Table 1.
Summary of study designs, settings, diseases under study, and characteristics of populations, interventions, and comparators
| Study design | Summary; name of app or platform | Study setting | Disease under study | Study population | Number of participants or app users | Automated (no recall or manual data entry) or partly automated | Technology involved (eg, GPS, Bluetooth) | Intended device (eg, web browser, basic mobile phone, smartphone, multiple) | Contact definition used or nature of contacts (eg, recall based, proximity based) | Comparator | |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Fully automated contact tracing | |||||||||||
| Bulchandani et al (2020)28 | Modelling study | Branching-tree mathematical model with derivations of mean-field equations for transitions to so-called digital herd immunity (ie, R0<1 because of automated contact tracing) | N/A | COVID-19 | Hypothetical modelled population | N/A | Automated | Not specified | Not specified | Not specified | None |
| Ferretti et al (2020)14 | Modelling study | Estimated the proportion of transmission from presymptomatic (from a series of 40 case pairs), asymptomatic, and symptomatic individuals and the environment; and quantified the effect of intervention (case isolation, contact tracing, and quarantine) at different delay periods and for different intervention success rates | N/A | COVID-19 | Hypothetical population network and contact structures, detailed in Fraser et al (2004)10 | N/A | Automated | Not specified | Smartphone app (standalone) | Location or proximity based | Non-automated contact tracing (comparisons based on delay to case isolation and contact quarantine) |
| Hinch et al (2020)29 | Modelling study | Multiple outcomes under different scenarios involving app-based contact tracing alongside non-targeted interventions, such as lockdowns and physical distancing | UK | COVID-19 | Hypothetical population network and contact structures selected to match age-stratified data reported in Mossong et al (2008)30 | N/A | Automated | Bluetooth | Smartphone app (standalone) | Proximity based | Other non-pharmaceutical outbreak control intervention |
| Kim and Paul (2020)31 | Modelling study | Effect of automated contact tracing to establish the minimum fraction of the population that needs to participate for R0<1 | Not specified | COVID-19 | Hypothetical population | N/A | Automated | Not specified | Not specified | Proximity based | Not specified |
| Kucharski et al (2020)32 | Modelling study | Effect of multiple interventions (eg, app-based contact tracing, limiting daily contacts to different extents [eg, four contacts per day] in settings other than work or school, and >1 intervention in parallel) on individual-level transmission events | UK | COVID-19 | Hypothetical population with contact structure based on data from the BBC Pandemic study of 40 162 UK participants33 | N/A | Automated | Not specified | Smartphone app (standalone) | Social contacts (conversational or physical contact), as per the contact definition of the BBC Pandemic study33 | >1 comparison (including with no contact tracing and manual contact tracing) |
| Xia and Lee (2020)34 | Modelling study | Effect of automated proximity-based contact tracing; derived formulae to estimate lower and upper bounds on the minimum adoption rate required for R0<1 | N/A | COVID-19 | Hypothetical population | N/A | Automated | Proximity-based GPS data | Smartphone app or standalone wearable device | Proximity based | None |
| Yasaka et al (2020)35 | Modelling study | Description of TrackCOVID, a decentralised Bluetooth-based contact-tracing app, including modelling of the population infected at different levels of uptake | N/A | COVID-19 | Hypothetical population | N/A | Automated | Checkpoints based on QR codes | Smartphone app (standalone) | Face to face with manual code scanning | Compared with no contact tracing |
| Partly automated contact tracing | |||||||||||
| Danquah et al (2019)36 | Proof of concept study with phased introduction | Observational study regarding the introduction of the Ebola Contact Tracing app; detailed the number of contacts identified compared with previous system | Sierra Leone | Ebola | Contact tracers and contact-tracing coordinators | 86 contact tracers, 26 contact-tracing coordinators | Partly automated | Manual data entry | Smartphone app (standalone) | Recall based | Other (paper-based system) |
| Li et al (2017)37 | Case study | Automated identification of contacts within an inpatient setting (lists generated based on user-defined parameters) | Singapore | Multiple, including influenza A | Hospital inpatients at Changi General Hospital; system used by infection control team | Not specified | Partly automated | Real-time integration of patient movement and laboratory data | Computer-based infection control management system | Shared room, concurrent contact, and duration of contact | Non-automated contact tracing |
| Schafer et al (2016)38 | Case study | App-supported contact tracing using Epi Info Viral Hemorrhagic Fever app; automation of tasks (eg, generation of daily follow-up lists); calculation of follow-up window; production of transmission chain diagrams | Seven African countries, two US states | Ebola | Used by contact tracers and contact-tracing coordinators in eight countries | Not specified | Partly automated | Manual data entry | Computer based | Recall based | Paper-based contact-tracing systems and Field Information Management System (programme developed by WHO) |
| Sacks et al (2015)39 | Case study | Introduction of, use of, and lessons learned from the CommCare contact-tracing system in Guinea (compared with previous paper-based and Excel-based systems) | Guinea | Ebola | Used by contact tracers | Used by 210 contact tracers (of 366 who were trained) to collectively monitor 9162 contacts | Partly automated | Manual data entry | Smartphone app (standalone) | Recall based | Paper-based contact-tracing system (used in parallel within Guinea by other contact tracers) |
| Tom-Aba et al (2015)40 | Case study | Use of Open Data Kit Collect app developed by use of Open Data Kit and FormHub to support contact tracing in Nigeria; automated alerting and SMS if any contact met the probable case definition; also refers to Ebola Sense app (supports follow-up of identified contacts, includes automated search functionality to assign contact tracers) | Nigeria | Ebola | Used by contact tracers | Not specified; all public health personnel carrying out contact tracing for Ebola cases | Partly automated | GPS (for accountability of contact tracers rather than directly within tracing efforts) | Smartphone (or tablet) app (standalone) | Recall based | Paper-based contact-tracing systems (data manually entered onto a single computer) |
| Automated contact detection in a relevant disease context (without subsequent contact tracing or contact notification) | |||||||||||
| Aiello et al (2016; iEpi substudy)41 | Substudy (descriptive observational study) within a cluster randomised controlled trial | iEpi substudy within a university-based trial; participants given smartphones that were used to detect other study devices and nearby Bluetooth-enabled devices to map proximity to contacts | USA | Influenza | Students aged ≥18 years | 103 (iEpi substudy) | Automated contact detection only | Bluetooth | Smartphone app (standalone) | Proximity based | None |
| Al Qathrady et al (2016)42 | Observational and modelling (simulation) study | Simulation of disease spread and contact tracing by use of a contact network recreated from WiFi traces; explored different approaches to prioritising investigation or follow-up of contacts at risk | Not specified | Hospital-based outbreaks | University faculty and students (contact data based on WiFi use on a university campus across six buildings) | WiFi trace data from 34 225 users of six university buildings | Automated contact detection only | WiFi network traces | Mobile devices (further detail not specified) | Proximity based | None |
| Voirin et al (2015)43 | Proof of concept observational study | Pilot study that combined micro-contact data from wearable proximity sensors and virological data to investigate influenza transmission within an elderly care unit of a hospital | France | Influenza A and B | Median ages in years were 24 (doctors), 30 (nurses), and 89 (patients); proportions of women were 67% (doctors), 78% (nurses), and 73% (patients) | Total 84 (32 nurses, 15 doctors, and 37 patients) | Automated contact detection only | Radio frequency identification | Radio frequency identification proximity sensors | Proximity based | None |
GPS=Global Positioning System. N/A=not applicable.