Search

IntroductionIn the Western Cape Province of South Africa, a wealth of patient-level data is collected through many separate electronic systems, which share a unique health identifier. However, clinicians primarily access paper folders, which can be unreliable, difficult to locate and are at risk of loss. Patients frequently attend multiple facilities and their information may not be accessible across facilities, hampering continuity of care. Objectives and ApproachFacilitated by the unique health identifier, a provincial Health Information Exchange (HIE), harmonises patient level data from routine systems into a health information exchange daily. The Single Patient Viewer (SPV) has been developed as a prototype web-based electronic health record and data access portal. SPV integrates clinical data for a single patient both longitudinally and cross-sectionally, in tabular and graphical views, to assist clinicians in rapid information discovery. The application is designed as a web application that calls a multi-purpose API that facilitates interoperability with the HIE. ResultSPV is in an extended pilot phase with over 200 clinicians using the portal, either for clinical care provision, or for clinical audit activities. The application has evolved to include referral, follow-up (voice call and messaging) and reporting functionality. In the past 6 months, over 17,000 patients have been viewed with an average daily search of 150 patients. An anonymised user survey with 52 participants showed that users felt that SPV was enjoyable to use, easy to learn, innovative, and supportive and valuable to their work. Conclusion / ImplicationsSPV has been developed as a global public good project and will be made freely available once matured. A unique feature of the development of SPV is that clinicians were using it while it was being built, enabling constant clinical user feedback. SPV demonstrates the value of integrating health data for clinical viewing while interoperable systems mature.

BACKGROUND Digital health programs, which encompass the subsectors of health information technology, mobile health, electronic health, telehealth, and telemedicine, have the potential to generate “big data.” OBJECTIVE Our aim is to evaluate two digital health programs in India—the maternal mobile messaging service (Kilkari) and the mobile training resource for frontline health workers (Mobile Academy). We illustrate possible applications of machine learning for public health practitioners that can be applied to generate evidence on program effectiveness and improve implementation. Kilkari is an outbound service that delivers weekly gestational age–appropriate audio messages about pregnancy, childbirth, and childcare directly to families on their mobile phones, starting from the second trimester of pregnancy until the child is one year old. Mobile Academy is an Interactive Voice Response audio training course for accredited social health activists (ASHAs) in India. METHODS Study participants include pregnant and postpartum women (Kilkari) as well as frontline health workers (Mobile Academy) across 13 states in India. Data elements are drawn from system-generated databases used in the routine implementation of programs to provide users with health information. We explain the structure and elements of the extracted data and the proposed process for their linkage. We then outline the various steps to be undertaken to evaluate and select final algorithms for identifying gaps in data quality, poor user performance, predictors for call receipt, user listening levels, and linkages between early listening and continued engagement. RESULTS The project has obtained the necessary approvals for the use of data in accordance with global standards for handling personal data. The results are expected to be published in August/September 2019. CONCLUSIONS Rigorous evaluations of digital health programs are limited, and few have included applications of machine learning. By describing the steps to be undertaken in the application of machine learning approaches to the analysis of routine system-generated data, we aim to demystify the use of machine learning not only in evaluating digital health education programs but in improving their performance. Where articles on analysis offer an explanation of the final model selected, here we aim to emphasize the process, thereby illustrating to program implementors and evaluators with limited exposure to machine learning its relevance and potential use within the context of broader program implementation and evaluation. INTERNATIONAL REGISTERED REPOR DERR1-10.2196/11456

Despite calls to address broader evidence gaps in linking digital technologies to outcome and impact level health indicators, limited attention has been paid to measuring processes pertaining to the performance of programs. In this paper, we assess the program reach and message exposure of a mobile health information messaging program for mothers (MomConnect) in South Africa. In this descriptive study, we draw from system generated data to measure exposure to the program through registration attempts and conversions, message delivery, opt-outs and drop-outs. Using a logit model, we additionally explore determinants for early registration, opt-outs and drop-outs. From August 2014 to April 2017, 1 159 431 women were registered to MomConnect; corresponding to half of women attending antenatal care 1 (ANC1) and nearly 60% of those attending ANC1 estimated to own a mobile phone. In 2016, 26% of registrations started to get women onto MomConnect did not succeed. If registration attempts were converted to successful registrations, coverage of ANC1 attendees would have been 74% in 2016 and 86% in 2017. When considered as percentage of ANC1 attendees with access to a mobile phone, addressing conversion challenges bring registration coverage to an estimated 83%–89% in 2016 and 97%–100% in 2017. Among women registered, nearly 80% of expected short messaging service messages were received. While registration coverage and message delivery success rates exceed those observed for mobile messaging programs elsewhere, study findings highlight opportunities for program improvement and reinforce the need for rigorous and continuous monitoring of delivery systems.

Information systems designed to support health promotion in pregnancy, such as the MomConnect programme, are potential sources of clinical information which can be used to identify pregnancies prospectively and early on. In this paper we demonstrate the feasibility and value of linking records collected through the MomConnect programme, to an emergent province-wide health information exchange in the Western Cape Province of South Africa, which already enumerates pregnancies from a range of other clinical data sources. MomConnect registrations were linked to pregnant women known to the public health services using the limited identifiers collected by MomConnect. Three-quarters of MomConnect registrations could be linked to existing pregnant women, decreasing over time as recording of the national identifier decreased. The MomConnect records were usually the first evidence of pregnancy in pregnancies which were subsequently confirmed by other sources. Those at lower risk of adverse pregnancy outcomes were more likely to register. In some cases, MomConnect was the only evidence of pregnancy for a patient. In addition, the MomConnect records provided gestational age information and new and more recently updated contact numbers to the existing contact registry. The pilot integration of the data in the Western Cape Province of South Africa demonstrates how a client-facing system can augment clinical information systems, especially in contexts where electronic medical records are not widely available.