Objective To determine the accuracy of axillary temperature relative to core rectal temperature during whole-body therapeutic hypothermia for moderate-to-severe hypoxic-ischaemic encephalopathy.

Design Retrospective audit.

Setting Single tertiary neonatal intensive care unit at The Royal Women's Hospital in Australia.

Patients Fifty-eight term newborn infants with moderate-to-severe hypoxic-ischaemic encephalopathy. Forty infants were treated with whole-body hypothermia between February 2001 and May 2010, 16 of whom were enrolled in the Infant Cooling Evaluation (ICE) trial, and 18 control infants randomised to normothermia in the ICE trial.

Intervention Comparison of simultaneous axillary and rectal temperatures measured between 0 and 84 h post randomisation or induction of cooling.

Results During the initiation of hypothermia (0–<6 h) axillary and rectal temperatures were similar (mean difference rectal-axillary =0.07°C), but with large variability (95% limits of agreement −1.18 to 1.33°C). There was larger variability in measurements between 6 and <72 h in the hypothermic infants (total SD 0.44) than in the normothermic group (total SD 0.24, p<0.001). In the hypothermic infants, the mean difference between the measurements during the rewarming phase (72–<84 h) was −0.19°C (95% limits of agreement −0.95 to 0.57°C).

Conclusion As there is wide variability in the difference between axillary and rectal temperatures at all stages of whole-body cooling, our data do not support the use of axillary temperature as a surrogate for core rectal temperature during therapeutic hypothermia.