Effective physician–patient communication increases collaboration, enhances mutual understanding, increases treatment adherence, and strengthens the therapeutic alliance (1, 2). Problematic communication impedes effective care and is the focus of most complaints about physicians to regulating bodies (1–3). Healthcare regulators and educational institutions require that physicians are competent in communication and interpersonal skills (4, 5). However, teaching effective communication is challenging, especially in difficult patient encounters where it is particularly important (6).
We define effective communication as therapeutic communication: interactions in which a physician 1) is attuned and responsive to a patient’s verbal and nonverbal communication; 2) acquires an understanding of the patient’s situation; and 3) responds in a timely, nonjudgmental, and empathic manner. We developed a brief intervention to teach therapeutic communication to family medicine trainees, Coaching Communication using Simulated Encounters (CCSE). CCSE uses standardized patients (SPs) to provide controlled exposure to common difficulties, which is followed by feedback and coaching by experienced psychiatry instructors.
CCSE aims to improve competence and self-efficacy in therapeutic communication. Self-efficacy is the physician’s confidence in his or her ability to handle complex physician–patient interactions (7). Physicians with low self-efficacy may experience greater anxiety, leading to communication errors. Since low self-efficacy and weak communication skills may not be apparent except under stressful clinical circumstances, physician training should include controlled exposure to clinically challenging experiences in order to enhance both competence and self-efficacy.
Research on difficult encounters has produced models of therapeutic communication (2, 8–12). For example, Zoppi and Epstein’s (13) patient-centered approach addresses barriers such as unexamined negative emotions or reactions to difficult patients. The Kaiser Permanente Four Habits Model fosters therapeutic alliance, facilitates information-exchange, and improves empathy through small- and large-group intensive courses that use SPs and role-plays (14). Others have experimented with various educational formats and tools (e.g., Boyle et al. (15)). Reviews indicate that communication instruction that uses experiential methods (e.g., simulated and/or real patients, feedback, and reflection) and is focused on engagement is superior to didactic approaches (16, 17). Experiential instruction emphasizes the trainer–trainee relationship (18). Several training programs have used these methods to prepare physicians to handle challenging clinical situations (14, 17).
The CCSE model is grounded in the principles of psychotherapy supervision and teaching (19, 20). The model assumes that physician–patient difficulty arises out of the interaction between the patient and the physician. In particular, difficulty arises from an interaction between the patient’s presentation regarding symptoms, and/or communication style, and the physician’s emotional response and communication skill. We further assume that clinicians can be trained to recognize and understand this interaction and to use this understanding to manage difficult situations more effectively. The CCSE model emphasizes the therapeutic alliance, which is fostered through seeking agreement on treatment goals and tasks, responding to patient emotions, physician recognition and management of his or her negative emotional reactions, and articulating an empathic understanding of the patient (21).
CCSE involves detailed exploration of trainee and SP interactions to enhance communication skills. Building an alliance between trainee and supervisor is essential in the experiential learning and becomes a platform for modeling effective communication. Key CCSE steps include viewing the videotaped trainee–SP interaction together, posing open-ended questions, active listening, providing constructive criticism, fostering reflective thinking, and modeling alternative responses (22, 23).
CCSE uses the psychotherapeutic concepts of priming and exposure (24). In training-as-usual, trainees often encounter difficult patient situations during emergencies without the opportunity for observation and reflection. In the CCSE intervention, trainees are exposed to clinical difficulties in a controlled, observed setting. The encounters are enacted by SPs, who exhibit 1) strong emotions, such as anger or sadness; 2) challenging interpersonal styles (e.g., being distant/avoidant); or 3) traumatic life circumstances.
Coaching focuses explicitly on understanding countertransference (the overt and covert emotional reactions a clinician has toward a given patient) and on trainee self-assessment. By understanding their emotional responses, trainees acquire useful information to better respond to patients (25). Self-assessment is emphasized because many physicians lack this skill (26). Some who do not perform well fail to recognize their deficits nor assimilate feedback, whereas others, who are highly competent, underestimate their performance (18, 27). In both cases, teaching opportunities to improve self-assessment occur via embedding self-observation and reflection into skills-teaching.
The objectives of the current study were to determine 1) whether family medicine trainees’ competence in therapeutic communication improves over the course of the CCSE intervention; 2) whether the trainees’ self-efficacy for therapeutic communication improves over the course of the intervention; 3) whether this improvement differs from changes resulting from residency training-as-usual; and 4) whether this improvement is maintained up to 6 months later.
+
Participants and Design
The study employed a single-subject AB research design across multiple subjects, with change measured during three time-periods: the 1-month control period (immediately preceding the intervention); the month of the CCSE educational intervention (four separate measurements); and two separate follow-up sessions (1 week after the end of the intervention, and an average of 6 months later). To determine whether change occurred during training-as-usual, a measurement was taken immediately preceding the 1-month control period and was compared with the Time 1 pre-intervention measurement.
Each trainee was assigned a coach, who was an experienced psychotherapy supervisor and faculty psychiatrist. All participants, SPs, and coaches provided informed consent. Trainees worked with their coach once weekly for four sessions over the month of the intervention. All coaches received the CCSE manual (28) and attended the orientation session. Regular review with the supervisors demonstrated that they administered the components of the training intervention as described below. The coach chosen for a given trainee did not have any administrative or faculty influence over that trainee’s evaluation in their respective residency or clinical programs and agreed to maintain confidentiality. The study was approved by the Research Ethics Board of Mount Sinai Hospital.
+
Components of the CCSE Intervention
The CCSE model is based on adult learning theory, performance coaching, principles of effective teaching (29), and recommended principles of psychotherapy supervision (30, 31). Supervisory principles encourage effective questioning, listening, dialogue on clinical reasoning, and providing meaningful and instructive feedback (32).
Feedback on the videotapes was used during the coaching sessions to provide an opportunity for comprehensive examination of the SP encounters (content, process, formulation, verbal, and non-verbal communication). During the coaching sessions, joint reviewing of the video engaged the trainee to reflect on action to develop their capacity to reflect in action (33).
+
Clinical Encounters With SPs
The SPs’ roles were partially scripted, to balance the flexibility required by SPs to be spontaneous with the necessary controlled exposure to a variety of commonly-encountered clinical difficulties. Difficulties in the encounter were implemented in two ways: 1) through the complexity of the problem domain; and 2) the manner in which the problems were communicated (e.g., intense anger). In addition, identity and content themes that typically cause discomfort in some trainees (e.g., suicide, terminal illness) were included. The simulated scenarios were scripted by two authors (PR and NM) and reviewed by a faculty member from the department of family medicine to ensure ecological validity. All SPs were trained by the University of Toronto Standardized Patient Program in the provision of constructive feedback. Also, “refresher” courses throughout the intervention were provided to reduce variation in SP performance over time.
The benefits of using SPs include 1) increased systematic control of the presentation of difficult treatment elements; 2) actual patients are protected from clinical errors; and 3) videotaping does not compromise confidentiality. One concern about using SPs is whether the skills learned in simulated situations are transferred into the clinical field. However, SPs have been extensively used in training and examinations, and interviewers report the experience to be very representative of real practice (34, 35).
+
Orientation Session (Time 0):
Each trainee received a copy of the CCSE Manual (30) and participated in a 1-hour orientation with their coach before the intervention. They discussed the learning objectives (e.g., improving the physician–patient alliance and communication skills) and logistics of the project (28). Trainees also completed baseline questionnaires, which included demographic information and a measure designed to assess self-perceptions of performance levels in counseling situations (a modified version of the Counseling Self-Estimate Inventory; 36).
+
Intervention (Time 1 to 4):
Over the course of the intervention, trainees had four weekly videotaped interviews with SPs that lasted 30 minutes each (see Figure 1). Each participant encountered the same set of four “intervention” and two “follow-up” SP scenarios. A maximum of two SPs performed each scenario.
The first scenario was designed to be a relatively easy “warm-up” to the process in order to help the trainees feel safe in the study. All subsequent SPs presented with differing kinds of difficulties, each with a specific context and a set of “empathy prompts.” Empathy prompts were expressions of distress, hopelessness, frustration, anger, anxiety, dismissiveness, or despair that are often difficult to respond to (e.g., “no one understands me.”). Trainees were instructed to engage with the patient, establish an alliance, and to discover as much as they could about the patient and what brought him or her to seek treatment. They were not required to make a diagnosis or treatment plan. After each interview, the trainee completed evaluation measures, and the SP, now out of “role,” provided the trainee with 15 minutes of constructive feedback. Trainees completed their measures before the coaching in order to obtain unbiased self-ratings of performance. While trainees completed their measures, the coach reviewed the videotaped encounter in full. Next, the trainee and coach met for 1 hour, during which they discussed the trainee’s experience and reviewed and discussed segments of the videotaped interview. Immediately after the coaching session, coaches completed evaluation measures.
+
Follow-Up (Time 5 and 6)
These videotaped SP interviews were done 1 week and an average of 6 months after the first interview. The average follow-up was 5.9 months, and the median was 4.25 months (range: 2 to 22 months). The follow-up sessions involved therapeutic communication challenges related to social determinants of health (e.g., unemployment, social isolation), unexplained medical symptoms, and disappointments over the quality of healthcare. No coaching was provided after these follow-up interviews. However, one of the authors (PR) met with the participants for a scripted debriefing. At the conclusion of each follow-up session, trainees completed evaluation measures.
+
Counseling Self-Estimate Inventory (COSE) (36)
This 37-item measure assesses counselors’ perceptions of their expected performance levels in counseling situations. It was completed by the trainees 1 month in advance of the intervention and after each SP encounter (four interventions, two follow-ups). The measure was modified by computerizing it, eliminating three items that referred to longitudinal counseling that did not apply to the single-session model of the CCSE, and modifying the wording of the anchors so that the trainee-rated scale was directed toward learning goals. Specifically, rather than rating items on a 6-point Likert scale (Strongly Disagree to Strongly Agree), in this modified version, trainees indicate that they 1) “need improvement in this area;” 2) “need only minor improvement in this area;” 3) are “satisfactory, no improvement needed;” are “knowledgeable with confidence;” 4) are “exemplary;” or 5) “no evaluation possible.” Higher scores reflect stronger perceptions of counseling self-efficacy. Larson and colleagues (36) report an internal consistency of 0.93 and a 3-week, test–retest reliability of 0.87. In the current study, internal reliability was 0.96. The correlation at baseline between the modified and original versions of the COSE was adequate (r=0.69; p <0.001).
+
Global Objective Structured Clinical Examination (G-OSCE) Rating Instrument (37)
This 5-item measure assessed communication skills overall and on 4 component subscales: empathy, coherence, verbal expression, and nonverbal expression. It was completed by each coach after each coaching session. Response options are tailored to each question and are rated on a 5-point scale. Higher scores indicate greater competency. This measure has good internal consistency (α=0.70) and construct validity (38).
Repeated-measures analysis of variance (ANOVA) was used to examine whether there were significant differences between mean ratings on the modified COSE and the G-OSCE Rating Instrument over time. Significant multivariate effects were examined through the use of repeated contrasts to determine the specific sources of the effects. Data were analyzed with IBM SPSS Statistics 18. All effects were tested at the p <0.05 levels.
Eligible participants were family medicine resident trainees. Participants from five University of Toronto-affiliated hospitals volunteered for the study between 2004 and 2007. Twenty-six participants (9 PGY-1, 11 PGY-2, 6 fellows) started the intervention, and none dropped out. Feedback during debriefing suggests that the trainees’ perception that they were learning valuable skills contributed to the high retention rate. Trainees were predominantly women (80.8%). Their mean age was 29 (SD: 5) years. The sample was culturally diverse, in that 73% were either first-generation Canadians or international medical graduates; 61% were Canadian-born.
+
Trainee-Rated Self-Efficacy
There were no significant changes in self-efficacy over the month before the intervention with training-as-usual (the pre-control period measurement compared with Time 1). In contrast, repeated-measures ANOVA for the intervention and follow-up periods revealed a significant improvement in trainee-rated self-efficacy over time (F[6, 150]=34.37; p <0.001; η2=0.58; see Figure 2).
+
Coach-Rated Communication Competence
Repeated-measures analysis revealed a significant improvement in trainee communication competence as measured by their coaches (F[3, 75]=3.914; p <0.05, η2=0.14; see Figure 2).
Therapeutic communication, particularly in difficult treatment situations, is a vital component of the physician–patient relationship and essential for the provision of effective care. The importance of teaching physician–patient communication is recognized in training curricula and standards for professional practice (4, 5). Our primary aim was to develop and test a brief, intensive intervention (CCSE) to improve physician communication with difficult patients. This intervention resulted in significant improvements in both communication self-efficacy and communication competence that were sustained an average of 6 months after CCSE. There was no such improvement during the training-as-usual pre-intervention control period. Coach-rated communication competence was correlated with trainee-rated self-efficacy. Moreover, most trainees viewed their CCSE experience positively, especially identifying the value of the skills learned and the opportunity to apply them to standardized clinical encounters.
The results converge with previous findings that experiential communication training for physicians improves confidence (12) and therapeutic skills (14). Such outcomes are associated with more effective consultations, reduced conflicts and complaints, improved patient and physician satisfaction, patient understanding and recall, and adherence to treatment plans (1–3). These results also lend support to the idea that therapeutic communication and awareness of one’s communicative skill are not simply innate; they can be taught and learned.
CCSE is based on recommendations of large-scale reviews of physician communication education (2, 11, 16, 17). Specifically, CCSE 1) uses a theoretical framework (therapeutic communication) to ground its intervention; 2) uses experiential teaching on subtle aspects of the interaction (e.g., formulating an understanding of the patient); and 3) has long-term follow-up. The CCSE is efficient, requiring approximately 9 hours of training over a 1-month period (see Figure 1). It is also novel in providing controlled exposure to clinically difficult interpersonal interactions without putting patients at risk of clinical errors.
In contrast with some previous studies (14), the CCSE intervention was used to teach residents, rather than practicing physicians. We believe that postgraduate training is a critical time for setting a foundation of practice attitudes, knowledge, and skills. However, residents rotate every 1–2 months during training and tend not to have longitudinal working relationships with large numbers of patients. This makes it difficult to teach and assess some dimensions of communication competence, such as the capacity to maintain trust over time.
Two limitations of the current study are the small sample size and the lack of randomization to control and experimental groups. Randomization is rarely possible in educational research, as trainees have little tolerance to being randomized to different educational programs or courses. A common solution to this problem is to match controls to trainees without randomization; however, key matching information is often unavailable, and critical variables (such as individual differences in communication self-efficacy) may be unknown at the time of matching. This study used an alternative design in which participants each act as their own control during the 1-month baseline period. This approach is often used in educational research because the chance of spontaneous benefit while waiting for an educational program is small (38).
An additional limitation is the possibility that improvement in communication competence may reflect a self-fulfilling prophecy (39); that is, coaches’ positive expectations may have driven this improvement. Although such an effect is theoretically possible, it is doubtful that coaches’ expectations alone could bring about improvement in a skill as complex and as difficult to teach as therapeutic communication. Similarly, this improvement may reflect a rating bias, where coaches expect improvement and rate their students accordingly. Although this possibility exists, research shows that leniency in ratings is most likely to occur when supervisors must communicate these ratings to long-time employees and experience the professional and psychological costs of imparting a negative evaluation (40). In the current study, coaches did not reveal their ratings to trainees, and no previous relationship existed between the two.
The results of this study provide a strong rationale for further research. For example, the use of qualitative methods could further elucidate coaching and supervisory processes that modify outcomes. Confidence in the validity of the findings would be strengthened by a larger, multisite study. The small sample size makes it difficult to generalize our results, which may be influenced by the fact that our trainees were volunteers invested in improving their skills. Moreover, the findings may not extend to difficult encounters of a type not targeted by the CCSE intervention. Despite these limitations, this study demonstrates the value of the CCSE intervention and lends support for future investigation.
Compared with training-as-usual, participation in a brief, intensive medical education intervention for family medicine trainees was associated with both short-term and sustained improvements in self-efficacy and communication competence. This result supports the idea that therapeutic communication can be taught and learned through systematic, user-friendly, and efficient methods. Future applications of this model in the training of physicians have potential to improve physicians’ communication competence and, in turn, patient care.
This research was supported through grants from Physicians Services Incorporated and the Royal College of Physicians and Surgeons of Canada. We also acknowledge and thank Dr. David Tanenbaum for his help in ensuring the face validity of the SP scenarios for the family practice setting, all the resident participants, the actors who portrayed the standardized patients, and the research assistants who helped to coordinate the study.