Temperature measurement in critically ill orally intubated adults: A comparison of pulmonary artery core, tympanic, and oral methods

Objective: Core temperature measurement using a pulmonary artery (PA) cathe ter is considered the gold standard for measuring temperatures in criticall y iii patients. The objective of this study was to compare oral and tympani c temperature measurements (in both the oral and core equivalence modes) ag ainst PA core temperature measurements to determine which method was the mo st accurate and reliable in the absence of a PA catheter. Design: Prospective, descriptive comparative analysis. Patients: Convenience sample of 102 critically ill orally intubated patient s with a PA catheter in place. Setting: A 24-bed medical/surgical/trauma intensive care unit in a universi ty-affiliate medical center. Interventions: Four experienced intensive care unit nurses were trained in the use of temperature measurement with the oral, tympanic (both core and o ral equivalence modes were used), and PA core methods. Simultaneous tempera ture measurements were then taken once in each subject using each method. T he potential covariates that were analyzed were mean blood pressure, patien t acuity using the Simplified Acute Physiology Score II, age, sex, ambient room temperature, and ventilator circuit temperature. Measurements and Main Results: The training period indicated that it took m ore time to train experienced nurses in the use of tympanic thermometry tha n oral thermometry. Descriptive statistics were the following: core, x = 37 .33 (SD = 0.89); oral, x = 37.18 (SD = 0.92); tympanic oral, x = 36.80 (so = 0.93); and tympanic core, x = 37.12 (so = 1.0). Bias averages were calcul ated and were significantly different from 0 for all three methods (oral-PA core, -0.15 [SD = 0.36]; tympanic core-PA core, -0.11 [so = 0.57], tympani c oral-PA core, -0.52 [so = 0.53]), indicating that there is some degree of decreased accuracy associated with each method when compared with PA core. However, scatter plots using the Bland and Altman methodology (Altman DG, Bland JM: Practical Statistics for Medical Research. London, Chapman and Ha lt, 1991) illustrate that the greatest variability is associated with the t ympanic method. Conclusions: Temperature measurement is an important piece of clinical data in a critically ill patient population. We found oral thermometry to be th e most accurate and reproducible method when a PA core measurement was not available.