Case

Consider the treatability statements in the following hypothetical case, based on an actual clinical case, and mirroring many cases reported during interviews with physician participants in our empirical work. In this case, we highlight two treatability statements (sentence 1 and sentence 2) that we will refer back to throughout our analysis.

Ms. P, an elderly woman with a history of recently diagnosed metastatic breast cancer, was found unresponsive and sedated in her apartment by her only daughter. After being brought to the hospital by emergency medical services, treated for possible opioid overdose, and evaluated by the intensive care unit (ICU) team, she is admitted to the ICU with severe sepsis of multiple possible sources. Standard aggressive treatment with IV fluids, vasoactive medications, and broad-spectrum antibiotics is initiated.

Due to lack of insurance and fear of cancer, Ms. P had waited an entire year before presenting to clinic three months ago with a breast lump and chest pain. After work-up, she was diagnosed with metastatic breast cancer and started on appropriate treatment, including cancer-directed hormonal therapy, pain treatment, and treatment to manage the metabolic derangements from her multiple lytic metastases.

Ms. P is currently disoriented and cannot communicate meaningfully with the clinical team about treatment decisions or prognosis. There is no indication in her medical chart that a broader goals of care discussion has taken place; she has just begun cancer-directed treatment. Ms. P’s daughter, her surrogate decision maker, knows that her mother has breast cancer, but has not yet spoken with anyone about prognosis or her mother’s treatment preferences.

Given Ms. P’s age and medical comorbidities, the ICU team wonders if she will survive this acute episode of sepsis. And even if she does survive, they predict she will return to the hospital with another decompensation. According to the oncologist, multiple forms of treatment for her breast cancer may be available if Ms. P stabilizes and survives to discharge, including further hormonal therapy, chemotherapy, or palliative radiation for symptomatic masses. However, her oncologist wonders if her poor functional status might mean that these treatments will have a poor benefit-burden ratio.

To discuss these matters, the ICU team and the consulting oncologist hold a conference with the patient’s daughter. The intensivist provides a summary of Ms. P’s ICU course, being careful to explain that Ms. P is elderly, critically ill, and may not survive to discharge. As part of explaining what the ICU team is currently providing, the intensivist says that, in addition to providing supportive care:

The conversation then turns to a discussion of the patient’s underlying cancer. The oncologist carefully explains that, if Ms. P survives this hospital stay, “she will still ultimately die from her cancer.” In order to reassure Ms. P’s daughter, the oncologist adds:

He clarifies that any treatment options “will need to be discussed” with Ms. P’s outpatient oncologist at a later time.

At the conclusion of the meeting, the daughter feels generally overwhelmed and understands that her mother is critically ill and may die during the hospital stay. She thanks the physicians for everything they are doing and asks that they continue to “do everything.” The physicians worry that Ms. P’s daughter does not understand her mother’s prognosis. Even though they feel they have communicated clearly, they worry about difficult goals of care conversations in the future, and how to help Ms. P’s daughter better understand her mother’s clinical situation.

Treatability Statements: Levels of Meaning

We have extricated sentence 1 and sentence 2 from their larger conversational context for two reasons. First, in our empirical research, ^{Footnote 5} we found that treatability statements are pivotal: patients tend to focus on them and derive a great deal of implicit meaning from them, often to the point of overwhelming other explicit information. Our physician participants also observed that patients often “cling onto” treatability statements in the context of serious illness, and derive unwarranted hope from them. Second, focusing on the treatability statements while still acknowledging their general conversational context allows us to analyze how a key conversational fragment may convey meaning pragmatically. Our aim is to illustrate that even a single sentence that conveys unintended pragmatic meaning may result in clinical miscommunication that impacts treatment decisions.

To unpack sentence 1 and sentence 2, we will define three levels of meaning for treatability statements. By literal sentence meaning, we refer to the conventional semantic meaning of spoken words. On this level, treatability statements simply mean that physicians have therapeutic interventions (treatments) they can use. Treatability statements communicate an essential aspect of a patient’s clinical situation: can physicians do something, or not? For example:

Note how lean the literal sentence meaning is. It does not convey whether the physician plans to use the treatment, whether the physician thinks the treatment is a good idea, or what the goal of the treatment is. It simply states that a treatment exists, and that physicians can use it.

However, our empirical work demonstrates that treatability statements convey meaning far beyond the literal sentence meaning.^{Footnote 6} These additional meanings are informed by contextual factors—the identity of the speaker, assumptions about the speaker’s intent, knowledge of the clinical situation, etc. Accordingly, we define the remaining two levels of meaning with these elements in mind: by physician’s intended meaning and patient’s received meaning, we refer to the meaning a physician intends to convey, or that a patient receives, respectively, from a treatability statement in a particular conversational context.

In the case above, we can unpack these levels of meaning for the different participants as follows:

The marked difference between the physician’s intended meaning and the patient’s received meaning illustrates that patients do not “successfully” arrive at what the physicians intend to convey. There are several key differences between the intended and received meaning. First, physicians are primarily concerned with clarifying the goals of treatment (e.g., reversal of an infection, modifying the cancer), while the daughter is primarily concerned with prognosis (e.g., is there good news for my mother?). Second, the physician’s intended meaning is framed in technical terms relevant for the physician’s work (specific clinical problems and interventions), while the patient’s received meaning is framed in everyday terms relevant for the patient’s life and experience (surviving, hoping, getting better). Lastly, the physician’s intended meaning is neutral with regard to treatment efficacy and impact, while the patient’s received meaning assumes that treatment is effective and will help the patient. These two levels of meaning are rarely if ever identical, and our empirical research demonstrates how far apart they often are.

We chose these particular intended and received meanings to illustrate that treatability statements come to have divergent meanings. These meanings are representative of some of our participants’ actual responses. Of course, an individual physician or patient might disagree with a meaning above or view these uses of treatability statements as unwise; however, from our empirical data, at least some physicians and patients report using or interpreting treatability statements in these ways.^{Footnote 7} The rest of this paper will explore how treatability statements come to have divergent meanings and the ramifications of these divergences for informed consent and shared decision making.

Treatability Statements as Conversational Implicature

One way to understand how treatability statements convey meaning beyond the literal sentence meaning is to view them as instances of conversational implicature. An implicature is an “act of meaning or implying one thing by saying something else.”^{Footnote 8} A conversational implicature occurs when that meaning is conveyed not by the literal meanings of the spoken words, but rather by the relationship between the words and their conversational context. For example:

Barb intends to convey, “No I cannot go to Paul’s party.” She does so by saying something entirely different: “I have to work.” Stripped from context, Barb’s intended meaning cannot be derived from the literal (semantic) meaning of her words; there is no way to recombine the dictionary definitions of Barb’s words (“I have to work”) to make them mean “I cannot go to Paul’s party.” In fact, Barb’s response could have conveyed the opposite meaning with the addition of a simple phrase: “I have to work…but I think I can still make the party.” Barb’s intended meaning is conveyed through the relationship between her spoken words and their conversational context.

H. Paul Grice (1989) proposed the concept of conversational implicature in his work exploring how speakers convey meaning in particular conversational contexts.^{Footnote 10} Grice’s work assumes that listeners are able to correctly calculate the intended meaning of the speaker, as in the example above. One of Grice’s key contributions is to highlight that we are only able to determine a speaker’s intended meaning by assuming that he or she participates in the conversation in a cooperative manner—that is, that the speaker intends to convey meaning that is relevant. In the example above, we assume Barb is communicating something relevant by: (1) not explicitly answering “yes”, and (2) saying, “I have to work.” In the conversational context, we conclude that she is answering the question, saying she cannot attend the party by providing the reason she cannot attend. We are able to reconstruct Barb’s intended meaning by combining an assumption of cooperativity and relevance with a wide array of contextual knowledge and the semantic meaning of the spoken words.

Extending Grice’s framework to treatability statements: a physician utters a treatability statement expecting to convey the physician’s intended meaning, even though it is never explicitly said. The role that contextual knowledge and assumptions play in deriving this meaning remains largely implicit and may not even be recognized. The issue of interest for this paper is our empirical observation that the physician’s intended meaning may diverge from the patient’s received meaning. Therefore, we now turn our attention to more deeply understand this divergence and its ramifications for informed consent and shared decision making.

The Importance of Shared Experience

William F. Hanks (2002) notes that successful conversational implicatures, in which the intended meaning and the received meaning align, depend on a “shared horizon of familiar experience.”^{Footnote 11} Listeners use this shared body of experience to derive the intended meaning of the speaker. Our empirical work illustrates that the unstated assumptions required for a shared understanding of treatability statements may not be shared even amongst physicians. For example, in the case above, the oncologist and the intensivist might disagree as to whether the patient’s incurable, metastatic cancer is ‘treatable’ or not. In our empirical work, many nononcologists claimed that incurable metastatic cancer is ‘not treatable’ by definition, and were surprised to learn that oncologists often reserved ‘treatable’ to describe such cancers.

This difference in perspective illustrates that the conceptual architecture employed in interpreting ‘treatable’ is historically and culturally specific; each physician sees ‘treatable’ as pointing to a distinction relevant from his or her vantage point in the healthcare system. These distinctions matter for conceptualizing the treatment decisions that each physician deals with on a day-to-day basis.

For the intensivist, the relevant distinction is between treating and supporting, because this makes a difference in determining the focus of ICU-level care: are we actually treating the patient’s core issue, or are we merely supporting them through the natural history of the disease? Since all ICU patients receive supportive care (physiologic support such as mechanical ventilation, vasoactive medications, fluids), treatment often refers to treatment beyond supportive care, which is disease-directed and intended to reverse a pathophysiologic process, returning the patient to their pre-ICU baseline.

For the oncologist, the relevant distinction is between treating and curing: do treatments aim to cure, or do they merely delay the patient’s inevitable decline and death? Therefore, by not explicitly stating that the cancer is potentially curable, the oncologist signals that treatment has some other goal. Since it is the cancer (and not the symptoms) that is “still treatable,” the treatment must still be cancer-directed. By saying that “the cancer is still treatable,” the oncologist conveys that this is an incurable cancer for which cancer-directed treatments are still available.

For both the intensivist and the oncologist, the intended meaning of the treatability statement only makes sense within a large body of clinical knowledge and experience, such as: (a) features of the disease at hand, (b) general categories of therapeutic interventions (e.g., disease-directed vs symptom-directed), (c) differences in scope and site of practice among physician specialties, and (d) general clinical concepts such as treatment, cure, or support.

Individuals may be completely unaware of the extent to which their communication depends on a large body of everyday knowledge and experience. In our empirical work, we observed that some intended meanings of treatability statements become so automatic that they are virtually idiomatic within a clinical specialty. For example, many of our oncology participants discussed that “treatable” means or implies “incurable” in an almost definitional sense. Similarly, many of our intensivist participants discussed that “treatable” means or implies “reversible” or “curable.” The pragmatic logic underlying these intended meanings was largely implicit, and only elicited through extensive interviews. To some degree, these intended meanings appeared to be automatic among certain groups of physicians who shared a particular clinical context and culture as their everyday environment. This matters because the discordances between the intended and received meaning may go undetected: the average physician who utters a treatability statement is likely not cognizant of the extensive clinical knowledge and analysis required to arrive at the intended meaning.

In everyday conversation, the unconscious and automatic nature of conversational implicature is an advantage: it signals pragmatic fluency and allows for greater efficiency of communication. But in the clinical environment, when differing parties arrive to high-stakes conversations with vastly different experiences and knowledge, there may be unintentional or undetected misfires. The intended meaning, though it is automatic or even obvious for the cultural insider (a physician of a particular specialty), remains opaque for the cultural outsider (a physician of a different specialty, or a patient). Thus, one reason treatability statements come to have divergent meanings is because the different parties do not have access to the same knowledge and experience.

This is particularly relevant for communication between physicians and patients. Physicians couch complex technical distinctions in the ostensibly everyday language of ‘treatment.’ In our empirical data, some physicians discussed that ‘treat’ is useful language precisely because it appears to be nontechnical—a convenient “middle ground” for discussions between expert and lay individuals. But as we have shown, physicians’ understanding of ‘treat’ relies on a complex network of clinical experience and knowledge. Consider that a dictionary definition of ‘treat’ is simply “give medical care or attention to.”^{Footnote 12} This definition, which seems to approximate patient understanding, might include all of the interventions we have discussed (physiologic support, antibiotics, curative chemotherapy, and palliative chemotherapy). Thus, based solely on the treatability statements above, a patient would have no reason to suspect that the speaking physician is drawing technical distinctions between types of therapeutic interventions and whether or not they constitute treatment. In treatability statements, technical concepts masquerade in everyday language.

Our empirical data suggests that even the most sophisticated patients do not possess the relevant contextual and background knowledge to successfully arrive at the physician’s intended meaning. Patients who become more familiar with the clinical context over time (e.g., through chronic illness) may be more familiar with the technical jargon of the healthcare environment. Though helpful, this semantic fluency does not mean we can expect that pragmatically conveyed ideas will be understood.

Divergent Assumptions About Conversations

Thus far, we have claimed that the divergent meanings of treatability statements stem from differing ideas and asymmetry in clinical knowledge and experience. This dynamic is further complicated by differing assumptions about the conversation itself—that is, assumptions about what physicians who utter treatability statements intend to accomplish. By comparing the physician’s intended meanings with the patient’s received meaning, we can elucidate these important, unstated assumptions.

Physicians assume that treatability statements are made to communicate that something can be done, and to clarify what that is; treatment is conceived of as a tool to accomplish particular clinical goals, defined in specific, technical terms. Hence, physicians use treatability statements to convey that they can “do something.” In contrast, patients assume that treatability statements are made to provide news about a patient’s future; treatment is conceived of as inherently beneficial, making a substantial difference for a patient’s life or experience in terms that matter to the patient. Hence, patients hear “good news.” This difference explains why the physician interpretation is inherently technical, while the patient interpretation is inherently prognostic. For the physicians, treatability statements emphasize what the physician can do; for patients, they emphasize how the patient will do.

We hypothesize that these differences in reasoning, which are clearly apparent in our empirical data, are grounded in differing understandings of the conversational context. Physicians tend to approach conversations in serious illness with informed consent in mind. In the standard bioethical narrative of informed consent, treatment options are presented, the risks and benefits of each option is discussed, and the patient or surrogate is allowed to select a treatment option according to their own autonomous preferences.^{Footnote 13} Therefore, it is somewhat common practice among physicians, especially trainees or junior attending physicians, to discuss and even offer treatments that they do not believe are beneficial, with the ultimate intent of persuading patients or surrogates against them.^{Footnote 14}

But patients are not privy to the shared implicit understanding of informed consent that physicians rely on. Patients generally do not exhibit awareness, in our data at least, that physicians might offer or even discuss treatment that they would not recommend or would consider to be inappropriate or nonbeneficial. In fact, some patients even expressed mistrust or suspicion when a physician described a terminal cancer as “treatable” when only palliative therapies were available.^{Footnote 15} Instead, our data suggests that patients assume an entirely different set of conversational purposes when providing updates regarding serious illness: to communicate good or bad news about the patient (prognostication), to express whether or not they will ‘help’ the patient (intentionality), or to communicate hope (emotional signaling). Because patients and physicians do not possess the same understanding of the conversation itself, they make different assumptions about the speaker’s intent. This unavoidably affects the pragmatic inferences drawn by physicians and patients.

Informed Consent and Shared Decision Making

Having unpacked the divergent meanings of treatability statements, we will now explore the stakes of this divergence. We will consider the ramifications of this divergence first for informed consent, then for shared decision making.

Disclosure, or physician communication of relevant clinical information, is an important element of informed consent.^{Footnote 16} For treatability statements to function effectively in disclosure, the physician’s intended meaning must align with the patient’s received meaning. But these meanings, as we have shown, can diverge—sometimes radically so. This divergence in meaning poses two distinct threats to the process of disclosure: treatability statements become (a) underinformative, and (b) falsely informative.

As physicians intend them, treatability statements are underinformative, since they primarily clarify whether therapeutic interventions of a particular sort exist, and what these interventions aim to accomplish. In interviews, physicians typically stated that treatability statements have no fixed implications for prognosis, quality of life, or probability of treatment effectiveness—all of which are vital pieces of information to be disclosed, and all of which are often what the patient wishes to know when hearing a treatability statement. These pieces of information are especially vital for “preference-sensitive” decisions, as the patient’s final choice may hinge on what information is presented and how it is presented.^{Footnote 17}

Our case illustrates this well: while Ms. P’s infection is ‘treatable’ by antibiotics, the physician does not mention that the antibiotics may have a low probability of success. The patient may, in fact, have a higher chance of dying from her infection than clearing it, yet it is still ‘treatable.’ And while the antibiotics aim to return the patient to her pre-ICU baseline, that baseline is still dying of metastatic cancer. Similarly, while the cancer is ‘treatable,’ in that cancer-directed therapies can be given, the oncologist has neither said nor implied that he thinks this treatment is a good idea; further cancer treatment may have a horrible benefit-to-burden ratio or may result in an unacceptable quality of life.

It is these unsaid elements that may be particularly discordant with the good news and encouragement to pursue treatment that a patient may derive from the treatability statements. In other words, for the patient’s daughter, the treatability statements are falsely informative, conveying positive information about prognosis, quality of life, and physician intention to treat—none of which were intended by the physicians.

These dynamics challenge conventional understandings of shared decision making, which tend to assume that requisite information transfer is possible and can be accomplished by stating the relevant information.^{Footnote 18} But as we have seen, lack of attention to the pragmatic features of communication can lead to loss of relevant information and the conveyance of unintended information. Whether or not these pragmatic misfires can be identified and/or overcome by physician communication strategies is a topic for further empirical research. Frankly acknowledging these communication challenges (and whether or not they may be ameliorated) may cause us to question the achievability of shared decision making as we currently conceive of it; at the very least, we need to consider how our current methods of operationalizing shared decision making—by attempting to convey information literally—may be ineffective.