SHBC1559

R.HAWKINS¹

Tan Tock Seng Hospital¹

Icterus index (IC) measurement is routinely performed on all clinical specimens in many laboratories. This studies examines whether IC values can predict hyperbilirubinaemia and can potentially be used to avoid unnecessary total bilirubin measurements.

In our laboratory, serum index measurements are automatically performed on all samples analysed on the 3 Beckman Coulter DxC-800 clinical chemistry analysers. Anonymised details of all simultaneous IC and total bilirubin (diazo, manufacturer supplied reagent) measurements for 6 months (Jan to June 2016) were extracted from the laboratory database.  The locally derived total bilirubin reference interval is 7-31 umol/L.  Excel 2003 and Analyse-it were used to prepare a ROC curve to assess the ability of IC to predict TBil > 31 umol/L.

There were 51111 paired IC and total bilirubin results. In 5562 (10.9%) cases, total bilirubin > 31 umol/L. The sensitivity and specificity, with 95% CIs, for IC to predict total bilirubin  > 31 umol/L were: for IC >0 (i.e >=1): sensitivity 1.000 (0.999-1.000), specificity 0.004 (0.003-0.005); IC >1 (i.e. >=2): sensitivity 0.992 (0.989-0.994), specificity 0.753 (0.749-0.757). Using an IC>0, 183 tests are avoided (at a cost of 2 false negatives) and IC>1, 34345 tests are avoided (at a cost of 47 false negatives).

Withholding total bilirubin measurement on specimens with icterus index (IC) of 0 or 1 would reduce total bilirubin test volumes by 67% with only a 0.08% false negative rate. The false negative rate could potentially be reduced further on analytical systems reporting smaller IC increments.