Communication Skills Training for Internal Medicine Trainees and Nurse Practitioners
Alison Bays, MD1
Ruth Engelberg, PhD2
Anthony L. Back3
Dee W. Ford4
Lois Downey2
Add other Codetalk faculty authors here
Stuart Alexander5
James Tulsky5
J. Randall Curtis2
Corresponding author:
J. Randall Curtis, MD, MPH
Division of Pulmonary and Critical Care, Box 359762
Harborview Medical Center, University of Washington
Seattle, Washington 98104
Phone: (206) 744-2256; Fax: (206) 744-8584
E-mail:
Abstract word count:
Word Count:
1Department of Medicine, University of Washington, Seattle Washington
2Harborview Medical Center, Division of Pulmonary and Critical Care, Department of Medicine, University of Washington, Seattle Washington
3Seattle Cancer Care Alliance, Division of Medical Oncology, Department of Medicine, University of Washington; Fred Hutchinson Cancer Research Center, Seattle Washington
4Medical University of South Carolina, Division of Pulmonary and Critical Care, Department of Medicine, Charleston South Carolina
5Duke University, Chapel Hill, North Carolina
Supported by the National Institute of Nursing Research: R01 NR009987.
ABSTRACT
Background:Good communication by physicians with patients and familiesis rated as an important component of quality care yet effective training inimportant communication skills, such as giving bad news and discussing end-of-life care, have not been fully evaluated. We designed a randomized trial of internal medicine trainees and nurse practitioner students to investigate the effect of an experiential communication skills-building workshop, called CodeTalk,on trainees’ ability to effectively communicate bad news and express empathy. In this report we describe the effect of CodeTalk on trainees’ communication skills as assessedwith standardized patientinterviews before and after the training. We also examined trainee characteristics that were associated with improvement.
Methods:We recruited nurse practitioner students andinternal medicine residents at the Medical University of South Carolina and the University of Washington.At baseline, trainees filled out a questionnaire on their characteristics as well as attitudes toward and experiences with end-of-life care. They then participated in eight workshop sessions over a one month period. Sessions included skills practice with simulated patients and feedback from experienced trainers. The first and last sessions were audiotaped sessions between the trainee and one of two standardized patients, which were coded for specific behaviors that have been shown to be effective communication tools. Trainees were used as their own controls, comparing post-intervention to pre-intervention scores. We used linear regression to identify predictors of change in communication skills.
Results:We identified XXX potentially eligible trainees of whom XXX agreed to participate (YY%). Of these, 73 (61%) trainees were able to be scheduled for the pre- and post-workshop standardized patients (XX physicians and YY nurse practitioners). Trainees significantly improved in 8 of the 11 coded behaviors (p<0.05). Trainee characteristics that were associated with greater improvement included lower pre-workshop performance, training site, and increasing year of training. We did not find an association with gender or with attitudes toward orexperience with end-of-life care.
Conclusions: CodeTalk was associated with improvement in trainees’ skills in giving bad news and expressing empathy. The association between lower baseline performance and greater improvement could represent a “ceiling effect” on the outcome measures or may reflect increased value of training for those with lower baseline performance. Later training year predicted improvement, suggesting that the intervention may be more effective later in training.
INTRODUCTION
Poor communication between clinicians and patients with serious and life-limiting illnesses, such as cancer, may result psychosocial distress and unnecessary treatment.1 By contrast, effective clinician-patient communication can result in better emotional health, improvement in symptoms and pain control, and reduced intensity of treatment at the end of life.2
Despite the importance of communication, especially in situations of deliveringbad news or discussing end-of-life care, training during medical school and residency is infrequent and junior trainees often find themselves in the position of giving bad news without supervision. A survey of residents at two universities in Boston revealed that, although 73% of residents first delivered bad news as a medical student or intern, 61% of these students had known the patient for only hours to days, a more senior trainee was available in only 11% of the cases, and an attending physician was present in only 5% of cases.3 Similarly, training for nurse practitioners, who are also responsible for conducting these discussions with patients is poor and studies suggest that nurse practitioners have similar shortcomings in this communication.4
Consensus guidelines for communicating bad news were released in 1995 and included the importance of assessing the patient’s understanding of the situation and encouraging patients to express their feelings.5 These and additionalrecommendations guiding best practices for communicating difficult news to patients formed the basis of a workshop, “Oncotalk”, a four day workshop for medical oncology fellows, The workshop emphasized skills practice in small groups with simulatedpatients, using five simulated cancer patients6-8This workshop resulted in significant improvement in communication skills with participants acquiring a mean of 5.4 skills in their ability to give bad news. It is one of few studies that has investigated interventions that teach clinicians how to discuss bad news and transitions to palliative care.9
We are conducting a randomized trial of an interdisciplinary workshop based on Oncotalk – which we call Codetalk – targeting internal medicine trainees and nurse practitioner students. As part of this trial, all trainees randomized to the workshop completed pre-intervention and post-intervention evaluative encounters with standardized patients in which participants were asked to give bad news to the patient. The primary objective of this reportis to evaluate improvementin participants’ skills giving bad news and expressing empathy to patients, as assessed by evaluation of performance with standardized patients. A secondary objective is to identifyfactors that predict improvement in giving bad news and expressing empathy.
METHODS
Intervention
CodeTalk is a randomized trial of an interprofessional, experiential workshop designed to teach end-of-life communication skills to nurse practitioner (NP) students and practitioners and internal medicine residents and fellows recruited from University of Washington (UW) and Medical University of South Carolina (MUSC).
Eligible trainees were contacted prior to the start of the academic year by mail or email; materials included an explanation of the study and a consent form. All internal medicine residents were eligible for the study. Subspecialty fellows were eligible if they were in the following fellowships: pulmonary and critical care, oncology, geriatrics,palliative medicine[j1], and nephrology. Nurse practitioner (NP) students were eligible if they were in training programs, or were already practitioners, for adult patients with cancer or other chronic life-limiting illnesses; older adults, and adult primary care patients. Participants were randomly assigned to either the workshop or the control group, stratified by site, year of training and physician versus nurse[j2]. The control group did not participate in the workshops or the standardized patient evaluations and are therefore not included in these analyses.
Workshops were composed of eight 4-hour sessions that were led by two trainers, one of whom was a physician and the other a nurse or nurse practitioner. Sessions included skill overviews, practice sessions and reflective discussions during which the following topics addressed: giving bad news, talking about advanced directives, interdisciplinary conflict, conducting a family conference, goals of care and Do Not Resuscitate (DNR) status, and transitions to hospice and end-of-life care.9,11 Skills practice was completed with simulated patients and families who appeared at each session at different points in their illness trajectory. Simulated patients and family members were actors or actresses trained to portray specific roles but, in contrast to standardized patients who follow unvarying scripts, they were trained to alter their reactions in response to differences in trainee communication and with faculty guidance, allowing trainees to practice new skills, try different approaches, and provide feedback The simulated patient actors/actresses received training through the University programs for standardized patients and also received Codetalk-specific training including XXX hours of training by Codetalk faculty.
Measures
To provide a process evaluation of the workshop’s effectiveness, participants completed an interview with a standardized, scripted patient at the first and last workshop sessions. We created two standardized patient scenarios for evaluation and each trainee had a different scenario for the pre- and post-intervention evaluation with an equal number of trainees seeing each of the two patient scenarios at pre and post-intervention. Standardized patients were actors or actresses trained specifically for evaluative purposes, behaving the same way in each interaction and were trained at both sites through University programs for standardized patients. In addition, we conducted an additional*** hours of training on the specific scenarios used for this evaluation.
All standardized patient sessions were recorded digitally and analyzed by trained coders who assigned scores based on specific communication behaviors. Coders were trained for *** hours. The coders [j3]were blinded to whether or not the encounter occurred prior to or after the intervention. This methodology was developed previously.9 Fifteen percent of the audiotapes were assessed for inter-rater reliability with к statistics ranging from 0.65 to1.00[j4].
Behaviors selected for coding were based on the communication skills presented in the curriculum and were represented by the acronyms of “SPIKES” and “NURSE”.SPIKES is a six-step protocol for giving bad news and includes: 1) setting; 2) assessing the patient’s perception of his or her illness; 3) obtaining an invitation from the patient to disclose information; 4) giving knowledge and information; 5) addressing emotion with empathic responses; and 6) strategy and summary of the next steps.12 Two of these steps, setting and summary, were not part of the scripted exchange and are not included in these analyses. NURSE is an acronym for verbal empathic expressions8,13,14 and includes the following: 1) naming emotion; 2) expressing understanding of a patient’s feelings or situation; 3) showing respect or praise for a patient’s behavior; 4) articulating support for the patient;and 5) exploring the patient’s emotional state. Each behavior shown by the trainee is coded only once.
In addition to the audiotaped interviews, participants completedan on-line questionnaireprior to the start of the workshop. The questionnaire included items to assess trainees’ experience (n= 6) and attitudes (n18) ([rae5]Block Arnold)towards palliative and end-of-life care.Based on a priori hypotheses about experiences that would be expected to contribute to readiness to learn palliative and end-of-life communication skills, we selected two experience items (i.e., the number of times the trainee personally delivered bad news (scale X, Y, Z), the number of times the trainee observed a more experienced clinician discuss end-of-life treatment options with their patients (scale X, Y Z)).The attitude items were selected using E/CRA analyses that enabled us to identify a scale that captured trainees’ self-efficacy for providing palliative and end-of-life care. The items included: 1) there is little that can be done to ease suffering or grief; 2) it is not possible to tell patients the truth about a terminal prognosis and maintain hope;3) depression is normal in patients with terminal illness; 4) talking about death tends to make patients with terminal illnesses more discouraged; and 5) depression is not treatable in patients with terminal illness. The response options ranged from “strongly disagree” (1) to “strongly agree” (4) and we created an average scoreranging from V to Z. The questionnaire includeddemographic questions (e.g., race, ethnicity years of training/experience, profession). In addition, residency programs and Schools of Nursing provided information on residency/school year and sex.
Hypotheses and Analyses
We hypothesized that, after the workshop, trainees would improve in their ability to give bad news and to express empathy. Trainees pre- and post-workshop code counts on each of the 11 measured skills were compared using the Wilcoxon signed rank test for paired data.
Additionally, we examined whether trainees’characteristics, experiences and attitudes were associated with a change incommunication skills after completing the workshop. We used linear regression with robust estimators to identify predictors of change in communication skills. We conducted two separate regression models, one with a SPIKES total score and the other with a NURSE total score. These total scores were a sum of the change scores (post-pre) for each set of codes. Although the outcome scores were limited in range (SPIKES, -6 to 6; NURSE, -5 to 5), they were relatively normally distributed. The predictor variables for both models included trainee characteristics, the two experience with end-of-life care items, and the average score for trainee attitudes toward end-of-life care. For trainee characteristics, we includedstudy site (UW, MUSC), professional training (MD, NP), sex and year of training[rae6]. Lastly, trainees’ baseline SPIKES or NURSE scores were included in the regression for that specific score, to examine the influence of pre-workshop performance.
SPSS was used to complete the Wilcoxon signed rank tests and STATA 10.0 to complete the robust linear regression models. Significance was set at p <=0.05.
RESULTS
We identified and approached a total of XXX potentially eligible trainees of whom XXX agreed to participate (YY%). Of the consenting trainees, XX were excluded because of scheduling conflicts; 119 were randomized to the intervention. Of these, 73 (61%) trainees completed audiotaped interviews from both the pre- and post-workshop sessions. The trainee characteristics are shown in Table 1.
Table 2 shows the scores of the trainees on the SPIKES and NURSE items before and after the workshop. Of the 6 skills belonging to the SPIKES framework for giving bad news, four skills were studied with six coded behaviors and trainees made statistically significant gains in fourof the six coded behaviors: [A7]1) addressing the patient’s understanding of their illness (perception, p<0.001); 2) requesting permission from the patient before giving bad news (invitation, p < 0.001); 3) using clear language when giving bad news to ensure that the patient understood the cancer diagnosis (knowledge, p=0.05); and 4) providing empathic statements directly following sharing the bad news (emotion, p=0.003). Trainees did not improve significantly on two “giving bad news” skills: 1) giving the patient time to consider the bad news (emotion, p = 0.74); and 2) exploring emotions related to receiving bad news (emotion, p=0.439). Of the five skills represented by the acronym NURSE, trainees improved significantly on the following four: 1) naming an emotion expressed by the patient (naming, p=0.01); 2) showing respect for or praising the patient (respecting, p=0.025); 3) reassuring the patient that the clinician will support them throughout their illness (supporting, p < 0.001); and 4) exploring the patient’s emotion at other times in addition to following the communication of bad news (exploring, p < 0.001). Trainees did not significantly improve in their ability to express understanding of an emotion (understanding, p=0.617).
Predictors of improvement in giving bad news (SPIKES) and expressing empathy (NURSE) are shown in Table 3. Of the eight predictors for improvement in the SPIKES skills, three were significantly associated with improvement after the workshop: trainees improved more at the University of Washington[A8](p<***); the further along clinicians were in training, the more they improved (p<***) and a lower baseline score was significantlyassociated with greater improvement (p<***). Of the eight variables examined for predicting improvement in skills represented by the NURSE acronym, only baseline score was significant: a lower the initial score was associated with the greater improvement (p<***).
DISCUSSION
Codetalk was modeled after Oncotalk and our findings similarly showed that the workshop was associated with significant improvement in trainees’ ability to give bad news and explore emotion, as assessed by standardized patient evaluation.9 Similarly, a randomized trial of a skills-building workshop has also been shown to improve communication skills of oncologists in clinical practice, as assessed by expert evaluation of videotapes with actual patients.16 This previous randomized trial showed no significant improvement with personalized written feedback for oncologists based on videotapes of their encounters with actual patients. These studies suggest that skills practice in a skills building workshop is effective at improving clinicians’ communication skills for giving bad news and talking about end-of-life care.
There were three specific skills that did not improve in our study. Two of three tasks – waiting ten seconds after disclosing bad news and expressing understanding for emotion – were performed by approximately 80% of trainees prior to the workshop and may not have had enough room to show improvement. Interestinglythe Oncotalk study had the same finding with oncology fellows.9The third skill, exploring emotion related to bad news, was performed by a minority of trainees and did not improve, perhaps because of the difficulty trainees have in exploring patients’ emotions related to bad news.17
Our secondary goal was to identify predictors of improvement to guide future studies and wehypothesized thatpriorexperience with end-of-life care would make the workshop more salient and provide trainees with motivation for learning a new skill.18 However, we found no association between performance improvement and either the number of times the trained had delivered bad news northe number of times trainees had observed a more experienced clinician present end-of-life options. In addition, we hypothesized that more positive attitudes towards end-of-life care would be associated with more improvement after the intervention, but found no significant association. It is possible that these associations exist, but our study was not powered to identify them. We did find that a lower baseline performance was associated with more improvement on both the SPIKES and NURSE scales. It is not possible to know whether this represents a “ceiling” effect for the better performers or whether the workshop was more effective for those with poorer baseline performance. Future studies will be needed to fully understand the effect of baseline performance on improvement. Interestingly, higher year of training was significantly associated with better performance on the SPIKES scale, but not on the NURSE. This supports the hypothesis that the intervention may be more beneficial to those further along in their training and also deserves further investigation. [A9]