• quality of life
• prisoner’s dilemma
• end of life chemotherapy
• end of life care

Author RSH (a palliative medicine physician) was asked to see a patient with a newly diagnosed metastatic cholangiocarcinoma and Eastern Cooperative Oncology Group (ECOG) performance status grade 3. The oncologist prefaced the consultation saying ‘[The patient] has no good chemo(therapy) options. I think he needs hospice.’ After consultation, RSH and the patient agreed on home hospice. The oncologist saw the patient later that day and told RSH, ‘Well, I’ve decided to give him chemo, because if I don’t give chemo, then someone else will.’

Great advances in the prevention, detection and treatment of cancer continue to improve survival. However, despite much attention focused on end-of-life chemotherapy, many patients continue to receive chemotherapy late into their illness, providing minimal benefit and potentially causing harm. In this article, we describe the prisoner’s dilemma, in the context of palliative oncology, as a framework to partially explain this finding. We believe that this novel perspective will provide valuable and interesting insights to practising clinicians.

Almost 10% of patients receive chemotherapy within 2 weeks of dying of cancer, and 60% receive treatment within 2 months of death.1 The majority of these patients mistakenly believe that this type of chemotherapy is curative rather than palliative.2 In fact, end-of-life chemotherapy has limited benefits and is associated with worse quality of life, more emergency and intensive care interventions and greater financial cost to patients and families. Consequently, the American Society for Clinical Oncology identified reducing end-of-life chemotherapy as one of the ‘top five’ practices for improving care and cutting costs. Despite these recommendations, large number of patients continue to receive aggressive treatment within their last few days of life.3

Why do patients continue to receive end-of-life chemotherapy? We believe that ‘inappropriate’ chemotherapy at end-of-life does not result from poor decision-making by oncologists. Instead, it is partially driven by an incentive structure leading to a phenomenon known as the prisoner’s dilemma. The prisoner’s dilemma, is a two-person interaction model involving cooperation or competition. Two hypothetical prisoners, Scott and Bill, are isolated in separate jail cells such that neither knows what the other is thinking. Each prisoner was offered a plea bargain: testify against his partner for a sentence of 1 year while his partner would receive a sentence of 10 years. If both choose to testify against each other, they would each go to prison for 5 years. If both keep quiet, then there would be no evidence with which to convict and both would go free (figure 1A). Clearly, Scott and Bill would do best by keeping quiet. However, because they cannot communicate with each other, testifying is the most rational option. Scott is thinking: (1) if I testify first and Bill keeps quiet, then I will get a lighter sentence and Bill will get the higher sentence, and (2) if Bill testifies, I’m better off testifying as well since I will get the 5-year sentence rather than the 10-year sentence. Bill will reach the same rational conclusion and testify as well. The prisoner’s dilemma is a paradox that forces the individuals away from the best collective outcome; by pursuing the best individual strategy, Scott and Bill will reach a worse collective outcome.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Classic prisoner’s dilemma depicting the paradox leading to the worst outcome. End-of-life chemotherapy shown as a prisoner’s dilemma often results to worse clinical and societal outcomes.

Although typically used to model self-interested choices between two players, the prisoner’s dilemma, when extrapolated over multiple players, can also show how these choices will lead to a worse societal outcome. This has been demonstrated in various medical contexts, including the prescription of antibiotics, the administration of vaccinations and the placement of percutaneous endoscopic gastrostomy (PEG) tubes. The detrimental impact of the prisoner’s dilemma on society is best demonstrated in the treatment of pain with opiates. For example, a patient may have real or fake pain and seek treatment. If he has real pain, the rational choice for the doctor is to treat him. If he has fake pain, it is still in the doctor’s best interest to treat. Otherwise, the doctor may have to deal with a dissatisfied patient. Given that evaluations of patients’ satisfaction are becoming more important than ever, concerns about poor ratings from dissatisfied patients can, in turn, lead to greater job-related distress and worse clinical care, including the negative societal outcome of perpetuation of inappropriate opiate usage.4

We believe an analogous prisoner’s dilemma is occurring with the usage of end-of-life chemotherapy. For example, a patient with incurable cancer may select to either accept or decline treatment when offered chemotherapy or no chemotherapy by an oncologist. The collective interest, in this case, is for oncologists to not offer and for patients to not accept end-of-life chemotherapy for incurable cancers due to small health benefits and major toxicity (figure 1B). However, as we know, end-of-life chemotherapy continues to occur, an outcome against society’s interest that leads to wasted resources and poor clinical outcomes.

There are several factors that contribute to this prisoner’s dilemma. Patient-related factors can be summarised as the expectation for treatment, the fear of death and the wish to maintain control and hope. Most patients often expect treatment rather than watchful waiting, and they are willing to pursue chemotherapy for small benefits with major toxicity. Accepting chemotherapy near end of life preserves false hope and reduces anxiety. Patients are reluctant to give up hope without trying all available options. Furthermore, they may view the transition from curative to palliative therapy as implying impending death, triggering fears of abandonment.5 Driven by dissatisfaction and anxiety, patients would seek a second opinion, hoping for a different recommendation and fulfilment of their unmet needs.

Oncologist-related factors are driven by the difficult discussion with patients, their own emotional distress, the patient’s preference for chemotherapy and other external factors. Oncologists often have difficulty discussing end-of-life options with patients; they feel a sense of failure, disappointment and guilt if they do not offer their patients any treatment. Such emotional distress makes it uncomfortable for an oncologist to refuse chemotherapy to patients near end of life. Oncologists, therefore, may continue to offer chemotherapy and forego end-of-life discussions in order to maintain the patient’s hope and satisfaction. In addition, an oncologist’s clinical decision is heavily influenced by the patient’s expectation and preference for end-of-life chemotherapy. As a result, patients with falsely optimistic beliefs about chemotherapy rate communication with the physicians more favourably. Another contributing factor is the inherent cognitive bias that favours offering chemotherapy when oncologists are faced with clinical unpredictability. When clinical factors do not give definitive information, oncologists tend to better recall successful cases that defied statistical expectations, and would offer treatment based on the notion that individual prognosis is difficult to predict and ‘you can never know’ how patients would respond.6 Finally, oncologists have a financial incentive to offer more chemotherapy; oncologists paid on a fee-for-service basis have a higher income than salaried oncologists.7

In short, the decision about chemotherapy is difficult for both patients and physicians. Because many patients expect treatment, oncologists will offer end-of-life chemotherapy to respect the patient’s wishes and to palliate emotional distresses. Patients will likely accept chemotherapy to maintain hope and control. Based on the factors previously mentioned, the default scenario is for the oncologist to offer and the patient to accept end-of-life chemotherapy, despite the negative outcomes. Each pursuing their own most rational option, both the oncologist and the patient will reach an outcome that is against the societal interest, thus establishing a prisoner’s dilemma.

Changing the incentives in this prisoner’s dilemma that realign individual and societal interests must involve oncologists, patients and payers. Oncologists need better guidance when determining which patients can benefit from chemotherapy and which patients will not. Second, more education is necessary to help practising oncologists improve their end-of-life conversations. Third, patients are often unaware of the extent of their disease and the minimal benefit of end-of-life chemotherapy. Better patient–oncologist communication about the risks/benefits of end-of-life chemotherapy therefore needs to be promoted. The routine use of simple interventions, such as question prompt sheets (ie, a structured list of questions), can encourage patient participation without negatively impacting patient satisfaction, consultation length or physician workflow.8 Fourth, early involvement of palliative medicine clinicians in the care of patients with advanced cancer must become the standard of care. Early integration of palliative medicine into routine oncological care is associated with improved patient outcomes, yet patients are still referred to palliative care in low numbers and late in their illness.9 Finally, the above will fail without significant financial reform. Financial incentives need to be restructured such that oncologists are rewarded for providing quality end-of-life care without encouraging the use of unnecessary and even harmful chemotherapy. Pilot studies have demonstrated that significant cost savings can coincide with the provision of high quality oncology care.10

The decision-making process for patients with advanced cancer illness is difficult for both patients and physicians, and can be described as a prisoner’s dilemma. As discussed, offering end-of-life chemotherapy to patients is the most rational choice for oncologists despite the negative repercussions for their patients and society. Significant reform using a multifocused approach will be needed to counter the use of aggressive end-of-life treatment and the rising cost in cancer care.