Only one in twenty-five Europeans infected report this through a COVID-19 app

The Dutch CoronaMelder app and its equivalents in other countries are used in around one in twenty-five COVID-19 cases. This has been shown by research conducted by journalists from four European media outlets led by Dutch research collective The Investigative Desk.

The journalists collected information about the use of 23 COVID-19 apps. The study shows that just over a million people in Europe used a ‘contact tracing app’ to warn others of a possible COVID-19 infection in a period in which almost 25 million people became infected with the coronavirus.

Dutch people used the app slightly more often than average to report an infection: between 1 December 2020 and mid-May 2021, more than 125,000 people did so. That amounts to 12% of all COVID-19 infections in the country. According to the Dutch Ministry of Health, Welfare and Sport, the CoronaMelder app therefore ‘strongly contributes’ to the regular source and contact tracing.

In most European countries, it is not possible to determine how many app users subsequently received a warning. It is also not known what people do when they receive a warning from the app: do they self-quarantine, do they report for a test? Data availability appears to be extremely limited and serious evaluations are not yet available.

Journalists from Die Zeit (Germany), VRT NWS (Belgium), Le Monde (France) and The Investigative Desk (the Netherlands) collected as much data as possible in recent weeks to provide insight into the use of the COVID-19 apps.

Low number of ‘active’ users

In countries where such an app is available, on average a quarter (24%) of citizens have downloaded it, but there are big differences between countries. Ireland (51%), Finland (41%) and Denmark (39%) have the most popular apps, while less than 2% of citizens in Croatia, Hungary and Cyprus downloaded a COVID-19 app.

The Dutch app was downloaded by 28% of the population and is thus in eighth place, just above France and Belgium (both 24%).

The number of people actually using the apps may be considerably lower. For the apps to work properly, they must not only be installed on the phone, but also set up correctly. The user must activate Bluetooth and ‘authorise’ the app to contact phones in the vicinity. Portugal, Ireland and Switzerland monitor remotely the extent to which this is the case. The number of ‘active users’ in these countries lies between 28% and 58% of the number of downloaders. It is not known how many ‘active users’ there are in other countries.

Lack of information

The fact that so little is known about the use and effectiveness of COVID-19 apps is partly due to design choices. COVID-19 apps have been introduced by governments to combat the spread of the virus. Shortly after the first wave, governments felt an enormous need to develop this type of app, says Aymeril Hoang, a member of the French National Committee for Scientific Research. He is involved in the introduction of the French corona app at the suggestion of French Minister of State for the Digital Transition and Electronic Communication Cédric O. ‘Given the huge health crisis, we simply had no choice but to try it.’

Initially, the French government also wanted to use the app to create epidemiological models, for example to map COVID-19 hotspots, but the public debate immediately turned to privacy concerns. ‘We could therefore only collect very limited information, which completely undermined the functionality of the system.’ When asked for a reaction, governments in other countries also mention ‘privacy by design’ as an important reason why there is so little information available about the use and effectiveness of the app.

The same situation applies in the Netherlands where only very limited data is collected that can provide insight into the use of the app, says former journalist and security expert Brenno de Winter, who was involved in the development of the CoronaMelder app. ‘I had all data removed that could possibly violate people’s privacy.’ De Winter wanted to prevent unrest among users as a result of statistics on the number of reports received.

There are, however, scientific articles that use mathematical models to calculate how many infections the apps may have prevented. In mid-May, for example, Oxford University published a study on the impact of the NHS COVID-19 app for England and Wales. It could possibly have prevented 914,000 COVID-19 infections.

It is also difficult to compare the effectiveness of the various apps, says Professor Wolfgang Ebbers of Erasmus University Rotterdam in the Netherlands. He was asked by the Dutch government to evaluate the CoronaMelder app.  ‘In each country, the app and source and contact tracing are set up differently.’ Because of the central role that the GGD Municipal Health Services play in the Netherlands, a relatively large amount of information is available there. For example, figures are available in the Netherlands on the number of people who get tested after being notified through the app: 158,000 people did so and 11,000 of them turned out to have COVID-19. The only other country to publish data on this is Denmark.

‘This shows that the apps do have some effect,’ says Belgian lung specialist and epidemiologist Wouter Arrazola de Oñate, who advises the Belgian government on COVID-19 policy. The apps at least identify a number of infected people. ‘But there is still not enough data available to really assess whether they are successful or not.’

Privacy used as a smokescreen

The call for accessible information is growing louder among scientists. According to Natali Helberger, professor of Law and Digital Technology at the University of Amsterdam, we should see contact tracing apps as an extension of government policy. ‘First we had source and contact tracing by the GGD Municipal Health Services, then we got digital apps.’ Such digital policies should be subject to the same rules as all other government policies. ‘The fact that this is not the case here creates problems, because it is now almost impossible to say whether those apps are good policy.’

App developers can gain insight into the current use of their apps through the Google and Apple app stores. Google confirmed to the journalists that this also applies to the developers of COVID-19 apps. This includes basic information such as the number of people who use the app on a daily basis and the number of people who have removed the app from their phones. The journalistic investigation shows that governments ignore this information. ‘We very deliberately decided not to look into it,’ says spokesman Frerick Althof of the Dutch Ministry of Health, Welfare and Sport. ‘We want to keep it as privacy-friendly as possible.’ The German and Belgian authorities gave similar answers. Bart Preneel, for example, who co-developed the Belgian app: ‘We decided not to look at the usage data to protect people’s privacy.’

Helberger: ‘That is really a remarkable argument. They use privacy as a smokescreen. I don’t see how users’ privacy can be violated if you look up in the app store how many people have deleted the app.’

Ebbers promises that a lot more information about the Dutch CoronaMelder app will be released soon. He is working on the ‘overall evaluation’, which will be published on 1 June at the latest. Ebbers will then also announce how many active users the Dutch app has. Brenno de Winter has devised a privacy-friendly method of calculating this.

As far as the Dutch Ministry of Health, Welfare and Sport is concerned, the CoronaMelder app will certainly not go away for the time being. According to spokesman Frerick Althof, the added value during the lockdown was limited. But now that the lockdown in the Netherlands has been lifted, the number of social contacts will increase. And that means the importance of the CoronaMelder app will increase again.

This investigation was carried out by journalists Dorien Vanmeldert, Tim Verheyden, Bart Aerts (VRT, Belgium), Markus Sehl (Freelance, Die Zeit, Germany), Simon Auffret (Le Monde, France), Manon Dillen and Daan Marselis (The Investigative Desk, the Netherlands).

A shortened version of this article was published at NOS. Other publications about this investigation appeared in VRT NWS (Belgium), Die Zeit (Germany) and Le Monde (France).