood sample was obtained, which was used for real-time testing with the iSTAT® (Abbott Laboratories, Illinois) and StatSensor® (Nova Biomedical Corporation, Massachusetts) POC devices. At the same time, a capillary blood sample taken from a fingerprick was also analysed on the StatSensor® POC device. There were no adverse events related to the intravenous administration of iohexol.

On the day of iohexol testing, for each participant, samples for creatinine and iohexol were refrigerated (2–8 °C) until all participants for the day had been processed (usually by 3 pm), after which all samples were centrifuged at 2,000 g for 10 min. Samples were stored and transported at −80 °C to a central laboratory for iohexol and laboratory creatinine measurements.

Plasma concentrations of iohexol were determined using ultraperformance liquid chromatography with tandem mass spectrometry using a published method [10] in an ISO 15189 accredited laboratory which participates in the EQUALIS external quality assurance programme for iohexol (Uppsala, Sweden). The calibration curve and internal standard were determined with certified reference materials for iohexol (CRM: USP H0J211) and ioversol (CRM: USP 34510F), respectively. Internal quality control was prepared using the iohexol certified reference material and EQUALIS samples were included in every run as an additional quality check. The slope-intercept method using the 120, 180 and 240 time points (during the second exponential phase of iohexol elimination) was used to calculate mGFR by plotting iohexol concentration against time [11]. mGFR results were then corrected for body surface area [12].

Laboratory creatinine was determined in a central laboratory using an isotope dilution mass spectrometry (IDMS) traceable compensated kinetic Jaffe method (Cobas c500®, Roche Diagnostics, Mannheim). POC creatinine was measured at the time of collection in venous whole blood on the iSTAT® and in venous whole blood and capillary blood on the StatSensor®, both POC devices make use of an enzymatic amperometric method with an enzyme cascade reaction. Laboratory creatinine was monitored with internal and external quality control schemes, while POC creatinine was monitored with internal quality control. Inter-day coefficients of variation (CV)s were determined by assessing the variation in quality control materials over the period of the study.

Statistical analysis was performed using Tibco Statistica® (version 13.5.0.17, United States) and MedCalc® (version 19.2.1, Belgium). POC creatinine was compared to laboratory creatinine using Passing–Bablok regression and Bland–Altman agreement. eGFR was calculated according to the IDMS-traceable MDRD equation and the 2009 CKD-EPI equation. All subsequent eGFR results are presented without the African-American coefficient applied, based on prior studies [13], [14] demonstrating that inclusion of the coefficient overestimates GFR in sub-Saharan Africans. eGFR was compared to mGFR with Passing–Bablok regression, Spearman rank correlation coefficient (r) and Lin’s concordance correlation coefficient were determined. Bias was defined as the median of individual differences between eGFR and mGFR, while imprecision was defined as the interquartile range (IQR) of the differences between eGFR and mGFR. Outliers were excluded if their residuals lay outside of 2.5 standard deviations on initial Passing–Bablok regression [15], [16]. Accuracy within 30% of mGFR (P30) was calculated. The ability of the eGFR equations to correctly classify mGFR: <60, 60–89, and ≥90 mL/min/1.73 m2 (corresponding to KDIGO GFR categories for CKD of G3–5, G2 and G1, respectively) was calculated. Differences between the laboratory and POC devices were tested using the exact McNemar test. Results were considered statistically significant if p<0.05.

Repeat analysis with laboratory Jaffe creatinine adjusted by a factor of +9.0 μmol/L was performed (Supplementary Table S1). This adjustment was based on an inter-site calibration study performed within the larger ARK study comparing laboratory Jaffe creatinine methodology performed in South Africa and Malawi with an enzymatic laboratory creatinine method in Uganda.

Results

Study population

Figure 1 depicts the flow of participants through the study. All participants were Black African adults with a broad spectrum of kidney function. eGFR equations for laboratory and POC creatinine overestimated GFR when compared to mGFR, except in the ≥90 mL/min/1.73 m2 group where a mixed picture was seen (Table 1). Mean POC eGFRs showed less overestimation compared to mGFR than the mean laboratory eGFRs.

Figure 1: Participant flow through study. aOutliers excluded on initial Passing–Bablok comparison between mGFR and POC eGFR, excluded based on residuals >2.5 SD.
Figure 1:

Participant flow through study. aOutliers excluded on initial Passing–Bablok comparison between mGFR and POC eGFR, excluded based on residuals >2.5 SD.

Table 1:

Characteristics according to mGFR.

CharacteristicmGFR (categorised by GFR CKD stage)a
All<60 (G3–5)60–89 (G2)≥90 (G1)
n, %674134 (20)286 (42)254 (38)
Age, years43 (14)50 (14)46 (14)37 (11)
BMI, kg/m228.6 (6.4)28.9 (5.9)29.7 (6.6)27.2 (6.2)
Body surface area, m21.91 (0.23)1.90 (0.22)1.94 (0.24)1.87 (0.22)
Male, n, %211 (31)38 (28)81 (28)92 (36)
mGFRa81 (24)47 (10)76 (8)105 (13)
Laboratory eGFR (MDRD)a116 (27)98 (26)112 (22)129 (25)
Laboratory eGFR (CKD-EPI)a111 (17)98 (21)109 (14)120 (11)
StatSensor capillary eGFR (MDRD)a87 (26)72 (19)85 (24)97 (28)
StatSensor capillary eGFR (CKD-EPI)a93 (21)79 (20)91 (20)102 (19)
iSTAT eGFR (MDRD)a97 (23)83 (22)94 (20)108 (22)
iSTAT eGFR (CKD-EPI)a102 (19)89 (22)100 (17)112 (13)
StatSensor venous eGFR (MDRD)a85 (21)b71 (27)c,f84 (19)d93 (21)e
StatSensor venous eGFR (CKD-EPI)a92 (19)b77 (32)c,f90 (16)d101 (17)e
  1. All data mean (SD) unless otherwise indicated. aIn mL/min/1.73 m2; bn=214; cn=31; dn=97; en=86; fmedian (interquartile range). All eGFR equations exclude the African-American coefficient.

POC vs. laboratory creatinine

All three POC creatinine devices showed positive bias compared to laboratory creatinine with Bland–Altman agreements of 17.4 ± 15.3, 9.4 ± 6.3, and 15.1 ± 13.0 μmol/L for StatSensor® capillary, iSTAT®, and StatSensor® venous respectively. Passing–Bablok regression analysis yielded proportional errors (slope) of: 1.45 (95% CI: 1.33–1.57), 1.16 (95% CI: 1.13–1.20), and 1.17 (95% CI: 1.03–1.33); constant errors (y-intercept) of: −7.68 (95% CI: −14.43 to −1.33), 0.16 (95% CI: −2.20 to 2.13), and 6.50 (95% CI: −2.33 to 14.58); and Spearman rank correlation coefficients (r) of: 0.57 (95% CI: 0.52–0.62) (p<0.0001), 0.92 (95% CI: 0.91–0.93) (p<0.0001), and 0.65 (95% CI: 0.57–0.72) (p<0.0001) for StatSensor® capillary, iSTAT®, and StatSensor® venous respectively (Figure 2).

Figure 2: POC vs. laboratory creatinine. Bland–Altman agreement charts of: (A) the StatSensor® capillary creatinine (µmol/L), (C) iSTAT® creatinine (µmol/L), and (E) StatSensor® venous creatinine (µmol/L) compared to laboratory creatinine (µmol/L). Passing–Bablok regression comparing: (B) StatSensor® capillary creatinine, (D) iSTAT® creatinine, and (F) StatSensor® venous creatinine to laboratory creatinine.
Figure 2:

POC vs. laboratory creatinine. Bland–Altman agreement charts of: (A) the StatSensor® capillary creatinine (µmol/L), (C) iSTAT® creatinine (µmol/L), and (E) StatSensor® venous creatinine (µmol/L) compared to laboratory creatinine (µmol/L). Passing–Bablok regression comparing: (B) StatSensor® capillary creatinine, (D) iSTAT® creatinine, and (F) StatSensor® venous creatinine to laboratory creatinine.

eGFR (using POC creatinine and laboratory creatinine) vs. mGFR

Passing–Bablok regression demonstrated that all eGFR equations using POC and laboratory creatinine measurements suffered from proportional and constant error when compared to mGFR. The largest constant error was seen in laboratory creatinine eGFR using the CKD-EPI equation whilst the smallest constant error was seen in StatSensor® capillary creatinine eGFR using the MDRD equation (y-intercept: 62.76 vs. 3.91). StatSensor® capillary creatinine eGFR using the MDRD equation showed the least proportional error, while laboratory creatinine eGFR using the CKD-EPI equation showed the most proportional error (slope: 1.01 vs. 0.59). MDRD eGFR showed less proportional and constant error than the CKD-EPI eGFR for all POC and laboratory creatinine measurements. All POC creatinine eGFR measurements, except for iSTAT® MDRD, showed less proportional and constant error than the respective laboratory creatinine eGFR equations. Additionally, Lin’s concordance correlation coefficient showed improved accuracy for the POC eGFR measurements over the respective laboratory eGFR equations (Table 2, Figures 3 and 4).

Table 2:

Analytical comparison of POC and laboratory eGFR equations compared to mGFR.

Passing–Bablok regressionLin’s concordance correlation coefficientd
Slope/proportional erroray-intercept/constant errorbrcp-value
Laboratory eGFR (MDRD)1.16 (1.05–1.27)20.47 (10.94–29.23)0.44 (0.38–0.50)<0.00010.25 (0.21–0.29)
Laboratory eGFR (CKD-EPI)0.59 (0.53–0.65)62.76 (57.60–67.88)0.50 (0.45–0.56)<0.00010.24 (0.21–0.28)
StatSensor capillary eGFR (MDRD)1.01 (0.89–1.14)3.91 (−6.10 to 12.04)0.34 (0.27–0.41)<0.00010.35 (0.28–0.41)
StatSensor capillary eGFR (CKD-EPI)0.88 (0.79–0.98)22.38 (13.77–30.18)0.40 (0.33–0.46)<0.00010.37 (0.31–0.42)
iSTAT eGFR (MDRD)0.92 (0.84–1.01)21.35 (13.35–28.75)0.45 (0.39–0.51)<0.00010.40 (0.35–0.45)
iSTAT eGFR (CKD-EPI)0.70 (0.64–0.77)45.18 (38.90–50.84)0.49 (0.43–0.54)<0.00010.34 (0.29–0.38)
StatSensor venous eGFR (MDRD)0.90 (0.74–1.08)10.24 (−4.41 to 22.10)0.37 (0.25–0.48)<0.00010.42 (0.31–0.53)
StatSensor venous eGFR (CKD-EPI)0.84 (0.70–1.00)22.10 (8.33–33.73)0.45 (0.33–0.55)<0.00010.47 (0.37–0.56)
  1. aSlope (95% confidence interval); by-intercept (95% confidence interval); cSpearman rank correlation coefficient (95% confidence interval); dLin’s concordance correlation coefficient (95% confidence interval). All eGFR equations exclude the African-American coefficient.

Figure 3: MDRD eGFR (without the African-American coefficient) vs. iohexol mGFR. Bland–Altman agreement charts of: (A) Laboratory MDRD eGFR, (C) StatSensor® capillary MDRD eGFR, (E) iSTAT® MDRD eGFR, and (G) StatSensor® venous MDRD eGFR compared to iohexol mGFR. Passing–Bablok regression comparing: (B) Laboratory MDRD eGFR, (D) StatSensor® capillary MDRD eGFR, (F) iSTAT® MDRD eGFR, and (H) StatSensor® venous MDRD eGFR compared to iohexol mGFR.
Figure 3:

MDRD eGFR (without the African-American coefficient) vs. iohexol mGFR. Bland–Altman agreement charts of: (A) Laboratory MDRD eGFR, (C) StatSensor® capillary MDRD eGFR, (E) iSTAT® MDRD eGFR, and (G) StatSensor® venous MDRD eGFR compared to iohexol mGFR. Passing–Bablok regression comparing: (B) Laboratory MDRD eGFR, (D) StatSensor® capillary MDRD eGFR, (F) iSTAT® MDRD eGFR, and (H) StatSensor® venous MDRD eGFR compared to iohexol mGFR.

Figure 4: CKD-EPI eGFR (without the African-American coefficient) vs. iohexol mGFR. Bland–Altman agreement charts of: (A) Laboratory CKD-EPI eGFR, (C) StatSensor® capillary CKD-EPI eGFR, (E) iSTAT® CKD-EPI eGFR, and (G) StatSensor® venous CKD-EPI eGFR compared to iohexol mGFR. Passing–Bablok regression comparing: (B) Laboratory CKD-EPI eGFR, (D) StatSensor® capillary CKD-EPI eGFR, (F) iSTAT® CKD-EPI eGFR, and (H) StatSensor® venous CKD-EPI eGFR compared to iohexol mGFR.
Figure 4:

CKD-EPI eGFR (without the African-American coefficient) vs. iohexol mGFR. Bland–Altman agreement charts of: (A) Laboratory CKD-EPI eGFR, (C) StatSensor® capillary CKD-EPI eGFR, (E) iSTAT® CKD-EPI eGFR, and (G) StatSensor® venous CKD-EPI eGFR compared to iohexol mGFR. Passing–Bablok regression comparing: (B) Laboratory CKD-EPI eGFR, (D) StatSensor® capillary CKD-EPI eGFR, (F) iSTAT® CKD-EPI eGFR, and (H) StatSensor® venous CKD-EPI eGFR compared to iohexol mGFR.

Bias and precision

All calculated eGFR equations by POC creatinine or laboratory creatinine measurements showed positive bias compared to mGFR when looking at the all mGFR values category. Less bias was observed for the MDRD equation compared to the CKD-EPI equation for all POC creatinine measurements, except in the mGFR >90 mL/min/1.73 m2 group where a mixed picture was seen. The observed biases for POC MDRD eGFR ranged from 1.9 ± 33.5 to 14.7 ± 32.0 mL/min/1.73 m2, and for POC CKD-EPI eGFR from 8.4 ± 31.0 to 19.9 ± 28.5 mL/min/1.73 m2 when looking at the all mGFR values category. For each calculated eGFR equation, the largest bias was observed with laboratory creatinine with all POC devices yielding less bias than the laboratory (Table 3).

Table 3:

Accuracy within 30% of mGFR (P30), bias and imprecision according to CKD stage.

BiasaImprecisionbP30c
AllmGFR <60d (G3–5)mGFR 60–89d (G2)mGFR ≥90d (G1)AllmGFR <60d (G3–5)mGFR 60–89d (G2)mGFR ≥90d (G1)AllmGFR <60d (G3–5)mGFR 60–89d (G2)mGFR ≥90d (G1)
Laboratory eGFR (MDRD)34.1 (31.1–36.3)52.7 (45.8–59.7)35.7 (30.7–38.6)22.6 (18.3–27.7)35.4 (30.4–40.7)38.1 (28.7–51.1)29.2 (21.9–37.0)33.3 (26.0–41.9)36 (32–40)10 (5–16)27 (22–33)60 (53–66)
Laboratory eGFR (CKD-EPI)28.6 (27.0–30.6)54.5 (51.2–58.8)33.8 (31.7–35.3)16.1 (14.2–17.3)26.8 (23.5–30.7)27.3 (18.1–38.6)19.4 (14.5–23.3)17.9 (13.7–23.5)43 (40–47)7 (3–12)25 (20–31)83 (78–88)
StatSensor capillary eGFR (MDRD)4.6 (1.3–7.7)28.0 (23.2–31.7)7.3 (3.7–10.4)−12.4 (−16.9 to −9.2)39.8 (34.2–46.4)26.5 (17.5–35.6)30.4 (21.6–38.6)38.8 (25.6–47.6)61 (57–65)29 (22–38)68 (62–73)70 (64–76)
StatSensor capillary eGFR (CKD-EPI)11.4 (10.0–14.3)35.3 (31.2–38.4)16.5 (12.8–19.5)−1.3 (−5.4 to 2.4)35.0 (29.8–41.8)29.3 (17.8–40.2)29.4 (22.0–36.9)28.7 (22.8–37.1)62 (58–65)21 (14–29)60 (54–66)85 (80–89)
iSTAT eGFR (MDRD)14.7 (12.1–16.5)39.1 (33.0–42.0)16.0 (13.8–18.6)3.1 (−1.1 to 5.4)32.0 (26.4–37.7)27.8 (21.4–40.1)23.0 (18.3–31.1)29.5 (21.7–36.5)62 (58–66)18 (12–26)61 (55–66)87 (82–90)
iSTAT eGFR (CKD-EPI)19.9 (17.7–21.4)47.6 (42.2–51.0)25.8 (21.3–27.8)8.5 (5.3–10.5)28.5 (23.9–33.9)28.9 (22.7–38.6)21.4 (16.8–27.3)20.1 (14.5–26.4)56 (52–60)14 (8–21)44 (38–50)92 (88–95)
StatSensor venous eGFR (MDRD)1.9 (−2.1 to 4.6)20.8 (11.0–31.6)6.4 (2.2–11.0)−14.0 (−22.8 to −8.6)33.5 (22.5–43.3)28.6 (10.0–42.5)27.9 (13.4–37.8)30.4 (18.1–41.0)71 (64–77)42 (25–61)74 (64–83)78 (68–86)
StatSensor venous eGFR (CKD-EPI)8.4 (5.1–12.4)24.5 (13.2–36.3)14.1 (10.9–18.0)−3.4 (−9.1 to 2.5)31.0 (21.0–37.8)33.0 (11.8–50.2)22.8 (17.0–35.1)28.0 (16.3–40.4)72 (65–78)35 (19–55)66 (56–75)92 (84–97)
  1. aMedian bias of individual differences between eGFR and mGFR (in mL/min/1.73 m2 with 95% confidence intervals); binterquartile range (IQR) of the differences between eGFR and mGFR (in mL/min/1.73 m2 with 95% confidence intervals); c% (95% confidence intervals); din mL/min/1.73 m2. All eGFR equations exclude the African-American coefficient.

Inter-day CVs varied for the different creatinine measuring devices, the iSTAT® had a CV of 5.9% at a level of 88 μmol/L, the two StatSensor devices had CVs of 7.4 and 10.2% at 84 μmol/L and CVs of 4.6 and 5.5% at 530 μmol/L, while the laboratory Jaffe assay had a CV of 3.5% at both levels of 89.3 and 342 μmol/L. When looking at IQRs, the MDRD equation showed less imprecision at a mGFR <60 mL/min/1.73 m2 for all POC measurements, at mGFR ≥60 mL/min/1.73 m2 the CKD-EPI equation showed the least imprecision. However, for laboratory eGFR the CKD-EPI equation showed less imprecision across all mGFR categories. When comparing POC eGFR imprecision to that of laboratory eGFR a mixed picture was seen, all methods suffered from imprecision; when the CKD-EPI equation was utilized the laboratory showed the least imprecision, however when the MDRD equation was utilized POC (generally the iSTAT device) showed the least imprecision. Capillary sampling with the StatSensor device showed the highest levels of imprecision except in the mGFR <60 mL/min/1.73 m2 category (Table 3).

Accuracy

Overall, the P30 for the POC and laboratory creatinine eGFR equations ranged from 36 to 72%. All P30s showed improved accuracy with increasing mGFR. Below 90 mL/min/1.73 m2 the MDRD eGFR showed greater accuracy (higher P30s) than the CKD-EPI eGFR in all POC and laboratory creatinine measurements, above 90 mL/min/1.73 m2 the CKD-EPI eGFR showed greater accuracy. In the different mGFR groups, POC P30s continued to show improved accuracy compared to laboratory P30s (Table 3). When serum creatinine was adjusted by a factor of +9.0 μmol/L improved P30s ranging from 62 to 71% were seen. Additional analysis using the revised Lund–Malmö equation yielded P30s ranging from 58 to 80% with improvement noted for POC devices and laboratory results (Supplementary Table S1).

The percentage of total samples correctly classified into the eGFR groups of <60, 60–89 and >90 mL/min/1.73 m2 ranged from 41.7% (95% CI: 38.0–45.6) to 52.8% (95% CI: 45.9–59.7). The POC creatinine eGFR equations correctly classified more samples into the correct eGFR groups than the respective laboratory creatinine equations, although the laboratory equations showed improved performance in the mGFR >90 mL/min/1.73 m2 group. Below 90 mL/min/1.73 m2 the MDRD equation correctly classified more samples than the CKD-EPI equation, above 90 mL/min/1.73 m2 the CKD-EPI equation correctly classified more samples (Table 4).

Table 4:

Percentage of samples correctly classified into mGFR groups <60, 60–89 and ≥90.

mGFR <60c (G3–5)mGFR 60–89c (G2)mGFR ≥90c (G1)Total correctly classifiedd
Laboratory eGFR (MDRD)8.2 (4.2–14.2)14.7 (10.8–19.4)95.6 (92.3–97.8)43.8 (40.0–47.7)
Laboratory eGFR (CKD-EPI)7.5 (3.6–13.3)7.0 (4.3–10.6)99.2 (97.2–99.9)41.7 (38.0–45.6)
StatSensor capillary eGFR (MDRD)24.6 (17.6–32.8)48.4 (42.5–54.4)52.0 (45.6–58.3)45.0 (41.2–48.9)
pa<0.0001<0.0001<0.00010.5451
StatSensor capillary eGFR (CKD-EPI)19.4 (13.1–27.1)38.2 (32.6–44.2)73.2 (67.3–78.6)47.5 (43.7–51.4)
pb0.0004<0.0001<0.00010.0028
iSTAT eGFR (MDRD)15.0 (9.4–22.2)39.6 (33.9–45.6)77.4 (71.7–82.4)49.0 (45.1–52.8)
pa0.0117<0.0001<0.00010.0048
iSTAT eGFR (CKD-EPI)12.0 (7.0–18.8)23.9 (19.0–29.2)93.7 (89.9–96.3)47.8 (43.9–51.6)
pb0.0703<0.00010.0002<0.0001
StatSensor venous eGFR (MDRD)35.5 (19.2–54.6)57.7 (47.3–67.7)53.5 (42.4–64.3)52.8 (45.9–59.7)
pa0.1250<0.0001<0.00010.4215
StatSensor venous eGFR (CKD-EPI)29.0 (14.2–48.0)45.4 (35.2–55.8)69.8 (58.9–79.2)52.8 (45.9–59.7)
pb0.3750<0.0001<0.00010.1249
  1. All data percentage (95% confidence interval). avs. laboratory eGFR (MDRD) using exact NcNemar test; bvs. laboratory eGFR (CKD-EPI) using exact NcNemar test; cin mL/min/1.73 m2; dtotal samples correctly classified into CKD stages: G3–5, G2 and G1. All eGFR equations exclude the African-American coefficient.

Discussion

In this study we used iohexol mGFR to evaluate the clinical utility of the iSTAT® and StatSensor® POC devices that measure creatinine and subsequently estimate kidney function using eGFR. Additionally, we evaluated the performance of the MDRD and CKD-EPI equations and compared the performance of the POC devices to that of laboratory Jaffe creatinine. Despite falling well short of the 90% of eGFRs within 30% of mGFR goal [17], POC eGFR showed improved accuracy compared to laboratory Jaffe eGFR when screening for CKD in a resource-limited setting. All current eGFR estimating equations performed poorly in our study sample.

A number of studies [18], [19], [20], [21] have shown the utility of the iSTAT® and StatSensor® POC devices for the diagnosis of acute kidney injury. Our study showed more bias compared to laboratory creatinine than these studies possibly due to differences in laboratory creatinine methods. Studies with less bias [18], [19], [20], [21] used the enzymatic method while studies which used the IDMS-traceable Jaffe method, including our own, showed more bias [22]. We found the iSTAT® device showed less imprecision than the StatSensor® device, consistent with previous studies [18], [19].

Comparison of POC eGFR to iohexol mGFR showed greater bias and inaccuracy than a previous study [23] from Belgium, which found a bias of 0.1 ± 17.6 mL/min/1.73 m2 with a P30 of 81% for StatSensor® capillary eGFR (CKD-EPI). Our P30 values for POC devices were low, and varied across the range of GFR, with the worst performance at low GFR. Bukabau et al. [13] reported a similarly low P30 of 31.3% in an African population with mGFR <60 mL/min/1.73 m2 for the MDRD and CKD-EPI equations, with a laboratory enzymatic creatinine method.

It has been shown that serum creatinine suffers from significant positive bias when there is a delay in sample separation of 24 h, this is true for Jaffe but not enzymatic methods [6], [24]. In our study POC creatinine measurements were performed in real-time, while laboratory creatinine measurements were performed on samples with a maximum delayed separation of between 8 and 9 h utilising an IDMS-traceable Jaffe method. This delay is not expected to have significantly influenced results, although if present a positive bias on laboratory creatinine with lowered eGFR would be expected (the opposite pattern from what was observed in our laboratory samples). Importantly in LMIC, these methodological and pre-analytical factors are common in clinical practice. For example, in South Africa, samples from rural areas are transported to central laboratories and only processed 12–72 h post collection [6] and the Jaffe method is used in all but a few large urban centres due to cost considerations [25] – despite an initiative for all laboratories to use the enzymatic method [26]. The IDMS-traceable Jaffe method has been shown not to reach the desirable specifications of the Laboratory Working Group of the National Kidney Disease Education Program [27] – nevertheless, it is commonly used with 93% of studies from sub-Saharan Africa which reported creatinine methodology making use of the Jaffe method [28].

In LMIC the potential public health benefits associated with POC screening for early detection and monitoring of CKD is an exciting possibility. Despite the bias and imprecision seen in POC eGFRs, these devices showed improved performance compared to laboratory Jaffe eGFR. The performance of POC devices to detect eGFR in the range 60–89 mL/min/1.73 m2 is of particular interest. With limited access to renal replacement therapy for severe kidney disease, it is advantageous to detect individuals with early disease who may benefit from renal protective measures. There was improved accuracy in this area compared to laboratory Jaffe measurements, as well as compared to measurements below 60 mL/min/1.73 m2. Nevertheless, the large bias, imprecision and misclassifications mean that results remain inaccurate with current eGFR estimating equations.

Enzymatic methods are more specific for creatinine than Jaffe methods which are vulnerable to interference from numerous substances including proteins, glucose and ketone bodies [29]. Various assay techniques such as kinetic reactions and compensated assays are used to improve specificity. The laboratory assay used in our study was an IDMS-traceable compensated kinetic Jaffe assay. Compensated assays assume a fixed concentration of non-specific interferences which are subtracted from the total creatinine value, bias will be introduced if this assumption is incorrect [30], [31]. The positive bias for creatinine which the enzymatic POC devices demonstrated compared to the laboratory Jaffe assay may reflect overcompensation of the Jaffe assay in our population. When laboratory creatinine results were adjusted by a factor of +9.0 μmol/L resulting eGFR equations showed similar P30s to that obtained for POC (Supplementary Table S1). Ideally all laboratories should move to enzymatic methods for the measurement of creatinine, while compensated Jaffe equations should be used with caution especially if this compensation has not been validated in a particular population. The poor performance of the laboratory Jaffe assay in our study may reflect overcompensation of the Jaffe method in our population.

Our study focussed primarily on accuracy; however, imprecision makes an equally important contribution to analytical variation. Jaffe methods are known with higher imprecision than enzymatic methods [27 32], [33]. The inter-day CVs of the POC creatinine devices were greater than that seen with the laboratory Jaffe assay. This high imprecision with POC assays, despite the use of enzymatic methodology, is an important distinguishing factor between laboratory and POC testing [34]. High levels of imprecision may result in total error that hinders the interpretation of creatinine and subsequent eGFR [32].

The ability of the CKD-EPI equation to quantify GFR above 60 mL/min/1.73 m2 offers significant advantages over the MDRD equation and avoids classifying patients based on a GFR cut-off which dichotomises all measurements [5], [35]. In keeping, we found the MDRD equation to suffer from greater imprecision above a mGFR of 60 mL/min/1.73 m2. Despite these apparent short-comings, we were surprised by the improved accuracy of the MDRD equation with higher P30s and sensitivities up to a mGFR of 90 mL/min/1.73 m2 compared to the CKD-EPI equation.

In keeping with other studies from sub-Saharan Africa [13], [14] we found a positive bias when using the African-American coefficient for the MDRD and CKD-EPI equations, and both led to substantial overestimation of GFR (Supplementary Figures S1 and S2). The revised Lund–Malmö equation outperformed the MDRD and CKD-EPI equations (Supplementary Table S1), suggesting that a new equation developed from an African population group may further improve performance. Despite the improved performance seen with the revised Lund–Malmö equation it remains rarely used in sub-Saharan Africa.

POC devices have the potential to provide immediately actionable results and mitigate pre-analytical errors, such as delay in separation commonly seen in LMIC. The iSTAT® utilises venous whole blood while the StatSensor® can be used with capillary and venous whole blood. Capillary sampling offers added benefit by eliminating the need for phlebotomy skills, and is preferred by patients [36]. Capillary sampling showed improved performance over corresponding laboratory creatinine, however it showed the highest levels of imprecision. Thus, venous sampling should remain the preferred option, with capillary sampling considered as an alternative only if venous sampling is unavailable and if the device has been validated for this purpose.

POC testing has limitations such as: refrigerated supply storage (both devices), susceptibility to lot-to-lot variations and a lack of monitoring by laboratory professionals (especially external quality control). Cost considerations need to be taken into account especially in resource-limited settings. POC testing has a significantly higher cost compared to laboratory measurements, especially when additional components such as the purchase of quality control materials and the need for refrigeration are considered. However, cost savings may be achieved if POC testing leads to improved detection, and reduced numbers of patients progressing to advanced CKD. Both devices, and others, have previously been reviewed regarding suitability in low-income countries, including estimated cost per test [37].

Our study had a number of strengths: (i) the large sample size drawn from a well characterised population-based study of rural South Africans; (ii) the study procedures were standardised and conducted in a well-controlled research environment; (iii) two different POC devices using venous whole blood and capillary blood were evaluated; (iv) participants had a wide range of iohexol mGFRs including 134 participants with a mGFR less than 60 mL/min/1.73 m2. Our study also had limitations: (i) the poor performance of laboratory creatinine likely reflects the Jaffe methodology used despite enzymatic methods being the preferred choice [27], [33], [38], [39], [40], however without a parallel analysis using a laboratory enzymatic method this assumption remains untested; (ii) although plasma clearance of iohexol is a well described method for measuring GFR, the MDRD and CKD-EPI equations were developed using urinary clearance of iothalamate and as such, methodologic bias may be introduced by using iohexol mGFR [4]; (iii) laboratory creatinine was processed after delayed sample separation (although less than the 24 h level at which changes become significant [6], [24]); (iv) POC testing was carried out by trained staff under controlled conditions, which may not be reproducible in a busy clinic environment; (v) roughly one-third of participants had venous measurements assessed with the StatSensor® due to cost considerations.

In conclusion, our study demonstrates poor performance of the compensated kinetic Jaffe method to measure creatinine in an under-resourced primary care setting and highlights the recommendation for laboratories to move to enzymatic methods. In these situations POC devices, despite their poor performance compared to iohexol mGFR, show improved accuracy in predicting GFR compared to Jaffe methodology. The inaccuracy and wide imprecision with current eGFR estimating equations mean that POC devices should be used with caution.


Corresponding author: Dr. Sean Currin, Department of Chemical Pathology, Faculty of Health Sciences, University of Witwatersrand, 7 York Road, Parktown, Johannesburg, 2193, South Africa; and Department of Chemical Pathology, National Health Laboratory Service, Johannesburg, South Africa, Phone: +2711 489 8499, E-mail:

Funding source: The International Society of Nephrology Clinical Research Program

Funding source: The South African Medical Research Council, with funds from the South African National Department of Health, MRC UK (via the Newton Fund) and GSK R&D

Award Identifier / Grant number: #074NEWTON NCD

Funding source: Faculty Research Committee Individual Research Grant, University of Witwatersrand

Award Identifier / Grant number: 0012548463101512110500000000000000005254

Acknowledgments

We thank Abbott Laboratories for the supply of the iSTAT® device, control materials and cartridges. We thank Nova Biomedical Corporation for the supply of the StatSensor® devices, control materials and biosensor strips.

  1. Research funding: We thank our funders: (i) the South African Medical Research Council, with funds from the South African National Department of Health, MRC UK (via the Newton Fund) and GSK R&D; (ii) Faculty Research Committee Individual Research Grant, University of Witwatersrand; (iii) The International Society of Nephrology Clinical Research Program.

  2. Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.

  3. Competing interests: Authors state no conflict of interest.

  4. Informed consent: Informed consent was obtained from all individuals included in this study.

  5. Ethical approval: Research involving human subjects complied with all relevant national regulations, institutional policies and is in accordance with the tenets of the Helsinki Declaration (as revised in 2013), and has been approved by the authors’ Institutional Review Board (Human Research Ethics Committee of the University of the Witwatersrand) or equivalent committee. (M160937 and M190434).

References

1. George, JA, Brandenburg, J-T, Fabian, J, Crowther, NJ, Agongo, G, Alberts, M, et al.. Kidney damage and associated risk factors in rural and urban sub-Saharan Africa (AWI-Gen): a cross-sectional population study. Lancet Glob Health 2019;7:e1632–43. https://doi.org/10.1016/s2214-109x(19)30443-7.Search in Google Scholar

2. Stanifer, JW, Muiru, A, Jafar, TH, Patel, UD. Chronic kidney disease in low- and middle-income countries. Nephrol Dial Transplant 2016;31:868–74. https://doi.org/10.1093/ndt/gfv466.Search in Google Scholar PubMed PubMed Central

3. KDIGO Working Group. Kidney disease: improving global outcomes (KDIGO). KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney Int Suppl 2013;3:1–150.Search in Google Scholar

4. Levey, AS, Inker, LA. GFR as the “gold standard”: estimated, measured, and true. Am J Kidney Dis 2016;67:9–12. https://doi.org/10.1053/j.ajkd.2015.09.014.Search in Google Scholar PubMed

5. Miller, WG, Jones, GRD. Estimated glomerular filtration rate; laboratory implementation and current global status. Adv Chron Kidney Dis 2018;25:7–13. https://doi.org/10.1053/j.ackd.2017.09.013.Search in Google Scholar PubMed

6. Chikomba, CE, Padoa, CJ, Tanyanyiwa, D. Evaluation of the impact of delayed centrifugation on the diagnostic performance of serum creatinine as a baseline measure of renal function before antiretroviral treatment. South Afr J HIV Med 2020;21:a1056. https://doi.org/10.4102/sajhivmed.v21i1.1056.Search in Google Scholar PubMed PubMed Central

7. Kazmierczak, SC. Point-of-care testing quality: some positives but also some negatives. Clin Chem 2011;57:1219–20. https://doi.org/10.1373/clinchem.2011.171538.Search in Google Scholar PubMed

8. Levin, A, Stevens, PE. Summary of KDIGO 2012 CKD Guideline: behind the scenes, need for guidance, and a framework for moving forward. Kidney Int 2014;85:49–61. https://doi.org/10.1038/ki.2013.444.Search in Google Scholar PubMed

9. Kalyesubula, R, Fabian, J, Nakanga, W, Newton, R, Ssebunnya, B, Prynn, J, et al.. How to estimate glomerular filtration rate in sub-Saharan Africa: design and methods of the African research into kidney diseases (ARK) study. BMC Nephrol 2020;21:1–12. https://doi.org/10.1186/s12882-020-1688-0.Search in Google Scholar PubMed PubMed Central

10. Annesley, TM, Clayton, LT. Ultraperformance liquid chromatography-tandem mass spectrometry assay for iohexol in human serum. Clin Chem 2009;55:1196–202. https://doi.org/10.1373/clinchem.2008.121533.Search in Google Scholar PubMed

11. Fleming, JS, Zivanovic, MA, Blake, GM, Burniston, M, Cosgriff, PS. Guidelines for the measurement of glomerular filtration rate using plasma sampling. Nucl Med Commun 2004;25:759–69. https://doi.org/10.1097/01.mnm.0000136715.71820.4a.Search in Google Scholar PubMed

12. Haycock, GB, Schwartz, GJ, Wisotsky, DH. Geometric method for measuring body surface area: a height–weight formula validated in infants, children, and adults. J Pediatr 1978;93:62–6. https://doi.org/10.1016/s0022-3476(78)80601-5.Search in Google Scholar

13. Bukabau, JB, Yayo, E, Gnionsahé, A, Monnet, D, Pottel, H, Cavalier, E, et al.. Performance of creatinine- or cystatin C-based equations to estimate glomerular filtration rate in sub-Saharan African populations. Kidney Int 2019;95:1181–9. https://doi.org/10.1016/j.kint.2018.11.045.Search in Google Scholar PubMed

14. Van Deventer, HE, George, JA, Paiker, JE, Becker, PJ, Katz, IJ. Estimating glomerular filtration rate in black South Africans by use of the modification of diet in renal disease and Cockcroft–Gault equations. Clin Chem 2008;54:1197–202. https://doi.org/10.1373/clinchem.2007.099085.Search in Google Scholar PubMed

15. Parrinello, CM, Grams, ME, Couper, D, Ballantyne, CM, Hoogeveen, RC, Eckfeldt, JH, et al.. Recalibration of blood analytes over 25 years in the atherosclerosis risk in communities study: impact of recalibration on chronic kidney disease prevalence and incidence. Clin Chem 2015;61:938–47. https://doi.org/10.1373/clinchem.2015.238873.Search in Google Scholar PubMed PubMed Central

16. Ricos, C, Ramon, F, Salas, A, Buno, A, Calafell, R, Morancho, J, et al.. Minimum analytical quality specifications of inter-laboratory comparisons: agreement among Spanish EQAP organizers. Clin Chem Lab Med 2012;50:455–61. https://doi.org/10.1515/cclm.2011.787.Search in Google Scholar

17. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:S1-266.Search in Google Scholar

18. Snaith, B, Harris, MA, Shinkins, B, Jordaan, M, Messenger, M, Lewington, A. Point-of-care creatinine testing for kidney function measurement prior to contrast-enhanced diagnostic imaging: evaluation of the performance of three systems for clinical utility. Clin Chem Lab Med 2018;56:1269–76. https://doi.org/10.1515/cclm-2018-0128.Search in Google Scholar PubMed

19. van der Heijden, C, Roosens, L, Cluckers, H, Van Craenenbroeck, AH, Peeters, B. Analytical and clinical performance of three hand-held point-of-care creatinine analyzers for renal function measurements prior to contrast-enhanced imaging. Clin Chim Acta 2019:13–9. https://doi.org/10.1016/j.cca.2019.06.025.Search in Google Scholar PubMed

20. Kosack, CS, De Kieviet, W, Bayrak, K, Milovic, A, Page, AL. Evaluation of the Nova StatSensor® Xpress™ creatinine point-of-care handheld analyzer. PloS One 2015;10:1–11. https://doi.org/10.1371/journal.pone.0122433.Search in Google Scholar PubMed PubMed Central

21. Snaith, B, Harris, MA, Shinkins, B, Messenger, M, Lewington, A, Jordaan, M, et al.. Point of care creatinine testing in diagnostic imaging: a feasibility study within the outpatient computed tomography setting. Eur J Radiol 2019:82–7. https://doi.org/10.1016/j.ejrad.2019.01.007.Search in Google Scholar PubMed

22. Griffin, BR, Butler-Dawson, J, Dally, M, Krisher, L, Cruz, A, Weitzenkamp, D, et al.. Unadjusted point of care creatinine results overestimate acute kidney injury incidence during field testing in Guatemala. PloS One 2018;13:1–12. https://doi.org/10.1371/journal.pone.0204614.Search in Google Scholar PubMed PubMed Central

23. Stojkovic, V, Delanaye, P, Collard, G, Ferrante, N, Le Goff, C, Lutteri, L, et al.. Estimated glomerular filtration rate using a point of care measure of creatinine in patients with iohexol determinate GFR. Clin Chim Acta 2019;499:123–7. https://doi.org/10.1016/j.cca.2019.08.032.Search in Google Scholar PubMed

24. Ford, L, Berg, J. Delay in separating blood samples affects creatinine measurement using the Roche kinetic Jaffe method. Ann Clin Biochem 2008;45:83–7. https://doi.org/10.1258/acb.2007.007109.Search in Google Scholar PubMed

25. Delanghe, JR, Speeckaert, MM. Creatinine determination according to Jaffe – what does it stand for? NDT Plus 2011;4:83–6. https://doi.org/10.1093/ndtplus/sfq211.Search in Google Scholar PubMed PubMed Central

26. National Institute for Health and Care Excellence. Chronic kidney disease in adults: assessment and management. NICE Guideline CG182; 2014.Search in Google Scholar

27. Boutten, A, Bargnoux, AS, Carlier, MC, Delanaye, P, Rozet, E, Delatour, V, et al.. Enzymatic but not compensated Jaffe methods reach the desirable specifications of NKDEP at normal levels of creatinine. Results of the French multicentric evaluation. Clin Chim Acta 2013;18:132–5. https://doi.org/10.1016/j.cca.2013.01.021.Search in Google Scholar PubMed

28. Fabian, J, George, JA, Etheredge, HR, van Deventer, M, Kalyesubula, R, Wade, AN, et al.. Methods and reporting of kidney function: a systematic review of studies from sub-Saharan Africa. Clin Kidney J 2019;12:778–87. https://doi.org/10.1093/ckj/sfz089.Search in Google Scholar PubMed PubMed Central

29. Panteghini, M. Enzymatic assays for creatinine: time for action. Clin Chem Lab Med 2008;46:567–72. https://doi.org/10.1515/CCLM.2008.113.Search in Google Scholar PubMed

30. Parry, DM Use of single-value protein compensation of the Jaffe creatinine assay contributes to clinically significant inaccuracy in results. Clin Chem 2008;54:215–6. https://doi.org/10.1373/clinchem.2007.091652.Search in Google Scholar PubMed

31. Cobbaert, CM, Baadenhuijsen, H, Weykamp, CW. Prime time for enzymatic creatinine methods in pediatrics. Clin Chem 2009;55:549–58. https://doi.org/10.1373/clinchem.2008.116863.Search in Google Scholar PubMed

32. Delanaye, P, Cavalier, E, Pottel, H. Serum creatinine: not so simple! Nephron 2017;136:302–8. https://doi.org/10.1159/000469669.Search in Google Scholar PubMed

33. Hoste, L, Deiteren, K, Pottel, H, Callewaert, N, Martens, F. Routine serum creatinine measurements: how well do we perform? BMC Nephrol 2015;16:1–9. https://doi.org/10.1186/s12882-015-0012-x.Search in Google Scholar PubMed PubMed Central

34. van Lint, CL, van der Boog, PJM, Romijn, FPHTM, Schenk, PW, van Dijk, S, Rövekamp, TJM, et al.. Application of a point of care creatinine device for trend monitoring in kidney transplant patients: fit for purpose? Clin Chem Lab Med 2015;53:1547–56. https://doi.org/10.1515/cclm-2014-0932.Search in Google Scholar PubMed

35. Dalton, RN. Serum creatinine and glomerular filtration rate: perception and reality. Clin Chem 2010;56:687–9. https://doi.org/10.1373/clinchem.2010.144261.Search in Google Scholar PubMed

36. Woods, K, Douketis, JD, Schnurr, T, Kinnon, K, Powers, P, Crowther, MA. Patient preferences for capillary vs. venous INR determination in an anticoagulation clinic: a randomized controlled trial. Thromb Res 2004;114:161–5. https://doi.org/10.1016/j.thromres.2004.05.013.Search in Google Scholar PubMed

37. Raimann, JG, Riella, MC, Levin, NW. International Society of Nephrology’s 0by25 initiative (zero preventable deaths from acute kidney injury by 2025): focus on diagnosis of acute kidney injury in low-income countries. Clin Kidney J 2018;11:12–9. https://doi.org/10.1093/ckj/sfw134.Search in Google Scholar PubMed PubMed Central

38. Piéroni, L, Delanaye, P, Boutten, A, Bargnoux, AS, Rozet, E, Delatour, V, et al.. A multicentric evaluation of IDMS-traceable creatinine enzymatic assays. Clin Chim Acta 2011;412:2070–5. https://doi.org/10.1016/j.cca.2011.07.012.Search in Google Scholar PubMed

39. Drion, I, Cobbaert, C, Groenier, KH, Weykamp, C, Bilo, HJ, Wetzels, JF, et al.. Clinical evaluation of analytical variations in serum creatinine measurements: why laboratories should abandon Jaffe techniques. BMC Nephrol 2012;8:133.10.1186/1471-2369-13-133Search in Google Scholar PubMed PubMed Central

40. Miller, WG, Bachmann, LM, Delanghe, JR, Inker, LA, Jones, GRD, Vassalotti, JA. Optimal use of biomarkers for chronic kidney disease. Clin Chem 2019;65:949–55. https://doi.org/10.1373/clinchem.2018.299073.Search in Google Scholar PubMed


Supplementary Material

The online version of this article offers supplementary material (https://doi.org/10.1515/cclm-2020-1882).


Received: 2020-12-29
Accepted: 2021-03-03
Published Online: 2021-03-15
Published in Print: 2021-07-27

© 2021 Sean Currin et al., published by De Gruyter, Berlin/Boston

This work is licensed under the Creative Commons Attribution 4.0 International License.

Downloaded on 28.2.2024 from https://www.degruyter.com/document/doi/10.1515/cclm-2020-1882/html
Scroll to top button