In their analysis, looking at coverage data from March to April, the study authors also defined risk based on factors such as age and the number of vaccinations people had received.
According to HSU data, there were approximately 718,948 unvaccinated residents, of whom 76 per cent were younger than 12 years.
While the roll-out achieved relatively high rates of coverage across most cohorts, their results showed the denominator mattered.
For instance, if Census data was used instead of HSU, the proportion of vaccinated Pacific females older than 25 would have risen by 6.3 per cent.
But at the same time, and more worryingly, the proportion of vaccinated Māori over 25 would have fallen by five per cent for females – and 12 per cent for males.
The study authors further pointed out that younger and marginalised populations were likely under-represented by HSU as they didn't frequently use health services - leading to over-estimates of coverage in these groups.
"However, if we use census data we may underestimate subpopulation sizes of Pacific peoples, which could lead to underestimates of unvaccinated, highlighting the difficulties in using census data to estimate accurate subpopulation counts in dynamic, migrant populations."
They concluded that, while true disparities in vaccinations remained unclear, ethnic minority groups should continue to be reached through targeted engagement.
"This is further evidence that the current way we do health statistics in this country aren't robust enough for an epidemic – and they aren't robust enough to inform timely responses to an epidemic," co-author and University of Auckland statistician Andrew Sporle said.
"We've known that for a while, but what's new about this particular study is it actually identifies the number of people at risk of a preventable, poor outcome from Covid.
"That's particularly important for South Auckland, where we're talking about priority populations which have been predicted – and demonstrated – to have worse outcomes."
The study, published in medical journal JAMA Network Open overnight, follows a Herald investigation into the issue – with experts suggesting inflated vaccination rates may have even led to higher rates of death and illness.
Some health statisticians told the Herald that the HSU – which Stats NZ has been reviewing the ministry's use of - may have undercounting Māori by as much as 20 per cent in critical age bands.
A ministry spokesperson acknowledged the new study – and also noted that any data set had limitations, "which has always been acknowledged by the ministry in the publication of its HSU data".
"This is why it is used in conjunction with a comprehensive range of sources to ensure the data is as robust as it possibly can be."
The ministry had continued to work to refine and improve the quality of its HSU data as the rollout had progressed, with Stats NZ's report soon to be published.
"Work is also underway to refresh our HSU dataset using the completed 2021 data, and we expect this to be published In line with our agreed HSU release cycle anticipated mid-year."
The spokesperson said we could still have confidence that "a very high proportion" of Kiwis were now vaccinated against Covid-19.
"This is because of the huge amount of work that has occurred across the motu by DHBS, health providers, the ministry and many others to vaccinate as many people as possible."
