Lower reference limits of quantitative cord glucose-6-phosphate dehydrogenase estimated from healthy term neonates according to the Clinical and Laboratory Standards Institute guidelines: a cross sectional retrospective study.

Background: Previous studies have reported the lower reference limit (LRL) of quantitative cord glucose-6-phosphate dehydrogenase (G6PD), but they have not used approved international statistical methodology. Using common standards is expecting to yield more true findings. Therefore, we aimed to estimate LRL of quantitative G6PD detection in healthy term neonates by using statistical analyses endorsed by the International Federation of Clinical Chemistry (IFCC) and the Clinical and Laboratory Standards Institute (CLSI) for reference interval estimation.

Methods: This cross sectional retrospective study was performed at King Abdulaziz Hospital, Saudi Arabia, between March 2010 and June 2012. The study monitored consecutive neonates born to mothers from one Arab Muslim tribe that was assumed to have a low prevalence of G6PD-deficiency. Neonates that satisfied the following criteria were included: full-term birth (37 weeks); no admission to the special care nursery; no phototherapy treatment; negative direct antiglobulin test; and fathers of female neonates were from the same mothers' tribe. The G6PD activity (Units/gram Hemoglobin) was measured spectrophotometrically by an automated kit. This study used statistical analyses endorsed by IFCC and CLSI for reference interval estimation. The 2.5th percentiles and the corresponding 95% confidence intervals (CI) were estimated as LRLs, both in presence and absence of outliers.

Results: 207 males and 188 females term neonates who had cord blood quantitative G6PD testing met the inclusion criteria. Method of Horn detected 20 G6PD values as outliers (8 males and 12 females). Distributions of quantitative cord G6PD values exhibited a normal distribution in absence of the outliers only. The Harris-Boyd method and proportion criteria revealed that combined gender LRLs were reliable. The combined bootstrap LRL in presence of the outliers was 10.0 (95% CI: 7.5-10.7) and the combined parametric LRL in absence of the outliers was 11.0 (95% CI: 10.5-11.3).

Conclusion: These results contribute to the LRL of quantitative cord G6PD detection in full-term neonates. They are transferable to another laboratory when pre-analytical factors and testing methods are comparable and the IFCC-CLSI requirements of transference are satisfied. We are suggesting using estimated LRL in absence of the outliers as mislabeling G6PD-deficient neonates as normal is intolerable whereas mislabeling G6PD-normal neonates as deficient is tolerable.