The use(lessness) of contact-tracing apps

No one is safe if not everyone is safe

COVID-19 is caused by the SARS-CoV-2 virus that spreads between people via droplets and contact transmission. In order to control the spread of COVID-19 — in the absence of immunity/an effective vaccine — interventions are needed to break the transmission chain so that the R number (effective reproduction number) remains less than 1. As part of an all-encompassing strategy, identification, isolation, testing and provision of cases, contact tracing and quarantine are among the actions critical to reduce transmission. (33) With the ‘Test and Trace’ pilot, Kendall et al. demonstrated that this strategy resulted in a smooth decrease in secondary infections — also for COVID-19 — and as such, a curtailment of the scope of the first wave of the pandemic. (34) These kinds of trial projects cannot be unreservedly translated to a different location, however. This is because the success of test and trace is influenced by a variety of factors, such as social involvement and culture, training and availability of personnel, logistical support and the availability of real-time data with expertise on hand for interpretation and reporting.

Pandemic preparedness

The speed of intervention is crucial (353) as also emerged in practice. For instance, in Taiwan (36) — supported by trained personnel and with the outbreak of SARS-CoV in 2003 still fresh in the collective memory (37) — every newly diagnosed case was rapidly isolated.

And yet we also saw countries outside of Asia that managed to smoothly get the virus under control. The MERS-CoV epidemic helped Saudi Arabia with a more alert public health system, infection control policy and measures (a number of extreme interventions were already taken in mobility, social and religious gatherings, travel and business even before the country’s first COVID-19 case). (38) But also Senegal, where experience with Ebola had resulted in a blueprint for combating disease via mobile testing locations and also widespread public trust in the government. Translating scientific and public health expertise into administrative policy quickly and clearly proved crucial. Considerations at the expense of combating the virus cost dearly: in victim numbers, more serious economic damage as a result of lockdown, and a decline in public trust in policymakers.

How do we place the gains?

Apps—mobile applications— could be seen as part of the larger whole of (digital) tracing, but the functionality and underlying technology does differ. This can include GPS tracking, mobile telephone data, Bluetooth, combination with other applications such as credit card data, selfies at locations. The objective of a contacttracing app is to (help) reduce the R number so that outbreaks can be prevented, while simultaneously limiting the number of people in quarantine. Reducing the time between isolation of the case (including pre-symptomatic contagious phase) and quarantine of the contact people by gains in tracing speed is the added value of a contact-tracing app.39 It can also serve as a supplement to the reach of the traditional (human form of) contact investigation, because even people who are not acquaintances of an infected person can be detected. In this way, the app is seen as a solution for lifting and preventing lockdown, in order to allow social life to resume (in part), with the reopening of public transport and public spaces, for instance. More fine-tuning of policy could also be possible through targeted insight on the local level, perhaps to be considered a form of precision epidemiology. This could then be generated by combining these app data with, for instance, data on changes in symptomatology, behaviour and other factors. Other advantages of digital contact tracing are, among other things, the low costs and its scalability, and the added value as assistance in guiding the deployment of costly resources to combat the virus.

Models

Simulations by Hinch et al. assume that 80% repression of epidemic is possible if 56% of the population is willing to participate in app-based tracing. (40) In an ideal setting (see infrastructure considerations), this should result in 36% of infections being identified. The percentage based on the voluntary download of the CoronaMelder app was at just over 4.36million downloads as of 29 December 2020. Approximately 25% of 17.5 million residents, this amounts to approximately an extra 6% of infections being identified with the help of digital tracing under the ideal conditions. (41) This number corresponds to the yield, based on the data available to date, from which around 4% of positive tests were traced with the CoronaMelder app among asymptomatic people.

In November 2020, the same group published in cooperation with Google Research, based on a model representative for the state of Washington, that 15% app use would result in 15% COVID-19 infection reduction if combined with the traditional tracing (in the most optimal utilisation) and in 8% infection reduction in the event of only digital tracing. (42) Based on Dignum et al., it emerged on the basis of unique simulations that included both an epidemiological model and behavioural effects that an app in a situation of even 60% adoption made no significant contribution to combating the virus. (43)

Not a stand-alone intervention

With all efforts to develop a reliable, well-functioning and secure system for digital tracing in the event of a pandemic, we must realise that the technology must be embedded throughout the testing and tracing system. The following matters must also be taken into account:

1. App reliability related to:
   1. ‘Testing’: proportion of people who test positive who are also reported via the app, the availability of testing at a reasonable distance, time between test and result, infection rate among the population (the higher the rate, the less reliable).
   2. Functionality: how reliably are contacts identified using the app? How seriously do people follow the instructions?
2. Adoption rate in the population (number of downloads and actual use).
3. Performance of the underlying technology, including:
   1. Overwhelming in the event of false alarms
   2. Privacy concerns
   3. Communication (broader than substantive, including things such as user experience, personalised notification, ongoing analysis for improvement and the use of specialised human interaction) of reports via the appSecondary to underperformance, behavioural change is another lurking
      factor.Which results not only in reduced adoption of the technology, but also
      a decrease in compliance with measures to combat the virus.
4. More widespread embedding in traditional contact tracing and government advice. A limited understanding of government measures in relation to lockdown and the corresponding effects on the decrease in COVID-19 emerged to be associated with limited willingness to participate in app-based contact tracing. (44) Other factors include to what extent and how guidance by the specialised team is provided via the app, whether people can contact someone if they have questions, how this is responded to, whether account is taken of and assistance provided in relation to quarantine advice, such as finding accommodation, arranging aftercare, and so forth.
5. Cultural embedding, public expectations and perception of digital tracing. A contact-tracing app could have more impact in a more totalitarian regime than the Netherlands, where it is voluntary. Is compliance on the part of the population affected by rapidly changing measures in relation to social distancing, the indirect effects of economic disruption or mental health deterioration and other indirect health consequences such as the deferment of oncological care? (45)
6. Potential long-term effects of the digital tracing technology in society. A collaboration between Google and Apple in relation to Bluetooth technology, as used for the CoronaMelder app, could be conducive for numerous other solutions — desirable or otherwise — at a later stage. (46)

Conclusions

Without pharmaceutical intervention, the success of combating infection lies mainly in a rapid and resolute response from the healthcare authorities and implementing supervisory services with a focused testing strategy. The contribution of digital tracing can be one component of that, but decidedly not one that is separate from the overall system of measures and factors. Based upon a few practical figures from the past months and epidemiological models, the CoronaMelder and similar apps make at best a very limited contribution, with a percentage of between 4 and 8% in terms of extra infections traced.

Even though this percentage is low, it could indeed be a supplement to the traditional testing and tracing, because this method enables the tracing of persons who may potentially become infected (for example in public transport) without being someone the COVID-19-positive person knows. Traditional contact tracing, which assumes specific knowledge and skill, is in the foreground. The personal approach is crucial to ensure that measures are and can be complied with. Based on the presumption that the population takes responsibility for health (their own health and that of others), an app can also yield time gains (by omitting the intermediary, such as a GGD).

In conclusion, the use or uselessness of an app such as the CoronaMelder, whereby privacy is safeguarded so that there is no automatic insight from the traditional test and trace system, depends entirely on the behaviour of the population. As such, the contribution of the app depends on various culturally related aspects, which was confirmed in the studies over the past year.

Because technology is anything but neutral in use and deployment, it is essential to also anticipate the least optimistic flip-side. Here one could consider, for instance, the possibility that a technical solution specially developed for this purpose might be used for other purposes, or that individuals may see use of the app as a sort of licence for less strict compliance with behavioural recommendations, born from a false sense of security.

Assessing the contribution of measures in combating the virus is extremely complex, since effects can also change over time. There is currently no scientific validation in randomised research of an app’s contribution, and it is therefore questionable whether this might not also be too complex and costly. What the app without any scientific evidence has indeed brought about is a discussion about fundamental values in our society, about how we relate to technology and our fellow man. Awareness needs to be raised in relation to the fact that technology is not a ‘no strings attached’ instrument. This applies for the use of technology, but also for coordination ahead of any implementation. Constant steering based on our fundamental human values, based on a multidisciplinary approach, is crucial.

Analyses

• An app as a stand-alone solution can, with reference to the analysis above, be placed in the category of tech solutionism or, as the case may be, useless. Focusing on an app that protects privacy, with the inclusion of multidisciplinary experts later in the process, is an undesirable situation because of the diversity of factors that influence the overall process of combating the virus, but also in particular related to the app. In this last case, it involves, for instance, cyber-physical interaction and the behavioural effects related to this.
• From the formulation of the terms of reference through to selection, continued development and possible replacement of an app as part of digital tracing, the starting point must be that multidisciplinary expertise must be represented with a transparent contribution, supported on a scientific basis, with an eye for any pitfalls or concerns that must be taken into account.
• Such an innocent app — even based on an open process — can cause effects that can manifest much more broadly in society. With all the implications this has for combating the virus, but also for numerous other domains in our society.
• Early acknowledgement or recognition of a crisis, daily public briefings and simple unequivocal messages in relation to health, and a transparent and respectful attitude towards scientific and medical/public health expertise on the part of policymakers deserve explicit attention. With a side note that for the last groups mentioned, training and guidance in (social media) communication could still use a significant boost for the benefit of public trust.