Implementation of a GPS-based location-tracking smartphone app in South Africa to improve engagement in HIV care: Lessons learned from real-world experience (Preprint)

Abstract

Mobile health (mHealth) interventions are common in public health settings in sub-Saharan Africa, and our preliminary work showed that smartphones are increasing in number in South Africa. We developed a novel smartphone app – CareConekta – that used GPS location data to characterize personal mobility in order to improve engagement in HIV care among pregnant and postpartum women living with HIV in South Africa. The app also used the user’s location to map nearby clinics.

We describe the feasibility, acceptability and initial efficacy of using the app in a real-world setting.

We conducted a prospective randomized control trial at a public-sector clinic near Cape Town, South Africa. We enrolled 200 pregnant (third trimester) women living with HIV who owned a smartphone that met the required specifications. All participants installed the app, designed to collect two GPS “heartbeats” per day to geo-locate the participant within a random one kilometer “fuzzy” radius (for privacy). We randomized (1:1) participants to a control arm to receive no additional support, or an intervention arm to receive supportive phone calls and/or WhatsApp messages from the study team when traveling more than 50 km out of the study area for more than seven days. In addition to mobility data collected daily through the phone, participants completed questionnaires at enrollment and follow-up (about six months postpartum).

Seven participants were withdrawn at the time of enrollment or shortly after due to app installation failure (n=6) or changing to a not-suitable phone (n=1). During the study period, no participant’s smartphone recorded at least one heartbeat per day, our primary feasibility measure. Of the 171 participants who completed follow-up, only half (53%, n=91) reported using the same phone as enrollment, with the CareConekta app still installed on the phone and GPS usually enabled. When staff contacted participants about lapses in GPS “heartbeats,” the top reasons stated were lack of mobile data, uninstalling the app or the participant no longer had a smartphone. Acceptability measures found positive responses, but participants at follow-up demonstrated a lack of understanding of the app’s purpose and function. The clinic finder was a popular feature of the app. Due to the lack of consistent GPS “heartbeats” throughout the study, we were unable to assess efficacy of the intervention.

We found that several key challenges impeded our study feasibility. While the app was designed to reverse-bill participants for any data usage, the lack of mobile data was a substantial barrier to our study success. Participants reported purchasing WhatsApp data, which could not support the app. Problems with the web-based dashboard meant we could not consistently monitor mobility. Our study provides important lessons learned about implementing an ambitious GPS-based study under real-world conditions in a limited-resource setting.

ClinicalTrials.gov NCT03836625

RR2-10.1186/s13063-020-4190-x