G214(P)\u2005The performance of the ABL90 blood gas analyser for jaundice testing

Abstract

Aims There is a growing interest in faster blood testing methods using smaller blood volumes, particularly for infants and neonates. Blood gas analysers are becoming increasingly popular for bilirubin measurement. We aimed to test the diagnostic validity of whole blood bilirubin measurement using the ABL90 blood gas analyser and its correlation to standard serum measurement. We also studied the time from blood gas to serum results and audited all London neonatal units for their routine bilirubin measurement methods. Secondary outcomes were the analyser’s performance on sodium, potassium and haemoglobin measurements. Methods We used 107 prospectively collected paired samples (serum and whole blood) taken for clinical indications, from jaundiced and non-jaundiced babies. Samples were analysed as per routine practice, using the ABL90 and the Roche/Cobas analysers. Results Data were analysed using GraphPad Prism v6. 91 paired bilirubin data were used. Analysis for jaundice diagnosis yielded: sensitivity 88%, 95% C.I. [51%,99%]; specificity 96%, C.I. [89%,99%]; PPV 72% C.I. [39%,93%]; NPV 98%, C.I. [93%,99%]. The two methods demonstrated excellent correlation (Spearman’s r=0.99, p<0.0001). Bland-Altman analysis yielded a clinically non-significant mean difference [MD=3.27 μmol/L, 95% limits of agreement (LOA): −28.29, 34.84]. There was a skewed positive bias in bilirubin values>200 μmol/L. Time from gas to serum analysis ranged between 0 and 240\u2009min, with a median of 59\u2009min [25%–75% percentiles: 42–79\u2009min]. 8/26 London neonatal units use gas analysers as their standard bilirubin measurement method, 7/26 use serum biochemistry measurements, 7/26 use centrifuged serum measurements and 4/27 use combinations of methods. Sodium measurement results were: MD=-2 mmol/L, LOA [−6.47, 2.47], r=0.81 (p<0.0001); Potassium: MD=0.07\u2009mmol/L, LOA [−1.23, 1.38], r=0.57 (p<0.0001); Haemoglobin: MD=−8.87\u2009g/L, LOA [−26.37, 8.61], r=0.98 (p<0.0001). Conclusions Our results suggest that the ABL90 gas analyser is a valid method for bilirubin measurement that could improve treatment and discharge processes in jaundiced babies. The excellent correlation could translate into serial measurement reliability. Clinical judgement should be exercised around high measurement levels, due to the possibility of false positives regarding the need for treatment. Haemoglobin data appear promising Further studies are required for sodium and potassium measurement performance.