Evaluation of a point‐of‐care serum creatinine measurement device and the impact on diagnosis of acute kidney injury in pediatric cardiac patients: A retrospective, single center study

Abstract Background and aims Agreement between measurements of creatinine concentrations using point‐of‐care (POC) devices and measurements conducted in a standard central laboratory is unclear for pediatric patients. Our objectives were (a) to assess the agreement for pediatric patients and (b) to compare the incidence of postoperative acute kidney injury (AKI) according to the two methods. Methods This retrospective, single‐center study included patients under 18 years of age who underwent cardiac surgery and who were admitted into the pediatric intensive care unit of a tertiary teaching hospital (Okayama University Hospital, Japan) from 2013 to 2017. The primary objective was to assess the correlation and the agreement between measurements of creatinine concentrations by a Radiometer blood gas analyzer (Cregas) and those conducted in a central laboratory (Crelab). The secondary objective was to compare the incidence of postoperative AKI between the two methods based on Kidney Disease Improving Global Outcomes (KDIGO) criteria. Results We analyzed the results of 1404 paired creatinine measurements from 498 patients, whose median age was 14 months old (interquartile range [IQR] 3, 49). The Pearson correlation coefficient of Cregas vs Crelab was 0.968 (95% confidence interval [CI], 0.965‐0.972, P < 0.001). The median bias between Cregas and Crelab was 0.02 (IQR ‐0.02, 0.05) mg/dL. While 199 patients (40.0%) were diagnosed as having postoperative AKI based on Crelab, 357 patients (71.7%) were diagnosed as having postoperative AKI based on Cregas (Kappa = 0.39, 95% CI, 0.33‐0.46). In a subgroup analysis of patients whose Cregas and Crelab were measured within 1 hour, similar percentage of patients were diagnosed as having postoperative AKI based on Cregas and Crelab (42.8% vs 46.0%; Kappa = 0.76, 95% CI, 0.68‐0.84). Conclusion There was an excellent correlation between Cregas and Crelab in pediatric patients. Although more patients were diagnosed as having postoperative AKI based on Cregas than based on Crelab, paired measurements with a short time gap showed good agreement on AKI diagnosis.


| INTRODUCTION
Point-of-care (POC) creatinine devices are capable of providing a quantitative estimation of a patient's serum creatinine concentration. 1 POC devices, including blood gas analyzers 2,3 , deliver quick results compared with a standard central laboratory, saving time and providing easy access to information to clinicians, especially in intensive care unit (ICU) settings. Some of the blood gas analyzers currently on the market can provide information on not only creatinine concentration but also biochemical, hematological, and acid-base parameters. 2,3 The creatinine sensor used in a Radiometer blood gas analyzer (ABL 800, Radiometer Co, Copenhagen, Denmark) follows the international recommendations by the National Kidney Disease Education Program (NKDEP) Laboratory Working Group 4 and has been shown to be a whole blood creatinine measurement method with calibration traceability to an isotope dilution mass spectometry (IDMS) standard. 3 Plasma creatinine concentration is proportional to the muscle volume, and the values at different time points soon after birth have different reference intervals compared with those in adults. 5 Furthermore, the required volume of a blood sample for a blood gas analyzer is only of 0.2 to 0.3 mL for neonates, which might cause errors in measurements. While some studies have shown good agreement between creatinine measurements using a POC device and measurements conducted in a central laboratory, [1][2][3] there have been no studies in which the agreement between measurements of serum creatinine concentrations by a POC analyzer and those conducted in a standard laboratory have been assessed for teenage or younger patients.
The rapid simultaneous measurements of creatinine, electrolytes, and acid-base parameters using a blood gas analyzer enable quick treatment and could change clinical practice, especially in a critically ill setting. We hypothesized that there had been more frequent creatinine measurements by a POC device in our pediatric intensive care unit (PICU) for postoperative patients than measurements in the central laboratory due to the easy access to the machine and the large range of information that can be obtained by a POC blood gas analyzer. Since worldwide guidelines such as Kidney Disease Improving Global Outcomes (KDIGO) consensus criteria 6 do not mention appropriate frequency of creatinine measurements for diagnosis of postoperative acute kidney injury (AKI), 6 numerous studies assessing postoperative AKI have adopted their own frequency of creatinine measurements within certain period of time after surgery. [7][8][9][10] If there are indeed more measurements by a POC device, these might provide additional opportunities to detect pathological changes in serum creatinine concentration, potentially improving the detection of AKI. In this sense, pediatric cardiac patients would be a good population in which to test this, because of the reported high incidence of postoperative AKI, especially after the use of cardiopulmonary bypass (CPB). 11 Our objectives were (a) to assess the correlation and the agreement between measurements of serum creatinine concentrations by a blood gas machine and those conducted in a central laboratory for pediatric patients and (b) to compare the incidence of postoperative AKI after cardiac surgery according to serum creatinine measurements by the POC device to that based on serum creatinine measurements in the central laboratory.

| Design
We conducted a single-center, retrospective study that was approved by the Okayama University Hospital Ethics Committee. The committee waived the need for informed consent for studies involving the use of the hospital's database. All regulations and measures of ethics and confidentiality were handled in accordance with the Declaration of Helsinki.

| Study population
This was a retrospective study of patients who underwent cardiac surgery with CPB in a tertiary teaching hospital (Okayama University Hospital, Japan) and were admitted to the PICU in the hospital during the period from December 2013 to January 2017. We included patients who were 18 years of age or younger and who were admitted to the PICU for the first time. Patients without data for initial and/or postoperative serum creatinine concentrations for the first 48 hours were excluded.

| POC creatinine testing and laboratory measurements
Blood samples were collected in standard prepared, heparinized blood gas syringes, and measurements were conducted without delay using a blood gas analyzer (ABL 800, 13B2X00079000003, Radiometer Co, Copenhagen, Denmark). The analyzer measured whole blood samples at 37 C. The laboratory in the hospital complies with standards of the National Association of Testing Authorities. The measurement of creatinine concentration by the blood gas analyzer (Cre gas ) is based on an enzymatic method, which uses an amperometric biosensor based on enzymatic conversion. 12 During the study period, blood gas analyses were performed at the discretion of intensivists or trained nursing staff in the PICU.
Perioperative serum creatinine concentration was also measured in a central laboratory (Cre lab ). Blood samples were taken for laboratory analysis, sent to the laboratory within 30 minutes, and analyzed immediately. Creatinine measurement by the central laboratory is based on an enzymatic method (BM8040, CA-Z13055TIJ, Japan Electron Optics Laboratory, Tokyo, Japan). The time of each analysis was recorded on our electronic medical chart.

| Primary and secondary objectives
The primary objective was the correlation and the agreement between Cre gas and Cre lab . Paired creatinine concentrations of Cre gas and Cre lab during a 5-day period after surgery for which the timings of the two measurements were the closest, and the time gap of their measurements was within 1 hour, considering recommended time gap of the two measurements from the same blood sample in our institution, were investigated. The secondary objective was the agreement between the incidence of postoperative AKI based on Cre gas and the incidence of postoperative AKI based on Cre lab during the first 48 hours after surgery. AKI was diagnosed and classified by KDIGO criteria. 6 According to this classification and staging system of AKI, serum creatinine concentration greater than or equal to 1.5 times the baseline or increase in serum creatinine greater than or equal to 0.3 mg/dL from the baseline creatinine concentration (both within 48 hours after surgery) constitutes "stage 1"; serum creatinine concentration greater than or equal to 2.0 times the baseline constitutes "stage 2"; and serum creatinine concentration greater than or equal to 3.0 times the baseline or initiation of renal replacement therapy constitutes "stage 3."

| Statistical analysis
Data are presented as n (percentages) or median (interquartile range [IQR]; 25% quartile, 75% quartile) as appropriate. The correlations between the two determination methods (Cre gas vs Cre lab ) were assessed by Pearson correlation coefficients (r) with 95% confidence interval (CI) and by linear regression. The linear regression is presented as a scatter plot, regression equation, and correlation coefficient. An issue of nonindependence from repeated measurements was solved by averaging the repeated measures data for each participant in sensitivity analyses. Additionally, the agreement between the two methods was analyzed, and Bland-Altman plots were made to visualize the agreement. The Bland-Altman plot shows the deviation between the two methods with different Cre lab concentrations. The agreement between the two methods on the diagnosis of AKI was analyzed by the Kappa statistic. The baseline creatinine level and postoperative Cre gas and Cre lab were investigated among patients with a discordant AKI diagnosis to understand their characteristics. In order to assess the agreement between the two methods with the same number of measurements and with a short time gap, a subgroup analysis was performed for patients who had both Cre gas and Cre lab where the time gap between the measurements was within 1 hour. All statistical tests were two-sided, and the significance level was defined as a P value of less than.05. All statistical analyses were performed using R 3.6.0 (R foundation for Statistical Computing, Vienna, Austria). F I G U R E 1 Scatter plots, regression analysis, and correlation coefficients for creatinine measurements between the blood gas analyzer and central laboratory. The result of the linear regression analysis shows the relationship: Cre gas = 1.021 × Cre lab + 0.010. Abbreviations: Cre gas , creatinine measuements by blood gas analyzer; Cre lab , creaninine measurements conducted in central laboratory Cre lab was Cre gas = 1.025 × Cre lab + 0.010. A Bland-Altman plot of Cre gas vs Cre lab is shown in Figure 2. The median bias between Cre gas and Cre lab was 0.02 (IQR -0.02, 0.05) mg/dL. The 95% limit of agreement was -0.137 to 0.178 mg/dL.

| Key findings
We found an excellent correlation between Cre gas and Cre lab for measuring creatinine, with r = .968 (95% CI, 0.965-0.972, P < 0.001) and median bias of 0.02 (IQR -0.02, 0.05) mg/dL. The 95% limit of agreement was -0.137 to 0.178 mg/dL in this group of pediatric patients (younger than 18 years of age). Serum creatinine was measured about seven times more frequently with the blood F I G U R E 2 Bland-Altman plot for the differences between blood gas analyzer serum creatinine and central laboratory serum creatinine. The 95% limit of agreement was −0.137 to 0.178 mg/dL. Abbreviations: Cre gas , creatinine measurements by blood gas analyzer; Cre lab , creaninine measurements conducted in central laboratory Abbreviations: AKI, acute kidney injury; Cre gas , creatinine measurements by blood gas analyzer; Cre lab , creaninine measurements conducted in central laboratory.
gas analyzer after cardiac surgery than in the central laboratory.
Despite the excellent agreement between the two methods, using  The impact of POC creatinine testing on clinical practice, however, is unclear. Although the shortened average time to diagnosis has been evaluated by using various POC devices in some studies, 15,16 there has been no study on the impact of measurements of creatinine by a blood gas analyzer, especially on the detection of AKI in patients.

| Comparison with prior studies
We showed that at our hospital, the frequency of creatinine concentration measurements by a POC analyzer was about seven times higher than that of creatinine concentrations measurements in a central laboratory. In addition, we found that the number of patients diagnosed as having AKI by Cre gas was about two times greater than the number of patients diagnosed as having AKI by Cre lab , and discordant diagnosis of AKI was found in around one-third of patients regardless of the good agreement between the two methods for creatinine concentration determination.

| Interpretation
According to the US Clinical Laboratory Improvement Amendments (USCLIA), the criteria for acceptable performance of measurements is the target value ±0.3 mg/dL for creatinine. 17 We demonstrated high levels of agreement, with a low median difference of Cre gas and Cre lab and 95% limit of agreement within the range specified by the USCLIA.
Our study, which is the first one to assess the performance of a POC device exclusively on patients younger than 18 years of age, supports the clinical usefulness of Cre gas for patients with a wide range of ages, when taken together with the results of the other studies mentioned above.
The reason for serum creatinine being measured much more frequently by the POC device than in the central laboratory might be related to the immediate availability of the results. 15 A second possible explanation for the discordant diagnosis of AKI despite good correlation between the two methods is due to a potential error of each Cre gas measurement. Although our study showed a high level of agreement of Cre gas , we could still see some differences between Cre gas and Cre lab , as shown in Figure 1. In addition, although the reported precision of the Radiometer ABL is relatively high, 1,3 there is a risk of obtaining incorrect creatinine concentration when the concentrations are measured many times by a blood gas analyzer.
Increased risk of errors due to multiple measurements of Cre gas could be one reason. Furthermore, low creatinine concentration in pediatric patients should also be considered in the discordant diagnosis of AKI between the two method. As the KDIGO criteria uses baseline creatinine concentration for reference and the multiplication of the baseline measurement for AKI diagnosis, low baseline creatinine con-

| Limitations/generalizability
There are several limitations in this study. First, it was a retrospective study and, thus, potentially subject to systematic error and bias. However, the clinical and electronic data were collected prospectively and were measured independently, and the data were, therefore, not subject to unintended manipulation. Second, this was a small, singlecenter study with a significant chance of a type I error and weak generalizability. Third, we used pairs of creatinine measurements for which the time gap between measurements of Cre gas and Cre lab was within 1 hour to assess the correlation and the agreement. This is because the two samples obtained at the same time were analyzed at different places, with an associated time gap, and because our electronic records only have information on the time of analyses, not the time of blood sampling. Although changes in creatinine are usually slow and unlikely to show a large difference within such a short period, this does not reflect a perfect assessment of the accuracy of Cre gas . In addition, dramatic change in volume status during postoperative period might affect our findings. Those limitations make it difficult to generalize our results. Furthermore, this study did not show any patients' outcomes. The impact of high frequent diagnosis of AKI by POC device on clinical outcomes remains unclear. Considering that the cost of one blood gas measurement is about 1.2 times higher than that of one ordinal chemistry panel in the Japanese public medical system, the advantage of frequent creatinine measurements with the POC device is unclear. However, it should be noted that our objectives were to evaluate the POC serum creatinine measurement device for pediatric patients and to assess the impact on diagnosis of AKI. In this regard, a future prospective study with scheduled measurements of Cre gas and Cre lab from a range of patients and evaluating clinical outcomes should be conducted.

| CONCLUSIONS
There was an excellent correlation between measurements of creatinine by the POC blood gas analyzer and those conducted in the central laboratory for patients who were younger than 18 years of age and underwent cardiac surgery. Although a discordant diagnosis of postoperative AKI based on KDIGO criteria was found in about one third of the pediatric patients between the two methods when using all the measurements, the paired measurements with a small time gap showed a good agreement on AKI diagnosis. Additional studies are needed to show the efficacy of frequent creatinine measurements by POC devices for diagnosis of AKI based on KDIGO criteria for pediatric patients.

TRANSPARENCY STATEMENT
Satoshi Kimura affirms that this manuscript is an honest, accurate, and transparent account of the study being reported that no important aspects of the study have been omitted and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

DATA SHARING STATEMENT
No original data are shared because of no informed consent for data sharing.