Can Australia fix its CovidSafe app and turn the pandemic into a ‘pingdemic’?

<span>Photograph: Dave Hunt/AAP</span>
Photograph: Dave Hunt/AAP

The federal government is in discussions to tweak its CovidSafe app to pick up more fleeting contacts due to the more infectious Delta variant, but is refusing to adopt technology used overseas that could speed up notification of close contacts.

As England and Wales experience what is being called a “pingdemic” – with more than 600,000 isolation alerts sent to users of its NHS app in one week earlier this month – Australia’s CovidSafe app has managed to identify just 17 close contacts who were not picked up by other means since it launched in April last year.

The recent outbreak of the Delta variant in Victoria resulted in more than 40,000 primary and secondary close contacts being forced into isolation after being identified largely through the state’s QR code check-in app managed by Service Victoria.

Related: Will Sydney’s Covid lockdown work and how different are restrictions to Melbourne’s ‘ring of steel’?

There were thousands of close contacts identified at large exposure sites at AAMI Park and the MCG, where people who did not know each other passed on the virus to one another. The massive contact tracing effort helped to bring the latest outbreak under control, but the $7m CovidSafe contact tracing app again played little role in finding close contacts.

“I would stand to be corrected but I don’t think we have had any cases in Victoria, if we had it would be an extremely small number that are picked up by the app,” the Victorian premier, Daniel Andrews, said when lockdown ended on Tuesday.

According to the Digital Transformation Agency, which built the app, since launching in April last year it has detected only 17 close contacts in New South Wales who were found directly through the app and were not otherwise identified through manual contact tracing methods. The agency bumps this figure up to 561 by including an additional 544 contacts linked to an exposure site at Mounties, a club in Sydney’s south-west after the CovidSafe app data revealed a positive case had been at the venue on a day not previously listed as an exposure date. However, rather than the contacts being identified using the app, tracers reviewed check-ins for that day.

A study published in Nature in May about how effective the NHS app in England and Wales had been between September and December last year found for every positive case who agreed to alert their contacts, one case of Covid-19 was averted.

More cases could be avoided while the UK vaccination rollout was under way if more people downloaded the app, the study suggested.

About 28% of the population (or 16.5 million people) were using the app at the time, and more than 1.7m notifications were sent to users in the four months, or 4.2 contacts per positive case.

The case numbers in the UK are on an entirely different scale to Australia’s, with more than 27,000 reported in the last day this week alone, compared with 256 on Thursday in Australia. But with more than 450 cases with unknown sources in NSW this week alone, a contact tracing app could prove useful in connecting cases where QR code check-in history and case interviews fail. It could also speed up notifying contacts to immediately isolate to reduce the period they are infectious in the community.

But Australian authorities have consistently ruled out moving to the same Apple/Google exposure notification framework used to develop the NHS app for CovidSafe.

The CovidSafe app uses Bluetooth wireless technology to register when someone with the app spends more than 15 minutes or more within 1.5m of someone else who also has the app.

A spokeswoman for the federal health department told Guardian Australia this week the position had not changed because, when the app pings a user, it puts the onus on that user to inform contact tracers, rather than giving contact tracers a list of close contacts.

“The Apple and Google design does not allow for tracing of the source of outbreaks or the follow-up function that contact tracers perform in order to ensure close contacts get tested and self-isolate if they have been exposed,” the spokesperson said. “This is crucial to quickly responding to and managing outbreaks.”

This claim has been disputed by tech experts, who say the code could easily be modified to ask people to allow contact tracers to contact them.

The department spokeswoman confirmed changes were being considered so that the app could record more fleeting interactions in response to the Delta variant, rather than the 15-minute threshold.

“The parameters were based off the medical advice. The government is undertaking consultation with the states and territories regarding amendments to the parameters of the app,” she said.

‘“If the medical experts believe that a shorter timeframe for a close contact should be considered, the government will look at that in consultation with the states and territories.”

But software engineer and one of the main critics of the Covidsafe app, Jim Mussared, said it would be difficult to measure whether any changes would make a difference because the government has not been transparent about the efficacy of the app.

Related: From ‘it’s not a race’ to ‘go for gold’: how Scott Morrison pivoted on Australia’s Covid vaccine rollout

He said it was bizarre to argue the app needed to be part of the centralised contact tracing system, when the app could function separately to get people to isolate.

“With the current outbreaks in Sydney and Melbourne, the two numbers that people care about are ‘total cases’ and ‘number already in isolation’ – the Google Apple exposure notification framework gets people into isolation automatically and quickly.”

The CovidSafe app continues to cost around $75,000 per month to run. More than 7m people downloaded the app last year.

The federal government has long been reluctant to admit the app has not lived up to expectations. When it launched in April last year, the prime minister said it was the key to getting out of lockdowns, and compared it to putting on sunscreen.

A long-awaited government report on the operation and effectiveness of the app released this week revealed the government now considers it “complementary to other tools such as QR code apps” and argues it has rarely been needed because of the “the relatively low number of cases in Australia and effectiveness of our contact tracing processes”.

“With the success of the suppression strategy, only 0.03% of the Australian population were infected at the peak of the pandemic with around 8,170 active cases on both 11 and 13 August 2020 (including those in hotel quarantine),” the report stated. “This has meant that the app’s ability to identify unique close contacts has not been called upon to a significant extent.”

Between April and November last year, there were 735 users who agreed to upload their data, with 2,579 potential close contacts identified, with almost all either excluded or already found through manual contact tracing.

This dropped to just 44 people uploading their records between November 2020 and May this year, and 1,729 potential close contacts.

The report said some close contacts identified may not be followed up for a variety of reasons, including because they might have been in contact with the person outside their infectious period.

The report doesn’t state how many of the more than 7m users are still using the app, and the DTA has refused freedom of information requests to reveal this information.

NSW Health reported in December last year just 21% of people who acquired Covid-19 in the community in the state had been using the app. A Victorian parliament committee report also found there was no evidence the app had been effective after the second lockdown.

In the meantime, the states have pushed mandatory QR code check-in apps as their best form of contact tracing. Just prior to Victoria’s fifth lockdown, on 5 July, the state recorded over 5.6m check-ins in a single day. The Service Victoria app will soon be updated to allow people to review their check-in history to compare against exposure site alerts.