Posted by Mike Clarke on July 29, 2026

This month’s Trials Tuesday answered this question 70 years ago and it continues to be cited by Cochrane Reviews when discussing the benefits of reducing heat loss in preterm babies.

Bill Silverman

The trial, published in Pediatrics in 1958 and led by Bill Silverman, compared babies placed in normothermic (31-32ºC) and hypothermic incubators (28-29ºC). It also tested two antibiotics, oxytetracycline and tetracycline, at two doses, leading to a factorial design with 8 intervention groups.

Babies under 72 hours old who were admitted to the premature nursery in the “Babies Hospital” in New York were recruited into the study, starting in March 1956. When they were five days old, all surviving babies were moved into uniform environmental conditions, at 30-31º C.

The Babies Hospital on Broadway at West 168th Street in New York City.

A matched-pairs sequential plan, devised by Agnes Berger and John Fertig, was used to analyse the primary outcome: death of the baby. Babies were placed in pairs such that a baby in one temperature group was matched with another of similar weight in the other group who had been allocated to the same antibiotic regimen.

By February 1957, the running analysis of accumulating evidence showed the effects of temperature. 182 babies were available for the analyses, of whom 44 had died: 76 of 91 (83.5%) babies placed in a normothermic incubator had survived, compared to 62 of 91 (68.1%) in a hypothermic incubator.

The graph of the matched-pairs analysis compared “tied” and “untied” pairs of babies and showed that the decision-line had been crossed. “Tied” pairs were those in which both babies had survived or died, while “untied” pairs had one that survived and one that died. There were 26 unmatched pairs, 20 in which the survivor was in the normothermic group and 6 in which they were in the hypothermic group.

The conclusions of the trial overturned a common belief that lower temperatures were better for premature babies and that trying to raise their body temperature might be harmful including increasing the risk of death. Its influence on practice was reinforced by other trials in the 1960s.

Bill Silverman wrote about this trial and others for newborn babies in a James Lind Library commentary the year before he died, aged 87 in 2004.

Later, in recognition of his habit of raising troubling questions about the evidence base and ethics of medical practice, the Cochrane Collaboration named a prize in his honour.