Pharmacovigilance of suspected or confirmed therapeutic ineffectiveness of artemisinin-based combination therapy: extent, associated factors, challenges and solutions to reporting
MALARIA JOURNAL
Authors: Kiguba, Ronald; Ndagije, Helen Byomire; Nambasa, Victoria; Manirakiza, Leonard; Kirabira, Elijah; Serwanga, Allan; Olsson, Sten; Speybroeck, Niko; Mukonzo, Jackson
Abstract
Background Therapeutic ineffectiveness of artemisinin-based combination therapy (ACT) increases the risk of malaria-related morbidity and mortality, and raises healthcare costs. Yet, little has been done to promote the pharmacovigilance (PV) of ACT ineffectiveness in sub-Saharan Africa, particularly in Uganda. This study aimed to determine the extent and associated factors of the past 6 months reporting of suspected or confirmed ACT therapeutic ineffectiveness by healthcare professionals (HCPs), and difficulties and potential solutions to the PV of ACT therapeutic ineffectiveness. Methods Survey of 685 HCPs conducted using a self-administered questionnaire from June to July 2018 in a nationally representative sample of public and private health facilities in Uganda. HCPs disclosed if they had spontaneously reported ACT therapeutic ineffectiveness to appropriate authorities in the previous 6 months. Multivariable logistic regression models were used to identify determinants of past 6-months, HCP-reported ACT therapeutic ineffectiveness. Results One in five (20%, 137/685; 95% CI 17-23%) HCPs reported ACT therapeutic ineffectiveness to an appropriate authority in the previous 6 months. HCPs commonly reported ACT therapeutic ineffectiveness to immediate supervisors (72%, 106/147), mostly verbally only (80%, 109/137); none had ever submitted a written report of ACT therapeutic ineffectiveness to Uganda's National Pharmacovigilance Centre. Common difficulties of reporting ACT therapeutic ineffectiveness were: unavailability of reporting procedures (31%, 129/421), poor follow-up of treated patients (22%, 93/421) and absence of reporting tools (16%, 68/421). Factors associated with reporting ACT therapeutic ineffectiveness in the past 6 months were: hospital-status (vs other; OR = 2.4, 95% CI 1.41-4.21), HCPs aged under 25 years (OR = 2.2, 95% CI 1.29-3.76), suspicion of ACT therapeutic ineffectiveness in the past 4 weeks (OR = 2.3, 95% CI 1.29-3.92), receipt of patient-complaint(s) of ACT therapeutic ineffectiveness in the past 4 weeks (OR = 2.9, 95% CI 1.62-5.12) and HCPs from northern (vs central; OR = 0.5, 95% CI 0.28-0.93) and western (vs central; OR = 0.4, 95% CI 0.17-0.77) parts of Uganda. Conclusion One in five HCPs reported ACT therapeutic ineffectiveness, mostly verbally to supervisors. The existing adverse drug reaction (ADR)-reporting infrastructure could be leveraged to promote the PV of ACT therapeutic ineffectiveness.
Deep learning enabled multi-wavelength spatial coherence microscope for the classification of malaria-infected stages with limited labelled data size
OPTICS AND LASER TECHNOLOGY
Authors: Singla, Neeru; Srivastava, Vishal
Abstract
Malaria is a life-threatening mosquito-borne blood disease and therefore, early detection is essential for health. The standard detection technique is a Giemsa-stained blood smear test, which requires a highly skilled technician. Automated classifications of different phases of malaria continue to be a challenge with limited label data, especially in the case of early trophozoite and late trophozoite or schizont phase where the sensitivity is lower. The aim of the study is to develop a fast, robust and fully automated malaria classification system with limited labeled data set by using a multi-wavelength based system and a pre-trained convolutional neural networks (CNNs). The multi-wavelength system improves the classification performance by increasing the training dataset. We also compare our customized CNN performance with other conventional CNNs and show that our network with a lesser number of layers has a comparable performance. We believe that our approach can be applied to other limited annotated biological datasets.