Abstract
For any decision in favor of performing an endoscopic procedure, its expected benefit should exceed its expected cost. Otherwise, one should not do the endoscopy. The choice between two options against, or in favor of, endoscopy is based on the underlying cost–benefit relationship. 1 Occasionally, situations arise when, besides these two options, the endoscopist faces yet a third option of “willful ignorance.” Rather than focusing on the potential benefit of endoscopy, the physician (or the patient) may, a priori, decide to remain ignorant of any of its potential outcomes. For instance, an abdominal computer tomography (CT) scan in an asymptomatic 85-year-old man with serious comorbid conditions reveals the incidental findings of a pancreatic head mass. Instead of subjecting the patient to further testing with endoscopic ultrasound and fine needle aspiration (EUS with FNA) or endoscopic retrograde cholangio-pancreatography (ERCP), the patient and his family decide against any additional diagnostic workup. Such decisions are relatively simple if the procedure is risky with few, if any, beneficial consequences. 2 The choice between competing options may become more complex and difficult to make if the procedural costs are less prohibitive and the diagnostic knowledge obtained through endoscopy comprises a mixture of beneficial, as well as harmful, aspects. The aim of the present article is to present an intuitive decision tool that would be applicable in clinical practice to resolve such diagnostic dilemmas without involving complex mathematical analysis.
A Simple Decision Tree with Complex Outcomes
A simple decision tree captures the main choices underlying the present decision analysis against or in favor of endoscopy (Figure 1a). The focal point of the analysis is the acquisition of beneficial or harmful knowledge through endoscopy compared to a state of blissful ignorance. The acquisition of knowledge is associated with costs. The two probability values p and q = 1 − p denote the expectation of acquiring beneficial or harmful knowledge, respectively, through endoscopy.
Decision tree of choices against, or in favor of, acquisition of knowledge through endoscopy. Costs are shown in red and benefits in blue. Costs are varied from 0 to −4 and benefits from 0 to +4. (a) General outline of the decision tree. (b) Asymptomatic 85-year-old man with serious comorbidities. (c) A 45-year-old man, who is adamant about his need to obtain a definitive diagnosis. (d) A 60-year-old man with possible autoimmune pancreatitis.
Costs, benefits, and harms associated with endoscopic knowledge or ignorance.
The term “benefit” is used to indicate all factors that could positively affect the decision in favor of acquisition of knowledge through endoscopy. Any endoscopy that reveals a treatable or curable disease is obviously associated with a beneficial therapeutic consequence. Even without immediate therapeutic consequence, a diagnosis can provide resolution of fears, comfort, and reassurance. In instances of inheritable disease, it can also help develop a plan for screening and surveillance that benefits the patient, as well as the patient’s offspring and relatives. On the one hand, knowledge about the limited life expectancy associated with an incurable or fatal disease is useful in that patients and their family can make plans for the immediate future, and maximize the limited time available to them. On the other hand, knowledge about one’s limited time to live can lead to psychological “harms”, such as despair and hopelessness. Burdened by their new knowledge, patients may forsake social interactions, or restrict activities that would still be within their means despite their underlying disease. A newly found diagnosis could change patients’ perceived health status and raise their insurance premiums. 3 It could also put a patient in a legal or moral bind to report his medical condition to his employer or social network. All these burdens could be avoided if a patient decided to remain in a state of “blissful ignorance”. Unburdened by diagnostic knowledge, patients could still perceive and act with absolute freedom of choice and action. They would be allowed to retain their hope for a happy ending.
How to Weigh Complex Outcomes
For acquisition of knowledge to be the preferred medical strategy, its benefit, minus costs and harms, needs to exceed the benefit of blissful ignorance:
The benefit of knowledge and the bliss of ignorance are counted as positive values, whereas the costs of endoscopy and the harm of knowledge are counted as negative values. The benefit and harm of knowledge are both weighted by their respective probabilities of occurrence, p and q. Rather than trying to assign absolute values to each of the items listed in Table 1, for each individual patient, one would try to assess their relative values by pairwise comparisons.
The example from above is illustrated in Figure 1b. The individual costs and benefits were ranked in magnitude between 0 and 4. The costs of interventional endoscopy in an elderly patient with serious comorbid conditions were considered as high (–4), and the benefits of knowing the correct diagnosis were considered as low (+1). The patient and his family were less concerned about the potential psychological harms of knowledge (–1) and the benefits (“bliss”) of ignorance (+1). If one assumes that p ≈ q ≈ 50%, then the benefit of ignorance of +1 always outweighs the cost of knowledge, because: 50%ċ1 – 50%ċ1 – 4 = –4. This outcome is largely independent of the particular values chosen for p and q.
Now consider a similar incidental CT finding in an otherwise healthy 45-year-old man with school-age children at home. The patient was absolutely adamant that he wanted diagnostic confirmation, even if the chances of subsequent surgical cure would be slim. In case of a dismal diagnosis, the patient also wanted his family to be psychologically prepared, and to start making financial arrangements for their wellbeing in case of his demise. From the onset, the patient was well aware of the risk of pancreatic cancer. A diagnostic confirmation would increase his pre-existing concerns only marginally. From the patient’s perspective, the benefit of knowledge by far exceeded its potential harms. All the costs of interventional endoscopy, including fear, inconvenience and possible adverse events, also appeared negligible compared with the benefit of knowing the truth (Figure 1c). For all values of p > 50%, the benefit of the upper branch is greater than +1, and exceeds the benefit of the lower branch.
The third example deals with yet another case of a 60-year-old man in whom, in addition to the pancreatic head mass on CT scan, serologic testing revealed elevated IgG4 levels suggestive of autoimmune pancreatitis. The patient saw for himself little benefit in ruling out a lesser possibility of pancreatic cancer. He was also very fearful of endoscopy and its impact on his busy professional schedule. He thought that he would become extremely upset by a positive diagnosis of pancreatic cancer and that he would rather maintain a positive attitude and wait for medical treatment with prednisolone to take effect. To him, the costs and the harms of endoscopic workup appeared to overwhelm the marginal benefit of diagnostic ascertainment. Rather than undergo an EUS with FNA for diagnostic confirmation, the patient decided to wait for the outcome of medical therapy. This scenario is depicted in Figure 1d. Again, assuming p ≈ q, a benefit of +4 associated with the lower branch would by far outweigh the cost (–5) of the upper branch. A similar result applies to all probability combinations of p and q.
In the three clinical scenarios from above, the cost and benefits of the endoscopic intervention and its outcome were assigned numerical values. This was done primarily to illustrate the thought process underlying the decision making. In the clinical setting, however, the physician may be able to weigh the different costs and benefits based on their perceived order of magnitude alone without even resorting to any numeric evaluation. Apart from a pancreatic head mass, other clinical examples abound. For instance, fearing the consequences of a positive diagnosis, patients may forgo surveillance endoscopy for Barrett’s esophagus or colon polyps.
Willful Ignorance as State of Mind
In general, any type of knowledge is considered a worthwhile commodity that should be pursued even if the acquired knowledge becomes initially painful, because in the long run its overall benefits will surpass its downsides. Such attitude towards knowledge has been especially advocated for any type of scientific knowledge, which should be embraced despite its potential risk for future unintended consequences. 4 Occasionally, one encounters situations where, besides two simple yes–no alternatives, a seemingly paradoxical third option presents itself of declining a definitive answer. The ensuing ambiguity allows one to retain hope and freedom of action, which may be valued higher than any decisive knowledge with its inherent implications. Not surprisingly, the psychological aspects of willful ignorance have drawn the attention of psychologists. 5 In their study of knowledge (epistemology), philosophers have also catalogued a wide variety of circumstances where it would be better not to know, many of which pertain to medical conundrums.6,7 On one hand side, geneticists have stressed the ethical issues involved in patients with heritable neuromuscular disease and their right to know, and not to know, the results of genetic testing.8,9 On the other hand, oncologists have advocated that physicians should always seek and tell the truth, even when the patient would rather not know it. 10
The present analysis is focused on the occurrence of willful ignorance in the realm of gastrointestinal endoscopy. Considering the multitude of possible scenarios, there is no clear-cut rule that could easily fit most conceivable situations. The personal and medical characteristics of individual patients encompass too large a variety. Different patients will assign different values to the same outcomes. Although the decision tree underlying medical decision making is deceptively simple, each decision can result in multiple, and complex, outcomes that are difficult to quantify. The present algorithm, nevertheless, may provide the endoscopist with some guidance on how to support patients in their decision making. In helping patients to make the best decision fitting their special needs and wishes, the first step is to consider all potential outcomes involved in the decision process as listed in Table 1. In a second step, the negative influence of harms needs to be balanced against the positive influence of various medical and psychological benefits. Finally, a sequential pairwise comparison of different outcomes will then help to sort the outcomes according to their relative magnitude. None of these steps involves any elaborate mathematics, and, in most instances, a solution can be found based on a small set of two or three thoughtful comparisons, or “common sense” alone. The analysis serves as reminder to consider, prior to endoscopy, all its potential “subjective” (emotional) and “objective” (financial, legal) consequences, including a patient’s desire to know or not to know.
