Introduction

Shared decision-making is a process whereby clinicians work collaboratively with patients to reach an agreed-upon treatment plan that aligns with evidence and patients’ preferences [1]. Many clinicians report that they consistently engage in shared decision-making when there is more than one reasonable care option [2]. When asked about shared decision-making explicitly, most clinicians support the approach [2, 3]. In practice, clinicians often share information and allow patients to ask questions, but few present true clinical equipoise, fully explore patient preferences [2, 4], or promote patient-clinician collaboration [5]. Researchers have yet to thoroughly examine the ways in which clinicians might influence patients’ choices through explicit or subtle recommendations during consultations [6].

Stivers and Barnes [7] created a taxonomy of recommendations, categorized across five actions: pronouncements, suggestions, proposals, offers, and assertions. These actions represent a gradient of recommendations based on the strength with which they are communicated. Pronouncements are the most authoritative recommendations, delivered without seeking input from patients [7]. Suggestions follow in this taxonomy, where a clinician maintains relative agency over patients’ decision, but might inform patients that they have a role. Proposals occur when a clinician endorses collaboration with a patient, but offers recommendations during the process. Offers change the dynamics of the patient/clinician interaction. The clinician invites the patient to share input that aligns with patients’ preferences and values, and makes a recommendation based on those preferences. Finally, assertions provide broad information about options and their benefit. Clinicians can present these options as relatively equal in terms of outcomes or can use assertions to bias choices, depending on the tone and strength of the assertions. The taxonomy does not have a category which would align with a neutral stance, such as acknowledging clinical equipoise while discussing more than one reasonable choice, but overall provides a useful framework for understanding how clinicians might approach conversations about care decisions.

It is often difficult to isolate how clinicians engage in shared decision-making [8]. Some communication scholars suggest that fostering collaboration can be even more important to shared decision-making than communicating information and encouraging patients to weigh trade-offs between options [5]. In the What Matters Most trial [9, 10], surgeons randomized to one of two intervention groups were trained to use shared decision-making, discuss clinical equipoise, and support shared decision-making with one of two breast cancer surgery conversation aids. Surgeons randomized to usual care provided care as they typically did. Breast-conserving surgery with radiation and mastectomy offer equivalent breast cancer-specific outcomes for patients with early-stage breast cancer [11, 12], representing a preference-sensitive decision [13] appropriate for shared decision-making [1]. Each surgery involves different potential harms, benefits and long-term consequences but have similar survival benefit [13]. National organizations encourage surgeons to ensure that women understand that a more conservative surgery (breast-conserving surgery with radiation) is as likely to lead to survival as mastectomy; some women mistakenly think a larger surgery will improve survival [14–16]. Given an individual patient, though, a surgeon might not agree that there is true equipoise between the options because of the tradeoffs between them. The low annual recurrence rate following breast conservation accumulates and may seem unacceptably high over time for younger patients [17, 18]. In this situation, a surgeon might encourage mastectomy over breast conservation. Mastectomy with reconstruction could be preferred by a surgeon who, based on experience, expects a patient to have a poor cosmetic result with breast conservation [19]. Or, because a mastectomy is more invasive than breast conservation, and requires longer recovery (especially if a woman opts for breast reconstruction with a potential risk of perioperative complications [20, 21], breast conservation could be a better option for some patients with comorbidities or increased baseline operative risk [22, 23]. Similarly, because of recovery time or baseline operative risk, breast conservation could be preferred for a patient without a postoperative caretaker at home or who has limited access to a radiation facility or a plastic surgeon [24–26]. The What Matters Most trial provided an opportunity to examine how surgeons talk to women about this nuanced decision, while the trial provided tools and training in shared decision-making for those in the intervention groups.

Using a purposive sample of transcripts of audio-recorded encounters from the What Matters Most trial, this study aimed to:

1. Examine how surgeons talked to patients with early-stage breast cancer about surgery options, categorizing their approach using the Stivers and Barnes’ taxonomy of recommendations.
2. Compare their approach and use of recommendations between usual care and the intervention groups.

Materials and methods

Results

In the parent trial, of 2057 patients screened, 818 patients were eligible and approached; 622 consented to participate in the trial. Of sixteen surgeons, fourteen had audio-recorded encounters available for the parent analysis (two had low patient volume and/or consent procedures at their respective site that occurred after the encounter began). Recordings lasted 26 minutes on average (range 4–83 minutes).

Table 1 displays descriptive characteristics of the patient and surgeon participants in this analysis 142 of 311 surgeon-patient recorded encounters of 13 surgeons. We collected limited demographic information about surgeons to protect confidentiality. Patients were 59 years on average (range 25 to 89). Slightly more than half identified as White, non-Hispanic (81/142, 57.0%). About one-third (46/142; 32.4%) had a 4-year college degree. Surgeons were mostly female (10/13; 76.9%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Participant characteristics.

https://doi.org/10.1371/journal.pone.0260704.t001

Table 2 displays actions in the framework and examples of how decisions were discussed within each category of recommendations. Surgeons using one of the two conversation aids displayed more shared decision-making than those in usual care, describing options and reviewing potential harms and benefits by acknowledging clinical equipoise in all encounters (usual care 64% vs. intervention 100%; p<0.001). Nevertheless, surgeons across study groups gave recommendations about treatments that used authoritative pronouncements (usual care 51% vs. intervention 36%; p = .09) which does not align with shared decision-making principles.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Representative quotations from coded transcripts of encounters.

https://doi.org/10.1371/journal.pone.0260704.t002

Discussion

In our analysis, many surgeons made authoritative recommendations within the categories of pronouncements and suggestions to patients with early-stage breast cancer making a preference-sensitive surgery choice. Those trained in shared decision-making and provided with a conversation aid were more likely than those practicing as usual to describe clinical equipoise and support deliberation during the encounter, describing this important concept in 100% of encounters in the intervention groups. However, across study groups, many surgeons used informational assertions to support strong pronouncements about what to choose. This study suggests that presenting comparable options does not necessarily lead to shared decision-making without support for collaborative partnership and consistent language acknowledging choice [5, 29].

Our study reinforces the notion that shared decision-making is nuanced and complex in the clinical environment. Elwyn and colleagues’ Three Talk Model of shared decision-making emphasizes Team Talk, where clinician and patient establish a partnership, Option Talk, where they discuss alternatives and risks, and Decision Talk, where the discussion turns to the patients’ preferences to reach an agreed-upon treatment plan [27]. Part of this shared decision making process can occur with or without true clinical equipoise. In fact, decision aids often support Option Talk, as they help patients deliberate and clarify their preferences through understanding information about options and thinking about how possible outcomes of options might impact their life [30, 31]. Likewise, the conversation aids in this study usually prompted surgeons to perform Option Talk, as they discussed risk and benefit information for each surgical choice. However, full shared decision making in the presence of more than one reasonable option should engage patients as partners through making this choice, exploring patient preferences rather than guiding the decision based on a clinician’s preference. As surgeons moved along in the consultation about breast cancer surgery choice, subtle or explicit recommendations were sometimes inconsistent with Team Talk and Decision Talk. Pronouncements (authoritative statements) in the intervention groups sometimes included Option Talk, but not Team Talk or Decision Talk. When Team Talk occurred with suggestions (more subtle but strong recommendations), it was often followed by strong endorsements of one surgical choice over another, suggesting that clinicians might not have perceived true equipoise about the decision. Even in proposals and offers that acknowledged patient autonomy while providing advice, surgeons occasionally oscillated between engaging in a deliberation and making recommendations. They often adapted their approach to shared decision-making as clinical visits progressed.

Given the complexity of practicing shared decision-making, even with training and tools, surgeons might need more support for how to elicit patients’ preferences and align care with those preferences before endorsing an option. Skills like active listening and responding to patient preferences might need to be paired with shared decision-making interventions. Most, if not all, surgeons in the study displayed humanistic elements of shared decision-making, like respect, compassion and empathy [32]. However, within their patient-centered approach, their own personal preferences came through either explicitly or subtly. Because of the natural tendency to provide recommendations, support in making recommendations after responding to patients’ preferences could enhance shared decision-making training. Training might need to include ways to respond to patients’ questions such as “what would you do?” while remaining consistent with the spirit of shared decision-making [33]. Training could include more specific role-playing about how to make a recommendation after engaging with patients to explore preferences (i.e., during Decision Talk), not at the start of an encounter or during an explanation of the pros and cons of options (i.e., during Option Talk). Finally, national organizations that might encourage surgeons to reduce mastectomy rates could instead support surgeons in documenting when they have engaged in shared decision-making to support patients’ informed, preference-consistent choices.

The What Matters Most trial provided a unique opportunity to explore the role that clinician communication styles play in surgical encounters [10] by providing shared decision-making training (including in communicating about comparable options, facilitating patient engagement and encouraging deliberation), tools and support to clinicians [2, 34]. Using the established communication framework about treatment recommendations was novel in the context of shared decision making research, and coding audio-recorded encounters built on this framework of recommendations to examine results. Most past work examining shared decision making behaviors during clinical consultations has explored whether clinicians acknowledge options and invite the patient to participate [35–38], rather than whether and how they make treatment recommendations. More recently, scholars recommend focusing on care that works for patients and makes sense given the context of their lives, above and beyond describing information about clinical outcomes [39]. The research presented here adds to this approach about how information is integrated into clinical encounters when formulating treatment plans and weighing reasonable options.

However, there are some study limitations that should be considered. These analyses were conducted across 142 patients of 13 surgeons discussing early-stage breast cancer surgery. Results might not apply across other contexts outside of surgery or breast cancer conversations. It is possible that clinicians did not support the overall shared decision-making model as they were faced with specific clinical scenarios or patient characteristics [3]. They might also believe that individual patients could benefit more from one treatment given baseline risk, age, or comorbidities. They might feel pressure from national societies to reduce the use of mastectomies when breast-conserving surgery with radiation is an option, if women mistakenly think a larger surgery will improve survival [14–16]. Thus, the surgeons might not have perceived true equipoise across surgery options. However, patients’ eligibility for both surgical options was confirmed by surgeons prior to inclusion. Almost all surgeons were interested in shared decision-making before the study started. An in-depth discourse analysis [40, 41] could provide more details on the quality of conversations, rather than focusing specifically on treatment recommendations and/or conversation descriptions.

Conclusion

Our study suggests that clinicians often make both subtle and explicit recommendations about choices. Conversation aids can facilitate offers of comparable options, but do not on their own support shared decision-making without an explicit acknowledgement of equipoise and an invitation to collaborate. More work should explore how language and framing of recommendations can influence choices.

Supporting information