Data Management and Analysis

All case record forms were collected, checked for completeness and double entered into a database using EpiInfo/EpiData version 3.5.1. (CDC, USA). Statistical analysis was done using STATA version 13.0. (StataCorp, USA). For visualization of results, the dot plot maker from Vanderbilt University (http://data.vanderbilt.edu/~graywh/dotplot/, last accessed on 17 September 2015) was used.

Spectrophotometry (Trinity Biotech quantitative assay, Procedure No. 345 –Trinity Biotech, Ireland) was considered as the reference method. G6PD deficiency was defined as described earlier [13] and calculated per 10¹² RBC and per gHb. In the absence of a standardized G6PD threshold activity, we calculated the median G6PD activity as measured by spectrophotometry of all male students and excluded all results with ≤10% G6PD activity of the derived median. We re-calculated the median based on the remaining samples of male participants and defined the resulting G6PD activity as 100%. Based on this reference value, G6PD cut-off activities for G6PD deficiency at 10%, 20%, 30% and 60% were calculated [13]. As hemoglobin content per dL blood and number of red blood cells do not match perfectly, the denominator for G6PD activity varies slightly. We considered the number of RBCs to be the more precise measure; test performance, therefore, was calculated based on the G6PD activity in IU/10¹² RBC.

The FST results were defined in two ways: 1) FSTdefint—FST deficient and intermediate test results were defined as G6PDd, and 2) FSTdef—only FST deficient results are defined as G6PDd and intermediate results as G6PD normal. The CSG was performed twice, each on capillary (CSGcap) and venous (CSGven) sample for comparison. G6PDd prevalence within the population was calculated per cut-off activity as measured by spectrophotometry and presented as percentage of the study population with activities at or below the respective cut-off activity.

For the purpose of analysis, a positive test result was defined as a test result that indicates G6PD deficiency. Sensitivity and specificity were calculated using a standard formula [13]. In SE Asia, point-of-care test will primarily serve to guide 8-aminoquinoline treatment of vivax malaria, sensitivities of all tests were compared at 30% G6PD activity [20]. A lower cut-off may apply for guidance of single low-dose primaquine therapy to prevent the transmission of falciparum malaria. Differences in proportions were calculated using Chi-square and Fisher’s exact test for the results of FST and WST, as appropriate. McNemar’s test was used for comparing differences in proportion for the CSG. Confidence intervals were calculated using the exact method.