Abstract
Neonatal jaundice is an important cause of morbidity and mortality worldwide, and neonates born in low and middle-income countries bear a disproportionate burden. We previously developed a low-cost, point-of-care system to measure total serum bilirubin (TSB) in neonates. This device was effective at detecting and monitoring jaundice; however, the disposable strips were difficult to produce at scale. Here, we report a new lateral flow cassette design, called BiliDx, that was produced at scale using traditional manufacturing techniques. We evaluated the performance of BiliDx at sites in Nigeria and Malawi. The lateral flow strip consists of plasma separation membranes, nitrocellulose, and a plastic cassette. We evaluated the performance of the strips and reader at two hospitals located in Nigeria and Malawi compared to reference standard TSB. We also assessed performance for samples with high direct bilirubin (DB) and high hematocrit (HCT). We collected 1,144 samples from 758 neonates (TSB ranged from 0.2 to 45.9 mg/dL). The mean bias of BiliDx measurements in the validation set was +0.75 mg/dL, and 95% limits of agreement were -2.57 to 4.07 mg/dL. The mean bias and limits of agreement were comparable for samples with HCT < 60% and HCT ≥ 60%, and for samples with low and intermediate DB levels; the samples with high DB levels had wider 95% limits of agreement (-4.50 to +3.03 mg/dL). Error grid analysis shows that 96.9% of samples measured with BiliDx would have resulted in the same clinical decision as the reference standard. This performance is comparable to previous results that used a handmade two-dimensional strip. Additionally, error grid analysis shows that all 20 samples with high DB levels would have resulted in the same clinical decision as the reference standard. This evaluation supports the use of BiliDx lateral flow cassettes to provide accurate point-of-care measurements in low-resource settings.