• Robert E Freundlich, Sara E Nelson, Yuxuan Qiu, Jesse M Ehrenfeld, Warren S Sandberg, and Jonathan P Wanderer.
• Department of Anesthesiology, Vanderbilt University Medical Center, 1211 21st Avenue South, MAB 422F, Nashville, TN, 37212, USA. Robert.e.freundlich@vanderbilt.edu.
• J Clin Monit Comput. 2019 Oct 1; 33 (5): 911-916.

AbstractThe prevention and treatment of hypothermia is an important part of routine anesthesia care. Avoidance of perioperative hypothermia was introduced as a quality metric in 2010. We sought to assess the integrity of the perioperative hypothermia metric in routine care at a single large center. Perioperative temperatures from all anesthetics of at least 60 min duration between January 2012 and 2017 were eligible for inclusion in analysis. Temperatures were displayed graphically, assessed for normality, and analyzed using paired comparisons. Automatically-recorded temperatures were obtained from several monitoring sites. Provider-entered temperatures were non-normally distributed, exhibiting peaks at temperatures at multiples of 0.5 °C. Automatically-acquired temperatures, on the other hand, were more normally distributed, demonstrating smoother curves without peaks at multiples of 0.5 °C. Automatically-acquired median temperature was highest, 36.8 °C (SD = 0.8 °C), followed by the three manually acquired temperatures (nurse-documented postoperative temperature, 36.5 °C [SD = 0.6 °C]; intraoperative manual temperature, 36.5 °C [SD = 0.6 °C]; provider-documented postoperative temperature, 36.1 °C [SD = 0.6 °C]). Provider-entered temperatures exhibit values that are unlikely to represent a normal probability distribution around a central physiologic value. Manually-entered perioperative temperatures appear to cluster around salient anchoring values, either deliberately, or as an unintended result driven by cognitive bias. Automatically-acquired temperatures may be superior for quality metric purposes.

31 keywords...

hide…