Training doctors briefly and in situ to involve their patients in making medical decisions—Preliminary testing of a newly developed module

Abstract Objective To carry out preliminary evaluation of a training module for doctors to enhance their ability to involve their patients in medical decision making. The training refers to the shared decision‐making (SDM) communication concept. Methods The training module includes a comprehensive manual, a corresponding video tutorial with communication examples and a 15‐minute face‐to‐face feedback session based on an SDM analysis of a consultation recording provided by the trainee. Ten trainees (four neurologists, three dentists, and three general practitioners) participating in the pretest each recorded four clinical consultations (total sample: N=40) and received three training components. After the training, doctors provided feedback on the module's feasibility in a questionnaire. Communication performance of doctors, patients and doctor–patient dyads was assessed by trained observers and self‐assessed by doctors and patients using the MAPPIN’SDM approach. Training effects were determined using Wilcoxon signed‐rank tests comparing baseline values with post‐intervention performance as assessed in the fourth consultations. Results The face‐to‐face training sessions were short and feasible with regard to clinical reality. Participants considered the training supportive for acquiring SDM skills and recommended more emphasis on the face‐to‐face feedback. Communication improved according to observers rating doctors (P=.05) and doctor–patient dyads (P=.07) and to doctors’ own judgements (P=.02). No improvement was observed in patients’ SDM behaviour (P=.11); accordingly, patients’ judgements did not indicate improvement (P=.14). Conclusions The training is designed to meet clinicians’ needs. Improvement of risk communication after training encourages optimization according to doctors’ feedback. Following this study, the efficacy of the training is now being examined in a randomized controlled trial.


| BACKGROUND
Three decades of research on shared decision making have not been enough to provide unambiguous proof that shared decision making (SDM) leads to patient-relevant outcomes, 1,2 such as better health outcomes, lower decisional conflict, and better adherence. According to a recent review, in 10 (out of 11) randomized trials which succeeded in increasing patient participation, only six (14%) of the 42 hypothesized outcomes were found to be positively influenced. 1 Perhaps, we should prepare ourselves to accept that shared decision making is justified by an ethical imperative only rather than empirical evidence. 3,4 Wouldn't this give enough reason to strive for its implementation? The absence of proof is, however, not proof of absence. Difficulties in demonstrating a benefit for patients from being involved in making their medical decisions could also be due to methodological deficiencies. If effects fail to appear after provision of interventions supposed to facilitate SDM communication, methodological deficiencies might arise in conjunction with insufficient quality of the interventions. Evidence shows neither SDM tutorials 5 nor patient decision aids 6 to reliably achieve the intended communication quality. If effects fail to appear although patient participation has demonstrably been achieved, methodological deficiencies might arise from existing measures' inability to really capture the concept's essentials. [7][8][9] Given the weakness of existing knowledge on the efficacy of SDM despite the voluminous literature of the last decades, consideration of such methodological deficiencies implies the potential to nevertheless discover the impact of the concept by use of newly developed or refined appropriate methods.
This study was carried out in response to both the communication deficits still present in health care 10 and the underdeveloped level of SDM training for health professionals. 5 A systematic review identified 54 training programmes in 14 countries and 10 languages. Only 17 of them had been evaluated, in most cases using trainee satisfaction or other subjectively reported outcomes. The authors concluded that more knowledge is needed on training didactics and on the training programmes' efficacy with regard to patient-relevant outcomes. 5 In addition, the vast majority of the training programmes listed in this review addressed decision making in specific medical domains. Broadscale implementation of shared decision making, however, requires generic training methods that can be used in any medical context. When we started developing our training module, no German-language SDM training for health professionals was available that complied with minimal criteria of feasibility, efficacy and generalizability.
To enhance physicians' ability to involve patients in the process of making medical decisions, we developed doktormitSDM, a short in situ training module. The new module was intended to comply with at least the following three requirements. (i) The training approach should entirely meet the essentials of the SDM concept. 11,12 To give examples, this includes encouraging professionals to develop willingness to rigorously share all relevant information with the patient and the corresponding competency to adhere to the criteria of evidence-based patient information. 13 (ii) To assure feasibility, the training should also comply with health professionals' practical needs and restrictions, particularly with regard to time resources. (iii) To be adaptable to varying contexts, the didactic concept of the new training module should also allow for a generic approach.
When we conducted this study, doktormitSDM had already undergone some unsystematic piloting and pretesting. Single components were evaluated either using in-depth interviews with physicians at our own unit or by administering a feedback questionnaire to participants in several conference workshops. 14-16 Particular attention was paid to practical issues, acceptance and subjective perception of usability. An initial rough draft was adapted, incorporating theoretical input, to produce an intervention module consisting of a 15-minute feedback session, a 20-minute video tutorial and an SDM manual, which we then considered ready for systematic pretesting.
This study aimed to explore the feasibility of the new training module. In particular, we investigated practicability in the context of the doctors' clinical practice and whether the doctors considered the course helpful. In addition, the study was intended to help in evaluating the appropriate training dose. Would the use of a minimal intervention be already enough enable doctors to change their communication behaviour? As, in our experience, health professionals' concerns about changing communication habits are often related to the length of consultations, this study also set out to explore the relationship between communication quality and the use of time.

| Design
The study used a one-cohort pre-post design consisting of an alternating sequence of decisional consultations and intervention components provided stepwise ( Figure 1). After the training, feedback on feasibility and perceived benefit was obtained from the participating doctors using a questionnaire. Communication quality was evaluated on the basis of four consultation recordings delivered by each participant. The first consultation was used for baseline assessment. After each of the first three consultations, a component of the intervention was provided. A fourth consultation (C4) was recorded to assess the training effect defined as improvement from baseline to the end of the training. To allow for exploratory elucidation of the learning curve, additional measurement points after the first two intervention components (C2 and C3) were included in the study design. Extent of patient involvement was evaluated from the third person observer's, the doctor's and the patient's perspective.
Patients participating in this study were blind towards the doctors' training level and were only recorded once within the study. For organizational reasons, the doctors were free to choose the consultations for the study. However, the selection was made before the consultations. The pauses between the consultations were supposed to be just long enough to receive the next intervention step, but as brief as possible.
This study was approved by the ethics committee of the Christian-Albrechts University Kiel, Germany, and was registered in the Current Controlled Trials register (ISRCTN78716079). All participating physicians and patients gave their written consent to inclusion in the study.

| The intervention
The doktormitSDM training module is minimized with regard to volume but comprehensive with regard to the didactic approach. This section describes the structure and content of the intervention used in this study. As we also wish to share our knowledge, which has grown over years, the purpose of the following section is threefold: to give a precise scientific description, to provide some narrative background about the genesis and to give insight into didactic considerations related to the SDM training approach evaluated in this study.

| Description of the training module
The doktormitSDM training module is included in the environmental review of health professional training in SDM by Légaré et al. 17 It has three parts:   Participants were invited to use the manual and the tutorial as preparation for their individual training. They were told that these sources provide background about the SDM approach and the particular system used, to structure the communication. 18 The manual explicitly states that it is written for both raters and clinicians and provides guidance indicating passages that are easy to read and others that are relevant to researchers only. In total, the manual has 40 pages; reading the entire text would have required approximately 2 hours.

| Origin and core operation
The training concept uses knowledge gained during the development of MAPPIN'SDM, a measurement inventory to assess SDM. 7,[18][19][20] MAPPIN'SDM had resulted from thorough consideration of the concept's assumptions 12 and of existing approaches. 8 The development of MAPPIN'SDM involved the definition of process indicators of a consultation strategy maximizing patient involvement and meeting the criteria of evidence-based patient information 19 and the definition of detailed criteria for scoring more or less skilled performance of these indicators. 18 Besides theory, this process was informed by insight into clinical practice via both an extensive pool of consultation ---recordings and intensive close co-operation with doctors and patients at our unit. 21 Basically, the doktormitSDM concept is a continuation of these discussions with the practitioners in which we shared our in-depth insight into their communication using terms of the MAPPIN'SDM method. Together, we strove to identify the optimal strategies to fairly and efficiently involve the patient in making medical decisions. At its core, doktormitSDM is a practitioner-trainer discourse on patient involvement, using an actual consultation recorded by the practitioner and its MAPPIN'SDM analysis by the trainer. The trainer provides measurement-based supportive feedback in line with the concept's assumptions. Using simple didactic methods, the doctor is then stimulated to see the doctor-patient dialogue from a third person's perspective This is intended to enable doctors to evaluate, refine and develop their own communication behaviour, and thus to take long-term control of their own learning.

Making the training feasible
The impact of the programme would be limited by its compatibility with the participants' daily routines. As regards practical issues, feasibility is challenged by the amount of time needed and the distance between the training site and the workplace. In addition, marginal discrepancies between the training content or context issues and the doctors' reality might, from the doctors' point of view, give reason to adopt a rejecting stance. For example, a doctor participating in a conference workshop might argue that an example provided in a roleplay is of no practical use because the consultation setting in his/ her clinical reality is slightly different. Such dynamics might indicate strategies to reduce cognitive dissonance. 22 The secret in preventing devaluation of training content caused by dissonance reduction lies in facilitating desirable strategies of dissonance reduction instead, in particular behaviour change. We presumed that behaviour change was most likely to happen when the interactive part of the training was provided at the participant's ward (in situ) and shaped and struc-

Overcoming barriers
During the modelling and piloting phase, we had learned about specific barriers clinicians typically face when intending to employ patient involvement behaviour. To achieve behaviour change, doktormitSDM has to address attitude-related beliefs as well as misconceptions regarding subjective social norms and individual behaviour control. 23,24 For instance, participants might be open-minded regarding the SDM approach but, as a result of their medical training, could expect to be quantitatively evaluated in the face-to-face feedback sessions. This misconception would mean a barrier to implementation of the target behaviour, as it would interfere with the autonomous role in the learning process intended for the trainee. This barrier adds to subjective social norm barriers arising from the way the information source, here the trainer and the training concept, is evaluated by the participants.

| Sample
The pretest used a convenient sample of 10 doctors and a consul-  To study feasibility and subjective usefulness, the participating doctors were asked to fill in an additional one-page questionnaire after finishing the training. Using 27 items, the questionnaire aimed particularly to explore the extent of the participants' use of the materials provided (two items), comprehensibility of the materials (three items), attitude and items used a four-point Likert scale. Emphasis on the components could be rated as "too much," "not enough" or "perfect," and suggestions for ordering could be provided by indicating one of three positions. Self-assessment of performance was rated on a scale from 0 "SDM not present" to 6 ="excellent" ( Table 1).

| Measurement
The extent of patient involvement was assessed from three measurement perspectives using five scales of the original German version of the MAPPIN'SDM inventory. 18 MAPPIN'SDM includes three observation scales for doctor, patient and the doctor-patient-dyad (which is the unit made up of doctor and patient). Additionally, the inventory includes two corresponding questionnaires for doctor and patient. All five scales address an identical set of 15 indicators of patient involvement 18 (Table 1). Each indicator is represented by one item, which is scored from "0" ("the indicator is not present") to "4" ("the indicator is present at an excellent standard").
The unit of analysis to assess patient involvement was the decision sequence within the medical consultation. To make sure that the measures of patient involvement were applied to the same unit, doctor and patient had to agree upon one decision as the index decision they would refer to when completing the questionnaire. To prepare the observer ratings, sequences including the index decision were coded a priori with regard to timeline and the set of available options. T A B L E 1 Questionnaire to evaluate the feasibility of the doktormitSDM training module If necessary, medical expertise was requested to affirm the set of available options. Sequences were selected in random order during the rating procedure.
The two raters coding the communication material in this study were successful finishers of a previous coder training. 25 Ratings were assisted by a comprehensive manual providing detailed examples for the scoring of each item. Both raters coded the material independently to allow for calculation of inter-rater agreement. In a second step, a consensus rating was agreed upon by discourse, which was used in the analyses in this study. By use of a pseudonym and randomized selection for coding, raters were blinded towards questionnaire data and the doctors' level of SDM training. MAPPIN'SDM questionnaires were completed by doctors and patients immediately after the consultations. To become familiar with the questions they had to answer afterwards, doctors using the questionnaire for the first time and patients were asked to read through it prior to the consultation.

| Analyses
Descriptive data characterizing the study sample were collected. All analyses were conducted using SPSS version 17.

| Descriptive results
Amongst the 40 patients participating in the study, 22 were male.
The 10 doctors (seven of whom were male) were specialists in neurology (4), dental (3) and internal and general medicine (3). A big variety of medical problems and treatment was discussed as well as the appropriate diagnostic medical options. This ranged from consideration of immunotherapy treatment or the procedure to confirm a diagnosis of multiple sclerosis to problems related to tooth loss, treatment of serious cardiovascular disease and questions related to prophylactic interventions such as vaccinations. The length of consultations ranged from 2.5 to 51 minutes (mean 20.5 minutes), while the length of decision sequences ranged from 2.5 to 38.8 minutes (mean 14.8 minutes).

| Feedback by doctors
According to subjective reports given by the participants after finish-  (Table 1).

| Quality of measurement
Inter-rater reliabilities were high to excellent in the observer scales in another paper. 18 The patients' judgements, which are presumably less biased owing to the fact that all patients only participated once in the study, do not reflect an improvement in communication.
Although the patients' tendency to give lower ratings with increasing training level might, considering the small sample size, be due to a random variation, it is hard to ignore (Figure 3) Meanwhile, another training curriculum for SDM has been evaluated with German doctors. 27 The curriculum turned out to be effective in improving skills classified with the OPTION scale, which partly covers essential SDM indicators. 28 Results were, however, gained within a highly selected and reduced sample remaining at follow-up. The dramatic loss to follow-up reflects the poor feasibility of the intervention.
In contrast to the authors' argumentation that the dosage of 12 SDM teaching sessions might have been too little, we consider that this involves too much effort to have a realistic chance of implementation. In this regard, the doktormitSDM training seems promising as a method for training specialist doctors in the German health system. Although developed together with medical doctors, it is applicable without limitation to other health professionals such as nurses or physical therapists. Moreover, the doktormitSDM approach is adaptable to various settings, media and time frames. The one-on-one didactic method has recently been translated into an online format. 29,30 Currently, much effort is being made to develop communication curricula for medical students. 31,32 However, the didactic strategies being used are not necessarily transferable to professional training. In a current trial, an adapted online tutorial application of the doktormitSDM curriculum is therefore being tested with medical students.
Training programmes on SDM require thorough consideration of current clinical communication routines. The "one doctor -one patient" setting is no longer the most common setting for making key decisions in many medical domains and has been replaced by virtual or real multidisciplinary settings such as tumour boards. 1 Didactic strategies of training have to respond to these new challenges.

| CONCLUSION
The new SDM training module has proven feasible and was considered important and supportive by the participating doctors. Used as a minimal intervention, this training approach has shown potential to positively affect the communication quality in terms of patient involvement. These results suggest that further evaluation will be worthwhile. The conclusions of this study have led to a revised version of the doktormitSDM training which is now being tested within a randomized controlled trial.