Volume 59, Issue 2 p. 166-172
Free Access

Improving anaesthetists' communication skills

C. Harms

C. Harms

Staff Anaesthetist

Search for more papers by this author
J. R. Young

J. R. Young

Biostatistician, Basel Institute of Clinical Epidemiology, University Clinics Basel, Kantonsspital, CH-4031 Basel, Switzerland

Search for more papers by this author
F. Amsler

F. Amsler

Consultant Psychologist

Search for more papers by this author
C. ZettlerD. Scheidegger

D. Scheidegger

Professor and Chairman, Department of Anaesthesia, University Clinics Basel

Search for more papers by this author
C. H. Kindler

C. H. Kindler

Staff Anaesthetist

Search for more papers by this author
First published: 16 January 2004
Citations: 47
C. H. Kindler
[email protected]


The attitude, behaviour and communication skills of specialised doctors are increasingly recognised as important and they have been identified as training requirements. We designed a programme to teach communication skills to doctors in a University Department of Anaesthesia and evaluated its effect on patient outcomes such as satisfaction and anxiety. The 20 h programme was based on videotaped reviews of actual pre-operative visits and role-playing. Effects on patient satisfaction and pre-operative anxiety were assessed using a patient questionnaire. In addition, all participating anaesthetists assessed the training. We provide evidence that the training increased patient satisfaction with the pre-operative anaesthetic visit. Training also decreased anxiety associated with specific aspects of anaesthesia and surgery, but the effect was rather small given the intense programme. The anaesthetists agreed that their interpersonal skills increased and they felt better prepared to understand patients' anxieties. Communication skills training can increase patient satisfaction and decrease specific anxieties. The authors conclude that in order to better demonstrate the efficacy of such a training programme, the particular communication skills of anaesthetists rather than indirect patient outcome parameters should be measured.

Effective communication between doctor and patient improves patient satisfaction, patient recall of information and medical outcome, and can even protect doctors against malpractice litigation [1]. Although interest in teaching communication skills in medical schools has increased over the years, most postgraduate medical education still focusses on the technical and biomedical aspects of medicine [2]. However, the importance of non-technical skills in the daily work of doctors is now increasingly recognised, even in subspecialties such as anaesthesia [3]. Anaesthesia residency review committees in different countries now demand documentation of training in communication skills. For example, the Royal College of Anaesthetists requires an assessment of communication skills, attitudes to patients and behaviour for its Certificate of Completion of Specialist Training [4]. The American Accreditation Council for Graduate Medical Education (ACGME) has endorsed six general competencies including interpersonal and communication skills for residents in all specialties [5]. The ACGME and the American Board of Medical Specialties are now collaborating to implement and evaluate these general competencies. The impact of doctors' personal skills has therefore become an area of clinical interest and research. Most medical training programmes that teach communication skills are designed for general practitioners, but a few have been designed for surgeons [1] or for Accident and Emergency Department doctors [6].

Apart from personal observations, there are no data available on how anaesthetists communicate with their patients [7, 8]. We designed a programme to teach communication skills to anaesthetists using videotaped reviews of their pre-operative visits and role-play. Communication skills that are effective for general practitioners are not necessarily effective for other specialists [1]. Our programme was therefore tailored to the particular situation of the anaesthetist and patient in the pre-operative setting. The objectives of the training programme were to improve the receptive and affective behaviour and interpersonal skills of anaesthetists and to increase patient participation in informed or shared decision-making with respect to the planned anaesthetic technique. Despite the resources invested in communication skills training, few programmes seem to have been assessed in terms of patient outcome [2, 9]. Using a patient questionnaire, we assessed whether our training programme increased patient satisfaction and decreased patient anxiety before surgery. In addition, the training programme was assessed by a questionnaire distributed to all participating anaesthetists.


The Local Research Ethics Committee of the University of Basel approved the communication skills training programme and its assessment by questionnaires. A clinical psychologist (F.A.) led the training. Anaesthetists in a University Department of Anaesthesia were trained in small groups of 7–10 with a similar mixture of residents and faculty within each group. Each group received 10 training sessions, one each month, and each session lasted about 2 h. Participation in the training was mandatory for all 59 anaesthetists involved in patient care during the study period, although not all received the full 20 h of training.

Training started with a short theoretical introduction to interpersonal communication and its effect on patient outcomes such as patient satisfaction. After this, training consisted firstly of reviewing videotaped pre-operative visits. Each trainee had to record two of his/her pre-operative visits and these videos were discussed and analysed within each group. Trainees were asked to observe themselves, recognise their own emotions, recognise their patients' expressions and listen actively to what their patients were saying. Second, role-playing was used to prepare for difficult situations. Trainees were asked to apply four behaviours of effective clinicians, adapted to the pre-operative situation:

1 establish a welcoming atmosphere for the pre-operative visit and agree with the patient on an agenda;

2 elicit the patient's concerns about anaesthesia and surgery;

3 demonstrate empathy both verbally and non-verbally;

4 actively involve the patient in making decisions about the planned anaesthetic technique whenever possible, and appropriately conclude the visit by reassuring the patient of ongoing care.

The training programme was assessed using a pre- and postintervention design. Patient satisfaction and patient pre-operative anxiety were defined as appropriate measures of improvement in communication skills. We constructed a questionnaire using a modified Delphi procedure [10] to measure patient satisfaction with the pre-operative visit, patient pre-operative anxiety [11] and patient perception of the anaesthetist [12]. The final version of the questionnaire contained 86 items. Inclusion criteria for patients were: age over 18 years; fluency and literacy in the German language; written informed consent. Patients with a seriously impaired mental status were excluded. Anaesthetists were informed verbally and in writing about the purpose of the study and asked to participate. They were told that they did not have to distribute questionnaires to patients if they did not wish to do this and they also signed a physician questionnaire containing patient and physician demographic data. Pre-intervention data were collected for a period of 3 months from all patients undergoing elective surgery, the intervention consisted of training in communication skills for a period of 10 months using videotape reviews and role-playing, and postintervention data were collected for a 3-month period after the training finished. At the end of the programme, all participating anaesthetists also assessed it.

Summary response variables

Six response variables were used to summarise patients' satisfaction with the pre-operative anaesthetic visit and pre-operative anxiety. Satisfaction with the pre-operative anaesthetic visit was summarised by ‘overall satisfaction’ in the form of each patient's response to a general question on overall satisfaction, and by ‘median satisfaction’ in the form of each patient's median response to 10 specific questions on satisfaction with different aspects of the pre-operative anaesthetic visit. All 11 questions used the same five ordered categories (insufficient, fair, appropriate, very good, excellent). Pre-operative anxiety was summarised by using the German version of the Spielberger-State-Anxiety-Score (STAI-G Form X1) [13], by ‘overall anxiety about anaesthesia’ and ‘overall anxiety about surgery’ in the form of each patient's response to two general questions on overall anxiety, and by ‘median anxiety’ in the form of each patient's median response to 10 specific questions on different aspects of anxiety about anaesthesia and surgery [11]. These 12 questions on anxiety all used a 10-cm visual analogue scale (VAS).

Predictor variables

The predictor variables used to model each summary response variable were: age, gender, level of education, prior experience of anaesthesia, whether or not patients felt they were involved in choosing the type of anaesthesia they would receive, planned duration of surgery, the anaesthetist and the number of hours of training the anaesthetist had received before the anaesthetic visit. Age, planned duration of surgery and the number of hours of training were continuous variables; all other variables were categorical. The patients' level of education was coded as one of three categories: at most primary or secondary school, apprenticeship or high school, college or university. Prior experience of anaesthesia was also coded as one of three categories: none, prior experience, prior bad experience.

Models and model fitting

The six summary response variables were modelled using analysis of covariance. ‘Mixed’ models were fitted with the anaesthetist as a random effect and all other predictors as fixed effects. As a fixed effect, each predictor variable or category of a predictor variable is represented by a single parameter. As a random effect, each anaesthetist is said to form a cluster; responses from patients with the same anaesthetist are correlated but responses from patients with different anaesthetists are independent. A single parameter, the between-cluster variance, describes the differences between clusters; cluster effects are assumed to be normally distributed with a mean of zero. In this way, variability between anaesthetists is modelled without needing additional predictor variables to describe the differences between them.

Different distributions were assumed for the six summary response variables. For both overall and median satisfaction, < 1% of responses were ‘fair’ and none was ‘insufficient’. The lower three categories were therefore combined and renamed ‘standard’. Overall and median satisfaction were assumed to follow underlying normal distributions, although each was measured on an ordinal scale [14]; the Spielberger score was assumed to follow a normal distribution; and the other anxiety responses were assumed to follow exponential distributions (Fig. 1).

Details are in the caption following the image

The six summary response variables for patient satisfaction with the pre-operative anaesthetic visit and pre-operative anxiety expressed as a percentage of patients in each category. Anxiety-surgery, anxiety-anaesthesia, and anxiety-median are represented on a 10 cm visual analogue scale.

Models were fitted using the NLMIXED, MIXED and GENMOD procedures in SAS version 8.2. (SAS Institute Inc., Cary, NC). The NLMIXED procedure fits non-linear models with both fixed and random effects; the MIXED procedure fits linear models with both fixed and random effects; the GENMOD procedure fits non-linear models with only fixed effects. Where possible, therefore, parameter estimates from NLMIXED were checked against estimates from other procedures: the MIXED procedure reproduced estimates for the Spielberger score with the random effect, the GENMOD procedure reproduced estimates for the other anxiety responses without the random effect.

When using NLMIXED, models were first fitted without the random effect to provide suitable starting values for fitting with the random effect. Optimisation in NLMIXED was by the trust region method or, if this failed to converge, by quasi-Newton methods. The trust region method uses both first and second order partial derivatives and often provides stable estimates when a model has a small number of predictor variables. When a model has more variables, the trust region method may fail to converge, but quasi-Newton methods may converge as these use only the first order partial derivatives [15]. All confidence intervals were approximate 95% confidence intervals based on the Wald statistic. All significance tests were based on the likelihood ratio statistic.


A total of 1338 patients completed the survey. Of these, 1228 patients were included in the statistical analysis: 905 were patients of anaesthetists without training in communication skills and 323 were patients of anaesthetists who received at least some training. Patients were included in the analysis if their data were available for all predictor variables and if their anaesthetist saw at least 10 surveyed patients. Missing values for response variables further reduced the sample size for specific models; the lowest sample size was 902 for the Spielberger score. Patients seen by anaesthetists with or without training were similar in age, education and prior experience of anaesthesia (Table 1). However, patients seen by anaesthetists without training were more likely to be male; those seen by anaesthetists with training were more likely to be female. Both the expected duration of surgery and the length of the pre-operative visit were similar for patients seen by anaesthetists with or without training.

Table 1. Patient characteristics, satisfaction with the pre-operative anaesthetic visit and pre-operative anxiety. Values are percentage or median [IQR].
Without training
(n = 905)
With training
(n = 323)
Gender (male/female) 53%/47% 44%/56%
Age; years 53 [39–66] 55 [40–68]
 Low 15% 14%
 Medium 66% 67%
 High 19% 19%
Prior experience of anaesthesia
 None 33% 33%
 Good 55% 55%
 Bad 12% 12%
Felt included in choice of anaesthetic
 Yes 50% 49%
 No 50% 51%
Expected duration of
surgery; min
90 [60–120] 90 [60–150]
Length of pre-operative
visit; min
20 [15–30] 20 [15–30]
Training received by
physician; sessions
0 [0–0] 5 [5–7]
Overall satisfaction
 Standard 22% 21%
 Very good 43% 39%
 Excellent 35% 40%
Median satisfaction
 Standard 22% 20%
 Very good 46% 41%
 Excellent 32% 39%
Spielberger anxiety score 37 [30–45] 37 [31–45]
Overall VAS anxiety score
for anaesthesia; cm
2.0 [0.5–4.8] 2.0 [0.6–4.7]
Overall VAS anxiety score
for surgery; cm
2.4 [0.8–5.1] 2.3 [0.7–4.7]
Median VAS anxiety score; cm 1.1 [0.4–2.6] 0.9 [0.3–2.1]
  • VAS = Visual analogue scale.

Overall satisfaction with the pre-operative visit was high. With and without training, 79% and 78% of patients, respectively, felt that their anaesthetic visit had been ‘very good’ or ‘excellent’. Overall pre-operative anxiety was low. On a 10-cm VAS, overall anxiety about anaesthesia scored a median [IQR] of 2.0 [0.5–4.8] both with and without training, while overall anxiety about surgery was 2.3 [0.8–5.1] with training and 2.4 [0.4–2.6] without (Table 1).

After adjusting for other predictor variables, there was some evidence that training increased patient satisfaction. The estimated effect of training on both overall and median satisfaction was an increase of 0.02 of a response category per training session (Table 2); an increase equivalent to a fifth of a response category if an anaesthetist attended all 10 sessions. To put this in context, overall and median satisfaction increased by 0.06 and 0.05 of a response category, respectively, for each 10-year increase in patient age. Both overall and median satisfaction were lower amongst those with little education (−0.11 and −0.24 of a response category, respectively) and amongst those with higher education (−0.23 and −0.07 of a response category, respectively) compared to those with an intermediate level of education.

Table 2. Mixed (fixed and random effects) models for the six summary response variables. Probability values were calculated with the Likelihood Ratio Test.
Response Sample
Within- cluster
Training effect Other significant
Estimate 95% Confidence
Overall satisfaction 1019 0.02 0.02 [−0.01–0.05] 0.13 Age, p = 0.01
Education, p = 0.03
Median satisfaction 1029 0.01 0.02 [0.00–0.05] 0.08 Age, p = 0.03
Education, p = 0.05
Spielberger anxiety score 902 0.00 0.04 [−0.24–0.31] 0.75 Age, p < 0.01
Gender, p < 0.01
Duration surgery, p = 0.02
Overall anxiety – anaesthesia 1013 0.01 0.00 [−0.07–0.07] 1.00 Gender, p < 0.01
Prior experience, p < 0.01
Overall anxiety – surgery 1011 0.00 −0.02 [−0.09–0.05] 0.75 Age, p = 0.01
Gender, p < 0.01
Duration surgery, p = 0.01
Median anxiety 1015 0.01 −0.05 [−0.09– −0.01] 0.01 Gender, p < 0.01

There was no evidence that training had any effect on overall anxiety when measured by the Spielberger score or by general questions about anxiety. However, having adjusted for other predictor variables, there was strong evidence that communication skills training decreased median anxiety about specific aspects of anaesthesia and surgery. On a 10-cm VAS, the estimated effect of training was −0.05 cm per training session, a decrease equivalent to −0.5 cm if an anaesthetist attended all 10 sessions (Table 2). To put this in context, median anxiety was significantly higher for females than males (0.7 cm) [11].

We then estimated the effect of training on each of the 10 specific aspects of pre-operative anxiety. The low within-cluster correlation (Table 2) and similar parameter estimates with and without the random effect suggest that the effect of different anaesthetists can be safely ignored. Models were fitted using the GENMOD procedure with the same eight fixed effects but without the random effect of different anaesthetists. Training seemed to decrease the anxiety associated with all 10 specific aspects of anaesthesia and surgery except anxiety about being aware during surgery (Table 3).

Table 3. Fixed effects models for the 10 specific aspects of anxiety about anaesthesia and surgery. Probability values were calculated with the Likelihood Ratio Test.
Response Sample
Training effect
Estimate 95% Confidence
Waiting 1002 −0.03 [−0.08–0.02] 0.29
At the mercy of staff 984 −0.04 [−0.09–0.01] 0.12
Awareness 969 −0.01 [−0.04–0.02] 0.56
Losing control 971 −0.04 [−0.07–0.00] 0.06
Emerging from anaesthesia 951 −0.04 [−0.08–0.00] 0.06
Harm from anaesthesia 958 −0.05 [−0.09– −0.01] 0.04
Not waking up 948 −0.04 [−0.08–0.00] 0.05
Postoperative pain 979 −0.05 [−0.09–0.00] 0.06
Postoperative nausea 972 −0.03 [−0.07–0.01] 0.15
Results of surgery 971 −0.06 [−0.11–0.01] 0.04

The participating anaesthetists rated the training programme on a four-point Likert scale as helpful and useful for their daily clinical practice (mean (SD) = 3.1 (0.7)), and they appreciated the video as a learning tool. The majority agreed that the value of the pre-operative anaesthetic visit is much underestimated both during medical school and postgraduate education and that the training programme increased their interpersonal skills.


Our results show some evidence that training anaesthetists in communication skills can increase patient satisfaction with the pre-operative anaesthetic visit, although this increase was not statistically significant. Programmes that teach communication skills to doctors often have a limited effect. A review of studies evaluating such training programmes showed that most studies used poor methodology and that in studies with the best methodology, training had the least effect [2]. A recent evaluation of an 8-h programme of workshops and audio-taped office visits designed to teach communication skills to primary care physicians and surgeons, found no increase in patient visit satisfaction [16]. Factors that may explain why training does not always lead to improved patient outcomes include:

  • environmental constraints such as changing practice volume;

  • appropriateness of outcome measures;

  • time point of outcome measurement;

  • high pre-intervention scores of the outcome measure;

  • differences in study groups;

  • bias of participants (i.e. volunteer participants may be more motivated and therefore already more skilled at communication than control subjects);

  • a genuine difficulty in or resistance to changing a fundamental behaviour such as a personal communication style [2, 17].

Randomised controlled studies often use sample sizes that are too small to give definitive results, although mean differences between study groups consistently suggest improved patient satisfaction and well-being [18]. Nevertheless, a few studies also have shown significant increases in patient satisfaction after such programmes [19, 20].

Of the six summary response variables used in the present study, those measuring satisfaction were the least sensitive to differences. Overall and median satisfaction were measured on an ordinal scale and only three of the five categories were commonly chosen. A VAS scale would have had more power to detect the effect of training on satisfaction. Patient satisfaction with the pre-operative anaesthetic visit was already high before the training programme. In a previous analysis of 10 811 patients, the overall level of satisfaction with anaesthetic care was also very high (96.8%) and only 0.9% of patients were ‘dissatisfied’[21]. Patient satisfaction questionnaires may therefore have a limited value as an evaluation tool in the doctor–patient relationship because patient satisfaction is usually very high before an intervention [22]. In such a situation, the modest improvements we have seen may be all we can reasonably expect.

In addition to a modest increase in patient satisfaction, our results show good evidence that teaching communication skills to anaesthetists significantly decreased patient anxiety associated with specific aspects of anaesthesia and surgery. However, the training had no effect on overall pre-operative anxiety. These two results are not necessarily inconsistent. General questions about overall anxiety may be more of a measure of ‘irrational anxiety’, whereas specific questions may be more of a measure of ‘rational anxiety’. The former is probably more difficult to influence and may reflect the ‘trait’ of the personality. Consistent with this idea is the finding that training had no effect on overall anxiety about surgery but decreased anxiety about the results of surgery. The general questions on overall anxiety about anaesthesia and surgery were asked before the 10 specific questions on anxiety; the specific question on the results of surgery was the last question asked. Therefore, general questions were asked before patients thought about specific issues; having thought about such issues, patients may have been more ‘rational’ about their anxiety.

The within-cluster correlation gives the ratio of random effect variation to total variation. Hence, the variability associated with different anaesthetists is at most 2% of the total variability in a response. This low percentage suggests that patient-dependent predictor variables are more important than anaesthetist-dependent predictor variables in modelling patient satisfaction with the anaesthetic visit and pre-operative anxiety. As a consequence, changing the anaesthetist's behaviour may have little impact on a patient's perceptions.

In the present study, essential environmental factors such as the anaesthetist's workload remained constant pre- and postintervention. The length of the pre-operative visit was also constant (median = 20 min). The training sessions were always held in the afternoon during regular working hours and the participants were relieved of their clinical responsibilities in the operating theatres by nurse anaesthetists and by those anaesthetists not in training on that day. While this format increased the cost of training, it maintained the mandatory aspect of the programme so as to minimise participant bias. It was also a change from clinical routine work and it was not surprising that most participants enjoyed these training sessions, although some faculty members complained about disruption to their non-clinical responsibilities and schedules. Anaesthetists found their empathy developed so that they could perceive the needs and anxieties of their patients better, although the transfer of the newly acquired techniques to the actual patient encounter seems for many to be difficult.

As a teaching hospital, we have a considerable turnover of residents and faculty. This makes a longitudinal study over a 16-month period difficult, and only 10 out of 59 physicians were present for the full 16 months. A randomised design with intervention and control groups might have been easier to organise and more successful in detecting significant differences. However, given the costs and effort involved, we wanted to offer communication training to all anaesthetists in the department. It also might have been easier to detect significant differences by measuring anaesthetists' behaviour directly, but we felt it was more relevant to measure behaviour indirectly in terms of patient outcome. Finally, while our anxiety measures have been validated [11], our measures of satisfaction with the pre-operative visit have unknown reliability and validity.

Despite efforts by the American Board of Internal Medicine to understand patient expectations of doctors' behaviour [23], it is still not clear which communication skills should be taught to the different medical specialties. Recently, a group of experts identified seven essential tasks of medical communication: building a relationship, opening the discussion, gathering information, understanding the patient's perspective, sharing information, reaching agreement on problems and plans, and providing closure. Our communication training programme included, at least in part, all seven of these elements, which are published as ‘The Kalamazoo Consensus Statement’[24]. Such effective communication between doctors and patients is an important factor in the quality of clinical care as well as a determinant of patient satisfaction, and both of these are important markers for health plans in a competitive health care environment [16]. Therefore, we encourage Departments of Anaesthesia to continue to teach communication skills to their anaesthetists and to investigate the effects of such programmes further.