The data has been presented in part by the first author at the First World TIVA-TCI Congress, Venice, 27–29 September 2007 under the same title.
The effect of pre-operative anxiety on induction of anaesthesia with propofol*
Article first published online: 11 APR 2008
© 2008 The Authors. Journal compilation © 2008 The Association of Anaesthetists of Great Britain and Ireland
Volume 63, Issue 5, pages 467–473, May 2008
How to Cite
Morley, A. P., Papageorgiou, C. H., Marinaki, A. M., Cooper, D. J. and Lewis, C. M. (2008), The effect of pre-operative anxiety on induction of anaesthesia with propofol. Anaesthesia, 63: 467–473. doi: 10.1111/j.1365-2044.2007.05402.x
- Issue published online: 11 APR 2008
- Article first published online: 11 APR 2008
- Accepted: 8 November 2007
In this prospective study, we investigated the effects of anxiety on the induction dose of propofol and subsequent cardiovascular changes in 197 patients. Pre-operative state and trait anxiety scores were measured using the State Trait Anxiety Inventory. Propofol was administered at 40 mg.kg−1.h−1. Propofol dose was recorded at loss of verbal response and when EEG Bispectral Index decreased to 50. Thereafter, propofol infusion rate was reduced to 8 mg.kg−1.h−1. Cardiovascular data were collected for 15 min after starting induction. Maximum percentage decreases in heart rate and mean arterial pressure, and the point at which the latter occurred, were recorded. On multivariate analysis, anxiety scores did not significantly affect propofol dose or cardiovascular end-points, although Bispectral Index at loss of verbal response decreased with increasing trait anxiety (p = 0.02). Anxiety, measured using State Trait Anxiety Inventory, does not appear independently to affect the induction characteristics of propofol.
Anxiety has long been recognised as common in pre-operative patients  and has been previously investigated with respect to its impact on postoperative outcomes [2, 3]. In one American study by Maranets and Kain, pre-operative anxiety was shown to affect the propofol dose required to achieve and maintain a Bispectral Index (BIS) between 40 and 60 .
BIS monitoring is not as common in the UK as it is in the US. This being the case, we designed a prospective observational study to determine the influence of pre-operative anxiety on propofol requirements for induction, with unconsciousness judged by clinical as well as electroencephalographic criteria. Furthermore, as pre-operative anxiety is associated with a rise in plasma adrenaline concentrations , we also aimed to establish whether anxiety affects the haemodynamic characteristics of propofol induction.
Patients and methods
Local Research Ethics Committee approval and informed written consent from patients were obtained. A total of 197 ASA I or II patients, aged 18–65 years and scheduled for minor surgery under general anaesthesia, were enrolled in the study. Pregnant and deaf patients were excluded, as were those in whom rapid sequence induction was indicated. Patients on antihypertensive medication, or patients found to be hypertensive on pre-operative assessment, were excluded. Hypertension was defined as systolic pressure ≥ 140 mmHg or diastolic pressure ≥ 90 mmHg  on a single measurement. Patients were excluded if their alcohol consumption exceeded 21 units.week−1 for males or 14 units.week−1 for females, or if they admitted to taking recreational drugs, either regularly or within the last 48 h. Patients on regular therapeutic medication with central nervous system effects were also excluded.
On the surgical ward, patients were weighed and demographic data were recorded for each patient including age, gender, smoking status and ethnic group. Patients were asked to select their ethnic group from the list used by the hospital for ethnic diversity monitoring. This includes 70 individual designations in five broader categories – white, black/black British, Asian/Asian British, ‘mixed background’ or ‘other ethnic group’. The proposed sequence of events at induction was explained.
Patients were then asked to complete the State-Trait Anxiety Inventory (STAI) . This is a well-established instrument for the self-reporting of anxiety. It has been used in the pre-operative context . It comprises two sets of 20 statements. The first set relates to the immediate situation – state anxiety – and includes ‘I feel calm’ and ‘I am worried’. To each statement, the subject is required to select one out of four responses: not at all, somewhat, moderately so or very much so. The second set of statements is intended to reflect underlying longer-term, or trait, anxiety. Examples include ‘I feel pleasant’ and ‘I am a steady person’. Again, the subject is required to select a single response: almost never, sometimes, often or all the time. A score for each, ranging between 20 and 80, may then be calculated by an investigator using a scoring key.
Patients in our study were not premedicated. In the anaesthetic room, a 20-G intravenous cannula was inserted into one antecubital fossa. Non-invasive blood pressure (NIBP) monitoring was applied to the contralateral arm, except in patients with a hand injury, and a 3-lead electrocardiograph and pulse oximeter were attached. Baseline mean arterial pressure (MAP) and heart rate (HR) readings were recorded. A Hudson facemask was used to administer 4 l.min−1 oxygen. A BIS sensor was applied to the patient’s forehead and connected to an A-2000 electroencephalographic monitor with revision 3.31 software (Aspect Medical Systems, Norwood, MA, USA) for BIS measurement. A smoothing time of 15 s was selected.
A syringe filled with propofol 400 mg (Propofol-Lipuro 1%, B. Braun Melsungen, Melsungen, Germany) was seated in an Asena Mk III intravenous infusion pump (Alaris Medical Systems, Basingstoke, UK) and attached to a 100 cm infusion tubing This was primed with propofol using the infusion pump and connected to the patient’s intravenous cannula. The volume on the infusion pump was then set to zero and the infusion rate set to 40 mg.kg−1.h−1, with the pump on hold.
At induction, three devices were started at the same time. The first was the intravenous infusion pump, on whose display panel the total infused volume was continuously displayed. Simultaneously, NIBP and HR recording at 1-min intervals were commenced, except in those patients where the cuff was sited on the same arm as the intravenous infusion. A digital voice recorder was also switched on, with the volume standardised at a normal speaking tone. The same recording was played to each patient, comprising the question ‘Are you awake?’ repeated at 5-s interval. Patients had been instructed in advance to answer ‘yes’ to this question each time they heard it. We chose loss of verbal response as a clinical end-point because it is used by most anaesthetists in everyday practice, unlike some of the alternatives e.g. syringe dropping.
Immediately the patient first failed to respond to the question, the volume displayed on the pump was recorded (Propofol dose required for loss of verbal response (PDLV)), as was the BIS at this point. When the BIS first fell > 50, the infused volume was again noted (Propofol dose required for a BIS of 50 (PDBIS50)) and the syringe inspected to confirm as far as possible that the displayed volume was correct. The investigator recording this data was unaware of the STAI scores at the time. The propofol infusion rate was reduced to 8 mg.kg−1.h−1. At this point, minute-interval NIBP and HR monitoring was commenced in those patients with an NIBP cuff and intravenous infusion on the same arm. In all cases, NIBP and HR monitoring was continued until 15 min after the start of induction and the data manually transcribed to the patient’s research record.
Our infusion regimen was designed for simplicity. We believed that the chance of investigator error would be minimised with only one infusion rate change. We chose not to use a target-controlled infusion as this would have implicitly relied on the assumption that anxiety has no effect on propofol pharmacokinetics. While we wanted to reproduce routine anaesthetic induction as far as possible, initial bolus administration of propofol would not have allowed adequate discrimination of individual dose requirements. At a propofol infusion rate of 40 mg.kg−1.h−1, such discrimination is possible and furthermore, mixing of propofol in the central pharmacokinetic compartment is said to be complete . Our study maintenance rate of 8 mg.kg−1.h−1 was selected from the established ‘Bristol regimen’, in which it is the middle of the three sequential reducing infusion rates .
If necessary during the study period, the patient’s airway was supported without instrumentation. Clinically important hypotension, defined as a decrease in MAP from baseline of > 40% with MAP < 70 mmHg, or MAP < 60 mmHg , was treated with intravenous ephedrine 6 mg. Once 15 min had elapsed, observations were discontinued. Anaesthesia, thereafter, was administered according to clinical indications and personal preferences of the operating list anaesthetist.
From the 15 min of cardiovascular data in each patient, the maximal percentage decreases from baseline MAP and HR respectively were calculated. The point at which the maximal percentage decrease in MAP occurred was also noted. These end-points were chosen to reflect data routinely recorded during anaesthetic induction. The MAP in the first 10 min after induction has been shown to influence postoperative outcome  and summary variables have been recommended in preference to repeated measures .
Statistical analysis was performed using spss 14.0 for Windows (SPSS, Chicago, IL, USA). A p value < 0.05 was deemed significant. We used univariate linear regression to determine whether pre-operative state or trait anxiety scores were predicted by any of the demographic variables or baseline data illustrated in Table 1. Those variables for which univariate analysis demonstrated a significant effect at p < 0.05 were tested in a multivariate model. Co-linearity between variables was determined by measuring tolerance.
|Asian/Asian British||7 (3.6)|
|Black/black British||38 (19.3)|
|Not recorded||17 (8.6)|
|Age; years||37 (10)|
|Weight; kg||72 (14)|
|State anxiety score||39 (11)|
|Trait anxiety score||36 (9)|
|Baseline mean arterial blood pressure; mmHg||97 (10)|
|Baseline heart rate; beats.min−1||74 (14)|
In the same manner, we next conducted univariate and subsequent multivariate analyses on induction data from all patients. We tested, in turn, the effect of state and trait anxiety scores, gender (female = 1, male = 0), weight, age, smoking status (occasional and regular smokers = 1, non-smokers = 0), black ethnicity (black/black British = 1, non-black/black British = 0), white ethnicity (white = 1, non-white = 0), baseline MAP and baseline HR on the six end-points described above. For the three cardiovascular end-points, we also tested the effect of PDBIS50 to determine whether differences in propofol dosage might account for any cardiovascular findings – a total of 11 potential predictors.
Finally, using logistic regression, we also tested the same variables to see if they affected the need for airway support during the study period. This additional end-point was recorded as a binary variable (support required = 1, no support required = 0).
The power of multiple regression analyses depends on the number of predictors included in the model. We realized during study design that this number would not be available to us until we had collected the data and conducted univariate analyses on each of the 11 potential predictors. Our sample size was calculated so as to provide suitable power in the event of all 11 predictors meriting inclusion in a multivariate model. Under these circumstances, a sample size of 192 provides 95% power at α = 0.05 to detect the same r value (0.35) sought by Maranets and Kain .
We recruited 197 patients whose demographic details and baseline data are presented in Table 1. Cardiovascular data were incomplete in six patients. A further two patients required 6 mg ephedrine during the study period, after which point their cardiovascular data were not included in statistical analyses. In another patient, cardiovascular recordings after induction were abandoned and excluded from analysis as severe coughing necessitated supplementary propofol. One patient required insertion of an oropharyngeal airway and manual ventilation with oxygen and air via a Bains circuit for 2 min after induction. Transient BIS failure occurred during induction in 33 patients.
In multivariate models of pre-induction data, female gender (p < 0.001) and increased trait anxiety (p < 0.001) were significant predictors of increased state anxiety. Trait anxiety was higher in smokers (p = 0.002). For the calculation of multivariate predictive models of induction data, only patients with datasets complete for the variables tested were included. No colinearity was seen and the results were as follows.
Propofol dose required for loss of verbal response
On univariate analysis, neither state nor trait anxiety were significantly correlated with PDLV (p = 0.88 and 0.53 respectively). Significant univariate predictors of PDLV are listed in Table 2. All remained significant in a multivariate model. PDLV was higher in male patients and those of white ethnicity, increased with increasing weight and baseline HR and MAP, and decreased with increasing age.
|Univariate predictor entered in model||Unstandardised coefficient B||95% confidence intervals for B||p value|
|(a) Propofol dose for loss of verbal response|
|Gender (female = 1, male = 0)||−10.46||−17.00||−3.91||0.002|
|Age; years||−0.61||−0.90||−0.33||< 0.001|
|Weight; kg||0.62||0.41||0.84||< 0.001|
|Baseline MAP; mmHg||0.38||0.07||0.69||0.015|
|Baseline heart rate; beats.min−1||0.56||0.34||0.78||< 0.001|
|White ethnicity (white = 1, non-white = 0)||14.26||8.26||20.27||< 0.001|
|(b) Propofol dose for suppression of BIS to 50|
|Gender (female =1, male = 0)||−18.62||−29.74||−7.49||0.001|
|Weight; kg||0.86||0.50||1.23||< 0.001|
|Baseline MAP; mmHg||0.26||−0.23||0.76||0.299|
|Baseline heart rate; beats.min−1||0.79||0.42||1.16||< 0.001|
|(c) BIS at loss of verbal response|
|Trait anxiety score||−0.12||−0.23||−0.02||0.021|
Propofol dose required for a Bispectral Index of 50
On univariate analysis, neither state nor trait anxiety were significantly correlated with PDBIS50 (p = 0.40 and 0.86 respectively). Significant univariate predictors of PDBIS50 are listed in Table 2. In a multivariate model, baseline MAP did not retain significance. PDBIS50 was higher in male patients and increased with increasing weight and baseline HR.
BIS at loss of verbal response
State anxiety did not affect BIS at loss of verbal response (p = 0.26) on univariate analysis. Age and trait anxiety, however, did and the effect persisted for both in a multivariate model (Table 2). BIS at loss of verbal response was higher in older patients but decreased with increasing trait anxiety.
Maximal percentage decrease in MAP
State anxiety was a significant predictor of maximal percentage decrease in MAP on univariate analysis (p = 0.03) while trait anxiety was not (p = 0.27). This effect did not persist in a multivariate model. Other univariate predictors are listed in Table 3. The maximal percentage decrease in MAP was related positively to baseline MAP and negatively to weight.
|Univariate predictor entered in model||Unstandardised coefficient B||95% confidence intervals for B||p value|
|(a) Maximal percentage decrease in MAP|
|Weight; kg||−0.14||−0.21||−0.08||< 0.001|
|State anxiety score||0.06||−0.02||0.14||0.127|
|Baseline MAP; mmHg||0.32||0.24||0.41||< 0.001|
|(b) Maximal percentage decrease in heart rate|
|Gender (female =1, male = 0)||0.88||−3.17||4.94||0.668|
|State anxiety score||0.10||−0.05||0.25||0.174|
|Baseline heart rate; beats.min−1||0.57||0.43||0.70||< 0.001|
Time to maximal percentage decrease in MAP
On univariate analysis, neither state (p = 0.08) nor trait (p = 0.41) anxiety scores affected time to nadir MAP. Black ethnicity alone exerted a significant effect on time to maximal percentage decrease in MAP (p = 0.04), this event occurring earlier in black patients.
Maximal percentage decrease in HR
On univariate analysis, state (p = 0.03) but not trait (p = 0.23) anxiety predicted maximal percentage decrease in HR. Other univariate predictors are listed in Table 3. On multivariate analysis, the maximal percentage decrease in HR was related negatively to weight and age and positively to baseline HR.
Requirement for airway support after induction
Using univariate logistic regression, state (p = 0.04) but not trait (p = 0.72) anxiety predicted requirement for airway support. Other univariate predictors are listed in Table 4. Of these, gender, age and weight remained significant on multivariate analysis. The odds of male patients requiring airway support were about three times greater than those of females. The odds of requiring airway support increased by 10% for each year in age and by 7% for each kg in body weight.
|Univariate predictor entered in model||Odds ratio||95% confidence intervals||p value|
|Gender (female =1, male = 0)||0.31||0.13||0.74||0.008|
|Age; years||1.10||1.05||1.15||< 0.001|
|Weight; kg||1.07||1.03||1.10||< 0.001|
|State anxiety score||1.00||0.97||1.04||0.885|
|Baseline MAP; mmHg||1.01||0.97||1.05||0.792|
In this study, we have found that anxiety independently affects neither the propofol dose required for unconsciousness nor the cardiovascular events that follow propofol induction.
Two recent studies [4, 12] have also addressed susceptibility to the hypnotic effect of propofol using the STAI, which has been described by some as the ‘gold standard’ for the measurement of pre-operative anxiety . In the first of these studies, Maranets and Kain administered propofol to 57 patients to induce and maintain loss of consciousness, defined solely in electroencephalographic terms as a BIS of 40–60. Using multivariate analysis, they found that trait, but not state, anxiety affected propofol requirements to attain BIS 40–60. Anxiety scores in their study were higher than those in our own but for a similar target BIS, the mean (SD) dose of propofol required by our patients was very close to theirs at 1.91 (0.49) and 1.85 (0.45) mg.kg−1 respectively. This is despite differences in the precise method of propofol administration and the fact that we used a different formulation of the drug.
It is difficult to understand why our results differ from those of Maranets and Kain. Our study was considerably larger than theirs (n = 197). Power calculations with our regression analyses showed that the models gave 80% power to detect r2 between 0.054 (maximal decrease in MAP) and 0.076 (maximal decrease in HR). This compares favourably with Maranets and Kain , whose sample size resulted in 80% power to detect an r value of 0.35.
Aside from sample size, there are a number of methodological differences between the two studies. All patients in their study were female, unlike ours. Female patients in our study had higher state anxiety scores than males and gender was a significant predictor of both PDLV and PDBIS50. Gender is known to influence the clinical response to propofol, a subject covered in a recent review . However, the multivariate analysis performed on our data took into account any gender effect. Furthermore, when we conducted a subset analysis on our 123 female patients alone, state and trait anxiety still showed no effect on PDLV (p = 0.98 and 0.64 respectively) or PDBIS50 (p = 0.94 and 0.76).
In both studies, ethnicity was a significant predictor of propofol requirement and there is evidence that this factor affects clinical aspects of propofol anaesthesia [15, 16]. While multivariate analysis should account for any ethnic effects, it may be relevant that a third of the patients in Maranets and Kain’s study were Hispanic – a group almost entirely absent from our own study and therefore impossible to include specifically in our analyses.
We chose to study a clinical end-point of consciousness in addition to an EEG one. This was for two reasons. The first was that every anaesthetist uses clinical observation in daily practice. This is not yet the case with BIS. In addition, it seemed possible to us that the relationship between BIS and clinical end-points of consciousness might itself be altered by anxiety. Relaxation techniques affect both BIS and state anxiety in the absence of anaesthetic or sedative drugs . During propofol administration, we reasoned, an identical BIS could represent a different clinical conscious level in a highly anxious patient than in a calm one. We subsequently found some evidence to support this hypothesis. In our patients, BIS at the point of loss of verbal response decreased significantly with increasing trait anxiety, and increased with age. The p values for these relationships were modest and it is possible that they represent the effect of multiple testing.
Other than our own, the only recent study to address anxiety, propofol dosage and clinical indices of consciousness is by Hong et al. . They investigated propofol requirements not for anaesthesia but for conscious sedation, a state defined as ‘eyes closed, responds promptly to verbal commands’. As in our study, they were unable to show any relationship between state or trait anxiety and propofol dose. This was despite apparently recording STAI scores in the anaesthetic room, at a time closer to propofol administration than either Maranets and Kain  or ourselves. State and trait anxiety scores in the study by Hong et al. were similar to those in our own.
It is possible that pre-operative anxiety does affect propofol requirements but is inadequately measured by STAI. Hong et al. found that when anxiety was assessed by visual analogue scale rather than STAI, it did correlate weakly with propofol dose. We found baseline HR to be a highly significant predictor of both PDLV and PDBIS50 – tachycardia is a well-known feature of anxiety and also a determinant of cardiac output, which in turn affects propofol induction dose . PDLV also increased significantly with baseline MAP, another potential indicator of anxiety. Baseline cardiovascular variables were not reported or analysed by Maranets and Kain .
We investigated the effect of anxiety scores on three cardiovascular end-points during induction with propofol: nadir MAP, nadir HR and time to nadir MAP. None of these were significantly predicted by state or trait anxiety on multivariate analysis.
Other risk factors for hypotension and bradycardia after induction have been identified in two very large retrospective studies of clinical practice. The design of our own study minimised the impact of many of these factors in our patients but there are some interesting comparisons to be made. In keeping with both Hug et al. , and Reich et al. , we found that blood pressure changes were not associated with propofol induction dose. Hug et al.  found that Caucasian patients were more likely to become hypotensive after induction. We found that neither white nor black ethnicity affected the magnitude of the maximal decrease in MAP, but that this nadir occurred earlier in black patients.
Our findings differ from those in the two retrospective studies in some other respects. In both of them, higher arterial blood pressures appeared to be protective against hypotension after induction – in our study, the opposite was true. Baseline MAP was positively correlated with maximal decrease in MAP. This may be related to the fact that we specifically excluded hypertensive patients. We also failed to detect an association between MAP effects and female gender  or age [10, 19], although Hug et al.  only identified age as a risk factor at > 65 years – the upper age limit for recruitment in our own study.
Turning to HR changes, we found that patients with a higher baseline HR had a greater percentage decrease after induction. Hug et al.  found instead that low pre-operative HR was predictive of postinduction bradycardia, but this can be explained by the fact that they used an absolute definition for bradycardia – an HR of ≤50 beats.min−1 while we investigated changes relative to baseline. They also found female gender had an effect. We did not, but identified age and weight as predictors instead.
Analysis revealed an interesting incidental finding related to airway obstruction in our patients. In a multivariate model, male gender, weight and age independently predicted the need for airway support during the study period. While we were unable to find any other published studies demonstrating the effect of these factors on the likelihood of airway obstruction during propofol infusion, they are all known to increase the risk of critical respiratory events in the postanaesthesia care unit . Propofol is now widely used as a sedative agent outside the operating theatre. Non-anaesthetic staff may be involved in the monitoring of sedated patients or even in propofol sedation, particularly outside the UK. Under these circumstances especially, knowledge of factors likely to lead to airway problems is valuable.
In conclusion, we were unable to demonstrate any effect of anxiety, measured using STAI, on propofol requirements for loss of verbal response or BIS suppression. Furthermore, no effect of anxiety was apparent on cardiovascular variables following propofol administration. This may be because of the failure of the STAI adequately to reflect anxiety in the particular circumstances of our study. In the absence of evidence for this hypothesis, however, our study provides some reassurance that anxious patients behave similarly to their less anxious counterparts on induction of anaesthesia with propofol.
- 7State-Trait Anxiety Inventory (Form Y). Redwood City, CA, USA: Mind Garden, 1983..
- 9Induction and maintenance of propofol anaesthesia. A manual infusion scheme. Anaesthesia 1988; 43 (Suppl.): 14–7., , , , .
|ANAE_5402_sm_v63p467-473.pdf||610K||Supporting info item|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.