During the initial consultation with a patient to communicate a diagnosis of late-stage cancer, the oncologist may refrain from giving survival statistics, redirecting the conversation from the bad news (incurability) to the practical aspects of the patient’s care (treatments, timetables, appointments, and testing to monitor response to treatment).1, 2 Whether conscious or unconscious, this diversion—call it “the stage IV shuffle”—helps cushion the impact of the disturbing news.

This paper will show that clinicians’ gingerly handling of harsh facts when they talk with patients also applies to health educators when they prepare materials about advanced cancer for the general public and to researchers when they report cancer research findings in the medical literature. While the intentions may be benevolent, the combined effect is to distort perceptions about what medical science has to offer patients with late-stage cancer, possibly compromising the ability of patients and families to make informed choices.3

CLINICIANS AND HEALTH EDUCATORS

On receiving a diagnosis of advanced cancer, most patients especially want to know the answer to a single question, typically phrased as “What are my chances” or “How long have I got?”1, 4 These formulations of the question mirror the two standard formats for presenting statistical data on prognosis: five-year survival (or survival using a different interval) and median survival time. Although pertinent data are available,5 the patient does not always receive a straight answer. Instead, the oncologist may offer, “it is extremely difficult to give any indication of the general prognosis because each patient is unique.”1 Such a response may represent a well-meaning effort to mitigate the shock of the bad news, but the patient may sense that critical information is being withheld—that the clinician is ignoring the elephant in the room.

If the patient fails to elicit a meaningful response from the clinician and turns to a cancer research or advocacy organization, obtaining the data from that source will be no easier (Box 1). Discussions about treatment efficacy are even less common.6,7,8, 10 The conversation focuses on what will be done, not how well it will work. What could explain this communication blackout?

Box 1 Hiding survival data in educational materials
Full size table

First, most treatments for late-stage cancer offer such modest benefits that clinicians and health educators may prefer to gloss over the facts of efficacy. It is difficult to inform a patient not only that the prognosis is poor, but that treatment is likely to offer only marginal survival gains.10,11,12 Instead, the clinician may state, “Your cancer is incurable, but it is treatable”13 and “We never know how an individual patient will respond to this therapy.”1 The UK’s National Institute for Health and Care Excellence systematically omits efficacy data from lay educational materials about advanced cancer.10 Since patients are not likely to appreciate the distinction between “treatable” and “effectively treatable,” they mentally substitute the latter for the former.1 This substitution may explain why so many patients believe their metastatic cancer can be cured3 and why so few patients with advanced cancer decline chemotherapy.14

Second, the usual way of presenting efficacy data in the medical literature can be difficult for clinicians to interpret. Randomized controlled trials and meta-analyses typically report results using the statistical measure, hazard ratio, often accompanied by survival curves comparing treatments.15 The hazard ratio summarizes in a single number the magnitude of the difference between alternative treatments over a study’s entire period of follow-up. While this mathematical construct does incorporate the essential information, it does not effectively communicate the meaning of the information in a way that can help clinicians and patients compare alternatives and make informed choices. A non-statistician gains little insight from knowing that, for colorectal cancer, the hazard ratio for chemotherapy vs. no chemotherapy is 0.65 (95% confidence interval, 0.56 to 0.76).16

CANCER RESEARCHERS

When reporting hazard ratios, researchers also frequently report a related statistical measure containing the same information: reduction in the risk of death. This measure is easily misinterpreted in a way that exaggerates the benefit of a treatment (Box 2). The widespread use of the measure in the medical literature may contribute to a blurring of the truth—an academic version of the stage IV shuffle.

Box 2 Confusion associated with “reduction in the risk of death”
Full size table

Another practice, commonplace for decades, involves a shift in focus from survival to a different endpoint, such as “tumor response” or “progression-free survival.” In a clinical trial, these surrogate outcomes have a practical advantage over survival: they require a shorter period of follow-up for data collection. In everyday practice, the rationale for measuring these outcomes is that they indicate if and how well a treatment is working. If a tumor shrinks or fails to grow for 6 months, that must be a good thing. It shows that the treatment has suppressed the cancer and gained the patient some time.

The problem with this narrative is that indicators of disease progression have not shown a consistent correlation with patient survival—the uncontested scientific gold standard for measuring efficacy.18 When chemotherapy does suppress tumor growth, this does not necessarily mean that the patient will live longer. Nevertheless, researchers often report tumor response or progression-free survival in cancer trials, sometimes designating one or the other as the primary outcome in place of survival. In fact, in recent years, more than half of all clinical trials leading to approval of a drug by the U.S. Food and Drug Administration for oncologic indications have used surrogate outcomes.19 Since chemotherapy almost always has a greater effect on these outcomes than on survival, the shift in focus tends to make results look more favorable.

GETTING TO STRAIGHT TALK

The first obstacle to straight talk about advanced cancer—uncertainty about how to balance candor with tact when communicating with a profoundly distraught patient—presents a greater challenge than the second obstacle—confusion caused by opaque methods of presenting data. Solving the first problem will require a divergence from entrenched professional and social norms. Patients, themselves, exacerbate this problem by colluding with physicians to keep the focus on the treatment calendar and steer the conversation away from the illness trajectory.1 Solving the second problem requires a better data display.

A Better Data Display

Survival curves are easier to interpret than hazard ratios. A survival curve offers visually compelling answers to questions about prognosis. A pair of survival curves representing the two arms of a clinical trial (e.g., chemotherapy vs. best supportive care) conveys information about efficacy in a straightforward manner. Adding one more component to this graphic—a stacked bar that interprets the displayed curves—converts a set of survival statistics to a “benefit analysis,” a proposed framework for straight talk and informed decision making (Fig. 1).

Fig. 1
figure 1

Survival benefit of conventional chemotherapy for stage IV cancer. These smooth exponential survival curves were constructed using median survival times reported in the pertinent literature, relying on the fact that survival curves for most stage IV cancers closely approximate negative exponential functions

Full size image

Figure 1 illustrates benefit analyses for selected neoplasms typically treated with chemotherapy in advanced stages of disease: breast, lung, and colorectal. These 3 malignancies account for about 35% of all new cases of cancer.20 The graph for breast cancer was constructed using data from an observational cohort study that compared survival in patients diagnosed before and after the introduction of fluorouracil chemotherapy in the mid-1950s.21 The graphs for lung and colorectal cancers are based on survival data derived from meta-analyses of chemotherapy trials from the 1970s and 1980s involving each type of malignancy.15, 16 The reason for relying on older studies is that recent chemotherapy trials rarely include an untreated control arm, which is needed to compare the benefit of chemotherapy vs. no chemotherapy. A suggested method for updating results from earlier studies is discussed later in this section.

A benefit analysis can answer these questions: (1) What is the probability that I will benefit from treatment? and (2) How much time will I gain? From Fig. 1, a patient with stage IV colorectal cancer has a 13% chance of benefitting from conventional chemotherapy, assuming a 2-year time horizon. This information yields a straight answer to the first benefit question (Box 3). The figure also shows the gain in median survival time—4 months in the case of colorectal cancer—providing an answer to the second question. The two ways of expressing benefit (probability and time gained) correspond to the vertical and horizontal distances between the survival curves measured at the 2-year and 50% survival milestones, respectively. These ideas can help engage patients in shared decision-making (Box 4) and facilitate clear thinking about the observed effects of treatment (Box 5).

Box 3 Straight talk about chemotherapy for metastatic cancer
Full size table
Box 4 Using the “benefit” concept to engage patients in shared decision-making
Full size table
Box 5 “We fought aggressively and the strategy paid off”
Full size table

The “benefit” in benefit analysis is identical to the “absolute risk reduction” sometimes used to report results of clinical trials. Scientific reporting guidelines22 and leading medical journals encourage authors to use this measure rather than the alternative “relative risk reduction,” which can make modest gains seem to be more substantial than they are.

Prolongation of life may not be the only goal of chemotherapy. If other goals such as symptom relief influence decision-making, then the efficacy of chemotherapy in achieving those goals should be quantified and incorporated into a benefit analysis. The patient needs an evidence-based answer to the question, “What am I going to gain from this treatment?”

To the extent that contemporary chemotherapy—and newer approaches such as immunotherapy—outperforms or enhances the effects of conventional chemotherapy, the survival benefit based on earlier placebo-controlled trials must be adjusted accordingly. Since present-day cancer trials usually lack a placebo or null control group, it is difficult to determine the survival benefit of the new treatment compared with supportive care alone. One possible approach is illustrated in Box 6. Developing rigorous methods for estimating treatment-vs.-supportive care differences in the absence of an appropriate control group deserves high-priority attention as a topic for cancer research methodology. Such methods would make it possible to determine if improvements in treatment over the past 4 decades have increased survival enough to alter the essential message of Fig. 1—that only a minor fraction of patients with stage IV cancer derive any survival benefit from treatment.

Box 6 Estimating the benefit of a contemporary treatment by combining recent and older data
Full size table

A Candid Conversation

If a patient agrees to undergo chemotherapy without a basic understanding of treatment goals and efficacy, the minimum requirements for informed consent have not been satisfied.10 Since patients receiving chemotherapy for advanced cancer typically lack this understanding,2, 3, 10 they have most likely given uninformed consent. This may happen in spite of earnest efforts by the health care team to communicate risks and benefits of therapy to patients who cannot or do not want to hear the truth. While it is reasonable to draw attention to the ways in which clinicians, health educators, and researchers might contribute to a communication gap, it is also important to acknowledge the extraordinary complexity of the problem and fruitlessness of laying blame.

It might seem cruel to convey harsh facts about life expectancy and treatment benefit to patients with incurable cancer. While an insensitive presentation of the facts would, indeed, be cruel, clinicians who adhere to best practices for delivering bad news should be able to communicate compassionately and honestly.24 Pointing out that treatment is unlikely to help does not “take away all hope,” as some defenders of the stage IV shuffle might argue. A candid discussion can give the patient realistic hopes instead of false hopes that are apt to betray later on.25 A patient with stage IV breast cancer can reasonably hope to live 2 years or longer. Hoping for 10 years is less realistic: the chances are currently about 8%—not impossible, but unlikely.5 While it is not the oncologist’s job to challenge a patient who hopes to beat the odds, it is the clinician’s job to disclose the odds when a patient wants to know. A patient who has asked for a prognosis is entitled to the truth. A patient who has asked for a prognosis may prefer to see the statistics rather than to be left in the dark.

While cancer patients are entitled to the truth, they are also entitled not to have the truth forced upon them. Individuals vary in the information they want and the format that is most meaningful to them. A disarmingly straightforward and respectful approach to accommodating this variation26 gives patients no more information than they want and no less (Box 7).

Box 7 The “preview, ask, tell, ask” approach to communicating with patients
Full size table

Much has been written about the need to improve doctor-patient communication about end-of-life issues. Given the legal and ethical implications of starting a course of treatment when the patient lacks an understanding of key facts, maintaining the status quo is not an option. It is time to put the stage IV shuffle to rest.