Feasibility of digital contact tracing in low-income settings - pilot trial for a location-based DCT app

Abstract

Background: Data about the effectiveness of digital contact tracing are based on studies conducted in countries with predominantly high- or middle-income settings. Up to now, little research is done to identify specific problems for the implementation of such technique in low-income countries.

Methods: A Bluetooth-assisted GPS location-based digital contact tracing (DCT) app was tested by 141 participants during 14 days in a hospital in Monrovia, Liberia in February 2020. The DCT app was compared to a paper-based reference system. Hits between participants and 10 designated infected participants were recorded simultaneously by both methods. Additional data about GPS and Bluetooth adherence were gathered and surveys to estimate battery consumption and app adherence were conducted. DCT apps accuracy was evaluated in different settings.

Results: GPS coordinates from 101/141 (71.6%) participants were received. The number of hours recorded by the participants during the study period, true Hours Recorded (tHR), was 496.3 h (1.1% of maximum Hours recordable) during the study period. With the paper-based method 1075 hits and with the DCT app five hits of designated infected participants with other participants have been listed. Differences between true and maximum recording times were due to failed permission settings (45%), data transmission issues (11.3%), of the participants 10.1% switched off GPS and 32.5% experienced other technical or compliance problems. In buildings, use of Bluetooth increased the accuracy of the DCT app (GPS + BT 22.9 m ± 21.6 SD vs. GPS 60.9 m ± 34.7 SD; p = 0.004). GPS accuracy in public transportation was 10.3 m ± 10.05 SD with a significant (p = 0.007) correlation between precision and phone brand. GPS resolution outdoors was 10.4 m ± 4.2 SD.

Conclusion: In our study several limitations of the DCT together with the impairment of GPS accuracy in urban settings impede the solely use of a DCT app. It could be feasible as a supplement to traditional manual contact tracing. DKRS, DRKS00029327 . Registered 20 June 2020 - Retrospectively registered.

Keywords: App; Digital contact tracing; GPS tracking; Location based; Low income; Public health.

Fig. 1

Course of study with participants successively designated as infected showing their cumulative number and the number of possible contacts for each day; the timing of the intervention and the visits with the respective tasks was mapped on the timeline. * Permission for the DCT app to use location, storage and phone (Bluetooth, mobile data) of the smartphone

Fig. 2

Measurement of accuracy during use in an outside urban environment. Distance of logged GPS coordinate (violet) to location of the participant (grey) in m. Time of logged GPS coordinate in hh:mm

Fig. 3

Flow of participants and results of the contingency table for effectiveness

Fig. 4

True Recorded Hours (tHR) in hours per participant over 14 days (total 469.3 h). * Poor adherence (GPS), battery consumption, energy supply, smartphone quality, technical experience

Fig. 5

Maximum Hours Recordable (mHR) broken down into total hours recorded (tHR) and causes for non-tracking

Fig. 6

Percentage of participants logging GPS coordinates per day from study group of 141 participants; Mean time of tHR from participants logging GPS coordinates per day

Fig. 7

Accuracy in buildings. Contact of 22 participants with an index smartphone within a distance < 5 m in multi-story building. Green dots represent distances derived from both GPS and Bluetooth. Orange dots represent distance derived from GPS only

Fig. 8

GPS accuracy in public transportation (mean 10.35 m); tracked population was 23 participants

Fig. 9

GPS accuracy outdoors in urban environment (mean 10.4 m); tracked population was 18 participants