Study population and objectives
We performed a prospective, single-centre study of routine clinical practice in the cardionephrology outpatient unit at Ambroise Paré University Hospital (Boulogne-Billancourt, France).
Between October 2015 and June 2016, we included all consecutive patients aged 18 or over attending the outpatient unit for diagnostic and/or interventional procedures involving the injection of iodinated contrast agent (contrast-enhanced computed tomography, peripheral arteriography, coronary angiography alone, or coronary angiography combined with percutaneous angioplasty) or for nephrological or cardiovascular examinations not involving the use of iodinated contrast agent. There were no exclusion criteria.
The study’s primary objective was to evaluate the diagnosis of kidney failure on the basis of the capillary blood creatinine level (Ccap) versus the plasma creatinine level (Cpl) in consecutive patients attending the cardionephrology outpatient unit. The study’s secondary objective was to evaluate the diagnosis of kidney failure on the basis of the glomerular filtration rate calculated from the Ccap (eGFRcap) versus that calculated from the Cpl (eGFRpl) in the same patients.
Assay methods
The Ccap was measured with a POC test (Stat Sensor X-press, Nova Biomedical Corporation, Waltham, MA, USA) used routinely in the outpatient unit. After the patient had fasted overnight, a nurse collected a drop of capillary blood from the finger, placed it on the device’s test strip, and inserted the strip into the reader. The result (expressed in μmol/l) was given 25 seconds later. The assay method is based on the transformation of creatinine into hydrogen peroxide, as catalysed by three enzymes in the following reactions:
The hydrogen peroxide was then detected electrochemically; the redox reaction generates a current that is directly proportional to the creatinine concentration in the sample (12). Although there are no analytical standards for POC tests, the National Kidney Disease Education Program’s Laboratory Working Group has stated that the measurement imprecision should be less than 8% (relative to calibration with isotope dilution mass spectroscopy (IDMS)) and the bias should be less than <5% for a Cpl above 88.4 µmol/L(13). The device used in the present study met these criteria.
The eGFRcap was estimated from the Ccap, using the Chronic Kidney Disease–Epidemiology Collaboration (CKD-EPI) equation (14).
In parallel, the Cpl was measured in a two-point, kinetic enzyme assay (VITROS® CREA, Ortho Clinical Diagnostics, Raritan, NJ, USA that had been calibrated with IDMS. The fasted patient’s blood sample was collected by venepuncture of the arm with a 22 gauge needle and sent to the hospital’s central laboratory. The laboratory estimated the eGFRpl from the plasma creatinine level, using the CKD-EPI equation.
Statistical analysis
The thresholds for kidney failure were a creatinine level of 110 µmol/l for men and 96 µmol/l for women, and an eGFR below 60 ml/min/1.73m2, according to the CKD-EPI equation. Quantitative variables were expressed as the mean ± standard deviation (SD), and qualitative variables were expressed as the frequency (percentage).
The degree of correlation between the capillary blood measurements (Ccap and eGFRcap) and the plasma measurements (Cpl and eGFRpl) was assessed by calculation of Pearson’s correlation coefficient (r). The level of agreement between the capillary and plasma measurements was evaluated by calculation of the intraclass correlation coefficient (ICC) and graphically in a Bland-Altman plot.
The performance indicators (sensitivity (Se) and specificity (Sp)) for the diagnosis of kidney failure on the basis of the capillary blood measurements (Ccap and eGFR cap) with the Stat Sensor X-press versus the plasma measurements (Cpl and eGFRpl) were calculated, and a receiver operating characteristic ROC curve was used to determine the eGFRcap threshold for a diagnosis of kidney failure.
All statistical analyses were performed with R Core Team (version 3.2.3. 2013). R a language and environment for Statistical computing. R Foundation for Statistical Computing, Vienn, Austria. URL http: //www.R-project.org/