Premature babies are at risk of a range of health problems
A treatment thought to improve a premature baby's chance of fighting infection does not actually provide any benefit, a UK study suggests.
A protein is given to stimulate growth of white blood cells and boost immune systems of babies born very early.
But research in 280 babies born at 31 weeks or under found it did not prevent blood poisoning - a major cause of newborn deaths.
Experts said the findings in The Lancet medical journal were unexpected.
Premature babies are particularly vulnerable to infection and those that survive can face developmental delay and neurological problems due to brain damage.
Those who have a lower than normal birth weight for their age are particularly prone to low white blood cell counts - or neutropenia - and this further increases the risk of infection.
Neonatal specialists have increasingly been using a protein known as granulocyte-macrophage colony stimulating factor (GM-CSF) in the hope of increasing white blood cells and preventing infection.
GM-CSF has already been shown to be effective in cancer patients whose immune systems have been damaged by chemotherapy.
But the trial of babies born at 31 weeks or under in 26 centres across England and Wales found no significant difference in deaths from blood poisoning - or sepsis - due to infection in those who had GM-CSF or those who had standard management.
Researchers did find that the number of white blood cells increased after treatment as expected.
Study leader, Dr Robert Carr from Guy's and St Thomas' Hospital NHS Trust, said there had been small studies which had been encouraging and they had been extremely optimistic about the outcome.
But the results showed treatments that are effective in adults cannot be assumed to be effective in premature babies and doctors should be wary of "extrapolation".
"The majority of medications used in premature babies have not been specifically evaluated in them.
"One of the problems is that drug companies are much less interested in funding research in smaller babies," he said.
He added that the fact white blood cell counts went up yet did not affect survival shows immunity in premature babies is a complex process and this treatment was dealing with just one part.
Professor David Field, an expert in neonatal medicine at the University of Leicester, said it highlighted the importance of doing trials.
"In theory you would have thought this was an important intervention.
"But it shows that in these babies there is some other factor which is important."