Impact of a Novel Goals-of-Care Communication Skills Coaching Intervention for Practicing Oncologists

Abstract

Background: Oncologists routinely have opportunities for goals-of-care (GoC) discussions with patients. GoC discussions increase the likelihood that patients receive care consistent with their values. However, oncologists often feel ill-equipped to discuss end-of-life care.

Objective: To assess the impact of a communication training and coaching intervention (INT) for oncologists during GoC discussions.

Design: We randomized oncologists to usual care (UC) or a communication skills training INT, which consisted of an interactive training session and four joint visits with communication coaches.

Setting/Subjects: Solid tumor oncologists seeing advanced cancer patients at four hospitals in New York and Connecticut.

Measurements: Three blinded coders evaluated recorded encounters before and after INT using a validated tool to assess skill attainment.

Results: Oncologists (n = 22) were 32% female and averaged 46 years of age. In baseline visits, INT oncologists (n = 11) and UC oncologists (n = 11) had no difference in the number of mean skills employed out of 8 GoC skills (INT 3.5, UC 2.4; p = 0.18). Post-INT, INT oncologists were significantly more likely to elicit patient values (55% vs. 0%; p = 0.01). There was no significant difference in overall mean skills employed (INT 3.4, UC 2.2; p = 0.14). Assessing for understanding, offering “I wish” statements, and providing prognosis were the least utilized skills among all oncologists.

Conclusion: Our real-time communication skills coaching INT resulted in a significant increase in oncologists' ability to elicit patient values during GoC discussions, suggesting that skill acquisition can occur in the face of less intensive training. Future studies can highlight gaps leading to the lack of differences in utilization of other skills.

Introduction

Skilled communication is central in providing high-quality care to patients with advanced cancer diagnoses. Goals-of-care (GoC) discussions allow physicians and patients the opportunity to define together which medical therapies align with patient preferences. Important components include provision of prognosis, elicitation of patients' goals and values, and explanation of treatment options.^1,2 GoC discussions are associated with higher rates of patient acceptance of terminal illness and higher quality of life ratings.^3,4 These GoC discussions are also associated with lower rates of intensive care unit admissions, decreased caregiver stress, earlier hospice utilization, and even increased survival.^3–6

One major barrier to effective GoC discussions is that many physicians, including oncologists, feel uncomfortable or ill-equipped when discussing end-of-life care.^7,8 They report little training in end-of-life communication: in a national survey of 402 oncology fellows, only 26% of them reported explicit teaching during fellowship on how to help patients and families with reconciliation and saying good-bye, and only 55% reported teaching on how to discuss stopping antineoplastic therapy.⁹ One study of >6000 practicing oncologists found that <33% of them felt that their training was “very helpful” in communicating with patients who were dying or coordinating their care.¹⁰ Furthermore, 56% of them reported “trial and error in clinical practice” as an important source of learning about end-of-life care.

To address deficiencies in oncologist communication in approaching end-of-life care, various communication skills training courses have been developed during which learners practice skills with simulated scenarios and receive structured feedback.^11,12 VitalTalk (VitalTalk.org), originally named OncoTalk, provides interactive clinician development courses that improve serious illness communication skills on an individual and institutional level.¹¹ The Serious Illness Care Program (SICP) triggers trainees to have conversations and prepare families and patients with structured training, e-mail reminders, and coaching from palliative care faculty.¹³ However, these and other courses are resource intensive, costly, and require learners to forfeit a significant amount of time from their existing clinical responsibilities. Given these barriers, the PCORI-funded “Improving Advanced Cancer Patient-Centered Care by Enabling Goals of Care Discussions” randomized control trial (RCT) tested an abbreviated communication skills training and coaching intervention (INT), based on the VitalTalk model. We assessed our INT's impact in increasing utilization of core communication skills by comparing audio-recorded clinical encounters before and after the INT.

Materials and Methods

Study design and setting

We conducted a single-blind RCT testing a GoC communication skills coaching INT. Coders performing assessment of INT impact were blind to the INT status of the oncologist. The study was conducted in four different settings in New York and Connecticut: (1) academic medical center, (2) community hospital, (3) municipal hospital, and (4) community-based rural hospital.

Oncologist sample and survey administration

The study population consisted of solid tumor oncologists who see at least two newly diagnosed advanced cancer patients per month with a prognosis of less than two years. We recruited the oncology division chiefs at each site to identify solid tumor oncologists and encourage participation. Ninety-two percent of eligible oncologists participated and signed a written consent. After informed consent, and before the INT, oncologists completed a self-administered survey. Oncologists were then randomized to the INT or the usual care (UC) group.

Survey instrument

The baseline survey included questions, used in the literature, that assessed oncologists' beliefs about end-of-life discussions,¹⁴ and their prior communication training and level of comfort with specific communication skill topics.¹⁵ Oncologists were asked to indicate their comfort level on a 5-point scale (from “not at all comfortable” to “very comfortable”) with statements about how comfortable they feel talking with patients about specific topics (e.g., giving bad news) and their prior training in these topics. For analyses, we collapsed the “not at all,” “slightly,” and “somewhat” comfortable responses into a “not comfortable” category, and the “comfortable” and “very comfortable” responses into a “comfortable” category. On a 5-point scale from “completely disagree” to “completely agree,” the oncologists also rated statements such as “all patients with advanced cancer should know the stage of their cancer.” The survey also collected information about sociodemographic characteristics (age, gender, race/ethnicity, and marital status), religious preference, year of medical school graduation, field, and status of board certifications.

Intervention

Oncologists were block randomized within each hospital to either INT or UC groups. All oncologists received a one-hour didactic training during oncology grand rounds, presented by a trained VitalTalk facilitator. The grand rounds included evidence about the importance of GoC discussions on shared decision making, which aims to align clinical care with patient preferences, and focused on the five critical components of SPIKES (Setting, Perception, Invitation, Knowledge, Emotion, and Summary), NURSE (Naming, Understanding, Respecting, Supporting, and Exploring) statements, and eliciting patient values.^11,16 The session acknowledged the changing nature of GoC discussions and emphasized the importance of starting these discussions at the first visit after imaging to assess response to first-line treatment. It included role-playing GoC discussions for both when imaging did and did not show disease progression. A full schema of our trial design has been previously published.¹⁷

The INT was an experiential learning model that used a two-hour role-play and four individualized coaching sessions to further practice specific SPIKES and NURSE components. These real-time joint sessions, conducted during oncologists' clinic sessions with their actual patients, were condensed from the four-day VitalTalk model.¹¹ Although communication skills, including SPIKES and NURSE statements, are increasingly integrated into medical training, the INT was aimed at practicing oncologists who receive no communication training. In addition, the INT allowed for practice with individual observation and feedback. The primary skill that the curriculum focused on was eliciting patient values, during both the role-play and the coaching sessions, to best promote alignment of treatment received and patients' goals and values. Exemplars of eliciting values statements include “What gives you meaning and strength” and “When you look to the future, what matters most to you?” When the patient's cancer was not responding to treatment, the following value statements were practiced: “What are your biggest concerns” and “Given this news and when you think about your cancer, what's important to you moving forward?” Coaches focused oncologists on eliciting values during the first joint visit to facilitate building rapport and identifying goals.

Data acquisition

Pre- and post-INT, we audio-recorded a clinical encounter of each enrolled oncologist to assess communication skill utilization. Post-INT encounters were recorded within one year of the role-play session and within six months of the last joint session. For the control group, we recorded encounters within nine months of baseline encounter. Each of these encounters was with a patient who had undergone imaging to evaluate disease progression after completion of the first cycle of first-line cancer treatment. Based on that imaging, the cancer was classified as having progression of disease or no progression.

Measurement of skill utilization

Audiotapes were reviewed by three blinded coders, who were each trained as a VitalTalk communication skills teacher. Each coder evaluated performance using a validated assessment tool, a checklist of the most important communication skills (as identified by communication skills experts when communicating with patients) that was shown to be feasible and reliable in another study (Figs. 1 and 2) and incorporates the most commonly used communication skills tools, SPIKES and NURSE statements.^11,16,18 Core skills identified by consensus of communication skills experts were (1) assessing patient/family understanding, (2) giving information about current condition, (3) avoiding use of medical jargon, (4) responding to emotions, (5) checking for understanding, (6) providing a summary, and (7) eliciting patient values. In our study, “responding to emotions” was measured by the use of empathic expressions that followed the NURSE acronym.¹¹

FIG. 1.

Assessment tool used by coders to score audiotapes of patient encounters, using communication skills identified by communication skill experts to be important in discussions with patients and families. Adapted from the Family Meeting Communication Assessment Tool used in another study.¹⁹ NA, not applicable; NURSE, Naming, Understanding, Respecting, Supporting, and Exploring¹¹; SPIKES, Setting, Perception, Invitation, Knowledge, Emotion, and Summary¹⁶; POD, progression of disease. C.B.S., E.L., and L.P.G. were the coders' initials.

FIG. 2.

Criteria for the assessment tool in Figure 1, with explanations for scoring “yes,” “no,” or “N/A” (or not applicable) to assist coders in scoring audiotapes of patient encounters.

We also defined six additional skills: using silence, providing a treatment plan, providing prognosis, using empathic terminators (which demonstrate a less advanced level of communication), goal setting, and providing “I wish” statements¹⁹ (Table 1). Except for “I wish” statements, the additional skills were not included in the communication skills training course.

Table 1.

Definitions and Criteria of Additional Skills

Skill	Definition
Silence	Allowing patients and/or family members to respond to questions and allowing for natural pauses in conversation.
Treatment plan	Verbalizing the follow-up plan with next steps and arranging for the next meeting.
Providing prognosis	Delineating the prognosis over a time range, for example: hours to days, days to weeks, or weeks to months.
Empathic terminatorsa	Statements that avoid a patient's expressed emotion or that change the topic without responding to emotional cues with empathic expressions. For example, when a patient says, “The cancer is back,” an empathic terminator response would be: “It's okay. We will manage it,” rather than using an empathic statement such as “It must be scary.”
Goal setting	Attempts to elicit a patient's or family member's goals and values in the context of ongoing or future care.
“I Wish” statements	These statements allow the expression of empathy while conveying simultaneously that this expectation is unrealistic.16 For example, “I wish we had more effective treatment for your condition” or “I wish I had better news.”

^aUtilization of empathic terminators demonstrates a less advanced level of communication.

We created a summary score consisting of the seven core skills and the skill of providing “I wish” statements. We present the summary of eight skills as these skills were reviewed during the coaching sessions.

Three blinded coders evaluated pre- and post-INT audio-recordings. Two coders reviewed each recording. Each skill, if exhibited, was counted only once (e.g., only one point would be given for “I wish” statements, even if two such statements were made). The majority (73%) of oncologists' tapes were reviewed by the same two coders. Of the three coders, intercoder agreement was substantial (K = 0.68) for one pair of coders and almost perfect (K = 0.89) for the other pair.

Each site's institutional review board approved the study. The trial was listed on clinicaltrials.gov—NCT02374255.

Statistical survey analysis

Bivariate comparison between pre- and post-INT skill ratings of INT and UC oncologists was performed using the Wilcoxon–Mann–Whitney test for the summary score, Wilcoxon signed-rank sum test for longitudinal measures, and Fisher's exact or McNemar's tests for rate comparisons between specific skills. Using SAS (version 9.4), a generalized estimating equation (GEE) model was applied to account for the correlation between oncologists' repeated measurements, with a cutoff for significance of p < 0.05. The GEE controlled for oncologist delineation (INT vs. UC) as well as any baseline survey items that differed between trial arms. The primary outcome for the model was the change in oncologists' ability to elicit patient values.

Results

Participant demographics and communication skill comfort

We enrolled 22 of 24 (92%) eligible solid tumor oncologists. Of these, 11 were randomized to the INT group and 11 to the UC group. Results on oncologist beliefs and prior training are listed in Table 2. Between trial arms, there was only one significant difference in comfort level with certain communication skills: comfort level with discussing prognosis with advanced cancer patients, with 73% (n = 7) INT oncologists and 18% (n = 8) of UC oncologists reported being comfortable doing so (p = 0.03). There were no significant differences in prior communication skills training (Table 2).

Table 2.

Comparison of Baseline Characteristics of Oncologists

Oncologists' characteristics	% of INT n = 11	% of UC n = 11	p
Mean age (SD)	44.3 (8.9)	43.5 (10.7)	0.86
Years in practice (SD)	17.3 (10.2)	17.0 (9.3)	0.99
Female	36%	27%	0.65
Race
White	64%	82%	0.63
Asian	27%	18%
Black	9%	0%
Ethnicity
Hispanic	9%	9%	0.95
Agreement about goals-of-care topics
“Moderately” or “completely” agree that all patients with advanced cancer should know the stage of their cancer	91%	100%	0.99
“Completely” agree that they would like to know their own situation if they had advanced cancer	100%	73%	0.21
Comfortable in performing the following skillsa
Eliciting patient or family's concerns	36%	55%	0.67
Expressing empathy	91%	100%	0.99
Discussing treatment options including comfort care	100%	100%	0.99
Discussing prognosis	73%	18%	0.03
Discussing discontinuing anticancer-directed therapies (chemo, immune, etc.) that are no longer appropriate?	82%	73%	0.99
Responding to advanced cancer patients or family members who want treatments that you believe are not indicated?	82%	55%	0.36
Prior training in the following skills
Eliciting patient and family concerns	55%	45%	0.99
Expressing empathy	75%b	83%c	0.99
Discussing treatment options including comfort care	73%	55%	0.66
Discussing prognosis	55%	45%	0.99
Discussing discontinuing anticancer-directed therapies (chemo, immune, etc.) that are no longer appropriate?	75%b	83%c	0.99
Responding to advanced cancer patients or family members who want treatments that you believe are not indicated?	63%b	67%c	0.99

Bold values indicate statistical significance.

^aIncluding respondents who rated their comfort level as “comfortable” or “very comfortable” with these abilities in encounters with advanced cancer patients on a 5-point scale in the initial survey.

^bn = 8 respondents.

^cn = 6 respondents.

INT, intervention; UC, usual care; SD, standard deviation.

Skill utilization: Pre- and post-INT

Our summary score combines the seven core skills and the skill of providing “I wish” statements, representing all of the skills that were included in our training course. Of note, there was no significant difference between the summary scores evaluating seven or eight skills. There was no significant difference in oncologists' skills pre-INT between trial arms (Table 3). In the post-INT audiotaped visits, INT oncologists were significantly more likely to elicit patient values (55% vs. 0%; p = 0.01). Examples of value statements that they made included “knowing that you have stage IV cancer, what do you want,” “given what I've told you [about the scan results], what do you hope for,” and “what are you most concerned about?” There was no significant difference in the use of any other skills between pre- and post-INT assessments among UC and INT oncologists.

Table 3.

Comparison of Core Skill Utilization Pre- and Postintervention

	Pre-INT INT n (%)	Pre-INT UC n (%)	p	Post-INT INT n (%)	Post-INT UC n (%)	p
Core skills
Assessing patient/family understanding	2 (18)	1 (9)	0.53	1 (9)	0 (0)	1.00
Giving information	10 (91)	11 (100)	1.00	11 (100)	11 (100)	1.00
Avoiding medical jargon	5 (46)	5 (46)	0.83	9 (82)	6 (55)	0.36
Responding to emotions (NURSE statements)	5 (46)	3 (27)	0.20	5 (46)	1 (9)	0.15
Checking for understanding	3 (27)	1 (9)	0.31	1 (9)	1 (9)	1.00
Providing summary	5 (46)	4 (36)	0.65	3 (27)	5 (46)	0.65
Eliciting patient values	3 (27)	1 (9)	0.31	6 (55)	0 (0)	0.01
Additional skills
Silence	4 (36)	2 (18)	0.36	5 (46)	1 (9)	0.15
Treatment plan	10 (91)	9 (82)	0.48	8 (73)	10 (91)	0.21
Providing prognosis	1 (9)	0 (0)	0.45	1 (9)	1 (9)	1.00
Empathic terminators	7 (64)	9 (82)	0.58	7 (64)	9 (82)	0.64
Goal setting	2 (18)	1 (9)	0.59	3 (27)	0 (0)	0.21
“I Wish” statements	1 (9)	0 (0)	0.48	1 (9)	0 (0)	1.00
Summary scorea
Mean summary score per oncologist	3.5	2.4	0.18	3.4	2.2	0.14

Bold values indicate statistical significance.

^aSkills included in the summary score include eight skills: the seven core skills and the skill of providing “I wish” statements.

NURSE, Naming, Understanding, Respecting, Supporting, and Exploring.

Multivariate model

We used a GEE model that assessed the relationship of oncologists' pre- and post-INT skills summary scores, which includes the seven core skills as well as the use of “I wish” statements, all of which were included in the training course. The model controlled for their INT or UC delineation, as well as comfort level in discussing prognosis, which was the only baseline survey item that differed between trial arms. Using this model, the INT was found to be significant in its effect on oncologists' ability to elicit patient values (β = 0.34; p = 0.04).

Discussion

Effective communication is a critical physician tool when caring for patients with advanced cancer. Despite the many benefits of GoC discussions, they happen infrequently: one study found that oncologists of patients with advanced cancer documented GoC discussions only 27% of the time.²⁰ In the face of these deficiencies, oncologists often feel ill-equipped when discussing end-of-life care,^7,8 and attending skills training courses can be infeasible. We aimed to study the effectiveness of a shortened training course coupled with coaching in actual practice settings on oncologists' utilization of key communication skills by comparing clinical encounters before and after INT. Our training and coaching communication model resulted in improvements in certain skills and highlights gaps for further research.

The biggest difference in skill utilization before and after INT was in eliciting patient values, which was found with significance in bivariate analysis as well as the GEE model. Eliciting patient values is a critical step that allows oncologists and patients to identify together which potential next steps in treatment best align with patient goals. Although all seven core skills and “I wish” statements were practiced in our communication skills coaching model, the INT focused specifically on imparting the skill of eliciting patient values. Honing in on this skill in the INT did improve oncologists' ability to ask about patients' values, but did not lead to a significant change in the use of other core skills. This indicates that focusing more time on certain skills during the INT leads to increased retention over the several months between INT and post-INT audiotape. Offering “I wish” statements and providing prognosis were the least utilized skills both before and after INT (despite the former being practiced in our INT), possibly because of oncologist lack of experience in “I wish” statements or discomfort with delineating prognosis. Furthermore, providing prognosis was not practiced in our INT. All oncologists in both trial arms reported being comfortable in discussing treatment options at baseline. We believe that the low rates of utilization of related skills (eliciting values, avoiding jargon, and checking or assessing for understanding) in both groups before INT demonstrate the need for training courses despite this comfort level.

The INT did not significantly increase oncologist assessment of patient or family understanding, or their use of NURSE statements. This finding is not consistent with results from VitalTalk or the SICP, which found significant improvements in these skills.^11,13 Perhaps in the shortened skills training, the bulk of learning happens during the hands-on coaching sessions, which focused on eliciting patients' values. In addition, the longer VitalTalk INT involved facilitated group practice sessions, whereas our INT employed a brief coaching method for skills practice. In the latter method, debriefing occurs after the encounter is completed. Facilitators in the former method are able to debrief during the discussion to illustrate an important point, which may allow for greater skill retention. The VitalTalk study used standardized patient encounters that occurred within four days after INT, which may also have allowed for greater retention than in our study in which post-INT encounters occurred around eight weeks on average from the joint coaching sessions. Future work can aim to study skill retention at different time intervals, both close to the time of INT and several weeks after. The SICP showed improvements when using a systems-based methodology with self-documentation of communication skills in the electronic medical record (EMR). Our findings suggest that oncologists report greater comfort in core communication skills than observed in practice at baseline (e.g., in discussing prognosis and eliciting patient and family concerns). Perhaps the smaller increase in skill utilization in our study is from using different standards for efficacy than studies such as VitalTalk and the SICP. Future studies may compare the systems-focused or standardized patient methods to ascertain skills with the more direct assessment of patient encounters used in our study.

The existing gaps in skill acquisition could be due to various potential limitations in our training. Our analysis did not account for time elapsed since INT, though all efforts were made to ensure similarly timed tapings of INT and UC oncologists. Time to audiotape was variable and depended on the availability and willingness to be audiotaped of newly advanced cancer patients returning for first postimaging visit. We assessed skill utilization through only one encounter before and after the INT. Results cannot predict consistency in oncologist performance across different encounters and cannot predict growth or retention of skills over time. Our sample size also did not allow for accounting of communication performance based on the type of encounter (progression of disease vs. stable disease). Oncologists may also perform differently knowing that they are being audiotaped (though this should exert a differential effect on INT vs. UC oncologists). As with similar studies such as VitalTalk, our validated assessment tool focused on verbal skills as our training did not emphasize nonverbal techniques, which can be another important facet of effective communication. In addition, this study did not assess the impact of communication skills on patient outcomes.

Further research is needed to evaluate the retention of skill acquisition over time. There may be a possible growth of skill utilization or a return to baseline with subsequent encounters. Future studies may investigate the use of reminders or repeat training to increase skill utilization over time. Training concepts could be reinforced with systems changes, such as the use of a conversation documentation template in the EMR, as used in the SICP trial.¹³ Other courses may also use our results to give more focus to certain skills such as checking for patient understanding and providing a prognosis, which were infrequently performed by all groups in our study. These skills could take greater focus than some commonly used skills among oncologists, such as providing the treatment plan.

Conclusion

This study has various strengths and implications for future practice. Our results show that some communication skill acquisition (e.g., eliciting patient values) can occur even during less time-intensive training, and oncologists do not need to attend a course spanning multiple days to show improvement in skill utilization when discussing end-of-life care. However, the limited scope of improvement suggests that more reinforcement and focus on specific skills may be required to achieve a breadth of core communication skills. Our innovative training model can enhance feasibility to train oncologists in leading more patient-centered GoC discussions in real time with their own patients.

Funding Information

This study was funded by a grant from PCORI (PCORI/IHS-1310-06444), granted to N.B. L.P.G. received additional support from the National Institutes of Health, National Institute on Aging (K23AG049930).

Outside of the submitted study, C.B.S. serves on the speaker bureau for Teva Pharmaceuticals. Outside of the submitted study, S.H.B., A.S.K., and C.B.S. were paid for teaching with the VitalTalk organization.

Author Disclosure Statement

No competing financial interests exist.