Babies close to death are less likely to get life saving treatment if they're Māori, Pacific or Indian - and experts partly blame racial bias.
A Weekend Herald investigation can reveal the ethnic divide in resuscitation attempts on very premature infants.
A top-level health body is now calling for all maternity and neonatal workers in New Zealand to be put through compulsory anti-racism training.
The number of infants born near the "edge of viability" - 23 to 26 weeks - is small at about 170 a year. However, 10 years of records reveal a disturbing ethnic divide.
Resuscitation was tried on 92 per cent of Māori babies, 89 per cent of Pacific and 86 per cent of Indian.
That compared to 95 per cent for "other" - mostly Pākehā and non-Indian Asians - which medical experts say is a statistically significant difference.
"Institutional bias or implicit biases are likely to play at least some part," concluded the Perinatal and Maternal Mortality Review Committee, a taxpayer-funded panel tasked with reviewing deaths of babies and mothers.
• Major differences in survival rates across neonatal intensive care units. Some don't usually resuscitate at 23 weeks while others will try in almost every case.
• At least one family moved cities because of such regional differences. Their daughter was born at 23 weeks and survived.
• A ground-breaking study has observed ethnic bias among final-year medical students at universities of Auckland and Otago.
• Auckland DHB has brought in policy to fast-track all eligible Māori and Pacific job applicants straight to interview.
Extremely premature infants cannot survive without resuscitation. Doctors always resuscitate from 25 weeks, but never before 23 weeks gestation.
The in-between "grey zone" presents an agonising decision for parents, made after counsel from the neonatal paediatrician and obstetrician.
Chances of survival can be slim. Babies that live can have severe disabilities including cerebral palsy and blindness.
Keri Thompson, whose son was born at Waikato Hospital at just 24 weeks and is now at law school, said she and her whanau felt "like aliens" at times.
One doctor bluntly told her that her baby had less chance of making it because he was Māori, she said.
Thompson later interviewed others about their experience for a thesis, and said relatively small actions could demoralise or lift up patients. For example, strict visitor policies which did not recognise that whanau went beyond the nuclear family.
"You are away from your own family. You may not have a lot of money and have to come from out of town. And you add the cultural differences - missing out on karakia, all that sort of stuff that would naturally happen."
The mortality review committee analysed 10 years of data for babies born from 23-26 weeks.
The committee's reports go to the Health Minister and DHBs. It wants workforce training to "reduce and minimise the impact of implicit bias and racism".
Dr John Tait, chairman of the mortality review committee and chief medical officer at Capital & Coast DHB, said no baby would not be resuscitated because of the colour of its skin.
Rather, bias could be found earlier in the chain of events leading up to the birth, and help explain why more Māori, Pacific and Indian women have babies whose condition, such as low birth weight, makes resuscitation less viable.
"Is enough effort going into Māori and Pasifika health to prevent them going into labour early? If people don't come to antenatal clinics...should we be going to them?"
Asked if doctors' biases could affect discussions with parents about whether to try keep a baby alive, Tait said that was possible - and a reason why his committee wants cultural competency training.
"We are still looking through the lens of Pākehā. Whereas the lens of Māori may be quite different."
Figures obtained under the Official Information Act show Māori, Pacific and Indian families take the burden of extremely premature birth.
They accounted for 85 per cent of births at 23 weeks in Counties Manukau, for example, with fewer than one-in-five surviving.
At Wellington Hospital NICU, which has a more active approach to resuscitation, over 40 per cent of Māori, Pacific and Indian babies born at 23 weeks survived.
Tia-Jane McVeigh was one of those survivors, and the country's smallest and most premature baby to live.
She was born in 2015 and just hours after the 23 week mark, weighing 516 grams and with eyes fused shut and skin translucent.
Her mother Alysha said Wellington NICU staff were wonderful during her tumultuous 137 day stay.
Clear communication was essential, and age came into that as much as culture, she said. Aly had her son at 19, and Tia-Jane 20 years later.
"My maturity level was a lot different ... my entire focus and time was on Tia-Jane.
"It's about support networks. I saw a few young ones in there who had a lot of family around, almost every day. And you can see the outcome for those families was different."
A working group is now considering aligning practice across the country, including resuscitation from 23 weeks.
DHBs told the Weekend Herald it wasn't fair to compare survival statistics, which aren't adjusted for factors like poverty or the proportion of mothers who arrive with birth imminent.
Dr John Tait of the mortality review committee agreed, and said the picture was complex, much more so than simply attempting resuscitation.
However, he supported a national approach. The committee has asked NICUs to "investigate and address" the regional differences in survival rates.
"You would hope, in a country this size, there wouldn't be much variation and they would all manage and look after people in a similar way."