Effect of ethnicity on the hypnotic and cardiovascular characteristics of propofol induction


  • Presented in part at the Association of Anaesthetists of Great Britain and Ireland Annual Congress, Liverpool, September; 2009

Andrew P Morley
Email: andrew.morley@gstt.nhs.uk


We compared the propofol dose causing loss of verbal response and suppression of bispectral index to 50, between 50 white and 50 black patients, aged 18–65 years. Propofol was administered at 40 mg.kg−1.h−1and reduced to 8 mg.kg−1.h−1 when bispectral index fell to 50. We recorded heart rate and mean arterial pressure for 15 min in total and calculated, for this period, maximal percentage change from baseline for each. A statistician, blinded to patient ethnicity, found mean (SD) propofol dose for loss of verbal response in white and black patients to be 1.41 (0.37) mg.kg−1 and 1.16 (0.25) mg.kg−1, respectively (p < 0.001). Corresponding figures for maximal percentage change in heart rate were 14.1 (12.6) % and 7.5 (14.0) % (p = 0.015). Other differences were non-significant. The dose of propofol required for loss of verbal response, but not for suppression of bispectral index to 50, is lower in black than in white patients.

In the last decade it has been suggested that ethnicity may affect the clinical response to general anaesthetics [1–4], with Caucasian subjects generally shown to be less susceptible to the anaesthetic effects of propofol than other races [1–3]. To date, however, the effect of ethnicity on the induction dose of propofol has not been fully investigated. In our earlier study on anxiety and propofol induction [5], we used multivariate analysis to investigate factors independently predictive of propofol dose requirement for loss of verbal response. In doing so, we established that white ethnicity was associated with greater propofol dose requirement than other ethnic groups.

To investigate this incidental finding further, we studied a new group of patients using a study design better suited to address specifically the issue of ethnicity. Our aim was to determine whether the induction dose of propofol differs between British patients describing themselves as white and black. Our null hypothesis was that ethnicity does not affect the dose of propofol required to produce loss of verbal response. We also sought to establish whether the initial cardiovascular characteristics of propofol anaesthesia differ between white and black patients.


With Research Ethics Committee approval and written informed consent, we recruited 269 patients of ASA physical status 1-2, aged 18–65 years, scheduled for surgery under general anaesthesia. We did not study patients who were of neither white nor black/black British ethnicity and those who were diabetic, pregnant, deaf, obese (body mass index > 35 kg.m−2) or hypertensive on a single pre-operative ward reading (systolic pressure ≥ 140 mmHg or diastolic pressure ≥ 90 mmHg [6]). Those with conditions requiring rapid sequence induction were excluded, as were those taking antihypertensives or medication with central nervous system effects. We also did not study patients using recreational drugs, either regularly or within the preceding 48 h, and female and male patients consuming > 14 or > 21 units of alcohol per week, respectively. No patient had been enrolled in our previously published studies.

For each patient, age, sex and weight were recorded pre-operatively. Patients were also asked to select their ethnic group from the hospital’s diversity monitoring list. This includes 70 individual designations in five broader categories –‘white’, ‘black or black British’, ‘Asian or Asian British’, ‘mixed background’ or ‘other ethnic groups’. Self identified ethnicity correlates well with ancient geographic ancestry, a major determinant of genetic structure [7].

No pre medication was given. In the anaesthetic room, a 20-G intravenous cannula was inserted into one antecubital fossa. A pulse oximeter, 3-lead electrocardiography and non-invasive blood pressure monitoring were applied, the latter to the contralateral arm unless precluded by surgical reasons. We recorded baseline mean arterial pressure (MAP) and heart rate. 4 l.min−1 oxygen was administered through a Hudson mask. A bispectral index (BIS) sensor was applied and connected to an A-2000 electroencephalographic monitor with software revision 3.31 (Aspect Medical Systems, Norwood, MA, USA) and a smoothing time of 15 s.

The rationale for our induction protocol is presented in full elsewhere [5] but, in short, it allowed us to discriminate individual dose requirements and allowed for complete mixing of propofol in the central pharmacokinetic compartment [8]. 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 connected to the patient’s intravenous cannula. At induction, propofol 40 mg.kg−1.h−1 was administered by the infusion pump. Blood pressure and heart rate recordings were also commenced at 1-min intervals, except in those patients where the cuff was ipsilateral to the intravenous cannula. Finally, the same digital voice recording was played to each patient at a standardised volume, comprising the question ‘Are you awake?’ repeated at 5-s intervals. Patients had been instructed in advance to answer ‘yes’ to this question each time they heard it.

The dose of propofol was noted at the first loss of verbal response (PDLV) and also when BIS first fell below 50 (PDBIS50). The propofol infusion rate was then reduced to 8 mg.kg−1.h−1 and minute-interval blood pressure and heart rate monitoring was commenced in patients with ipsilateral blood pressure cuff and intravenous cannula. In all patients blood pressure and heart rate monitoring were continued until 15 min after the start of induction, all data being recorded manually by the investigator.

Airway support was provided during induction when required, without instrumentation. Intravenous ephedrine 6 mg was given in the event of clinically important hypotension, defined as a decrease in MAP from baseline of > 40% with MAP < 70 mmHg, or MAP < 60 mmHg [9]. Fifteen minutes after starting propofol, data collection ceased. Routine anaesthesia was administered thereafter by the operating list anaesthetist.

From the 15 min of cardiovascular data from each patient, the maximal percentage decreases from baseline MAP (max%ΔMAP) and heart rate (max%ΔHR) were calculated. The point at which the maximal percentage decrease in MAP occurred was also noted (tmax%ΔMAP).

Statistical analysis

The nature of the study meant that blinding of the investigators at induction was deemed impractical. Instead, data from the 50 patients selecting an ethnic designation in the black/black British category were entered into database A and data from the 219 patients selecting a white/white British designation were entered into database B. A statistician, unaware of the nature of the study and the significance of the two databases, used stratified randomisation to select at random 50 patients from database B, to balance age and sex with those of patients in database A. These two groups were designated ‘white group’ (n = 50) and ‘black group’ (n = 50), respectively. In this way, we attempted to introduce an element of blinding at the patient selection stage.

Blinding was sustained throughout data analysis, the statistician still being unaware of the factor distinguishing the two groups. spss 14.0 for Windows (SPSS, Chicago, IL, USA) was used to compare the two groups in terms of our primary endpoint, PDLV, and secondary endpoints PDBIS50, max%ΔMAP and max%ΔHR with unpaired t-tests. Since Tmax%ΔMAP did not follow a Gaussian distribution it was compared using a Mann–Whitney U-test. Categorical variables were compared with chi-squared tests. Values of p < 0.05 were deemed significant.

Data from our earlier study informed the power analysis, in which mean (SD) PDLV in patients of white ethnicity was 1.4 (0.34) mg.kg−1. We calculated that a sample size of 100 would provide 80% power to detect a 16% reduction in PDLV, using two-tailed α = 0.05. We felt that a difference of this size would be relevant clinically, as well as provide evidence for the general principle that genetic factors influence anaesthetic susceptibility. We recruited according to the criteria described until we had a cohort of 50 black patients. By that time, we had also recruited 219 white patients.


Data from 100 patients, selected as described, were analysed initially. Patients’ characteristics and baseline data are presented in Table 1 and did not differ significantly between groups. During induction, no patient required ephedrine for hypotension. Airway support was necessary in 11 (22%) white patients and 16 (32%) black patients. One white patient required brief manual ventilation of the lungs. Apart from sex distribution (p = 0.011), baseline data from the white group (male:female = 10:40) were not significantly different from the other 169 white patients (male:female = 67:102) in database B.

Table 1.   Characteristics and baseline data of white and black patients (n = 100). Values are number or mean (SD). No significant differences between groups.
 White group (n = 50)Black group (n = 50)
Sex; male:female10:409:41
Age; years37 (11)36 (8)
Weight; kg72 (12)73 (11)
Baseline mean arterial blood pressure; mmHg98 (8)97 (10)
Baseline heart rate; min−175 (13)75 (12)

Between-group comparisons of PDLV and PDBIS50 are illustrated in Table 2. Mean PDLV was significantly lower in the black group than in the white (p < 0.001). Table 3 shows the cardiovascular effects observed in the two groups during the 15-min observation period. Mean maximum percentage decrease in heart rate was greater in the white group than that in the black group (p = 0.015). No other differences were observed.

Table 2.   Induction dose of propofol in white and black patients. Values are mean (SD) or number (95% CI).
 White (n = 50)Black (n = 50)Difference between meansp value
  1. BIS, bispectral index.

Propofol dose for loss of verbal response; mg.kg−11.41 (0.4)1.17 (0.25)−0.24 (−0.37 to −0.12)< 0.001
Propofol dose for suppression of BIS to 50; mg.kg−11.91 (0.44)1.77 (0.44)−0.14 (−0.32 to 0.03)0.11
Table 3.   Cardiovascular effects of propofol induction and subsequent infusion in white and black patients. Values are mean (SD), median (IQR[range]) or number (95% CI).
 White group (n = 50)Black group (n = 50)Difference between meansp value
Maximal percentage decrease in MAP; %27 (7)25 (6)−2 (−6 to 1)0.72
Maximal percentage decrease in heart rate; %14 (13)7 (14)−7 (−11.9 to −1.3)0.015
Time to maximal decrease in MAP; min13 (7–15 [3–15])11 (6–14 [3–15])−20.20


In this study, the induction dose of propofol was lower in black than in white patients, when unconsciousness was defined as loss of verbal response. However, when unconsciousness was defined using an electroencephalographic threshold, namely BIS < 50, no ethnic effect was observed.

While ours is the first study specifically to address propofol induction and ethnicity, ethnic factors are known to influence intra-operative and recovery characteristics of propofol anaesthesia. To maintain a similar intraoperative BIS, Malaysian Indians require less propofol than their Chinese or indigenous counterparts, or than Italian Caucasians [3]. Black African patients recover more slowly from propofol anaesthesia than Caucasians [1, 2]. Within the last few years, a study has shown ethnicity to affect the minimum alveolar concentration of sevoflurane [4], a finding discussed in an associated editorial [10].

The clinical effect we observed may be the result of genetic and/or environmental factors. We attempted to reduce the impact of the latter by not studying patients misusing drugs or alcohol, which may itself be affected by ethnicity. Without total abstinence as an inclusion criterion, however, it is still possible that the greater PDLV in white patients might have been linked, for example, to greater alcohol consumption. Chronic alcoholism is known to increase propofol requirements [11].

Even if we assume that the effect of ethnicity on PDLV in our patients was a direct genetic one, the study did not allow us to determine whether pharmacokinetic or pharmacodynamic factors are responsible. This might have been clearer had we measured plasma propofol concentrations. Racial pharmacokinetic differences have long been recognised in a range of medical specialties [12, 13]. As for pharmacodynamics, many polymorphisms in receptor genes exert clinical effects. For example, patients with the A118G polymorphism in the mu-opioid receptor gene, OPRM1, require more postoperative morphine than those without [14]. Other OPRM1 polymorphisms may also affect pain-related endpoints [15]. The frequency with which the A118G polymorphism occurs differs markedly between racial groups [16] and it seems reasonable to suggest that this may account in some degree for the well-known inter-ethnic differences in analgesic requirements. [17]. It is conceivable that polymorphisms in genes coding for central nervous system receptors that respond to propofol, such as gamma-aminobutyric acid receptor subunits, might similarly contribute to ethnic variations in propofol susceptibility.

Is it possible that the effect we observed was due to some other confounding influence, entirely unrelated to ethnicity? In our earlier article [5], we used multiple regression analysis to identify factors independently associated with PDLV. These were sex, age, weight, baseline heart rate and MAP, and white ethnicity. In the current study, the randomisation method ensured no significant differences between groups in terms of age or sex distribution. The propofol dose was indexed to weight and, although the size of our white database did not allow weight matching between groups, mean group weight was not significantly different between white and black patients. The same was true of baseline MAP and heart rate. While our study design minimised the effect of many known confounders, others may have affected our results. In particular, body mass index was used solely as a recruitment criterion – data were not retained for analysis. It is possible that our findings would have been different had propofol dose been indexed to lean body mass, rather than to weight.

We chose to study two hypnotic endpoints of practical importance during propofol administration. In clinical terms, PDLV might be deemed more relevant to sedation and PDBIS50 to anaesthesia. Our findings indicate that ethnicity affects PDLV but not PDBIS50; this is not wholly unexpected as it is consistent with our earlier work [5]. Our failure in the current study to detect an effect of ethnicity on PDBIS50 is not likely to be due to sample size. Post hoc calculations indicate that this was sufficient to detect a 13% reduction in PDBIS50 at β = 0.2, although we may, of course, have failed to detect a genuine effect smaller than this.

Bispectral index monitoring effectively quantifies a combination of EEG-derived variables, previously shown to correlate with a clinical hypnotic state. Though EEG variables specifically relevant to loss of verbal response were used to develop the BIS algorithm [18], they were not the only ones. It might be reasonably argued, then, that BIS effectively measures something more than, or even different from, the likelihood of verbal response during anaesthesia. This may account for the disparity between ethnic effects on PDLV and PDBIS50 in our patients. Furthermore, the relationship between the PDLV and PDBIS50 may be non-linear or have different population distributions.

These potential explanations would equally apply to other anaesthetic endpoints, had we chosen to study them and observed similar discrepancies in ethnic effect. Therefore, we believe no conclusions should be drawn from the current study about the influence of ethnicity on other propofol actions such as respiratory depression, the suppression of airway reflexes or the prevention of awareness.

We also demonstrated an effect of ethnicity on heart rate changes at induction. Despite similar baseline heart rates, the decrease in rate during propofol induction was greater in white than in black patients. Our earlier study employed similar patients and identical methodology but on that occasion, we did not find an ethnic effect on heart rate changes [5]. It is hard to account for this disparity. Ethnicity affects many aspects of cardiovascular physiology, pathology and pharmacology [19], and in the current study we may have identified a genuine effect, though at p = 0.015, the level of significance is modest.

One aspect of our study design that merits comment is blinding. It is very difficult to devise a study in which observers are unaware of participants’ ethnic group. Strategies adopted by other authors include the recruitment of different groups by different investigators in different countries [3]. Partial blinding was achieved by Ezri et al. in another, single-centre study of ethnicity and the minimum alveolar concentration of sevoflurane [4]. Observers were not apparently blinded to patients’ ethnic group but were unaware of sevoflurane concentration when recording whether movement on incision occurred. Our approach of blinding the statistician performing the analysis mitigated these potential biases.

In conclusion, our finding that the dose of propofol required for loss of verbal response is lower in black patients than in white ones adds to the existing literature on the relevance of ethnicity to anaesthetic effect. The implications have been explored at length by other authors [10] but they bear reiterating in brief. Genetic differences between individuals have been shown to influence anaesthetic sensitivity, and from this emerges the prospect of investigating specific genes in studies that may shed light on molecular mechanisms of general anaesthesia.

Competing interests

AM has previously received an honorarium for speaking at an educational event from Aspect Medical Systems, who provided without charge some (< 10%) BIS sensors for this study. Otherwise, this study was funded from departmental resources.