A comparison of three methods to estimate G6PD activity in the Chittagong Hill Tracts, Bangladesh

Published: January 25, 2017


Ley B, Alam MS, O’Donnell JJ, Hossain MS, Kibria MG, Jahan N, Khan WA, Thriemer K, Chatfield MD, Price RN, Richards JS. A comparison of three methods to estimate G6PD activity in the Chittagong Hill Tracts, Bangladesh. PLoS One, 2017;12(1):e0169930.  doi: 10.1371/journal.pone.0169930.



Glucose-6-phosphate-dehydrogenase-deficiency (G6PDd) is a major risk factor for primaquine-induced haemolysis. There is a need for improved point-of-care and laboratory-based G6PD diagnostics to unsure safe use of primaquine.


G6PD activities of participants in a cross-sectional survey in Bangladesh were assessed using two novel quantitative assays, the modified WST-8 test and the CareStart G6PD Biosensor (Access Bio), The results were compared with a gold standard UV spectrophotometry assay (Randox). The handheld CareStart Hb instrument (Access Bio) is designed to be a companion instrument to the CareStart G6PD biosensor, and its performance was compared to the well-validated HemoCue method. All quantitative G6PD results were normalized with the HemoCue result.


A total of 1002 individuals were enrolled. The adjusted male median (AMM) derived by spectrophotometry was 7.03 U/g Hb (interquartile range (IQR): 5.38–8.69), by WST-8 was 7.03 U/g Hb (IQR: 5.22–8.16) and by Biosensor was 8.61 U/g Hb (IQR: 6.71–10.08). The AMM between spectrophotometry and WST-8 did not differ (p = 1.0) but differed significantly between spectrophotometry and Biosensor (p<0.01). Both, WST-8 and Biosensor were correlated with spectrophotometry (rs = 0.5 and rs = 0.4, both p<0.001). The mean difference in G6PD activity was -0.12 U/g Hb (95% limit of agreement (95% LoA): -5.45 to 5.20) between spectrophotometry and WST-8 and -1.74U/g Hb (95% LoA: -7.63 to 4.23) between spectrophotometry and Biosensor. The WST-8 identified 55.1% (49/89) and the Biosensor 19.1% (17/89) of individuals with G6PD activity <30% by spectrophotometry. Areas under the ROC curve did not differ significantly for the WST-8 and Biosensor irrespective of the cut-off activity applied (all p>0.05). Sensitivity and specificity for detecting G6PD activity <30% was 0.55 (95% confidence interval (95%CI): 0.44–0.66) and 0.98 (95%CI: 0.97–0.99) respectively for the WST-8 and 0.19 (95%CI: 0.12–0.29) and 0.99 (95%CI: 0.98–0.99) respectively for the Biosensor. Hb concentrations measured by HemoCue and CareStart Hb were strongly correlated (rs = 0.8, p<0.001, mean difference = 0.09 g Hb/dL, 95% LoA: -2.15 to 2.34).


WST-8 and the CareStart G6PD Biosensor represent advances in G6PD diagnostics in resource poor settings, but will require further development before clinical deployment. The CareStart Hb instrument produced a precise measure of haemoglobin concentration.