Incidence of Connected Consciousness after Tracheal Intubation: A Prospective, International, Multicenter Cohort Study of the Isolated Forearm Technique

EXPLICIT recall of intraoperative events after intended general anesthesia is rare (0.1 to 0.2%),^1–3  implying that anesthesia is a remarkably effective therapeutic intervention with a number needed to treat approaching 1 (1.002).⁴  However, since anesthetics are effective amnesic agents, reliance on postoperative report of intraoperative events is unlikely to capture all events of intraoperative consciousness⁴  and thus overestimate anesthesia’s therapeutic effectiveness. Indeed, patients expect to be unaware of surgery during general anesthesia. The isolated forearm technique (IFT)⁵  does not depend on explicit postoperative recall of events as a surrogate of consciousness,⁴  providing real-time information about the presence of consciousness using the best available method: behavioral report.⁶  The IFT captures evidence of intraoperative consciousness of sensory stimuli (so-called connected consciousness as the experiences are connected to the environment).^4,6  Hence, the IFT collects data on clinically relevant connected consciousness,⁷  as subjects need to be aware of their sensory environment to hear the command. The IFT does not discriminate between disconnected consciousness (e.g., dreaming) or unconsciousness, both of which may be considered acceptable states under anesthesia because they are not associated with awareness of surgery. One limitation of the IFT is that patients with connected consciousness may still not respond to the command despite hearing it (for example, due to impaired motivation or anesthetic actions on motor responses).⁴  As such, the IFT represents a conservative estimate of connected consciousness.

Our review of previous studies of the IFT after a clinically relevant noxious stimulus, such as tracheal intubation or skin incision, identified that approximately 37% of patients report connected consciousness of the stimulus under general anesthesia (based on 13 studies).⁴  If this estimate is correct, it would imply a number needed to treat of 1,587 for anesthesia, representing a very different therapeutic effectiveness compared to calculations based on explicit recall. However, many of the studies included in our review were small, single-center studies, with variable—and sometimes contrived or controlled—dosing of anesthetics reflecting only one hospital or provider or experimental protocol. To overcome these limitations, we conducted an international, multicenter, pragmatic study to establish the incidence of IFT responsiveness after tracheal intubation in routine practice (hence, the anesthetic technique was left to the discretion of the anesthesiologist). This is the first multi-institution collaboration investigating the use of the IFT and is the largest single study on the topic.⁴  Our study benefits from greater internal validity than many previous studies with the methodologic advantage of having an adjudicator, who is independent of the clinical team, judge the IFT response. External validity is conferred by a cohort design recruiting from six international centers. We collected data on the incidence of IFT responsiveness before and after intubation and on variables that may be associated with the likelihood of responsiveness including patient and anesthetic factors.

Ethics board approval was obtained locally at each site, and the study was registered at clinicaltrials.gov before commencement (NCT02248623). All patients provided written informed consent to participate in the study after a careful discussion of risks and benefits. All adult patients (older than 18 yr) undergoing general anesthesia with endotracheal intubation were considered eligible if they could follow the commands for the IFT test. Exclusion criteria included a contraindication to the IFT, such as the inability to have tourniquet on arm for the IFT (e.g., lymphedema or operative site). Patients were recruited from August 2014 to August 2015 from six centers: University of Wisconsin (Madison, Wisconsin), Waikato Hospital (Hamilton, New Zealand), University Hospital RWTH Aachen (Aachen, Germany), University of Michigan (Ann Arbor, Michigan), University Medical Centre Groningen (Groningen, The Netherlands), and CHR Citadelle Hospital of Liège (Liège, Belgium).

The primary endpoint of this study was to observe the incidence of IFT responsiveness after intubation in a routine clinical setting. All clinical decisions, such as drugs, approach to intubation, and so forth, were left to the discretion of the attending anesthesiologist. The attending anesthesiologist was not blinded to the IFT results. A blood pressure cuff was placed on the forearm (preferably the dominant arm) and inflated to 50 mmHg above systolic blood pressure during injection of induction agents but before administration of neuromuscular blockers. The cuff remained inflated for the duration of the IFT (typically less than 5 min) and was deflated after the final command.

IFT responses were recorded by an observer, who was not the attending anesthesiologist and, where possible, verified by a second witness. A positive cognitive IFT response was defined as a contraction of the hand in response to a verbal command. The first command was “[NAME], squeeze my hand.” If there was a positive response, a second command was given “[NAME], squeeze my hand twice if you have pain.” Given the pragmatic nature of this study, attempts were not made to control ambient noise in the operating room.

These questions were asked immediately before laryngoscopy and after securing the endotracheal tube (typically within 1 min of intubation). For the IFT to be considered movement to command, the hand had to be still immediately before the command. In situations where the hand was spontaneously moving, and hence response to command was difficult to confirm, the IFT was considered negative giving us a conservatively biased estimate of the incidence of connected consciousness. Lack of paralysis of the hand was confirmed using a train-of-four monitor with four twitches present in all cases.

Secondary endpoints included the following:

1. Establish the incidence of IFT responsiveness with pain after laryngoscopy (prespecified outcome).

2. Establish the incidence of IFT responsiveness before laryngoscopy defined as after induction of anesthesia but before laryngoscopy (prespecified outcome).

3. Establish the combined incidence of IFT responsiveness before and after laryngoscopy (post hoc outcome).

4. Establish the incidence of postoperative explicit recall of events in IFT responders versus nonresponders identified using the modified Brice questionnaire⁸  within 24 h of the operation (prespecified outcome).

5. Identify any alterations in patient satisfaction, measured by the Bauer satisfaction scale,⁹  associated with IFT responsiveness compared to nonresponders (prespecified outcome).

6. Identify risk factors for IFT responsiveness after intubation (prespecified outcome).

Collected variables included surgical site, age, sex, body mass index, race, American Society of Anesthesiologists physical classification score, significant comorbidities, Mallampatti score, observer-predicted difficult intubation (yes/no), drug exposure before intubation, observer-rated difficult intubation (yes/no), number of attempts at intubation, estimated duration of intubation, depth of anesthesia monitor use (yes/no) and numerical values, hemodynamics before and after intubation, response to the IFT before and after intubation, and whether witnessed by a second observer. For the depth of anesthesia monitoring data, we combined data from the Bispectral Index (BIS) and Neurosense monitors as both are on a scale of 0 to 100 with a clinical range of 40 to 60; hence, clinicians will use the information in the same way for either monitor. Postoperatively, within 24 h of emergence, participants were asked the modified Brice and Bauer questionnaires.^8,9  Data were entered locally into a REDCap database managed at the University of Wisconsin.

The primary outcome was the incidence of IFT responsiveness after intubation, and hence, we were unable to conduct a power calculation for the study. The study sample size was estimated based on previous data suggesting an event rate of up to 40% after intubation¹⁰  and review of IFT responses after noxious stimuli during anesthesia.⁴  Based on an upper limit estimate of a 40% response rate, recruitment of 260 patients would provide approximately 100 events. If achieved, this would allow multivariable modeling of up to 10 risk factors for postintubation IFT responsiveness. In this instance, the risk factors we considered were age, gender, preoperative midazolam, preoperative β-blocker, propofol dose, opioid dose, volatile anesthetic before intubation, duration of laryngoscopy, number of attempts at laryngoscopy, and depth of anesthesia monitor value at IFT response. We planned to competitively enroll at eight centers with a minimum recruitment rate of 25 patients per center but two centers had to withdraw before patient recruitment. Consequently, we enrolled at six centers.

Continuous data are presented with mean ± SD, and categorical data are presented as frequencies and percentages. Univariate comparisons of continuous variables were made with Student’s t test assuming equal variances, and comparisons between categorical variables were made with Pearson chi-square or Fisher exact test. As age may act as a confounder for both incidence of IFT response and selection of induction medications, we conducted logistic regression with bivariate models including age and drug variable. All P values were set at 0.05 with two-tailed hypothesis testing. All analyses were performed using SAS software (SAS 9.3; SAS Institute, USA).

Of the 260 patients enrolled (fig. 1), 4.6% (12) were IFT responsive after intubation (table 1). Eighteen patients moved spontaneously after postlaryngoscopy prohibiting the IFT and therefore were designated as nonresponders because spontaneous movement compromises IFT assessment. We assumed that these movements were probably some manifestation of spinal reflex response but cannot exclude that at least some of these subjects were also conscious. Seven subjects were lost to postoperative follow-up (fig. 1). Of the 12 responders after intubation, five reported pain through a second hand squeeze (42% of IFT responders; 1.9% of total cohort). There were no reports of explicit awareness of intraoperative events by either responders (n = 12) or nonresponders (n = 241) when asked within 24 h of the operation.

We conducted post hoc sensitivity tests for the incidence of IFT responsiveness. The incidence of IFT responsiveness after intubation did not vary significantly by center (range, 0 to 12%; P = 0.27 by chi-square test). We excluded the lowest (0/25) and highest (3/25) responding centers; the remaining centers’ response rates were 2% (1/50), 4% (1/25), 4% (4/100), and 9% (3/35), respectively. The mean response rate in this sensitivity analysis was 4.3%. We also conducted a sensitivity analysis in participants who responded after a protocol deviation of being asked the first question more than twice. On this basis, two IFT responders were excluded, leaving a mean response rate of 3.9%. Hence, our sensitivity analyses suggest that the response rate is approximately 4%, which is similar to the rate at the largest single site (4/100).

Five (1.9%) patients were IFT responsive before intubation, but only one patient was responsive before and after intubation. Of these five patients who responded before laryngoscopy, BIS values were only available in one patient. In this patient, the BIS was 46 at the time of prelaryngoscopy IFT responsiveness, followed by 66 at intubation, and 38 at the time of the postintubation IFT. This individual did not respond to the IFT questions after intubation. As a secondary post hoc outcome, we calculated the total IFT response rate, including both the preoperative and postoperative IFT response rate, as 6.2%.

The characteristics of the responders and nonresponders are reported in tables 2 to 4. The age range for the whole cohort was 18 to 88 (51 ± 16) yr old. On average, the responders were younger than nonresponders but otherwise were not different in terms of baseline characteristics including American Society of Anesthesiologists score and body mass index (table 2; fig. 2). There were no differences between anesthesiologist rating of the difficulty of intubation, number of intubations, or estimated duration of intubations between the responders and nonresponders (table 3).

Total intravenous anesthesia (TIVA) was used in 31 cases (12.7%) with the majority achieved by a bolus of propofol, followed immediately with a continuous infusion. Only one patient (a nonresponder) received an α-2 adrenergic agonist as part of induction. There was no difference in exposure to TIVA or volatile anesthesia, fentanyl equivalents, midazolam, ketamine, propofol, or β-blockers between responders and nonresponders (table 4). As age may influence the administration of different medications, and also the risk of IFT responsiveness, we then performed logistic regression to adjust for the effect and interaction of age on the putative effect of relevant drug variables. After adjustment for age, in patients not receiving TIVA, exposure to volatile anesthetic before intubation was associated with a reduced odds of responding (odds ratio, 0.2 [0.1 to 0.8]; P = 0.02). However, induction doses of propofol (1.4 mg/kg [0.6 to 3.5]; P = 0.47) and ketamine (3.75 [0.9 to 15.7]; P = 0.07) were not associated with altered odds of responsiveness after adjustment for age. Of the responders, there was no difference in fentanyl equivalents between those subjects who reported pain versus no pain (P = 0.81). Interestingly, all three responders who were administered ketamine (25 mg) as part of their induction reported pain despite coadministration of opioids.

No adverse events related to the IFT were reported. In 154 patients, a depth of anesthesia monitor was placed (Neurosense, n = 25; BIS, n = 129); use of depth of anesthesia monitoring did not differ between the groups (table 2). Depth of anesthesia monitoring values showed a wide range in both groups; however, mean values were higher for responders before (54 ± 20 vs. 42 ± 14; P = 0.02) and after (57 ± 25 vs. 43 ± 16; P = 0.02) intubation (fig. 2; table 3). The effect was not pronounced enough for receiver operator characteristic (ROC) calculations to demonstrate the ability to predict IFT responsiveness after laryngoscopy for either preintubation (ROC area = 0.69 [0.49 to 0.89]; P = 0.08) or postintubation (ROC area = 0.66 [0.42 to 0.89]; P = 0.11) values.

Responders had a higher incidence of observer-rated signs of sympathetic activation (signs of lacrimation, tachycardia, or hypertension) after intubation (50% vs. 2.4%; P = 0.03). Three of the five responders who reported pain demonstrated observer-rated signs of sympathetic activation (table 1). However, there were no differences in prelaryngoscopy or postlaryngoscopy heart rate between responders and nonresponders. There was also no difference in blood pressure between the groups before and after laryngoscopy.

Given the low event rate of IFT responsiveness, we urge caution with the following findings due to the potential for confounding and our lack of statistical power to adjust for multiple comparisons. These data must be only considered hypothesis generating. In addition to conducting the modified Brice questionnaire for explicit recall (available in 253 patients only), we also conducted the Bauer patient satisfaction survey to understand if IFT responsiveness was associated with reduced postoperative satisfaction.⁹  One IFT responder (8.3%) reported dreams (which were reported as disturbing), while 24 out of 241 nonresponders (10.3%) reported dreams (P = 1.00) with one patient reporting disturbing dreams. There were no univariate differences in patient-reported drowsiness (P = 1.00), pain at the surgical site (P = 0.14) or injection site (P = 0.64), thirst (P = 0.59), hoarseness (P = 0.88), sore throat (P = 0.29), nausea or vomiting (P = 0.45), or feeling cold (P = 0.60). We previously hypothesized that overlapping mechanisms of connectedness may link IFT responsiveness and postoperative delirium.³  However, there was no difference in patient reporting of feeling confused or disoriented (responders 33%, nonresponders 14%; P = 0.20). IFT responders appeared more likely to report shivering (P = 0.04; n = 253) at the univariate level: severe shivering (responder 17% vs. nonresponder 2%), moderate shivering (responder 0% vs. nonresponder 10%), and no shivering (responder 83% vs. nonresponder 88%). Univariate differences in patient satisfaction were not observed regarding the information they received preoperatively (P = 1.00), their wake up (P = 0.63), their nausea (P = 0.10), or their general impression of the anesthesia department (P = 0.79). However, IFT responders reported being less satisfied with their pain after surgery (P = 0.02 by Fisher exact test; n = 252): very satisfied (responder 42% vs. nonresponder 57%), satisfied (responder 33% vs. nonresponder 38%), dissatisfied (responder 16% vs. nonresponder 4%), and very dissatisfied (responder 8% vs. nonresponder 0.4%).

In this prospective, multicenter, pragmatic, international cohort study, we established that the incidence of IFT responsiveness after intubation was 4.6%. An estimate of 4% was supported by sensitivity analyses. Five patients (1.9% of the whole cohort) reported pain at the time of postintubation IFT. Although our previous review suggested that only 14% of IFT responders report pain,⁴  42% reported pain in this study. This may reflect the intensity of the nociceptive stimulation of laryngoscopy.¹¹  Our data argue for the clinical importance of the IFT, as few patients would regard intraoperative pain as acceptable. Consistent with previous data,⁴  and the known clinical effectiveness of anesthetic drugs to suppress memory at concentrations well below hypnotic doses,¹²  IFT responsiveness was not associated with postoperative explicit recall of the event. Our data suggest that the incidence of intraoperative connected consciousness (with responsiveness) after intubation may be 25 times higher than the incidence of explicit recall of events.

As we have previously discussed, a positive IFT response represents the standard in consciousness research—behavioral report.⁶  Interpretation of a negative response is more complicated because behavioral confirmation of unconsciousness is impossible. Our parsimonious explanation is that these subjects are either disconnected from their environment or unconscious. However, it is possible that these subjects are conscious and either choose not to respond or are unable to respond. Further discrimination between these states is not possible with current methodologies in the field. It is possible that administration of a volatile anesthetic may affect any/all of these endpoints through affecting motor responsiveness, motivation, or consciousness itself.⁴ 

We concentrated on providing a methodologically robust incidence of connected consciousness after intubation; however, several factors make our estimate conservative. We excluded patients with spontaneous movement as it made interpretation of the IFT difficult. Without an unambiguous behavioral response to command, the conscious state of subjects moving spontaneously is unclear. We also cannot exclude that further episodes of intraoperative awareness occurred after intubation (during surgery). Nonetheless, while our estimate should be considered conservative, it is methodologically robust as we used observers who were independent from the clinical team caring for the patient to judge the IFT responses.

We studied intubation because it is a clinically relevant stimulus, and the IFT tourniquet can easily be applied for less than 20 min. Longer durations of IFT testing are complicated by the need to reperfuse the hand to prevent ischemic paralysis. Importantly, our study was inclusive, to allow a broad population of the patients undergoing general anesthesia, and we did not seek to control the anesthesia to allow insights about routine clinical practice in six centers. This pragmatic design increases the external validity of our study though we cannot exclude a Hawthorne effect, whereby observation decreased the actual incidence of responsiveness.

The incidence rate in our study is lower than the rate we predicted based on our systematic review. Some studies identified in our review⁴  and published after it¹³  conducted the IFT during intubation, which we avoided in this study due to the high incidence of spontaneous movement at that time. Furthermore, it is not always clear from previous studies how spontaneous movement affected methodology. However, when we focus on studies that concentrated on postintubation IFT testing, a few notable, but small, studies deserve mention. St Pierre et al.,¹⁴  testing three different doses of etomidate, found that 5 of 30 patients (17%; aged 29 to 80 yr old), responded to the IFT in 120 s after intubation. Similarly, a study of ketamine and succinylcholine anesthesia (1.5 mg/kg) showed that 0 of 13 patients responded in the 10 min after induction for cesarean section.¹⁵  Using the LMA-Fastrach insertion technique and remifentanil-propofol anesthesia titrated to a BIS of 40 to 65, an IFT response rate of 7/51 (13.5%; aged 22 to 75 yr old) was observed.¹⁶  In another small study, Schneider et al.¹⁰  found that 8 of 20 (40%) patients responded to the IFT after intubation with propofol-alfentanil anesthesia titrated to a preintubation BIS of 50. A key difference with our study was that the preintubation BIS values were typically lower than 50. Our estimate of IFT responsiveness after intubation in real-world practice, falls within the range of response rates (0 to 40%) of these heterogeneous previous studies. Nonetheless, the wide range in the previous studies emphasizes the need for the current investigation.

Our data do not suggest a clear way of identifying those at risk despite responders being younger than nonresponders on average. The wide range of ages, drug doses, and depth of anesthesia monitoring values associated with IFT responsiveness (table 1; fig. 2) indicate these variables are poorly predictive. However, given that age will bias the selection of anesthesia medications, we also performed logistic regression to adjust for this variable. After adjustment for age in those patients undergoing inhalational anesthesia, nonexposure to volatile anesthetic before laryngoscopy was associated with significantly increased odds of responsiveness. This implies that early administration of a volatile anesthetic drug, while waiting for the muscle relaxant to take effect, may reduce the odds of responding to the IFT questions after intubation. Another interpretation is that a single bolus of intravenous anesthetic may be insufficient to ensure anesthesia through to intubation. Consistent with this, in patients who had depth of anesthesia monitoring placed, prelaryngoscopy monitoring values tended to be higher in responders versus nonresponders. However, the wide interpatient variance suggests caution in relying on the monitors: values below 40 occurred in both responders and nonresponders, as has been observed previously.^13,17,18  Our receiver operator curve data appear consistent with previous studies showing that BIS^13,18  and Narcotrend¹⁷  values are poorly predictive of IFT responsiveness. Furthermore, recent estimates for the BIS value with 100% sensitivity to prevent IFT responsiveness to intubation are disproven¹³ : three subjects responded with BIS values below the previously identified threshold of 37.¹³  Hence, while using population-level summary statistics can discriminate the groups in some instances (such as age or depth of anesthesia monitoring values), this was not possible at the individual level. A much larger study will be required to identify the individual risk factors that have useful prognostic value.

Overall, our data indicate that in patients who ultimately receive volatile anesthetics for maintenance of general anesthesia, commencing the vapor before intubation may reduce IFT responsiveness (connected consciousness) after intubation. Biologically, this appears plausible if a single bolus of intravenous anesthetic is used to induce anesthesia. This suggests that continual administration of an anesthetic to the time of laryngoscopy may be a prudent strategy to reduce connected consciousness. However, we recognize that in some situations, such as rapid sequence induction, this strategy may be inappropriate.

We consider intraoperative-connected consciousness, particularly with pain, an important clinical problem. However, the longer term effects of this state, especially given the lack of recall of the events, is unclear. Postoperative questionnaires were conducted to begin to probe some possible associations but we urge caution against overinterpreting our postoperative secondary endpoints; these should be considered hypothesis generating only. Responders appeared to be more likely to be dissatisfied in their pain management and had an increased incidence of shivering postoperatively. If this is a real effect, it remains unclear whether patients who are more likely to be dissatisfied with these outcomes are prone to intraoperative connected consciousness or whether patients who experience intraoperative connected consciousness are more likely to report adverse outcomes. Furthermore, we do not have data on important confounders related to the propensity of each endpoint, such as temperature.

We report a large prospective, multicenter cohort study that establishes the incidence of IFT responsiveness after intubation in present clinical practice, but there are limitations to our study. First, while it is reassuring that the incidence of IFT responsiveness after intubation is approximately 10-fold lower than previous data suggested, this left us underpowered for multivariable analysis of the data. Hence, our secondary analyses should largely be considered hypothesis generating. Future studies should investigate how to reduce the incidence of, and identify electroencephalograph biomarkers of, connected consciousness during anesthesia. Nonetheless, we were able to establish a simple message: younger patients who are not exposed to continuing anesthesia (e.g., volatile anesthesia) after an induction bolus appear more likely to respond on the IFT. Finally, while we had no episodes of explicit recall in the study, the cohort was small and our questioning was performed within 24 h of the operation to keep the study pragmatic; this is not the optimal time point for capturing these events (which is typically better done at 3 to 7 days or later).

In this international, multicenter, prospective cohort study, IFT responsiveness occurred in 4.6% of subjects with 1.9% of subjects reporting pain. At the population level, age and depth of anesthesia monitoring values distinguished responders and nonresponders. Our data also support the continued administration of anesthesia before intubation to reduce the odds of connected consciousness.

The authors are grateful to Misha Perouansky, M.D., and Kirk Hogan, M.D., Ph.D. (University of Wisconsin, Wisconsin, Madison), and Ian Russell, F.R.C.A. (Hull Royal Infirmary, Hull, United Kingdom [retired]), for forming the Data Safety Monitoring Board and for helpful discussions.

Support was provided solely from institutional and/or departmental sources.

The authors declare no competing interests.

Full protocol available from Dr. Sanders: robert.sanders@wisc.edu. Raw data available from Dr. Sanders: robert.sanders@wisc.edu.