SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Case scenario 1: hepatic abscess
  5. Case scenario 2: Schatzki's ring
  6. Patient vs. physician utility
  7. Discussion
  8. References

Aims : To illustrate the characteristics of situations in gastroenterology when patients and physicians harbour different perspectives of medical costs and benefits, and how such different perspectives affect the outcome of medical decision-making.

Methods : Two exemplary scenarios are presented, in which threshold analysis yields different results depending on the varying values assigned to identical medical events. The occurrence of varying values is subsequently phrased in economical terms of varying utility functions that characterize patient vs. physician behaviour.

Results : Safety and therapy are the two major preferences that determine patient and physician utility functions. Patients and physicians make medical decisions based on two different utility functions. In comparison with their patients, gastroenterologists are more concerned with safety and inclined to spend more health care resources on safety than therapy because safety and the occurrence of medical complications affect their own professional status. In trying to maximize their own utility, gastroenterologists tend to spend more resources on safety than the patient him/herself might have spent given a free choice of management options.

Conclusions : In instances of potential complications associated with risky medical interventions, patients may receive less medical therapy in exchange for more procedural safety.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Case scenario 1: hepatic abscess
  5. Case scenario 2: Schatzki's ring
  6. Patient vs. physician utility
  7. Discussion
  8. References

Management of patients with gastrointestinal disease frequently involves the use of invasive procedures. The ultimate power about all medical decisions and the execution of various diagnostic or therapeutic procedures rests with the patient. The traditional role of the gastroenterologist is to present the patient with the diagnostic and therapeutic options that will help the patient to decide on the best approach to manage the disease. Ideally, the physician does not harbour or is able to suppress any financial or personal gains in favour of the patient's very own best interests. Occasionally patient and physician disagree on the best type of medical management to treat the patient's ailment. Even if all the facts are being explained to the patient in a coherent and easily understandable fashion both parties seem to come to different conclusions based on the same set of facts. The present article is aimed at understanding the factors that contribute to the occurrence of such decisional discrepancies. Decision analysis has been developed to utilize existing medical knowledge in the most efficacious way and to derive medical choices that maximize the benefits and minimize the costs to patient and society.1–3 Decision analysis formalizes the decision process, highlights the factors that influence the decision, and applies mathematical rigour to quantify decision-making. By phrasing a controversial medical situation in a quantitative fashion, mechanisms that underlie a discordant decision become more readily revealed.4–6 The present article illustrates the characteristics of situations when patient and physician harbour a different perspective of medical costs and benefits and how such different perspectives affect the outcome of a decision process related to gastrointestinal disease. The exemplary scenarios of two actual patients are presented, in whom decision analysis yielded different results depending on the varying values assigned to identical medical events. The occurrence of varying values is subsequently phrased in economical terms of two different utility functions that characterize patient vs. physician behaviour.

Case scenario 1: hepatic abscess

  1. Top of page
  2. Summary
  3. Introduction
  4. Case scenario 1: hepatic abscess
  5. Case scenario 2: Schatzki's ring
  6. Patient vs. physician utility
  7. Discussion
  8. References

A 66-year-old man was admitted to the hospital for further evaluation and care of a large mass in the posterior right lobe of his liver, measuring 7 × 9 cm in diameter. The patient had noted decreased energy levels, night sweats, hiccups and weight loss. A computed tomography (CT)-guided aspiration returned whitish purulent material mixed with blood. Cultures from the aspirate, as well as multiple blood cultures, failed to identify any specific organisms. Serological tests for Entamoeba histolytica and Echinococcus were negative. The CT scan revealed a normal appearing biliary system without evidence of gallstones or other structural abnormalities. A negative barium enema ruled out any potential colonic source of infection.7 A percutaneous drain placed by interventional radiology produced initially 10 ml per day. One week after he was started on i.v. ceftriaxone and oral metronidazole his general symptoms markedly improved, and the hepatic drainage dropped to <4 ml per day. The drain was pulled and the patient was ready to be discharged from the hospital to continue outpatient therapy with oral antibiotics for additional 3 weeks. One day prior to discharge, a two-dimensional transthoracic echocardiogram was performed to rule out endocarditis as yet another potential aetiology for the liver abscess. The mitral valve leaflets appeared thickened, but opened well. The anterior mitral valve leaflet showed diffuse thickening extending from mid leaflet to tip, but no mass suggestive of vegetation was found. A similar finding had already been described on an echocardiogram 3 years before. The medical consults from cardiology and infectious disease stated that endocarditis could not be excluded with certainty and that the patient might, therefore, benefit from a 6-week course of i.v. antibiotics.8 The patient did not meet the requirements for an outpatient i.v. treatment, but he also refused to remain hospitalized for another 6 weeks. He felt that the low probability of endocarditis did not warrant another 6 weeks of inpatient treatment with i.v. antibiotics.

Figure 1 depicts the decision tree of whether to treat the patient with antibiotics or discharge him home.9 Six-week treatment with i.v. antibiotics is associated with costs −A, such as prolonged stay in the hospital. In the present context, the term ‘cost’ is meant to reflect not only monetary expenditures but also time, pain or discomfort, and adverse medical events. Conversely, discharging the patient home is associated with the benefit of saving all these costs of antibiotic therapy (−A = 0). The two decisions in favour of or against antibiotic therapy are followed by the probability P for the presence of endocarditis or 1 − P for its absence. In case of endocarditis, antibiotic therapy would prevent future complications and costs of endocarditis (−E = 0). Discharging the patient without antibiotic therapy would entail the risk of future flare-ups and new complications, as indicated by the cost value −E.10 All costs are counted as negative numbers. If the patient harbours no risk for endocarditis, antibiotic therapy confers no benefit nor does withholding it result in any costs, that is E = 0. For the treatment to represent the preferred strategy, the benefit of the upper branch needs to exceed the benefit of the lower branch: −A > −PE or P > A/E. In other words, the threshold probability of endocarditis needs to exceed the cost ratio of antibiotic therapy to endocarditis for prolonged antibiotic therapy to represent the preferred (more beneficial) management option. It was explained to the patient that the probability of endocarditis did not exceed 10%.

image

Figure 1. Decision tree regarding treatment of possible endocarditis in a patient with liver abscess. −A and −E represent the costs of antibiotic therapy and complications of endocarditis, respectively; P represents the probability of endocarditis.

Download figure to PowerPoint

The patient understood the medical risks associated with an underlying endocarditis, but felt that in case of recurrent symptoms or complications he could still easily return to the hospital and undergo antibiotic treatment then. In his assessment, the presence of endocarditis and its potential complications did not amount to more than two or, possibly under worst conditions, three additional hospitalizations (−E, two or three hospitalizations) as opposed to one prolonged hospitalization needed for i.v. administration of antibiotics (−A, 1 hospitalization). Without stating this explicitly, his ‘calculation’ yielded a cost ratio ranging between A/E = 33% and 50% that well exceeded the threshold value of P = 10%. In contradistinction with the patient's own assessment, his physicians felt that the cost of antibiotic therapy paled in comparison with the costs of potential complications and that A/E ≪ 10%. The patient decided to leave the hospital nevertheless, and he has remained asymptomatic and disease-free for the past 16 months.

Case scenario 2: Schatzki's ring

  1. Top of page
  2. Summary
  3. Introduction
  4. Case scenario 1: hepatic abscess
  5. Case scenario 2: Schatzki's ring
  6. Patient vs. physician utility
  7. Discussion
  8. References

A 45-year-old male patient complained about dysphagia that occurred about once per 6 months. The dysphagia occurred only with solid food and mostly with meat. The patient had experienced these symptoms of mild dysphagia for more than 6 years. He was moderately obese and he denied any recent weight loss. During endoscopy, the oesophageal mucosa showed normal whitish-pink mucosa without signs of inflammation and a sharp Z-line at the gastro-oesophageal junction. No stricture was noted and the endoscope could be easily passed into the stomach. Upon prolonged inspection of the gastro-oesophageal junction, the occasional formation of a subtle ring-like structure was noted. The ring would not remain constant and frequently efface with oesophageal motility. The physician wondered whether the oesophagus should be dilated with a Maloney or Savary dilator.

Figure 2 depicts the decision tree of whether to dilate the oesophageal ring or not. It has been recommended to burst such rings with a single dilation using a bougie of 20 mm diameter.11, 12 Again, costs are counted as negative and benefits as positive numbers. The decision in favour of dilation is associated with costs (−C), whereas the decision against dilation results in cost savings (−C = 0). In addition to procedural expenses, these costs also include patient discomfort, pain and potential complications of oesophageal dilation, such as bleeding or perforation.12, 13 If the patient's symptoms are truly related to the ring-like stricture and respond to successful dilation (with a probability of P), the patient will subsequently experience the benefit (−B = 0) of improved deglutition. There is a (1 − P) probability that the patient's occasional dysphagia is totally unrelated to any oesophageal narrowing and that dilation provides no benefit whatsoever (−B = 0). If the physician withholds dilation in a patient who could have benefited from dilation, as shown by the lower branch, the option to improve deglutition is forgone (−B) at the benefit of avoiding the costs of dilation including any major complication (−C = 0). For dilation to represent the preferred strategy, the outcome of the upper branch needs to exceed the outcome of the lower branch: −C > −PB. This equation can be solved to yield a threshold probability of P > C/B. In other words, the threshold probability of successful dilation needs to exceed the ratio of dilation cost over dilation benefit. The medical literature is ambiguous with respect to the outcome of empirical dilation in the absence of a clear-cut oesophageal stricture. Several authors have reported a benefit associated with empirical dilation; however, a larger randomized clinical trial failed to show any significant benefit.14–16 Therefore, the a priori chances of improved vs. unchanged deglutition after oesophageal dilation are estimated as a toss-up of 50:50.

image

Figure 2. Decision tree regarding dilation of a Schatkzi's ring in a patient with mild dysphagia. −B and −C and represent the costs of unimproved deglutition and the costs of dilation, respectively; P represents the probability that the patient's symptoms would be resolved by oesophageal dilation.

Download figure to PowerPoint

Without going into a detailed calculation of cost and benefit, one can try to estimate their respective magnitudes on a Likert scale or rank them between 0 and 10. The patient, for instance, fully realized the implications of any potential perforation or bleeding, but even in the light of complications the costs of dilation still appeared to him much smaller (C = 2) than the perceived benefit of improved swallowing (B = 8): P = 2/8 = 25%. As 50% chance of successful dilation surpasses this threshold, from the patient's perspective dilation appeared a worthwhile procedure.13, 14 The physician had past experience with perforation secondary to oesophageal dilation, and he had gone through considerable amounts of administrative and legal hassle associated with such an adverse event. To the physician, on the one hand, the overall costs of an empirical dilation appeared higher than to the patient (C = 4). On the other hand, the potential benefit of dilation in a moderately obese patient without obvious stricture appeared somewhat lower (B = 6). Therefore, P > C/B = 4/6 = 67%. As this probability value exceeded the a priori chances of symptom improvement, the physician decided against the procedure.

Patient vs. physician utility

  1. Top of page
  2. Summary
  3. Introduction
  4. Case scenario 1: hepatic abscess
  5. Case scenario 2: Schatzki's ring
  6. Patient vs. physician utility
  7. Discussion
  8. References

The previous two examples illustrate that although patients and physicians may have a clear understanding of the underlying decision process and although they may agree on the issues at stake, they may end up with different decisions. This discrepancy arises because they assign different values to the same types of outcome. In essence, the decisions made by the patient or the physician both concern achieving the best balance between safety and therapy. In the present context ‘safety’ will serve as short form for avoidance of medical complications or side-effects and for reducing the probability of therapeutic risks to the lowest possible level. The short form ‘therapy’ is used to indicate therapeutic efficacy or successful therapy that will allow the physician to alleviate the patient's complaints. Safety and therapy represent the two preferences that determine patient and physician utility. In economic terms, utility functions serve to describe decision behaviour. If one combination of safety (S) and therapy (T) is chosen preferentially over another one, then the preferred combination is said to have a higher utility value.17 The Cobb-Douglas function represents a commonly used utility function of economics:

  • image(1)

where a and b are positive constants characterizing the utility function. The utility function determines the shape of indifference curves as shown in Figure 3. Each utility value is associated with a different indifference curve. The indifference curves describe in graphical terms how one preference, for instance safety, can be exchanged in favour of a second preference, for instance therapy, without changing the overall utility value. The mathematical concept of ‘marginal rate of substitution’ describes to what extent a patient (or physician) would be willing to substitute safety for therapy without changing the overall utility value.17

image

Figure 3. The straight line represents the budgetary constraints that apply identically to patient and physician. The upper and lower indifference curves represent the patient and physician utility function, respectively. The optimal utility is achieved at the two points where the two indifference curves become tangential to the budget line.

Download figure to PowerPoint

Safety and therapy are both associated with costs, and the total amount of health expenditures that can be spent on both is limited by the available budget (b). If the prices of therapy (v) and safety (w) are given, the budgetary constraint can be phrased as:

  • image(2)

A common problem of microeconomics relates to maximizing the utility function (u) given the budgetary constraint (b). This problem can be solved, for instance, by using equation (2) to express S as function of T. The substitution of S by T in equation (1) changes the utility function into a function of a single variable T. The maximum value of T is then found by differentiating the utility function with respect to T and setting the result equal to zero.17

Figure 3 illustrates the variation in decision-making between the patient and the physician in terms of varying utility functions. The straight line represents a budget line with v = w = 1. The patient and the physician face the same budgetary constraints. The upper curve represents the patient utility function with a = b = 0.5. For the physician utility function, constants such as a = 0.8 and b = 0.2 were chosen. Whereas the patient places equal emphasis on safety and therapy, the physician values safety higher than therapy and its associated efficacy. The point where the utility function meets the budget line represents the best combination of safety and therapy given the budgetary constraints. From the patient's perspective, in the present example, the best choice is achieved when equal amounts of health care resources are spent on therapy and safety, that is, S = T = 0.5. From the physician's perspective, however, the best choice is achieved when S = 0.8 and T = 0.2.

The exact price of safety or therapy (v or w) and the exact values of the two constants associated with safety or therapy (a or b) are of secondary relevance to the overall outcome of the present analysis. Similar results can be obtained with numerous sets of other values, as long as, in comparison with the patient's preferences, the physician's preferences are shifted in favour of safety. The key insight of the analysis relates to the fact that patient and physician operate according to two different utility functions. In comparison with the patient, the gastroenterologist performing invasive procedures is more concerned with safety and tends to spend more health care resources on safety than therapy to maximize his/her very own utility. In maximizing his/her own utility, the gastroenterologist spends more on safety than the patient would have spent given a free choice of management options.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Case scenario 1: hepatic abscess
  5. Case scenario 2: Schatzki's ring
  6. Patient vs. physician utility
  7. Discussion
  8. References

Two examples were chosen to illustrate the difference in patient vs. physician perspectives underlying gastroenterological decision-making. In the first scenario, the patient rejected a strategy favoured by the physician. In the second scenario, the physician rejected a strategy favoured by the patient. Other such examples abound, as they have become a common feature of interventional medicine and gastrointestinal procedures associated with potentially severe side-effects. For instance, physicians are reluctant to refer and surgeons are hesitant to operate patients with high preoperative risks, such as old age, comorbid conditions, or generally reduced health status.18–20 Similarly, gastroenterologists are reluctant to engage in risky procedures without guaranteed benefit to the patient.21 Although technically feasible, endoscopists may occasionally shy away from snaring large colonic polyps, place expandable stents for tumour palliation, or dilate benign and malignant strictures of the gastrointestinal tract.

The values assigned to different outcomes of a given medical decision vary between the patient and his/her physician. Although these values may also vary among individual patients as well as physicians, the difference between the two separate groups is probably larger than the variation within each group itself. Other analyses of patient and physician preferences, for instance, in the realm of cancer treatment, have revealed similar differences in utility assessment by the two groups.22, 23 Compared with their patients, physicians tend to assign a higher preference to safety than successful therapy. In general, physicians weigh errors of commission more heavily than they do errors of omission.24 The difference between patients and physicians does not reflect an underlying difference in medical knowledge or experience. Ignorance usually leads people to overestimate risks associated with medical or non-medical interventions, especially, if they can draw a mental picture of the risk, but lack any personal experience to assess its incidence or health impact.25 Because their own professional existence and well-being depends on an impeccable track record, not to mention the costly administrative hassle and demoralizing effect of litigation, physicians are inclined to assign a relatively large weight to complications.26–28 Although gastroenterologists may be able to evaluate endoscopic interventions solely from a patient's perspective and weigh safety concerns against therapeutic benefit, unlike the therapy, the safety issue is always loaded with implications that affect the physician personally and directly. Gastroenterology is nowadays practised within an administrative and legal environment that forces upon physicians concerns on how a complication might affect not only their patients’ health but also their very own professional well-being. In managing patients, gastroenterologists have to maximize the therapeutic benefit to their patients and minimize the risk to their patients, as well as to themselves.

The present analysis utilizes mathematical concepts borrowed from decision theory and microeconomics to phrase in theoretical terms some unique aspects of patient–physician interactions that apply especially to invasive gastrointestinal procedures.9, 17 The analysis is limited by providing a descriptive rather than prescriptive model of the interactions between patients and physicians. The model does not allow one to calculate the outcome of a specific clinical situation. Because there is no generally accepted scale to reliably measure safety or efficacious therapy in economic terms, it would be difficult to quantify the preferences of individual patients or physicians. Although demands by insurance companies, hospital policies, governmental regulations and professional guidelines affect the overall available resources, the exact budgetary constraints are often not spelled out and remain unknown. However, the descriptive analysis helps to shed light on some general principles that underlie many medical decisions in gastroenterology. The model shows that patients and physicians often assign varying values to identical medical events and that such variation can markedly alter the outcome of a formal decision analysis. The concept of a medical utility composed of safety and therapy explains how, in trying to maximize their varying utility functions, patients and physicians end up with different choices. Lastly, the analysis reveals that a patient may receive less medical therapy in exchange for more procedural safety. On the surface, the approach to medical management focused on safety may seem unquestionably virtuous. However, the choice of safety at the expense of efficacious therapy occurs frequently according to the physician's own preferences and turns out more conservative that the patient might have chosen for him/herself.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Case scenario 1: hepatic abscess
  5. Case scenario 2: Schatzki's ring
  6. Patient vs. physician utility
  7. Discussion
  8. References
  • 1
    Weinstein MC, Fineberg HV. Clinical Decision Analysis. Philadelphia, PA: Saunders, 1980.
  • 2
    Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology. A Basic Science for Clinical Medicine, 2nd edn. Boston, MA: Little Brown and Company, 1991.
  • 3
    Chapman GB, Sonnenberg FA. Decision Making in Health Care. Cambridge, UK: Cambridge University Press, 2000.
  • 4
    Lusted LB. Decision-making studies in patient management. N Engl J Med 1971; 284: 41624.
  • 5
    Pauker SG, Kassirer JP. Decision analysis. N Engl J Med 1987; 316: 2508.
  • 6
    Kassirer JP, Moskowitz AJ, Lau J, Pauker SG. Decision analysis: a progress report. Ann Intern Med 1987; 106: 27591.
  • 7
    Huang CJ, Pitt HA, Lipsett PA, et al. Pyogenic hepatic abscess. Changing trends over 42 years. Ann Surg 1996; 223: 6009.
  • 8
    Wilson WR, Karchmer AW, Dajani AS, et al. Antibiotic treatment of adults with infective endocarditis due to streptococci, enterococci, staphylococci, and HACEK microorganisms. American Heart Association. JAMA 1995; 274: 170613.
  • 9
    Pauker SG, Kassirer JP. The threshold approach to clinical decision making. N Engl J Med 1980; 302: 110917.
  • 10
    Mansur AJ, Grinburg M, da Luz PL, Bellotti G. The complications of infective endocarditis. A reappraisal in the 1980s. Arch Intern Med 1992; 152: 242832.
  • 11
    Eckardt VF, Kanzler G, Willems D. Single dilation of symptomatic Schatzki rings. A prospective evaluation of its effectiveness. Dig Dis Sci 1992; 37: 57782.
  • 12
    Spechler JS. AGA technical review on treatment of patients with dysphagia caused by benign disorders of the distal esophagus. Gastroenterology 1999; 117: 23354.
  • 13
    Silvis SE, Nebel O, Rogers G, Sugawa C, Mandelstam P. Endoscopic complications. Results of the 1974 American Society for Gastrointestinal Endoscopy survey. JAMA 1976; 235: 92830.
  • 14
    Marshall JB, Chowdhury TA. Does empiric esophageal dilation benefit dysphagia when endoscopy is normal? Dig Dis Sci 1996; 41: 1099101.
  • 15
    Colon VJ, Young MA, Ramirez FC. The short- and long-term efficacy of empirical esophageal dilation in patients with nonobstructive dysphagia. A prospective randomized study. Am J Gastroenterol 2000; 95: 9103.
    Direct Link:
  • 16
    Scolapio JS, Gostout CJ, Schroeder KW, Mahoney DW, Lindor KD. Dysphagia without endoscopically evident disease: to dilate or not? Am J Gastroenterol 2001; 96: 32730.
    Direct Link:
  • 17
    Varian HR. Microeconomic Analysis, 3rd edn. New York: WW Norton & Company, 1992: 94115.
  • 18
    Ryynanen OP, Myllykangas M, Kinnunen J, Takala J. Doctors’ willingness to refer elderly patients for elective surgery. Fam Pract 1997; 14: 2169.
  • 19
    Esposito TJ, Maier RV, Rivara FP, Carrico CJ. Why surgeons prefer not to care for trauma patients. Arch Surg 1991; 126: 2927.
  • 20
    Mavrantonis C, Wexner SD, Nogueras JJ, Weiss EG, Potenti F, Pikarsky AJ. Current attitudes in laparoscopic colorectal surgery. Surg Endosc 2002; 16: 11527.
  • 21
    Theodossi A, Spiegelhalter DJ, McFarlane IG, Williams R. Doctors’ attitudes to risk in difficult clinical decisions: application of decision analysis in hepatobiliary disease. Br Med J 1984; 289: 2136.
  • 22
    Stiggelbout AM, de Haes JCJM. Patient preference for cancer therapy: an overview of measurement approaches. J Clin Oncol 2001; 19: 22030.
  • 23
    de Haes H, Koedoot N. Patient centered decision making in palliative cancer treatment: a world of paradoxes. Patient Educ Couns 2003; 50: 439.
  • 24
    Cohen BJ, Pauker SG. How do physicians weigh iatrogenic complications? J Gen Intern Med 1994; 9: 203.
  • 25
    Moore PG. The Business of Risk. Cambridge, UK: Cambridge University Press, 1983.
  • 26
    Eeckhoudt L, Lebrun T, Sailly JC. Risk aversion and physicians’ medical decision-making. J Health Econ 1985; 4: 27381.
  • 27
    Shapiro RS, Simpson DE, Lawrence SL, Talsky AM, Sobocinski KA, Schiedermayer DL. A survey of sued and nonsued physicians and suing patients. Arch Intern Med 1989; 149: 21906.
  • 28
    Franks P, Williams GC, Zwanziger J, Mooney C, Sorbero M. Why do physicians vary so widely in their referral rates? J Gen Intern Med 2000; 15: 1638.