Abstract
BACKGROUND: There are a variety of approaches being used for malaria surveillance. While active and reactive case detection have been successful in localized areas of low transmission, concerns over scalability and sustainability keep the approaches from being widely accepted. Mobile health interventions are poised to address these shortcomings by automating and standardizing portions of the surveillance process. In this study, common challenges associated with current data aggregation methods have been quantified, and a web-based mobile phone application is presented to reduce the burden of reporting rapid diagnostic test (RDT) results in low-resource settings. METHODS: De-identified completed RDTs were collected at 14 rural health clinics as part of a malaria epidemiology study at Macha Research Trust, Macha, Zambia. Tests were imaged using the mHAT web application. Signal intensity was measured and a binary result was provided. App performance was validated by: (1) comparative limits of detection, investigated against currently used laboratory lateral flow assay readers; and, (2) receiver operating characteristic analysis comparing the application against visual inspection of RDTs by an expert. Secondary investigations included analysis of time-to-aggregation and data consistency within the existing surveillance structures established by Macha Research Trust. RESULTS: When compared to visual analysis, the mHAT app performed with 91.9% sensitivity (CI 78.7, 97.2) and specificity was 91.4% (CI 77.6, 97.0) regardless of device operating system. Additionally, an analysis of surveillance data from January 2017 through mid-February 2019 showed that while the majority of the data packets from satellite clinics contained correct data, 36% of data points required correction by verification teams. Between November 2018 and mid-February 2019, it was also found that 44.8% of data was received after the expected submission date, although most (65.1%) reports were received within 2 days. CONCLUSIONS: Overall, the mHAT mobile app was observed to be sensitive and specific when compared to both currently available benchtop lateral flow readers and visual inspection. The additional benefit of automating and standardizing LFA data collection and aggregation poses a vital improvement for low-resource health facilities and could increase the accuracy and speed of data reporting in surveillance campaigns.