The contact tracing CovidCard wearables failure

Testing the $100 million CovidCard for contact tracing...without checking it worked first...

Contact tracing was critical to New Zealand’s implementation of, and maintaining, Covid zero aka the elimination strategy from early 2020.

Officials from the Ministry of Health - spurred on by famous entrepreneurs pursued wearable technologies, 1 of those - the CovidCard - was for contact tracing.

How CovidCard started

Initially spearheaded by Sam Morgan - he proposed a wearable Bluetooth enabled card that would record all close contacts so that contact tracing efforts were faster.

His key push for the cards was apparently an attempt at avoiding lockdowns and he thought any issues with uptake by the public would be swiftly put aside saying “…once their relatives start dropping dead they will be even more willing to wear it…”

In May of 2020 a week long trial was run supported by the Department of Internal Affairs (DIA) where 42 Nelson Hospital staff were given a Bluetooth enabled contact tracing card. The results were…okay. The card definitely found that users didn’t always recall via memory all their close contacts vs the cards.

Meanwhile on the 20th of May the government’s official Covid Tracer app was released. But use of the app was shown to fall when Alert Levels fell - when there was no community transmission - people scanned less. This was a key driver to Morgan to continue to push the card to government.

The card was seen as potentially useful as it meant people didn’t need to have a smart phone on them and then have to manually scan the QR codes. Officials called it reducing the ‘hassle’ of signing in but really it tracked and signalled a person’s movement 24/7.

Covid Tracer app scanning was studied alot. Later in 2020 a big 4 consultancy, PwC, cropped up to do research into “behavioural barriers that prevent people from consistently using the App to scan” and noted that behavioral science indicates people need to make the habit automatic. (I have not found the actual research only references to it so I suspect it has not been proactively released.)

The Research Agency (TRA) had already been commissioned to study attitudes to Covid restrictions by DPMC and a later study - from May 2021 - showed the low scanning reasons to be:

Yes. People couldn’t grasp why they needed to scan to prevent Covid - when there was no Covid.

Complaining about “dilly-dallying” on the card Morgan and a fellow pusher of the card, Animation Research founder Ian Taylor appeared to think a mandatory requirement to wear the card could overcome the drop in app scans.

The cost of the card to procure, distribute and advertise 6 million cards was roughly estimated as well. It looks like cards could be assigned to people with everyone’s personal details ‘pre-stored’ on the cards from this Bluetooth presentation in an OIA on the Covid Tracer app (I’ll be covering the app in a seperate post too).

Sustainability clearly went out the window when it came to Covid - all those masks littering the streets, single use PPE gear - they wanted to add discarded Bluetooth cards quietly emitting into the ether too!

In July 2020 then Health Minister Chris Hipkins and Finance Minister Grant Robertson were asked to sign off on a $1 million trial to improve contact tracing through the use of the CovidCard.

The CovidCard was led by Kris Faafoi, then Minister for Government Digital Services who considered “ …the CovidCard should be investigated further as part of looking at multiple technologies to optimise contact tracing capability.”

The money was designed to set up a community-based trial and to fund further research into public sentiment - the first trial showed the only adequate way for the cards to work was to wear them on a lanyard around your neck out in the open. Discretely hidden in your pocket didn’t work. The cards were also non-rechargeable (estimates thought the battery could last a year although this was disputed).

At the same time the CovidCard trial was pushing ahead the Privacy Commissioner raised many practical concerns around potentially mandating people to wear CovidCards.

In response to criticisms then Minister Hipkins stated “The CovidCard won’t store location data or track users. It simply blindly records the length and distance of the interaction with others who have cards who are up to five metres away…”. Wait but the presentation said….nevermind….lets move on.

The Productivity Commission also raised concerns calling the Covid Tracer app a failure and that the proposed use of the CovidCard:

..share some characteristics: both are “skunkworks” projects, developed out of the public eye and reflect deep (and untested) assumptions about what the public will, and will not, accept. And both appear to embody a “New Zealand should go it alone” attitude that, in the midst of a global pandemic, may be unhelpful.

This is an interesting criticism I haven’t seen raised elsewhere - although at the time the amount of behavioral research and social media listening the government was doing may not have been publicly known.

Cabinet reviewed the card on 3 August 2020 and the trial was signed off.

The Ministry of Health and the Government Chief Digital Officer (GCDO) in DIA established the Contact Tracing Technologies Prototype Research Programme to do discovery and run the trial.

However by the time the trial had start Sam Morgan had walked away from advocating for the card in disgust that it had become “buried among people who are useless, deep inside ministries who just can’t deliver this.”

Rotorua trial

The University of Otago led the study “…examining the efficacy and acceptability of the card among a community in Ngongotahā, Rotorua” - ie. if what people said (sure, they’d wear it) matched how they acted (they actually did wear it).

The location appeared chosen to fufill the study’s “guiding principle of commitment to upholding the Te Tiriti O Waitangi and the alignment to key elements of Whakamaua (Maori Health Action Plan)”.

1,191 people received 1 card each. 35% of people didn’t use their CovidCard at all.

The study estimated that when used correctly the cards could identify 55% of close contacts. As long as they were wearing the cards that is.

Overall many people used it, as would be expected when you’re part of a short trial, but even so there was less usage among younger trial participants and people expressed privacy concerns.

In an exit survey the majority of people who responded were happy to wear a Blueetooth enabled card but they preferred to wear it when there were active Covid cases in the community rather than in the existing Covid zero environment.

The privacy concerns were valid - the CovidCard did not even have a decentralised and anonymous data model (unlike the Covid Tracer App which did) and how would the cards be connected to people? They’d have to store some personal data to link it to you.

Remember this data was going to be used for contact tracing.

Contact tracers at the time were made up of staff from public health units and from a central unit based out of the Ministry of Health.

The CovidCard trial had gone ahead without adequately addressing the issue of how CovidCard contacts would be fed into existing processes and technology in use by contact tracers.

Contact tracers feedback was the technology would not replace any of their processes in contact tracing and could increase their workload if used (as it sent all contacts to be sorted through rather than allowing contact tracers to assess which contacts they needed to gather from a person).

At the same time a technical review done by the Defence Technology Agency (DTA) identified the same issues. The interoperability of the card was dismal.

Alongside the trial (and issues with uptake of the Covid Tracer app) Colmar Brunton also heard the sweet sound of invoices to government as they were commissioned to report on New Zealander’s attitudes to contact tracing.

At the conclusion of the trial the University of Waikato reported on the co-design of the community trial and rated the success against an evaluation framework.

The end of the trial

The final report on the Rotorua trial noted ”…however it has been the approach rather than the deliverables that have been its greatest success.” While they puffed up with pride over engagement with the community - the actual outcome was a failure
”…this is because mandated use of the Card by all New Zealanders would be required for it to be effective, which would likely erode our current social licence, and it is not interoperable with other digital technologies. Additionally, the cost of rolling out the Card outweighs the contact tracing benefits we would gain.”

How much would the cost be?

$100 million.

Not only that - but it was surprising they did a trial for $1 million - because an earlier April 2020 Cabinet paper clearly stated “Technology choices must be guided by the principles of: public health efficacy, respect for privacy, freedom of movement, and technical feasibility.”

If the CovidCard was never technically feasible for use with contact tracers and their systems - which also means it couldn’t meet the public health efficacy principle either - then it did not meet those principles prior to the trial. Why did they not simply check within their own Ministry - before the trial - if it could even be used?

Just to contrast this - the Covid Tracer app and associated work with QR posters cost $6.5 million in 2020 to set up and maintain - which was integrated with the system contact tracers used.

In a marae based announcement the following year that the CovidCard wouldn’t go ahead Ministry of Health Deputy Director General for Data and Digital Shayne Hunter said while they do worry about their ability to scale for contact tracing “…the Ministry's main focus was on vaccinations.”

A verbal report to Cabinet was provided on the 7th of December 2020 - officials did not recommend pursuing it further.

Bizarrely in early 2021 the co-founder of the Bluetooth card used in the CovidCard trial was stating they could roll out cards in as little as 2 weeks and simply needed to work on the “finer details” of interoperability and that the trials were a “huge success.”