Effective improvement of doctor–patient communication: a randomised controlled trial

Subjects

All GPs (n = 1066) in three regions in the south of the Netherlands (Zuidoost Brabant, Noord-Brabant Noordoost, Limburg) received a letter informing them about the study and asking whether they would be interested in participating (November 2001). Subsequently, they were invited to enrol (May 2002). We aimed at a maximum of 100 and a minimum of 80 participants. Power calculations using historical data showed these were sufficient to detect a 10% difference in the MAAS-global score with 80% power and a 5% risk of a type 1 error.17 The first 100 subscribing GPs enrolled in the study and a questionnaire was used to obtain participants' personal and professional characteristics (for example experience with modern educational methods).

Randomisation

Randomisation was performed after sampling professional characteristics and before baseline measurement. GPs participated individually, but randomisation was done at practice level. Practices were allocated as a whole to either the intervention or the control arm of the study. This was done to avoid the possibility of closely collaborating colleagues, randomised to different arms of the study, benefiting from mutual educational experiences. Practices were stratified by region and experience in working with modern educational methods; these were identified in an expert panel prior to the study as possible effect modifiers. Subsequently balanced randomisation of practices was done. Within strata, practices were allocated to either the intervention or the control condition by the principal researcher and the projects research assistant using the ‘allocate random data’ option in SPSS.

Educational interventions

Intervention group. The intervention consisted of assessment, selection of global topics for improvement, feedback and revalidation activities. Assessment was the first step to identify aspects of GPs' performances where improvement was likely to be beneficial.18^,19 Three topics were assessed. One of these topics, for example, doctor–patient communication, is the focus of this paper. For each participant two topics for further improvement were selected. This was done by relating assessment results to standards for desired care.20

GPs with scores above the standard focused on other topics than communication, but did not drop out of the study. The performance of GPs was assessed by video observation in daily practice.17 Participants received individual feedback on their assessment results and the selected topics from a GP. These were regional GPs who were invited to participate only for this specific task in the project, and who were trained by the authors. Feedback was given on each participant's mean communication score in relation to the criterion-referenced standard and mean scores of 10–15 colleagues. Also, detailed feedback was on the scores on items and the standards for these items, thus enabling GPs to identify aspects requiring improvement. Within 2 weeks after feedback, a series of three to four small group meetings started. The meetings lasted 2 hours and the groups consisted of four to six participating GPs and a tutor. The tutors were practising regional GPs, without specific expertise in doctor–patient communication, and were prepared with an 8-hour training course in supporting their colleagues' learning process. The training consisted of an introduction in the background and rationale of the intervention and of educational exercises that focused especially on the first two meetings. Participants also received a manual with information about quality improvement, examples, checklists and diagrams to fill out. The manual contained no information about doctor–patient communication. The content of the meetings was the participants' own responsibility. GPs started by defining personal goals for improvement, then analysed barriers to change, developed practice development plans including time schedules, and finally evaluated their results.14^,21^,22 Meetings were held within a fixed time frame of 7 months; participants were free to choose the time and place of the meetings. GPs received 2 hours of credits for performance assessment and feedback and 2 hours for each attended meeting.

Variables and instruments

Statistical analysis

Analyses were performed for item scores and total score. For each participant the pre and post MAAS-Global scores were calculated as the mean of all items over all observations. The item scores were calculated as the means of items over all consultations. The effect of the intervention was analysed by linear regression, using the post score as dependent variable and the pre score and treatment (intervention or control condition) as independent variables. Possible outliers affecting the analyses were identified by means of scatter plots and excluded from analyses. To identify the contribution of the educational intervention in the post score, we calculated the 95% confidence interval (CI) of the treatment effect. Analyses were done as ‘intention-to-treat’ analyses if data were available: all participants, whether they participated in educational activities on doctor–patient communication or not, were included in the analysis.

The effect size (d-value, a commonly used term to compare the effects of educational interventions) was defined on the basis of the post versus pre scores (post-pre-score) for the participants in the intervention and control group.24 For the two groups the average and standard deviation of the post-pre-score were calculated. The effect size was obtained by subtracting the average scores for the two groups, and dividing it by the pooled standard deviation.24

To check whether participation in the small group meetings influenced the effectiveness, a second regression analysis was performed. As participants with high scores were allocated to other topics in the study, this analysis was performed only for the participants in intervention group that were allocated to the small group meetings on communication. In this analysis, the post score was used as dependent variable, and the pre score and the percentage of meetings attended were used as independent variables.

Summary of main findings

The educational approach involving assessment of communication in daily practice and personalised learning activities guided by structured small group work proved to be an effective way of improving doctor–patient communication. The size of the effect was moderate to large. A dose-response effect was seen as the attendance to the intervention significantly contributed to the improvement of the scores. The participants showed most improvement on patient-centred items. This study showed assessment of GPs performance in the real process of care delivery to be an effective first step of an educational programme. GPs considered the assessment outcomes as an acceptable basis for educational activities, and selected aspects of their performance with low scores for improvement. GPs were also found to be capable of drawing up effective personal development plans with support only for the process of change, given the improvement that was seen on the reported improvement goals. The participating GPs had to get used to the approach investigated in this study, as the majority reported having no experience with the methods used in the intervention. We expect that repeated use of the programme may yet enhance its effectiveness.

Strengths and limitations of the study

Participants in the intervention group undertook more study-related activities to improve doctor–patient communication than their colleagues in the control group. This may be interpreted as bias, but in our view it should be regarded as an effect of the intervention, that is, the assessments to drive learning activities. Educational activities not related to the study may have influenced the results, but, as GPs in the control arm attended more of those activities than those in the intervention arm, this only emphasises the effectiveness of the intervention. Selection of randomised groups by attrition may cause bias: although we cannot rule out this effect, no differences in available pre scores were found between those who did and did not drop out of the study.

An unexpected finding is the deterioration of the scores of the control group. As the baseline scores of the control group are very near the post-intervention scores of the intervention group, it may appear that the effects must be attributed to chance, instead of the intervention. Although we have no satisfying explanation for the scores in the control group, we have arguments to attribute the effect to the intervention itself. First, the effects in the intervention group are seen in the participants that actively participated in the small groups. Second, the effects are seen in those items that were reported by participant as goals for improvement. And third, the regression analysis used is the preferred method in case of baseline differences.25

The effects of the intervention may be temporary and fade over time. Therefore, we planned the second measurement 2–6 months after the intervention. Whether improvements are permanently embedded in GPs' communication behaviour is still not clear.

Comparison with existing literature

This study provides evidence that supports the current view on continuing professional development, which stresses the importance of learning activities being learner-centred, focused on day-to-day clinical practice and integrating several approaches.5^–8 Convincing evidence on educational interventions that improve the performance on doctor–patient communication towards desired standards is lacking.12^–15 Therefore, the study also provides an approach for learning and improving in practice for those interested in the specific domain of doctor–patient communication.

Implications for future research

The findings of this study cannot easily be generalised to other topics or to the general population of GPs. The participants in our study were not representative as they were motivated to experience this new approach. Effectiveness may be different when GPs are not really interested in this educational method. Also, the assessment method of video observation in daily practice is ideal for this purpose, as it is personalised and performancebased, and yields videotapes that can be used in the small group meetings. Further research will have to shed light on the effectiveness of this approach for other topics, assessment formats and participating doctors. This approach may be implemented as an evidence-based educational approach to support continuing professional in the domain of doctor–patient communication.