Skip to main content
. 2016 Nov 9;14:123–130. doi: 10.1016/j.ebiom.2016.11.011

Fig. 1.

Fig. 1

Description of subjects and flow chart of sample collection and assays performed. Blood samples for hematological, biochemical, and parasite growth analyses were drawn at Day 0, as well as Day 49 and Day 84 for those taking iron. A full hematology panel was measured in EDTA-stabilized blood (Medonic M20M GP). We also assayed plasma ferritin, soluble transferrin receptor (sTfR), serum iron, transferrin saturation (TSAT), C-reactive protein (CRP), alpha 1-acid glycoprotein (AGP) (Cobas Integra 400 plus); and hepcidin (Hepcidin-25 (human) EIA Kit (Bachem)). Genotyping for hemoglobinopathies was performed using hemoglobin electrophoresis. Glucose-6-phosphate dehydrogenase (G6PD) enzyme activity was measured by commercial kit (R&D Diagnostics Ltd). For malaria assays, 2.5 ml of venous blood was drawn directly into microvette tubes containing CPDA-1 (Sarstedt, Germany). Unavailable donors include safety exclusion (Hgb < 7 g/dl or positive malaria test, RDT pos) or general loss to follow up (withdrawal and travel). Failure to collect blood from subjects (e.g. from phlebotomy failure, subject moved or withdrew, or became significantly ill) was 7.8% (32/407) at Day 0, 17.0% (23/135) at Day 49, and 20.7% (28/135) at Day 84. RBCs from study subjects were evaluated with in vitro P. falciparum growth assays (using strain FCR3-FMG) as a proxy measure for malaria susceptibility. In order to standardize the growth assays, control for inter-assay variability and variability between parasite preparations, assays on clinical samples were run in parallel with and reported relative to growth assays done using RBCs from non-anemic donors. Each available blood sample at every time point was subjected to growth assays but not all produced growth data, as some blood was unusable (e.g. clotted, hemolysed, contaminated). Further growth data exclusions (e.g. parasites died or control blood did not provide a readable output for comparison) do not represent population sampling bias, as subject characteristics are the same between those with and without corresponding growth data (Supplemental Table 2).