Background: Although core temperature can be measured invasively, there are currently no widely available, reliable, noninvasive thermometers for its measurement. We thus compared a prototype zero-heat-flux thermometer with simultaneous measurements from a pulmonary artery catheter. Specifically, we tested the hypothesis that zero-heat-flux temperatures are sufficiently accurate for routine clinical use.
Methods: Core temperature was measured from the thermistor of a standard pulmonary artery catheter and with a prototype zero-heat-flux deep-tissue thermometer in 105 patients having nonemergent cardiac surgery. Zero-heat-flux probes were positioned on the lateral forehead and lateral neck. Skin surface temperature probes were attached to the forehead just adjacent to the zero-heat-flux probe. Temperatures were recorded at 1-minute intervals, excluding the period of cardiopulmonary bypass, and for the first 4 postoperative hours. Zero-heat-flux and pulmonary artery temperatures were compared with bias analysis; differences exceeding 0.5°C were considered to be potentially clinically important.
Results: The mean duration in the operating room was 279 ± 75 minutes, and the mean cross-clamp time was 118 ± 50 minutes. All subjects were monitored for an additional 4 hours in the intensive care unit. The average overall difference between forehead zero-heat-flux and pulmonary artery temperatures (i.e., forehead minus pulmonary artery) was -0.23°C (95% limits of agreement of ±0.82); 78% of the differences were ≤0.5°C. The average intraoperative temperature difference was -0.08°C (95% limits of agreement of ±0.88); 84% of the differences were ≤0.5°C. The average postoperative difference was -0.32°C (95% limits of agreement of ±0.75); 84% of the differences were ≤0.5°C. Bias and precision values for neck site were similar to the forehead values. Uncorrected forehead skin temperature showed an increasing negative bias as core temperature decreased.
Conclusions: Core temperature can be noninvasively measured using the zero-heat-flux method. Bias was small, but precision was slightly worse than our designated 0.5°C limits compared with measurements from a pulmonary artery catheter.