Dreaming during Anesthesia and Anesthetic Depth in Elective Surgery Patients: A Prospective Cohort Study

This prospective cohort study received prospective ethics committees approval (Royal Melbourne Hospital, Parkville, Victoria, Australia; Royal Perth Hospital, Perth, Western Australia, Australia; and Women's and Children's Health Service, Perth, Western Australia, Australia). With written informed consent, 300 patients, aged 18–50 yr and of American Society of Anesthesiologists' physical status I–III, presenting for elective surgery during relaxant general anesthesia, were recruited. Exclusion criteria included (1) inadequate English comprehension due to a language barrier, cognitive deficit, or intellectual disability; (2) diagnosis of a psychotic disorder, major affective disorder, or major drug dependence disorder; (3) inability to monitor BIS because of the site of surgery; and (4) planned postoperative ventilation or anticipation of unavailability for postoperative interviews (fig. 1).

The primary endpoint was the difference in median BIS values during maintenance of anesthesia between patients who reported dreaming and those who did not report dreaming at the first postoperative interview. Secondary endpoints included duration of BIS greater than 60 during maintenance of anesthesia and preoperative and anesthetic factors associated with dreaming.

Intravenous access was established and premedication was administered if indicated. Routine monitoring and BIS monitoring (BIS-XP® Version 4.0, A2000 monitor, 15-s smoothing; Aspect Medical Systems Inc., Newton, MA) were commenced. Relaxant general anesthesia was induced and maintained with the drugs of the anesthesiologist's choice (see table 2for drugs chosen). Combined general and regional anesthesia was permitted. After tracheal intubation, intermittent positive-pressure ventilation was commenced. Anesthesiologists were encouraged to titrate anesthesia with the aim of a rapid recovery (i.e. , using clinical signs and a BIS of 40–60 during surgery22). At the conclusion of surgery, neuromuscular blockade was reversed, the patient's trachea was extubated, and the patient was transferred to the postanesthesia care unit. BIS monitoring continued until the first postoperative interview was completed.

Patients were interviewed by a blinded observer as soon as they were orientated to time, place, and person and again at 2–4 h postoperatively. At both interviews, patients were asked,1“What was the last thing you remember before going to sleep?”“What was the first thing you remember when you woke up?”“Can you recall anything between?” and “Did you have any dreams during your anesthetic?” If dreaming was reported, a narrative report was collected, and the characteristics of the dream were assessed (see Data Collection). If any evidence of awareness was encountered, this was managed according to hospital protocol, and an awareness report was completed for subsequent assessment by three independent and blinded anesthesiologists. Awareness was defined as “unlikely,”“possible,” or “probable” by each adjudicator. Patients were also interviewed in person or by telephone 24 h after anesthesia about quality of recovery (see Data Collection).

Baseline data included demographic and surgical details, preoperative medications, preoperative quality of recovery scores (a validated nine-item questionnaire on quality of recovery with a minimum score of 0 and a maximum score of 18; collection of a preprocedure baseline score is also validated23), hospital anxiety and depression scores (HAD; range: each section 0–21, total 0–4224), educational level (1 = all or some of primary school; 2 = all or some of secondary school; 3 = some university education), home dreaming recall frequency (0 = never; 1 = less than once a month; 3 = two or three times a month; 4 = about once a week; 5 = several times a week; 6 = almost every morning25) and risk factors for awareness, including a history of awareness, heavy alcohol use, the use of jet ventilation, and anticipated difficult intubation.4 

Intraoperative data included times of anesthetic induction, first surgical incision, and completion of wound closure; drug doses and concentrations; and clinical signs of inadequate anesthesia. Postoperative data included times to eye opening, eligibility for postanesthesia care unit discharge (Aldrete score ≥ 926) and each postoperative interview (all defined from the time of completion of wound closure), quality of recovery score, satisfaction with anesthetic care (100-mm visual analog scale at 2–4 h postoperatively), and HAD score at 24 h postoperatively.

Dreaming during anesthesia was defined as any experience that was described by the patient as dreaming and was thought by the patient to have occurred between the induction of anesthesia and the first moment of consciousness after anesthesia.27Awareness was defined as postoperative recall of intraoperative events. All patients who reported dreaming were considered to be “dreamers” for the purpose of the analyses, whether or not they could remember the narrative of the dream. However, only dreaming reports where the narrative was remembered were classified using five-point Likert scales as follows28:

• Emotional content (1 = very negative; 5 = very positive)

• Memorability (1 = can't remember narrative of dream; 5 = most memorable ever)

• Visual vividness (1 = not at all vivid; 5 = most vivid ever)

• Amount of sound (1 = no sound; 5 = most sound ever)

• Emotional intensity (1 = not at all intense; 5 = most intense ever)

• Meaningfulness (1 = not at all meaningful; 5 = most meaningful ever)

• Amount of movement (1 = no movement; 5 = most movement ever)

• Strangeness (1 = not at all strange; 5 = most strange ever)

Bispectral Index and signal quality data were recorded from anesthetic induction until completion of the first postoperative interview. The data downloaded from the BIS® monitor's memory were the averages of all the displayed BIS values (i.e. , values with signal quality > 15) over the minute before the time stamp. The phases of anesthesia were then defined in each patient as follows (fig. 2):

• Induction: from induction of anesthesia to the lowest BIS value following the initial rapid decrease in BIS after induction

• Maintenance: from the end of the induction phase to the time of completion of wound closure

• Recovery: from the time of completion of wound closure to the time of the first postoperative interview

• Following phase definition, recordings with signal quality below 50 were excluded. The percentage of total data excluded from the maintenance and recovery phases was calculated for each patient.

A sample size of 300 allowed an adequate sample of dreaming patients at both interviews, based on an incidence of dreaming of 5% at 2–4 h4and with anticipation of a higher incidence at the immediate postoperative interview, giving 96% power to detect a difference of 5 BIS units and 85% to detect a difference of 4 BIS units (SD = 5) between dreamers and nondreamers.

The data analyses were planned prospectively and were performed using Stata 8.2 (Stata Corporation, College Station, TX). Continuous variables were graphed to determine their distribution. Normally distributed variables were described using mean and SD and compared using the Student t  test. Skewed variables were described using median and range (or interquartile range) and compared using the Wilcoxon rank sum test. Survival data (time to an event) were described using median and range (or interquartile range) and compared using the log-rank test. Categorical variables were described using number and percentage and compared using the chi-square or Fisher exact test. Clinically significant predictors of dreaming from the univariate analyses with P  values less than 0.2 were included in multivariate logistic regression models. Clinically significant interactions were tested, and significant interactions were included in a final parsimonious model. P < 0.05 was considered statistically significant.

Three hundred patients finished the study with complete BIS data (fig. 1). The signal quality of the BIS data was above 50 for 95% of recordings. The median amount of data removed for signal quality less than 50 from the maintenance phase in each patient was 1.7% (0–62%), and that from the recovery phase was 0.001% (0–100%). The baseline, intraoperative, and postoperative characteristics of these patients are presented in tables 1–3.

Sixty-five (22%) of the 300 patients reported dreaming during anesthesia at the first interview, and 74 (25%) of the 300 patients reported dreaming during anesthesia at the second interview. Fifty-three patients reported dreaming at both interviews, 33 patients reported dreaming at one interview, and 214 patients did not report dreaming at either interview. A dream narrative was provided by 47 patients (16%) at the first interview and 53 patients (18%) at the second interview. Of the 35 patients who reported dream narratives at both interviews, 31 patients reported the same dream at both interviews, and 4 patients reported different dreams at each interview.

Median BIS values (median of medians, 37; range of medians, 20–59; interquartile range of medians, 32–41) during anesthesia were lower than recommended in the protocol. However, BIS values during maintenance of anesthesia in dreamers and nondreamers (37 [23–55]vs.  38 [20–59]; P = 0.68) and the time at BIS values greater than 60 (0 [0–7]vs.  0 [0–31] min; P = 0.38) were similar (fig. 3and table 4). There were no significant differences in BIS values between patients under general anesthesia (n = 280) and those under combined general–regional anesthesia (n = 20), except that the BIS values at eye opening were higher in the combined general–regional anesthesia group (83 [60–98]vs.  87 [78–98]; P = 0.02).

Most dreams were pleasant, were not strange, and were meaningful to the patient (i.e. , the dream involved family, friends, work or recreation) (fig. 4and table 5). Few dreams were unpleasant (identified by bold type in table 5). Often, patients reported that they were dreaming just before they “woke up” and some said that their dream was interrupted by their anesthesiologist “trying to wake them up.” Although only a few dreams concerned hospitals and surgery, many dreams contained people talking or standing around the patient.

Only two dreams resulted in awareness reports. A female patient remembered dreaming about “driving on a road. The road just swallowed her up. The doctor said she was okay but the car was wrecked. She couldn't move—she was trying to tell the driver to stop but he couldn't hear her…” This patient moved and developed tachycardia during abdominal closure, coinciding with less than 1 min of BIS values near 60 (fig. 2). At this time, the anesthesiologist administered propofol and told the patient, “Everything is okay.” Although the patient believed that she had been dreaming, an awareness report was completed, and all three adjudicators believed that awareness was “possible.” Another female patient thought that she woke up briefly during the surgery and was asked a question. All three adjudicators believed that awareness was “unlikely.” A further female patient dreamed about being a fish. For most of the operation, the surgical team had been discussing fishing trips. This episode did not result in an awareness report.

Univariate factors associated with dreaming during anesthesia at the first interview were younger age, lower American Society of Anesthesiologists physical status, higher home dream recall frequency, and lower HAD scores (table 1). The differences in HAD scores were not considered to be clinically significant. Dreamers were more likely to have received propofol maintenance and regional anesthesia but were no more likely to have displayed intraoperative signs of inadequate anesthesia than nondreamers (table 2). Dreamers opened their eyes sooner than nondreamers, but there were no differences in the time to the first interview or the time to postanesthesia care unit discharge. There were no clinically significant differences between dreamers and nondreamers in satisfaction with anesthetic care, postoperative quality of recovery or HAD scores, or changes in quality of recovery or HAD scores (table 3).

In a multivariate model, factors associated with dreaming were younger age, male sex, lower American Society of Anesthesiologists physical status, dreaming at home almost every day, receiving propofol maintenance or regional anesthesia, and having a shorter time to eye opening after anesthesia (table 6). Interactions between variables were assessed, and a significant interaction between sex and time to eye opening was found: 20% of women reported dreaming regardless of the time at which they opened their eyes after anesthesia. In contrast, 36% of men who opened their eyes less than 14 min after surgery reported dreaming, whereas only 12% of men who opened their eyes more than 14 min after surgery reported dreaming.

Our study is the first large-scale and detailed investigation of the relation between depth of anesthesia and dreaming, using a widely validated monitor of anesthetic depth.22Dreaming during anesthesia was not usually related to BIS values indicating light anesthesia, nor was it associated with clinical signs of light anesthesia.9In fact, dreaming occurred despite relatively deep anesthesia, and there were few arousals to BIS levels usually associated with awareness during anesthesia.4 

The incidence of dreaming among patients interviewed immediately upon emergence from anesthesia was 22%. This is consistent with previous studies in which patients were interviewed immediately after emergence.13,16–19We attribute the high incidence of dreaming at our 2–4 h interview3,9to reinforcement of memories of dreaming produced by the first interview, because dreams are usually hard to remember unless actively recalled.25,29Dreams reported by patients at the second interview, and not the first, may have occurred between the two interviews and not during anesthesia at all, although residual anesthetic effects at the first interview may have impaired memory retrieval in some cases. We believe that, to most accurately assess the incidence of dreaming during anesthesia, patients should be interviewed immediately after they emerge from anesthesia.

Despite the high incidence of dreaming, patients do not often spontaneously disclose that they have been dreaming when they emerge from anesthesia.30This may be because patients equate anesthesia with sleep and therefore do not find dreaming remarkable; or because they quickly forget their dreams, are embarrassed by their content, or are preoccupied with the outcome of the surgery, pain, or vomiting.

The form of anesthetic dreams has not been assessed in detail before. Most dreams in our study were similar in form to the dreams of rapid eye movement sleep, except that they were shorter and less strange.28,29In this respect, they more closely resembled the dreams of sleep onset and slow-wave or non–rapid eye movement sleep.29Most dreams were pleasant,14,15,17,31unpleasant dreams were unusual, and dreams that included surgical themes or events occurring during anesthesia were rare. Some patients could not remember the narratives of their dreams, which is consistent with previous studies13,16and the dreams of sleep.29 

Our findings suggest that there are two types of dreams associated with anesthesia. The first are rare “near-miss awareness” dreams, and the second, which are much more common, are “recovery” dreams that occur during emergence from anesthesia. Both can be identified in this study.

The first type of dream is well illustrated by the patient whose dream included the anesthesiologist's voice and a sensation of paralysis. Just as dreams of sleep can incorporate contemporaneous sensory input (such as an alarm clock), near-miss awareness dreams may also incorporate auditory and sensory stimuli, as both the primary and secondary auditory cortex remain responsive to auditory stimuli during anesthesia.32This type of dream may have decreased in incidence with improvements in anesthetic care (in a similar way to awareness).1,2,33Although rare, this type of dream probably still has a higher incidence than awareness3and may be prevented by clinical vigilance and depth-of-anesthesia monitoring.9However, in this study, BIS monitoring did not prevent the occurrence of near-miss awareness dreams.

The second type of dream may occur when patients are recovering from drug-induced hypnosis and may have entered physiologic sleep.11Evidence supporting “sleep” dreams during recovery from our study includes the following: (1) Patients who dreamt during anesthesia were similar with respect to age and health to patients who recall dreaming a lot during sleep. (2) Dreamers often reported that they were dreaming just before they “woke up.” (3) Dreams were short, simple, and not very strange, resembling dreams of sleep onset.29They were not hallucinations occurring during consciousness, like some drug-induced states.34(4) Dreaming during maintenance seems unlikely as deep levels of anesthesia were present without differential clinical signs of inadequate anesthesia in dreamers and nondreamers.

Patients who reported dreaming were younger and healthier and dreamed more often at home than patients who did not report dreaming. Our result is supported by sleep studies reporting that older, sicker patients experience difficulty falling asleep and recall fewer dreams,35,36and by previous studies of dreaming during anesthesia.3,9,31However, home dream recall frequency has not been reported to influence dreaming during anesthesia in previous studies.13,16,18,19If our hypothesis about recovery dreams being dreams of sleep is correct, patients who remember dreams at home should also remember their dreams after anesthesia.

Women often report more dreams after anesthesia than men.9,20,31A possible explanation is the propensity to higher dream recall after sleep in women25and the fact that women emerge from anesthesia faster than men37,38and hence are able to communicate their dreams sooner, before they are forgotten. However, in our study, men who emerged more quickly from anesthesia than women were more likely to report dreaming, whereas the opposite was true for men who emerged more slowly. This result is difficult to explain.

Propofol maintenance was a predictor of dreaming even after multivariate adjustment. Propofol maintenance was associated with higher incidences of dreaming in previous studies when compared with enflurane,18,19isoflurane,13,16or isoflurane, sevoflurane, or desflurane.9A possible explanation for such findings is that patients receiving propofol maintenance emerged more rapidly from anesthesia than patients receiving isoflurane or enflurane.19However, our current and previous studies have supported an increased incidence of dreaming in the propofol group even when compared with desflurane.9Differences in the pharmacologic actions of propofol and the volatile agents in this respect require further study. Regional anesthesia was associated with increased odds of dreaming in this study, but this result should be interpreted with caution because the numbers were small. This result awaits confirmation by a larger study.

In our study, patients who opened their eyes on command sooner were more likely to report dreaming. However, whereas times between completion of wound closure and eye opening were shorter in dreamers, times between completion of wound closure and the first interview were similar (and therefore our calculated BIS gradients were similar). In addition, dreamers had higher BIS values at the first interview. Dreaming patients therefore spent a longer time in a lightly unconscious or perhaps sleep state11between opening their eyes and being interviewed, during which time they might have dreamed. The exact physiologic state of these patients is speculative because their raw electroencephalograms have not been inspected for drug-induced or physiologic sleep features and BIS monitoring cannot accurately differentiate between sleep and anesthesia.39 

Dreaming is a subjective experience, both for the dreamer and for the dream interpreter. It was difficult for the interpreter to determine whether dreaming truly represented near-miss awareness, especially if the patient believed that he or she was dreaming. Similarly, it was difficult for the interpreter to determine when dreaming occurred, even if the patient was sure that dreaming occurred during or after anesthesia. In addition, we may have modified the incidence, content, or form of dreams by informing patients about dreaming during the consent process.

We chose to use BIS as measure of anesthetic depth in this study because (1) BIS monitoring is widely available in our centers, (2) BIS monitoring is validated as a reasonable measure of the depth of hypnosis,22and (3) we used BIS monitoring in our previous study on dreaming and wanted to explore the issue further.9However, BIS monitoring may not have captured changes in the electroencephalogram that may be associated with dreaming, in particular sleep-like features during recovery. As mentioned above, collection of raw electroencephalographic data would be necessary for this purpose.

Although a BIS range of 40–60 was recommended, the median BIS during maintenance was 37 (interquartile range, 32–41). This deep level of anesthesia provided us with a narrow range of anesthetic depths with which to make predictions and may have minimized the number of near-miss awareness reports. We were also not able to identify whether higher median BIS values predicted a faster recovery, as reported previously.38Finally, because of the numerous multivariate analyses, an inflated type I error cannot be excluded. However, only significant univariate predictors of clinical significance were highlighted or included in multivariate models.

In conclusion, dreaming during anesthesia is a common side effect of anesthesia that is harmless, pleasant, and unrelated to the depth of anesthesia in the majority of cases. The similarities between most anesthetic dreams and the dreams of sleep onset suggest that anesthetic dreaming occurs during recovery when patients are in a lightly sedated or physiologic sleep state. Our results are therefore reassuring to patients and their carers, who occasionally may equate dreaming with inadequate anesthesia.

The authors are grateful for assistance with recruitment and data collection by David Saunders, M.B., B.S. (Research Fellow, Department of Anaesthesia and Pain Medicine, Royal Perth Hospital, Perth, Western Australia, Australia); Karen Scott, M.B., B.S., and Sebastian Chua, M.B., B.S. (Research Fellows, Department of Anaesthesia, King Edward Memorial Hospital for Women, Perth, Western Australia, Australia); Susan Bugler, R.N., and Martin Feckie, R.N. (Research Nurses, Department of Anaesthesia and Pain Management, Royal Melbourne Hospital, Parkville, Victoria, Australia); Shauna Fatovich, R.N., and Susan March, R.N. (Research Nurses, Department of Anaesthesia and Pain Medicine, Royal Perth Hospital); and Desiree Cavill, R.N., and Tracy Bingham, R.N. (Department of Anaesthesia, King Edward Memorial Hospital for Women).