Babies in intensive care are 50 per cent more likely to die if the unit is full, according to a wide-ranging new study.

The workload, rather than the number of nurses or consultants or even the size of the unit, was the key factor determining survival rates, the study said.

The findings – which were carried out by a team led by Dr Janet Tucker of the University of Aberdeen, and are reported in The Lancet today – dispel the standard thinking that "bigger is better" for the units, known as neonatal intensive care units.

"We found a direct linear association between increasing occupancy of all sizes of units and the risk-adjusted mortality," Dr Tucker said. But no such association emerged when looking at the size of the units, she said.

Dr Tucker suggested that the findings meant lives could be saved by having designated nurses in the hospital who could be drafted into the units as they got busier. Another possibility would be to formalise a "network" that already exists informally, through which units in the same region transfer ill babies elsewhere to even caseloads out.

The study was carried out in 1998-99, and looked at mortality rates across 54 randomly-chosen units, which admitted 13,500 infants.

The units are generally busy; between one and two out of every 100 babies require an hour or more of neonatal intensive care.

The study was the first to undertake wide-ranging research into the factors that affected mortality rates.

There had been earlier studies, but they were smaller in scale and had employed different criteria to define the care given. Some researchers had described it as ongoing if the infant left the unit but stayed in the hospital, while others defined leaving the unit as the end of the treatment.

However, some researchers suggested that the latest findings might already be outdated. "Things have moved on since this study was started," Murray Pollack and Kantilal Patel, of the George Washington University School of Medicine, wrote in an accompanying commentary.

They said that it was now time to study the quality of care. "Factors such as volume of patients and total number of staff no longer adequately represent what is really happening at the bedside," they wrote. "The details of delivery of care can be examined."