Sample application technique affecting whole blood creatinine point of care test stick results

In 2018, NICE published a technical briefing ¹ on the use of point of care creatinine testing in radiology departments. It included data from studies of costs, interferences and correlation with laboratory results (Nova Biomedical, Runcorn, UK). ² Locally, we use five Nova StatSensor-i® meters, which measure creatinine in venous blood or finger-prick samples. Overall correlation between meter whole blood creatinine and laboratory serum creatinine values is high (R = 0.93), but individual sample agreement is sometimes poor, with differences of more than 20% for 18 of 68 paired samples, and more than 40% for six of these pairs.

Imprecision with the StatSensor-i® meter is most likely responsible for the large differences occasionally seen between meter and laboratory results. Variable agreement between our five meters is also observed for EQA samples. Better precision is obtained with the internal quality control samples, but these are analysed in a ‘QC mode’ and have a different matrix.

We considered whether sample application technique could affect the meter results. Creatinine was measured on paired serum and whole blood samples, by kinetic Jaffe reaction (Roche Cobas) and StatSensor-i®, respectively. StatSensor-i® measurements were carried out using three different application techniques:

Technique 1: dipping the test strip vertically into an aliquot of blood to just below the first reagent well.

Technique 2: sideway application (180°) to the edge of the test strip through a capillary tube.

Technique 3: sampling from a bead of blood placed on a clean glass slide.

Mean and CV was calculated for each of 15 sets of duplicates for each technique, and compared using Student’s paired t-test (Analyse it for Excel version 2 Leeds, UK). The patient group mean for laboratory serum creatinine was 130 μmol/L.

Technique 1 (dipped) (mean 140 μmol/L) gave closest agreement but precision was the worst of the three techniques (overall CV 12.8%). Technique 2 (capillary application) (mean 153 μmol/L) had a CV of 10.1%. CV was 6.1% with technique 3 (glass slide) (mean 158 μmol/L). Meter results were statistically higher than laboratory results for techniques 2 and 3, and also significantly different from technique 1. This introduces the possibility that the technique with which the sample is applied may affect the result.

The sample flow rate within the test strip is reported to be crucial for successful testing; indeed, if it is outside set limits, the meter displays an error. In our hands, at least 10% of patient tests are rejected in this way. The reason for the flow rate rejection is often not visually obvious. Possibly application method (e.g. depth of immersion or angle of application) affects the sample flow rate. Use of these meters may help to reduce cancellations of scans due to lack of a laboratory evaluation of renal function, but our findings indicate that results must be interpreted with caution.