Health New Zealand believes the technology behind its failure-prone cervical screening register is largely OK - but not entirely.

The register that was rushed in three years ago failed to trigger almost a million notifications to women about cancer screening.

Experts who investigated the problems last year told HNZ in February to look at whether the register's IT platform was viable and, if not, find other tech.

The agency's action plan says it will both assess and enhance the tech.

HNZ told RNZ it began that work in June.

"We are confident that the programme's IT system is generally fit for purpose, although we recognise there is work to do to ensure that invitations to screen are generated for all women that should get them," said national prevention screening director Alana Ewe-Snow.

"In short, we anticipate improvement to the current IT system, not an overhaul."

All 48 recommendations, including the tech to improve the system, would be completed within two years and it was monitoring for any harm meantime, HNZ said. No harm had been found.

Women's health advocates said the register failure was "appalling" and workforce churn, including at the tech level, had contributed to it.

As a result of the national screening unit being shut down as a standalone agency, data analysts had been shifted to HNZ's digital services side, helping create "exactly the kind of workforce churn that our organisations warned would be a recipe for disaster", said the Auckland Women's Health Council, Federation of Women's Health Councils and the Cartwright Collective in a statement.

The register was built to store cervical screening test results, track screening paths - how often women should be screened, according to risk - and send out notifications, including recalls, reminders and invitations to women.

Despite the failures, it had boosted screening numbers from just under 10 percent in 2023 to 77 percent in 2026, Ewe-Snow said.

"The register is working well for most women, so there is no intention to start again.

"We know, however, we have more work to do tighten the register to ensure invitations are getting sent to all eligible women."

Its launch in September 2023 was rushed, and HNZ knew it had limited functions and several problems, including incomplete data, yet did not monitor it properly after it went live, the expert investigation report's 118 pages found.

It was sending out only a couple of hundred notifications a week, instead of the expected few thousand.

By April 2024, an internal report showed the register's design was not up to expectations and the tech's 'logic' system "was not built as intended". Poor data quality made it worse.

That sparked the external review, done by mid-2025, although its report was released only this month, after reporting by Stuff.

"We are deeply concerned by reports that nurses and other frontline staff have said they tried to raise the alarm about these failings before the system went live," said the women's health advocates.

Initially, the project contract was signed in 2019, with Deloitte using software from Salesforce, although actual work began in 2022. By early 2023, it still needed funding.

The goalposts on technical requirements were shifted, as the screening platform was being built.

"The register is not yet functioning as a comprehensive clinical safety net," the review said in 2025, after 18 months of work trying to fix it.

It was, however, working "satisfactorily in many ways".

The advocates noted the failures included no screening histories properly transitioned onto the new register; and a "concerning level" of clinical risk around recommendation mismatches between screening laboratories and the register.

The experts gave Health NZ a list of five things to do on the tech front.

It "must critically assess the long-term viability of the register and the underlying National Screening Solution [NSS]", they said in February.

It needed to know from the IT supplier if and when this could be done, and at what cost - and if not, change tack.

"If this assessment determines that the required functionality cannot be implemented, a risk-assessment should be undertaken of procuring a new software solution in the long term, rather than continuing with the register and NSS."

Ewe-Snow said they began that in June, "assessing how the technology that sits in behind the register can be improved", and this was ongoing, although a timeline was not yet finalised.

"We need to be alert to opportunities for further improvement, especially in a world where technology is changing rapidly."

Meantime, primary care providers would continue to send out invitations to women for screening, as usual.