Why Many Oncologists Fail to Share Accurate Prognoses: They Care Deeply for Their Patients (2024)

A majority of cancer patients want information about their disease and prognosis [1]. Yet, oncologists do not routinely share prognoses. In a study of nearly 600 patients with advanced cancer, only 17.6% of the 71% who wanted to know their prognosis reported being told [2]. Further, patients’ preferences for information about their disease can change over time, so one discussion at one time point will not suffice [3]. We applaud Al-Samkari for articulating this gap and highlighting what has become a persistent problem in oncology: lack of timely and repeated discussions of prognosis. Physicians are much more comfortable sharing information about active treatment options than they are delivering prognoses [1]. When ignorant of their prognosis, patients can hold unrealistic expectations and elect to participate in futile treatment plans that ultimately lower their quality of life [4]. On the other hand, patients who understand their prognosis are more likely to pursue more concordant treatment steps, and handle advance care planning such as living wills and do-not-resuscitate orders in a more timely manner [2].

In the effort to encourage physicians to appropriately share prognostic information with patients who want this information at each major junction in care - diagnosis, progression, remission, limited treatment options - we need to carefully identify the factors contributing to physician reluctance to share a patient’s prognosis. This reluctance is not at its core old fashioned paternalism. Though it is easy to blame this old bugaboo, much more is involved here. Studies have identified multiple factors: dread of discussing poor prognoses, not wishing to destroy hope, struggling to balance sensitivity and honesty, fear of damaging their relationship with the patient, determining whether to share this information with statistics or generalities, determining how to assess what the patient wants to know, and being sensitive to cultural and family issues [5]. Some have tried to address these concerns [6]. For example, it has been shown that hope remains steady despite patients having truthful discussions that may contain less optimistic information than expected. With this information, patients are more likely to feel educated and empowered in their decision-making process for further treatment steps [7]. Moreover, honest prognoses discussions have not been observed to lead to higher rates of depression or worry [8] and are associated with more frequent and earlier hospice enrollment, lowering rates of emergency interventions such as ventilation and resuscitation [9]. A recent study found that prognostic disclosures were not associated with any measurable harm to either the emotional well-being of the patient or their relationship with their physicians [2]. However, despite these encouraging studies, a majority of oncologists still find sharing prognosis, especially when it is “bad news” and when they care deeply for their patients, to be stressful [6] [10]. This source of the problem has been confirmed. A most interesting study found no single physician characteristic correlated with failure to provide accurate prognoses other than how long the physician has known and cared for the patient [11]. This study is so telling of the physician’s hope: rather than serving as cold, paternalistic parent figures, these oncologists grow close to their patients and care so deeply to cure them. We therefore agree with Al-Samkari’s senior colleague that, at its core, the failure to discuss prognosis is not primarily due to paternalism but is motivated by beneficence: physicians strive every minute to help their patients and may be subject to wishful thinking, which may in turn result in overestimated prognoses or avoiding the conversation altogether due to wanting to remain hopeful for their patients [12].

Given the complexities involved in sharing prognosis, we agree that adding evaluations might be helpful, but, and we are sure Al-Samkari would agree, oncologists need and deserve more. We suggest three ways to help: (1) communication techniques that can pave the way to this difficult discussion; (2) simple ways to assess the patient’s preferences for information; (3) tools to help with accuracy of prognosis.

We offer two helpful communication techniques to help the physician introduce the conversation about prognosis. The “I” step, obtaining the patients’ invitation, in the well-known SPIKES six-step protocol for delivering bad news [13] could be adapted to determine if the patient wishes to know his or her prognosis. Physicians could ask: “How would you like me to give you information about how your disease may develop and the possible outcome? Would you like me to give you all the information or briefly sketch out the how your disease is doing and spend more time discussing the treatment plan?” An advantage of using protocols like SPIKES is that their reflective nature may better equip physicians to handle their own distress of being the bearers of bad news [14].

Another simple communication technique suggested by Cortez et al. is to interrupt the normal flow of the discussion, which is to move directly from discussing lab and scan results to discussing the next available treatments. This discussion flow can be interrupted by asking a question after stating the results of the labs or scans: “would you like to discuss what this means?”[15] Asking this simple question opens the door to a discussion of prognosis, if the patient answers in the affirmative.

While the SPIKES six-step protocol and Cortez’s simple question encourage dialogue, simple patient preference tools, which consist of short questionnaires that the patient fills out and gives to the physician, can be helpful tools. Physicians have appreciated such tools in other contexts because they allowed the physician to quickly and easily assess patient preferences [16]. A 2011 study conducted by an end of life care research group asked nearly 200 patients how much they agreed with statements like “I want information such the chances of cure of the disease” and “I want information about my life expectancy with the disease [1].” Patients ranked their answers on a 6-point Likert scale (from totally disagree to totally agree). While the questionnaire was never given to the patients’ treating physicians, preference tools akin to this one could developed and tested. By explicitly determining patient preference either through guided conversation or patient preference forms, doctors can ensure they are giving the right amount of information to the right patients.

A second way of helping physicians share prognoses is to improve prognostic accuracy. In the palliative care setting, prognoses have been shown to be overoptimistic, with only 43% of the prognoses shared being accurate. This reality holds true for cancer patients, where there is rarely an agreement between official prognostic estimations and actual survival [17]. Christakis and Lamont noted a 20% accuracy rate in 365 different clinicians’ prognoses for over 500 hospice outpatients, with 63% of the total prognoses being overestimations[11]. These misjudgments can be responsible for poor clinical management, since undue optimism in terminal patients can result in the pursuit of fruitless treatments instead of opting for beneficial palliative care therefore voiding the possibility of quality of life improvement [11].

One way to improve prognostic accuracy is to identify which patient factors are most useful to consider when determining prognosis [18]. Downing et al. recently tested the validity of 15 either common or previously unvalidated prognostic factors to develop a tool applicable to palliative care patients, eventually named Prognostat. Only 5 of the 15 factors tested were significant in accurately estimating survival time in the multivariable model used, including clinician prediction of survival, Palliative Performance Scale value, primary illness, gender, and state of delirium. The tool for calculating Palliative Performance Scale itself has been utilized for well over a decade, and has been validated, updated, and incorporated into many practices [19]. Interestingly, several commonly used factors in other popular prognostic tools such as Charlson index, weight loss, and illness trajectory were all found to be insignificant factors. After being optimally formatted to fit into clinical practice, the final Prognostat tool incorporated all 5 significant prognostic factors and was found to have an accuracy of 69% over the clinicians’ accuracy of 42% [20]. Many prognostic tools have proven to be just as or more accurate than clinicians’ predictions alone, suggesting that prognostic tools should complement, not replace, physician estimation on prognoses [20]. As these important prognostic factors that improve accuracy will undoubtedly change throughout disease progression, prognostication of survival prediction should be treated as a process with regular update rather than a single event [12]. By utilizing these tools, physicians may be able to offer more accurate prognoses that empower their patients to make more realistic and better-informed decisions over the course of their disease, and, as some have found, have a better quality of life [9].

In sum, the problems of avoidance of discussing prognosis or overly optimistic prognoses are not due to paternalism or an attempt to rob patients of their autonomy. Oncologists are dedicated to curing and helping their patients, and this dedication makes sharing prognoses very difficult. We hope the two communication techniques and the two simple tools, one that lets patients state whether they want this information or not and one that can help with accuracy, will make this difficult conversation somewhat easier.

Acknowledgments

Funding Source: This research was supported by Winship Cancer Institute of Emory University and the NIH/NCI under award number P30CA138292. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Research was also supported by the Davidson College Impact Fellowship.

Footnotes

References