This study and consent forms were approved by our IRB and submitted to clinical-trials.gov as NCT02331108 by Jonathan V. Roth on November 20, 2014. Informed consent was obtained from all participating patients. The manuscript complies with the CONSORT requirements. This study was performed at the Albert Einstein Healthcare Network in Philadelphia, Pennsylvania during 2014 and 2015.
Four groups (described below and summarized in Appendix 1) of 50 patients each were studied. The major inclusion criteria were: age 18 to 55 years inclusive; supine or lithotomy positioning; scheduled for general anesthesia where 50% nitrous oxide would be used; endotracheal intubation or laryngeal mask airway insertion would be used; afebrile (preoperative oral or temporal scan temperature between 36.2 and 37.4°C inclusive); forced air warming would be used; and expected duration of anesthetic to be at least 60 minutes. A complete list of inclusion and exclusion criteria are presented in Appendix 2. After enrollment, random assignments were contained in opaque envelopes that were opened immediately before induction of anesthesia. Each of the envelopes contained one of the four group designations, 50 envelopes for each group. Randomization was achieved by putting the envelopes in a basket and mechanically mixing the envelopes within the basket. When a patient was entered into the study, an opaque envelope was selected arbitrarily from any location in the stack.
For all patients, operating rooms were kept between 21°C and 24°C with a target of 22°C. No patients were prewarmed. Upon entering the operating room, cotton blankets were placed on all patients covering their lower extremities, abdomen, and thorax. These blankets were removed after induction to allow for forced air warming (FAW) blanket placement and surgical positioning, preparation, and draping. All operating rooms had the same air flow design. Patients were administered 2 mg IV midazolam prior to entering the operating room. No opioid narcotics were administered until after the airway was secured with either a laryngeal mask airway (LMA) or endotracheal tube. Heat and moisture exchangers were used on all patients. Patients could receive up to 300 mL room temperature intravenous crystalloid before fluid was warmed (Ranger, Arizant Healthcare, Eden Prairie, MN) to 41°C. All inductions, nasal temperature probe placement, and application of a FAW blanket were performed in the same manner by the first author. (Nasal temperature was used as a surrogate for core temperature for all patients since it could be used for patients having either an LMA or endotracheal intubation.8) Either an upper or lower body FAW blanket (SW-2010 Snuggle Warm Small Upper Body Convective Warming Blanket, or SW-2001 Snuggle Warm Adult Full Body Convective Warming Blanket, Level 1, Smiths Medical ASD, Rockland, MA) was used. The face was not directly covered by the FAW blanket in order to avoid the possibility that a collection of warm air could affect the nasal temperature measurements. Cotton blankets were placed on top of the warming blankets. The FAW (Equator Convective Warmer, Level 1, Smiths Medical ASD, Rockland, MA) was turned on to 44°C as soon as the patient was prepped and draped; the time duration from the start of induction (T0) until the time the FAW was turned on was recorded. Neurophysiologic monitors to measure “depth of anesthesia” were not used. Pre-induction core temperatures were not measured. All doses of propofol were based on actual (not ideal) body weight.
Group INH/100 – Inhalation induction with sevoflurane in 100% oxygen (O2)
A baseline blood pressure was taken prior to induction. No formal preoxygenation regimen was performed. The patients were asked to breath for a few breaths via the face mask with 100% O2 just to confirm reservoir bag movement and capnograph detection of carbon dioxide. At time T0, with an unprimed circuit, the O2 flow meter was set at 6 LPM and the sevoflurane vaporizer was turned on at 8%. Blood pressures were recorded every minute starting one minute after T0 (T1) until airway intervention commenced. At the discretion of the first author, an LMA was inserted when the patient was assessed to be adequately deep, determined by masseter muscle relaxation, typically just two minutes after T0 (T2). Alternatively, if the patient was to be endotracheally intubated, muscle relaxant (vecuronium, rocuronium, or succinylcholine) was administered when the patient was assessed as being unconscious, typically at T1. Positive pressure ventilation was performed as required until endotracheal intubation. If necessary, to avoid hypotension, the Sevoflurane concentration was decreased while waiting for adequate muscle relaxation. If the systolic blood pressure dropped below 85 mm Hg prior to airway intervention, the patient would be treated immediately either with phenylephrine or airway intervention if ready. After securing either the LMA or endotracheal tube, anesthesia was maintained with sevoflurane in 50% nitrous oxide (1 LPM) and 50% O2 (1 LPM). Opioid narcotics (fentanyl, hydromorphone, methadone), neuromuscular reversal agents (glycopyrrolate, neostigmine), dexamethasone, and ketamine were administered as per the discretion of the attending anesthesiologist.
A skin temperature probe (Skin Temperature Sensor, 400 Series, DeRoyal Industries, Inc., Lane Powell, TN) was modified by removing the skin adhesive portion, bending the probe 90° 8 cm from the tip (to assure insertion depth would be 8 cm), and straightening the probe from the bend to the tip. Previous work has shown a close agreement between the nasal technique used in this study and distal esophageal temperature measurements.8 Within 10 minutes of T0, the blunt tipped nasal temperature probe was inserted 8 cm into one naris.8-10 This provided a minimum of 5 minutes for thermal equilibration of the temperature probe before the first measurement (T15), fifteen minutes after T0. Either naris was used arbitrarily. Starting at T15, nasal temperatures were recorded every 15 minutes (T15, T30, T45, T60). If the core temperature reached 37.5°C, the FAW was turned off. The patient’s data were included in the analysis if there were at least two temperature measurements (T15 and T30). If the anesthetic ended before 30 minutes or if there was a protocol violation, that patient’s data were not analyzed; a replacement envelope assigning another future patient to that group was generated and inserted randomly back into the envelope stack. All patients received 4 mg ondansetron within 15 minutes of emergence. Temperature data collection ceased at the initiation of IV acetaminophen administration or if there was any event that could have a substantial impact on patient temperature. All cystoscopy procedures were conducted with warmed bladder irrigation.
Group INH/50 - Inhalation induction with sevoflurane in 50% nitrous oxide (N2O) / 50% O2
The protocol was identical to group INH/100 except that induction was performed with 3 LPM N2O and 3 LPM O2 (instead of 6 LPM O2) with 8% sevoflurane.
Group PROP – Intravenous induction with 2.2 mg/kg intravenous propofol
The induction differed from group INH/100 in the following manner. Two mL of 2% lidocaine (40 mg) were added to 20 mL of 1% propofol. After preoxygenation with 100% O2 for a minimum of 2 minutes, three mL of 2% lidocaine (60 mg) was administered followed immediately by 2.2 mg/kg propofol (rounded to the nearest 5 mg) at T0. If the patient was to receive an LMA, one blood pressure was taken at T1 and then the LMA was inserted. If the patient was to be endotracheally intubated, muscle relaxant was administered immediately after propofol administration, blood pressures were measured every minute, and positive pressure ventilation with 100% O2 was performed as required. After securing the airway, the protocol continued in the same manner as in Group INH/100.
Group Phnl/PROP – Intravenous induction with 2.2 mg/kg intravenous propofol preceded by 160 mcg phenylephrine
The protocol differed from group PROP only in that 2 mL of 80 mcg/mL phenylephrine (160 mcg) was administered immediately after the administration of 3 mL 2% lidocaine but before the 2.2 mg/kg propofol.
STATISTICAL METHODS
To address the lack of pre-test core temperature measurements, we used the post-test only, single blind randomized trial. This is a “true experimental design”.11 The primary outcomes were the nasal (core) temperatures at 4 time points after induction (not changes from pre-induction baseline). In bivariate analyses, we compared differences in mean core temperature between the propofol only induction control group (PROP) and each of 3 groups administered alternative induction techniques (INH/100, INH/50, and Phnl/PROP). Specifically, analyses of the mean temperature differences (and 95%CIs) for 1) INH/100 vs. PROP, 2) INH/50 vs. PROP, and 3) Phnl/PROP vs. PROP were performed at each of 15, 30, 45, and 60 minutes (T15, T30, T45, and T60) after induction. These differences in mean core temperatures at T15, T30, T45, and T60 among groups were assessed using unpaired t-tests and corresponding 95% confidence intervals (95% CIs). Bonferroni’s correction was used to adjust for these 12 multiple comparisons. Core temperature data were tested for normality by the Shapiro-Wilk test.
The random assignment of 50 patients per group made it likely that the treatment groups would be balanced in both measured and unmeasured characteristics (including pre-induction core temperatures). However, imbalances did occur in BMI and sex. Those imbalances and the lack of pre-induction core temperature measurements necessitated a multivariable analysis comparing the average core temperatures at T15, T30, T45, and T60; the covariates were BMI, sex, age, ASA classification, and time to initiating FAW. (Upper vs lower FAW were not covariates because the rates of heat transfers are similar.12) This multivariable analysis was a linear mixed model with random intercepts and random slopes and unstructured covariance. This model fit better than a model with random intercepts alone nested within it (p<0.0001 by the likelihood ratio test).13 There was no statistically significant interaction between group and time (p=0.15). Since the results of the bivariate and multivariable analyses were similar and led to the same conclusions, we present the simpler bivariate results. The differences in the secondary outcomes, the percentages of patients who had at least one temperature <36.0oC (and ≤35.5oC) between the control group (PROP) and each of the other three groups were evaluated by Fisher’s exact tests. Although the resulting p values were exact; the corresponding 95% CIs were approximate.
Interval estimates of the percentages of patients that developed hypotension requiring treatment and of patients undergoing an inhalation induction who developed apnea were computed using exact binomial 95% CIs.
In a statistical power analysis, assuming alpha = 0.05 and beta = 0.2, 17 patients per group were needed to detect a 0.5 oC difference between means of any two compared groups as statistically significant in a two tailed test. We enrolled patients who were expected to have a surgical procedure lasting at least 60 minutes after the induction of anesthesia. However, we had no way to estimate the percentage of cases that would end early and therefore not provide data at T45 and T60. We increased the size of the study groups to 50 in order to be reasonably sure there were at least 17 patients in all groups at T60.
All statistical analyses compared patients as treated using two-sided tests with alpha=0.05 and were performed using Stata, version 14 (Stata Press, College Station, Texas).