Clinical Gastroenterology and Hepatology
Volume 8, Issue 3 , Pages 233-234, March 2010

Risk and Reason in Barrett's Esophagus

published online 18 December 2009.

Article Outline

 

Podcast interview: www.gastro.org/cghpodcast.

Most gastroenterologists have had the experience: The recently endoscoped patient meets with you for the postendoscopy debriefing. It is your duty to inform the patient of the presence of Barrett's esophagus (BE), a change in the lining of the esophagus that is associated with an increased risk of esophageal cancer. “But the good news is, the risk of cancer is low,” you report in an encouraging tone. The patient leaves with refills of the proton pump inhibitor prescription and a plan for follow-up endoscopy.

Although the conversation likely is similar in most doctors' offices, we do not usually stop to wonder what the patient really heard. How “low” is “low”? And what can I do to get it lower? Am I going to die of cancer?

Back to Article Outline

What Is the Cancer Risk in Barrett's Esophagus? 

In this issue of Clinical Gastroenterology and Hepatology, Sikkema et al1 report a meta-analysis of cancer risk in BE. This idea represents well-trod ground, being at least the fifth such attempt at this question in the past decade.2, 3, 4, 5 The reported cancer risk, about 6 per 1000 patient-years of follow-up evaluation, is remarkably similar to previous estimates.

In addition to incorporating the latest available data and being well performed, this meta-analysis is notable for 2 other reasons. First, the study gives the most complete estimate to date of the rate of progression to the combined end point of high-grade dysplasia (HGD) and cancer. Although such data were of less utility previously, HGD is increasingly becoming an actionable diagnosis. Given the increasing availability of relatively well-tolerated endoscopic ablative therapy, as well as the presence of 2 studies in the literature showing a decreased rate of progression to adenocarcinoma in subjects with HGD treated with ablative therapy,6, 7 progression to the combined end point of HGD or superficial cancer often precipitates the same intervention, endoscopic ablative therapy. Because the contraindications to endoscopic ablative therapy are much less restrictive than those for esophagectomy (previously our only other therapeutic option), more patients are eligible for intervention. Therefore, an estimate of this combined end point is important from the health services perspective because this population will require adequately trained and equipped teams to intervene.

A second notable feature of the current article is the estimate of esophageal cancer–related mortality. As might be expected from individual studies reporting this outcome, as well as clinical experience, this rate is low in the general BE population, at 3 per 1000 patient-years, and easily is outstripped by other causes of mortality, especially cardiovascular disease. Those seeking to use these data to make the argument against endoscopic surveillance in BE should do so with caution; after all, the low cancer death rate may be a product of the surveillance and not the benign natural history of the disease. Without a comparator group not undergoing surveillance (which is not available in this study), the impact of surveillance is unclear.

But perhaps more interesting than any given estimate from the article is the bigger question: what is the clinician to do with these data? How should we explain these risks to our patients, and how should they impact our practice?

Back to Article Outline

Patient Perceptions of Cancer Risk 

Patients are demonstrably poor at understanding risk in general. This is especially true of cancer risk in BE. Part of this inability to place risk in the appropriate context may be owing to suboptimal numeracy skills—many people do not really know what a percentage or proportion is, and translating these data into information consumable by our patients is a difficult task. However, part of the overestimation of cancer risk seen in subjects with BE also likely is secondary to the human tendency to overestimate the risk of extreme or catastrophic outcomes. Such tendencies hardly are limited to this disease or to our field. People commonly refuse vaccines over the concern of neurologic impairment or anaphylaxis, when the incidence of these conditions is dwarfed by the risk of morbidity or mortality from the infectious disease to be prevented. In gastroenterology, patients with erosive esophagitis are willing to forego the healing properties of PPI therapy because of the concern of hip fracture or vitamin deficiency, despite evidence showing that the increase in absolute risk of these events is very small or increased inconsistently.8 Indeed, relatively few people make decisions rationally, if, by rationally we mean choosing the outcome that is most likely, on average, to give the best outcome. It appears from empiric evidence that patients will sacrifice gains to avoid what they view as a tragic loss (ie, people fear losses more than they covet gains of a similar magnitude). For instance, people fear a loss of $500 more than they value a gain of $500. The overweighting of rare events and risk aversion are cornerstones of Prospect Theory, a hypothesis seeking to explain human risk-taking behavior. An entire science has evolved around the study of the irrational aspects of human decision-making behavior, which is fascinating to read. The work of the pioneers of this field, Daniel Kahneman and Amos Tversky,9 resulted in the Nobel Prize for Kahneman in 2002.

Although it is tempting to ascribe the irrational behavior in health care use to patient misunderstandings and human nature, to do so minimizes the physicians' role in the interaction. Simply put, many physicians do not convey risk data to the patient in terms that the patient can understand.

Back to Article Outline

Setting the Numbers in Context: Discussing Risk 

Consider the following 2 interactions.

Physician 1: You have Barrett's esophagus. Having this condition increases your risk of esophageal cancer by about 100 times, or 10,000%, compared with someone who does not have it. For that reason, we often perform an endoscopy every few years and take biopsy specimens to check on it. We do not really know if this maneuver helps. Would you like us to do it for you?

Patient 1: Where do I sign up?

Compared with the following interaction.

Physician 2: You have Barrett's esophagus. Having this condition increases your risk of esophageal cancer from less than 1 in a thousand every year to 5 in a thousand every year, compared with someone who does not have it. For that reason, we often perform endoscopy every few years and take biopsy specimens to check on it. We do not really know if this maneuver helps. Would you like us to do it for you?

Patient 2: Uh, let me get back to you.

These 2 vignettes illustrate the differential impact of relative versus absolute descriptions of risk for events of low likelihood. Although descriptions of relative risk are useful in assessing the strength of a relationship between a potential causative agent and an outcome (all other things being equal, we are more likely to believe an association with a relative risk of 100 than one of 1.5), they are of secondary importance in describing risk to the patient. Here, absolute measures of risk, such as the risk difference (the absolute risk of the outcome in the presence of the risk factor minus the absolute risk of the outcome in the absence of the risk factor) are most useful. Describing risk in this way, as in the second vignette, gives the patient a better idea of what the putative risk factor (or harmful health behavior) might really mean to him or her.

One useful, but underused, way of encapsulating absolute risk is with the number needed to treat, or the number needed to harm. This number, simply one divided by the risk difference, is easy for patients to understand and helps place the value of an intervention in context. For instance, if the lifetime risk of cancer death in the setting of nondysplastic BE is 3%, and endoscopic surveillance is perfectly effective in averting this outcome, the best the number needed to treat can be is 1/(0.03–0), or 33.3. That is to say that for every person who benefits from endoscopic surveillance, 32 will not. Some patients may look at such a number and consider it prohibitive. Others may wish the risk reduction it entails. Either way, this more quantitative approach to shared decision making gives a better framework for the patient to apply his or her personal values to the decision.

Applying this to the decision of endoscopic surveillance in BE, an initial discussion should include both the putative advantages of surveillance, placed in absolute terms, as well as risks and costs. Although a credible estimate of the risk reduction associated with surveillance is not available in the literature, we can frame it in absolute terms in a manner similar to that described earlier, by explaining the risk if the program is 100% effective, 50% effective, and so forth. For instance, if the risk of progression to adenocarcinoma in a nondysplastic BE patient is 0.5% per patient-year, and surveillance is 100% effective, the patient's annual risk decreases from 5 per 1000 to 0 per 1000. If it is 50% effective, the risk is 2.5 per 1000, and so forth. With respect to risks and costs, the risks of endoscopy, the considerable risk of misinterpretation of histology leading to inappropriate intervention, and indirect and direct costs of the procedures, should be considered. Although some patients wish to avoid such discussions, many will use it as a springboard for frank conversation about their condition.

Back to Article Outline

The Demise of Paternalism, and Understanding Risk 

As patients move toward a more active role in their health care decision making, it is likely that our ability to describe and quantify risk will become even more important. “Whatever you say, doctor,” is a much less common answer among subjects in their 30s than in older generations. Couple that with the ready access to medical information found on the Internet (of varying quality and accuracy), and one can imagine hundreds of future patient interactions, piecing through this information in an attempt to arrive at answers to complex decisions for the care of patients with conditions such as BE. Although this work sometimes can be frustrating, and often is time consuming, the pay-off, a patient-centered decision that is informed by the facts and best reflects the patient's values and concerns, is worth it, and will be demanded increasingly by our patients.

Back to Article Outline

Podcast 

Back to Article Outline

References 

  1. Sikkema M, de Jong PJF, Steyerberg EW, et al. Risk of esophageal adenocarcinoma and mortality in patients with Barrett's esophagus: a systematic review and meta-analysis. Clin Gastroenterol Hepatol. 2010;8:235–244
  2. Shaheen NJ, Crosby MA, Bozymski EM, et al. Is there publication bias in the reporting of cancer risk in Barrett's esophagus?. Gastroenterology. 2000;119:333–338
  3. Yousef F, Cardwell C, Cantwell MM, et al. The incidence of esophageal cancer and high-grade dysplasia in Barrett's esophagus: a systematic review and meta-analysis. Am J Epidemiol. 2008;168:237–249
  4. Thomas T, Abrams KR, de Caestecker JS, et al. Meta analysis: cancer risk in Barrett's oesophagus. Aliment Pharmacol Ther. 2007;26:1465–1477
  5. Chang EY, Morris CD, Seltman AK, et al. The effect of antireflux surgery on esophageal carcinogenesis in patients with Barrett esophagus: a systematic review. Ann Surg. 2007;246:11–21
  6. Overholt BF, Lightdale CJ, Wang KK, et al. Photodynamic therapy with porfimer sodium for ablation of high-grade dysplasia in Barrett's esophagus: international, partially blinded, randomized phase III trial. Gastrointest Endosc. 2005;62:488–498
  7. Shaheen NJ, Sharma P, Overholt BF, et al. Radiofrequency ablation in Barrett's esophagus with dysplasia. N Engl J Med. 2009;360:2277–2288
  8. Ali T, Roberts DN, Tierney WM. Long-term safety concerns with proton pump inhibitors. Am J Med. 2009;122:896–903
  9. Kahneman D, Tversky A. Prospect theory: an analysis of decision under risk. Econometrica. 1979;XLVII:263–291

 Conflicts of interest The author discloses no conflicts.

 Funding Dr Shaheen receives research funding from CSA Medical, BÂRRX Medical, AstraZeneca, Takeda Pharmaceuticals, and Oncoscope; and is a consultant for CSA Medical, AstraZeneca, and Oncoscope.

PII: S1542-3565(09)01303-2

doi:10.1016/j.cgh.2009.12.007

Clinical Gastroenterology and Hepatology
Volume 8, Issue 3 , Pages 233-234, March 2010