Accuracy of non-invasive sensors measuring core body temperature in cardiac surgery ICU patients - results from a monocentric prospective observational study.

PURPOSE: Temperature monitoring in the perioperative setting often represents a compromise between accuracy, invasiveness of probe placement, and patient comfort. Transcutaneous sensors using the Zero-Heat-Flux (ZHF) and Double-Sensor (DS) technology have been developed and evaluated in a variety of clinical settings. The present study is the first to compare the performance of both sensors simultaneously with temperature measured by a Swan-Ganz catheter (PAC) in patients admitted to the intensive care unit (ICU) after cardiac surgery. METHODS: In this monocentric prospective observational study patients were postoperatively transferred to the ICU and both sensors were placed on the patients' foreheads. Core body temperature measured by intraoperatively placed PAC served as gold standard. Measurements were recorded at 5-minute intervals and up to 40 data sets per patient were recorded. Bland and Altman's method for repeated measurements was used to analyse agreement. Subgroup analyses for gender, body-mass-index, core temperature, airway status and different time intervals were performed. Lin's concordance correlation coefficient (LCCC) was calculated, as well as sensitivity and specificity for detecting hyperthermia (≥ 38 °C) and hypothermia (< 36 °C). RESULTS: Over a period of six month, we collected 1600 sets of DS, ZHF, and PAC measurements, from a total of 40 patients. Bland-Altman analysis revealed a mean bias of -0.82 ± 1.27 °C (average ± 95% Limits-of-Agreement (LoA)) and - 0.54 ± 1.14 °C for DS and ZHF, respectively. The LCCC was 0.5 (DS) and 0.63 (ZHF). Mean bias was significantly higher in hyperthermic and hypothermic patients. Sensitivity and specificity were 0.12 / 0.99 (DS) and 0.35 / 1.0 (ZHF) for hyperthermia and 0.95 / 0.72 (DS) and 1.0 / 0.85 (ZHF) for hypothermia. CONCLUSION: Core temperature was generally underestimated by the non-invasive approaches. In our study, ZHF outperformed DS. In terms of agreement, results for both sensors were outside the range that is considered clinically acceptable. Nevertheless, both sensors might be adequate to detect postoperative hypothermia reliably when more invasive methods are not available or appropriate. TRIAL REGISTRATION: German Register of Clinical Trials (DRKS-ID: DRKS00027003), retrospectively registered 10/28/2021.

A comparison of the NeurOs® and the INVOS 5100C® cerebral oximeter during variations of the partial pressure of carbon dioxide and fractional inspiratory concentration of oxygen.

This prospective method comparison study compared cerebral oxygen saturation (ScO2) measurement performance of the new cerebral oximeter (NeurOs®, Mespere LifeSciences, Ontario, Canada) in comparison to the established INVOS 5100C® (Medtronic, Boulder, USA) cerebral oximeter. We performed measurements during different levels of carbon dioxide pressure (PaCO2) during hyper- and hypoventilation and different levels of arterial oxygen saturation (SaO2) induced by variation of the inspiratory fraction of oxygen (FiO2). 59 anesthetized cardiac and vascular surgical patients were studied during hemodynamically stable conditions. Two versions of the NeurOs® oximeter were used in 39 and 20 patients, respectively: an older version with one bi-hemispherical sensor attached to the midline of the forehead and a newer version with two sensors that were attached to the left and right forehead. Alternating measurements of ScO2 with the INVOS® oximeter (bifrontal sensors) and the NeurOs® oximeter were performed during baseline conditions and after PaCO2 had been randomly in- and decreased by changes in ventilation (constant FiO2) and SaO2 had been randomly modified by variations in FiO2 (constant PaCO2). Employing the most recent NeurOs® version, measurements were additionally performed in a default and a high penetration mode. Bland-Altman analyses revealed comparable bias and limits of agreement for INVOS® and NeurOS® measurements during baseline conditions when using the bi-hemispherical sensor and the version with two sensors, respectively. Consequently, further analyses were performed on the pooled data of 59 patients. Bland-Altman analysis for repeated measurements revealed a bias of - 0.5%, a lower limit of agreement of - 16.3% (95% CI - 19.6 to - 13.7%) and an upper limit of agreement of 15.4% (95% CI 12.8 to 18.8%) during variations of PaCO2. The respective analysis during changes in SaO2 induced by variation of the FiO2 revealed a bias of - 0.8%, a lower limit of agreement of - 16.3% (95% CI - 19.7 to - 13.6%) and an upper limit of agreement of 14.7% (95% CI 12.1 to 18.2%). Both analyses showed a proportional error. No significant differences in ScO2 were observed during measurements with the bi-frontal sensors in the default as well as the high penetration mode. The ScO2 measurement performance of the NeurOs® cerebral oximeter is not interchangeable with the INVOS® cerebral oximeter during variations of ventilation and oxygenation in elective cardiac or vascular surgical patients. The lack of reactivity to changes in ventilation (by variation of PaCO2) and oxygen delivery (by variation of FiO2) question the reliability of NeurOs® measurements to reflect changes in cerebral blood flow and cerebral oxygen balance. This holds true not only for different sensor positions at the forehead but also for different modes of penetration.