Volume 133, Issue 1, Pages 5-12 (January 2003)

2 of 28

ABSTRACT

FULL TEXT

FULL-TEXT PDF (166 KB)

Determining an appropriate threshold for referral to surgery for gastroesophageal reflux disease☆☆☆★★★

Accepted 16 July 2002.

Abstract

Background. Persistent symptomatic gastroesophageal reflux disease (GERD) can be treated with medication or surgery. The purposes of this study were (1) to determine how poor the quality of life on medication would need to be to justify assuming the risks of surgery, and (2) to estimate the proportion of patients currently on medication whose quality of life is below this value. Methods. We developed a Markov decision analysis model to simulate health outcomes (measured in quality adjusted life years [QALY]) over 10 years for medication and surgery in patients with typical GERD symptoms. We included probabilities of events obtained from a systematic literature review. Quality of life adjustments, expressed as utilities, were drawn from a survey of 131 patients 1 to 5 years after antireflux surgery. By using this model, we calculated what quality of life on medications would change the optimal strategy from medication to surgery (threshold). To determine the proportion of patients below this value, we prospectively surveyed 40 medically treated GERD patients at our hospital. Results. Surgery resulted in more QALYs than medical therapy when the utility with medication use was below 0.90. Sensitivity analysis showed this value to be relatively insensitive to reasonable variations in surgical risks (mortality, failures, reoperation) and quality of life after surgery. Among those surveyed on medications, 48% fell below this threshold and would be predicted to benefit from surgery. Conclusion. Our model suggests that surgery would likely benefit a high proportion of medically treated GERD patients. Individual assessment of quality of life with GERD should be considered to aid clinical decision making. (Surgery 2003;133:5-12.)

Article Outline

• Abstract

• Methods

• Overview

• Model:

• Variables used in the model

• Probabilities:

• Quality of life:

• Discounting:

• Analysis:

• Sensitivity analysis:

• Results

• Discussion

• References

• Copyright

Gastroesophageal reflux disease (GERD) can be treated with medication or surgery. Three randomized controlled trials have attempted to determine the relative effectiveness of these two alternatives. One early trial showed open fundoplication to be superior to medication.1 Another trial from the Department of Veterans Affairs initially suggested better quality of life after open fundoplication,2 but almost equivalent outcomes after long-term follow-up.3 A third trial, conducted after the introduction of proton pump inhibitors (PPI), suggested good symptom relief with medical therapy, with less morbidity compared with open fundoplication.4 Apart from the variable conclusions reached, inferences from these trials to current practice are limited because they do not directly compare the current two best therapies: PPI and laparoscopic fundoplication.

Debate about the best treatment of GERD has intensified since the development and improvement of techniques for laparoscopic fundoplication. Although PPIs are safe for long-term use and very effective in controlling heartburn symptoms, advocates for surgical treatment counter that these medications must be taken indefinitely and do not fully control symptoms of regurgitation. Case series suggest that laparoscopic fundoplication is very effective for most patients,5, 6 but the procedure entails risks of morbidity, short-term side effects, and late failure. To date, the relative effectiveness of medical therapy and surgery has not been assessed in randomized controlled trials. For this reason, the value of the two approaches continues to be debated, reflected by wide regional variation in the use of antireflux surgery.7

To explicitly consider the tradeoffs between medication and surgery, we used decision analytic techniques to compare long-term outcomes of PPI therapy versus laparoscopic surgical treatment of GERD. Our model was designed to determine what quality of life threshold would justify referral to surgery to optimize long-term outcomes. We then used a prospective survey of medically treated patients with GERD to determine the proportion of patients who fall below this threshold.

Methods

Overview

Model:

We developed a Markov model8 using DATA3.5 software (TreeAge Software, Inc, Williamstown, Mass)9 to simulate outcomes in a hypothetical cohort of patients with severe GERD. We estimated benefit over 10 years for patients receiving the best available medical therapy (PPI) or surgery (laparoscopic fundoplication). Benefit was expressed in quality adjusted life years (QALYs).10 Sensitivity analyses were performed for all model variables.

Our base case analysis simulated 45-year-old patients in otherwise average health with an established diagnosis of uncomplicated GERD with typical symptoms (heartburn and/or regurgitation). We assumed that all were successfully treated with medications and required them indefinitely to avoid recurrence of symptoms. In the decision model, patients moved between hypothetical health states over a 10-year period according to assigned probabilities in 3-month intervals (cycle length). In addition to treatment-related outcomes, all patients in the model were assumed to face age-adjusted risks of mortality from other causes.

A Markov model assumes patients are always in one of a finite number of states of health. Patients may transition from one state to another based on preassigned probabilities. The one exception is the absorbing state “dead” from which patients cannot leave. Each of the health states has a separate utility associated with it.

Patients in our model assigned to surgery at the start of the simulation had 3 possible short-term outcomes: (1) an uncomplicated perioperative course, (2) conversion to an open procedure, or (3) perioperative death. After surgery, patients would either be successfully treated (symptoms controlled and no longer requiring daily medications), require reoperation, experience persistent symptoms requiring medical management, or die from other causes (Fig 1).

View Large Image
Download to PowerPoint

Fig. 1. Overview of the model.

Surgery was assumed to be equally effective regardless of the preoperative quality of life with failure rates derived from the literature. We assumed that persistent side effects might resolve without intervention or require reoperation or another intervention (eg, endoscopic dilation). We also assumed that patients who require a second operation and develop recurrent GERD or have persistent side effects would not undergo a third operation, but would instead be maintained on medical therapy.

Hypothetical patients assigned to medical treatment were assumed to continue with medical management indefinitely.

Variables used in the model

Probabilities:

Probabilities for chance events were determined by using the best available data from the medical literature (Table I). We located peer-reviewed publications with a MEDLINE search for English-language articles on adult GERD from 1994 to 2000, or by cross-citation from review articles. For probabilities related to surgery, we reviewed articles that included more than 50 Nissen or Toupet fundoplications performed by using standard laparoscopic techniques. To estimate reoperation risk, we reviewed articles that described more than 20 laparoscopic attempts. For those patients whose procedures are converted to open surgery, probabilities of recurrent symptoms and side effects are taken from the open Nissen fundoplication literature (Table II).

Table II.

Laparoscopic antireflux procedures


Author (y)	N	Mortality (%)	Conversion to open (%)	Reoperation within 3 mo (%)
Hinder et al29	198	0.5	3.0	1.0
Collet and Cadiere30	734	0	3.8	Not reported
Rattner and Brooks41	74	1.4	10.8	Not reported
Medina et al31	74	1.4	1.4	Not reported
Cadiere et al32	224	0	1.3	Not reported
Fingerhut et al33	146	0	8.2	Not reported
Landreneau et al34	150	0	1.3	3.3
Peters et al35	100	0	2	Not reported
Ritter et al36	123	0	0	Not reported
Soper and Dunnegan39	292	0	0.7	1.0
Hunter et al37	758	0	Not reported	0.5
Rantanen et al38	1162	0.09	1.3	Not reported
Bais et al40	57	0	8.8	3.5
Open antireflux procedures
Rantanen et al38	3993	0.23		0.5
Bais et al40	46	0		Not reported

All probabilities in our model used were weighted by sample size in the original studies. The probability of dying from other causes was based on population mortality rates, adjusted for age and race, from US Vital Statistics data.11

Table I.

Probabilities and utilities used in the decision model for laparoscopic fundoplication


	Base case analysis	Range tested in sensitivity analysis	References
Cycle length	3 mo
Primary operation:
Perioperative death (open/laparoscopic) (%)	0.16	0-1.4	29-40
Early reoperation after laparoscopic surgery (%)	1.1	0-3.5	29, 34, 37, 39-40
Conversion to open (%)	4.0	0-10.8	29-36, 38-41
Early reoperation after open surgery (%)	0.5	0-0.5	38
Reoperation:
Perioperative death (%)	0.31	0-1.4	29, 34, 37-38, 42-47
Conversion to open (%)	15.4	0-100	29, 34, 37-40, 44- 47
Late outcomes:
Any symptom after surgery (eg, GERD, dysphagia) (%/y)	7	6.4-11.2	29, 33, 35-37, 39-40, 42
% patients with symptoms who require a nonoperative intervention	37	0-60	29, 33, 39, 42
% patients with symptoms who require a reoperation	50	25-76	29, 34, 37, 39-40
Long-term health states:*
After successful laparoscopic surgery (± 1 SD)	0.91	0.76-1.0	6
After unsuccessful surgery, on medication (± 1 SD)	0.85	0.72-0.98	6
Short-term health states†
Conversion to open surgery (d)	14
Open reoperation (d)	14
Postoperative nonoperative intervention (d)	½
*Value applied as multipliers per life year for being in that health state. †Period of time subtracted from the cumulative life expectancy.

Quality of life:

The Markov model incorporated adjustments for quality of life associated with undergoing surgery, and requiring medications after unsuccessful surgery. By using a mail survey, we quantified quality of life for patients who underwent surgery between 1 and 5 years previously.6 Quality of life (utility) is expressed by a value from 0 (death) to 1.0 (perfect health). To measure utilities, we chose the “feeling thermometer,” a visual analogue scale, for ease of administration and comprehension by the patients. Patients are instructed to make 2 marks on the thermometer. The first mark shows their health, managing their current level of reflux disease, including their symptoms and treatments. The second mark represents where they think their health would be without reflux disease. We asked patients to consider the following when making their marks on the thermometer: general mood and feeling of well being, energy level, ability to do daily activities, any physical problems or limitations, any emotional problems, bodily pain, and any physical or emotional problems that interfere with social life. By this method, we quantified the quality of life decrement attributable to their reflux disease alone. In addition to quality of life changes resulting from operation or medication, our model also accounted for short-term decrements in quality of life due to postoperative discomfort, operative complications, and any interventions needed to treat persistent side effects or recurrent symptoms (Table I).

Health outcomes are represented by QALYs that are a composite measure of time and quality of life in a particular health state. During any 3-month cycle, patients can accrue a maximum of 0.25 QALYs. However, if the quality of life is only 0.90 for the health state in which the patient is assigned, patients would accrue only 0.22 QALYs (0.90 times 0.25) for that cycle. Adding up the QALYs accrued for all cycles over the 10-year time horizon of our analysis yields the expected value of each strategy.

Discounting:

Discounting is necessary because all future health consequences should be stated in terms of their present value to the decision maker. All outcomes in the model for both medication and surgery were discounted over time by using a 3% annual discount rate.12

Analysis:

Our primary goal was to determine how poor the quality of life on medications would need to be to justify referral to surgery to maximize long-term benefit (threshold analysis). Using decision analytic software, we calculated the value for quality of life on medication, qualitatively expressed as utilities (between 0 = death and 1.0 = perfect health), which would result in equal benefit for the medical and surgical strategies (ie, the threshold value). With values for quality of life on medication above this threshold value, our model would predict greater benefit from a strategy of staying on medications. With values below this threshold, our model would predict greater benefit from the surgical strategy.

Our second goal was to estimate the proportion of people with stable GERD symptoms on medication who fall below this threshold. Stable GERD means that patients self-assessed that their medication was working well to control their symptoms. We surveyed a convenience sample of 122 patients newly referred to the gastroenterology clinic, who were on medications and had a confirmed diagnosis of GERD. We asked each subject: “How do you think your treatment is working?” and allowed responses using Likert-type choices of “perfectly,” “very well,” “well,” “not well,” and “not at all.” To establish our cohort with stable symptom control on medications, we selected those patients on PPI who responded to this question with “perfectly,” “very well,” and “well” (n = 40). For these medically treated patients, we measured quality of life using the same visual analogue scale that we used to measure quality of life for patients after surgery. This study was approved by the Institutional Review Board at Dartmouth-Hitchcock Medical Center.

Sensitivity analysis:

The probabilities used in a model are estimates and are not known with certainty. Therefore, we use sensitivity analysis to see how the decision is affected by changing 1 or more variables in the model. In sensitivity analysis, the model is rerun as each variable in the model is varied across a range of plausible values. For example, using the quality of life estimates for postoperative states, we tested values across a range of 1 SD above and below the mean obtained by survey.

Results

In our model, we compared outcomes over 10 years after initial medical or surgical treatment of severe GERD in a hypothetical cohort of 45-year-old patients. With the use of our baseline assumptions, surgery resulted in greater long-term benefit than medical treatment when the utility for life on medications was below 0.90 (Fig 2).

View Large Image
Download to PowerPoint

Fig. 2. Threshold value as determined by the quality of life on medications resulting in equivalent long-term benefit (over a 10-year time horizon).

We performed sensitivity analyses for each of the variables over the ranges given in Table I to examine how changing one of these baseline values might influence the threshold value for quality of life. We found the threshold to be relatively insensitive to reasonable variations in baseline risks of surgical mortality, failure, and reoperation (Table III).

Table III.

Sensitivity analysis for important baseline assumptions


	Variable	Threshold* for quality of life on medication
	Surgical mortality
	Best case = 0%	0.90
	Worst case = 1.4%	0.89
	Quality of life after surgery, on medications
	Best case = 0.98	0.91
	Worst case = 0.72	0.89
	Quality of life after successful surgery
	Best case = 1.0	0.98
	Worst case = 0.76	0.77
	Persistent symptoms after surgery
	Best case = 0.94%/y	0.91
	Worst case = 9.7%/y	0.90
	Reoperation rate
	Best case = 1.8%/yr	0.90
	Worst case = 5.3%/y	0.90
	*Threshold is the value at which the optimal treatment choice changes. Surgery is preferred when the utility is below the threshold; medical therapy is preferred when the value is above the threshold.

With the use of the worst case for surgical mortality (1.4%), the threshold value dropped only marginally from the baseline: 0.90 to 0.89. In other words, if the surgical mortality is 1.4%, people can have a slightly lower quality of life on medications before surgery would become the optimal treatment. Using the best case for surgical mortality (0%) does not change the threshold appreciably. Similarly, by using the worst case for surgical failure rate (9.7% per year), the threshold value does not change. With the use of the best case (0.94% per year), the threshold value increases only marginally to 0.91.

To evaluate where our population of medically treated patients with severe GERD rated their quality of life compared with this threshold value, we used data from 40 patients who reported their medications were working “well,” “very well,” or “perfectly.” Relative to the threshold value (0.90) generated by our model, 48% of those surveyed measured utilities below this threshold, 18% had a utility above this threshold, and 34% reported a utility equal to the threshold (Fig 3).

View Large Image
Download to PowerPoint

Fig. 3. Distribution of utilities for patients with adequate symptom control on daily medications (n = 40).

Discussion

Our analysis suggests that a substantial proportion of GERD patients who respond well to medication might still benefit from antireflux surgery. We calculated that a patient with a utility for quality of life on medications below 0.90 would be predicted to benefit from surgery in the long term. To help readers put this threshold value in a broader clinical context, Table IV lists average utility scores (quality of life) for several other chronic conditions, each similarly assessed with a rating scale.

Table IV.

Average utility scores (quality of life*) using a rating scale for chronic disease states


Health state	Preference score	Range	Reference
Postcholecystectomy syndrome	0.95	0.5-1.0	48
Interferon-α 2b treatment for hepatitis	0.93	0.90-1.0	49
Congestive heart failure	0.90	0.80-1.0	50
Moderate prostatism	0.90		13
Chronic lower extremity edema	0.90		51
Overt hypothyroid	0.74	0.50-0.90	52
Claudication	0.70		14
*10 is perfect health.

For example, patients with moderate prostatism rate their current quality of life at 0.90,13 but patients with claudication rate their current quality of life at 0.70.14 Among a sample of patients successfully treated with PPI at our institution, almost half of the patients had self-reported utility scores (quality of life) below our calculated threshold value of 0.90.

A previously published decision model comparing medical and surgical treatment of GERD found that both strategies were equally effective after 10 years.22 However, the authors represented both the long-term postoperative and chronic omeprazole (PPI) health states with utilities equal to 1.0 (ie, perfect health). In contrast, our survey data shows that utilities for these two health states are distinct, vary among patients, and not equivalent to perfect health.

Several limitations of our decision model should be considered when interpreting our findings. First, our probability estimates are limited by the available data in the literature. Publication bias can lower the estimates for key variables such as morbidity and mortality with surgery. Second, because laparoscopic antireflux surgery and PPIs are relatively new treatments, long-term outcome data are scarce. Although the open procedure has been shown to be quite durable, some may argue that laparoscopic antireflux surgery has not yet stood the test of time. Similarly, the safety and efficacy of life-long PPI use is unknown. For this reason, we limited the time horizon of our analysis to 10 years. Third, our model may be too simple to account for nuances in clinical decision making. For example, we did not account for risk aversion among patients faced with surgery. Although we found that almost half of those surveyed on medications fall below the threshold and may benefit from surgery over the long term, risk aversion may explain their real world decisions to pursue medical management. Finally, our model did not account explicitly for relatively unusual complications associated with GERD, including late stricture, Barrett's esophagus, and malignancy. There is currently insufficient data to judge the extent to which these outcomes may differ between patients treated medically and surgically.

Some may question our decision to use a visual analogue scale (the feeling thermometer) for utility assessment. We chose this measure based on its ease of administration and comprehension by the patients and because its use as for utility assessment (quality of life) has been validated by prior studies.15, 16, 17 A good correlation with a 5-point categorical rating scale for general health and the thermometer has previously been established.18 Comparison of our thermometer scores with our 5-point Likert scale for general health yielded similar results; low scores on the visual analogue scale (thermometer) correlated with the lower categories of self-assessment (fair/poor), whereas higher scores correlated with better self-assessment of health (Spearman rho = 0.38, P < .001).

We could instead have relied on other popular, but more complex, utility assessment techniques, such as the time tradeoff (TTO) or standard gamble (SG). In the former, patients decide how much of their remaining life expectancy they would give up to avoid their current health state with a disease/disability. A utility of 0.90 means that patients would give up 10% of their life expectancy to not have GERD anymore. With the standard gamble, patients decide which mortality risk they would accept to avoid their current health state with a disease/disability. A utility of 0.90 means that patients would assume a 10% mortality to avoid their GERD-related symptoms. Previous studies have suggested that utilities assessed by the TTO or SG tend to be higher than when measured with the visual analogue scale.19, 20, 21 For this reason, our threshold value of 0.90 should be interpreted in the context of the method used for utility assessment.

Although our analysis considered only the patient perspective, decisions at the population level would need to consider costs in addition to effectiveness. Some may argue that the small benefit predicted by our model may not be worth the cost of undergoing surgery. Previous studies estimating the relative costs of medical and surgical management of GERD have conflicting results.22, 23, 24, 25, 26, 27, 28 Future studies will need to account for potential reductions in the cost of PPIs as some medications come off patent protection.

Our analysis suggests that quality of life on medications does not need to be very poor to justify referral for antireflux surgery. Based on our survey results, we conclude that even among GERD patients who are successfully treated with PPI, many would gain greater long-term benefit from antireflux surgery. Because of the relative parity of medical and surgical treatment, individual assessment of quality of life should be considered to aid clinical decision making.

References

1. 1 Behar J, Sheahan D, Biancani P, Spinro H, Storer E. Medical and surgical managment of reflux esophagitis. N Engl J Med. 1975;293:263–268. MEDLINE

2. 2 Spechler SJ, for the Department of Veterans Affairs Gastroesophageal Reflux Disease Study Group . Comparison of medical and surgical therapy for complicated gastroesophageal reflux disease in veterans. N Engl J Med. 1992;326:786–792. MEDLINE

3. 3 Spechler S, Lee E, Ahnen D, Goyal RK, Hirano I, Ramirez F, et al. Long-term outcome of medical and surgical therapies for gastroesophageal reflux disease: follow-up of a randomized controlled trial. JAMA. 2001;285:2331–2338. MEDLINE | CrossRef

4. 4 Lundell L, Miettinen P, Myrvold H, Pedersen SA, Thor K, Lamm M, et al. Long-term management of gastro-oesophageal reflux disease with omeprazole or open antireflux surgery results of a prospective, randomized clinical trial. Eur J Gastroenterol. 2000;12:879–887.

5. 5 Bammer T, Hinder R, Klaus A, Klinger P. Five- to eight-year outcome of the first laparoscopic Nissen fundoplications. J Gastrointest Surg. 2001;5:42–48. CrossRef

6. 6 Liu J, Wolosin S, Laycock W, Schwartz L. Late outcomes after laparoscopic surgery for gastroesophageal reflux. Arch Surg. 2002;137:397–401. MEDLINE | CrossRef

7. 7 Finlayson S, Birkmeyer J, Laycock W. Trends in surgery for gastroesophageal reflux disease: the effect of laparoscopic surgery on utilization. In: Annual Meeting of the International Society of Technology Assessment in Health Care. Philadelphia, Pa. June 2001;.

8. 8 Sonnenberg F, Beck J. Markov models in medical decision making: a practical guide. Med Decis Making. 1993;13:322–338. CrossRef

9. 9 DATA 3.5. 3.5 ed. Williamstown, MA: TreeAge Software, Inc; 1999;.

10. 10 Torrance G. Utility apporach to measuring health-related quality of life. J Chronic Dis. 1987;40:593–600. MEDLINE | CrossRef

11. 11 The National Center for Health Statistics . Report of final mortality statistics. Vol 47. Washington, DC: US Public Health Service; 1996; no. 7.

12. 12 Lipscomb J, Weinstein MC, Torrance GW. In: Gold M, Siefel J, Russell L, Weinstein M editor. Cost-effectiveness in health and medicine. New York: Oxford University Press; 1996;p. 214–246.

13. 13 Baladi J, Menon D, Otten N. An economic evluation of finesteride for treatment of benign prostatic hyperplasia. Pharmacoeconomics. 1996;9:443–454. MEDLINE | CrossRef

14. 14 Sculpher M, Michaels J, McKenna M. A cost-utility analysis of laser-assisted angioplasty for peripheral arterial occlusion. Int J Technol Assess Health Care. 1996;12:104–125. MEDLINE | CrossRef

15. 15 Johnson J, Coons S, Ergo A, Szava-Kovats G. Valuation of EuroQol (EQ-5D) health states in an adult US sample. Pharmacoeconomics. 1998;4:421–433.

16. 16 Brooks R. EuroQol: the current state of play. Health Policy. 1996;37:53–72. Abstract | Full-Text PDF (1303 KB) | CrossRef

17. 17 Brazier J, Jones N, Kind P. Testing the validity of the Euroqol and comparing it with the SF-36 Health Survey questionnaire. Qual Life Res. 1993;2:169–180. MEDLINE | CrossRef

18. 18 Llach X, Herdman M, Schiaffino A. Determining correspondence between scores on the EQ-5D “Thermometer” and a 5-point categorical rating scale. Med Care. 1999;37:671–677. MEDLINE | CrossRef

19. 19 Nease R, Kneeland T, O'Connor G, Sumner W, Lumpkins L, Shaw L, et al. Variation in patient utilites for outcomes of the managment of chronic stable angina. JAMA. 1995;273:1185–1190. MEDLINE

20. 20 Llewellyn-Thomas H, Sutherland H, Tibshirani R, Ciampi A, Till J, Boyd N. Describing health states: methodologic issues in obtaining values for health states. Med Care. 1984;22:543–552. MEDLINE | CrossRef

21. 21 Read J, Quinn R, Berwick D, Fineberg H, Weinstein M. Preferences for health outcomes: comparison of assessment methods. Med Decis Making. 1984;4:315–329. CrossRef

22. 22 Heudebert GR, Marks R, Wilcox M, Centor RM. Choice of long-term strategy for the management of patients with severe esophagitis: a cost-utility analysis. Gastoenterol. 1997;112:1078–1086.

23. 23 Myrvold HE, Lundell L, Miettinen P, Pedersen SA, Liedman B, Hatlebakk J, et al. The cost of long term therapy for gastro-oesophageal reflux disease: a randomised trial comparing omeprazole and open antireflux surgery. Gut. 2001;49:488–494. MEDLINE | CrossRef

24. 24 Lundell L, Dalenback J, Janatuinen E, Hattlebakk J, Levander K, Miettinen P, et al. Comprehensive 1-year cost analysis of open antireflux surgery in Nordic countries. Br J Surg. 1998;85:1002–1005. MEDLINE | CrossRef

25. 25 Bloomston M, Zervos E, Gonzalez R, Albrink M, Rosemurgy A. Quality of life and antireflux medication use following laparoscopic Nissen fundoplication. Am Surg. 1998;64:509–514. MEDLINE

26. 26 Heikkinen T, Haukipuro K, Koivukangas P, Sorasto A, Autio R, Sodervik H, et al. Comparison of costs between laparoscopic and open Nissen fundoplication: a prospective randomized study with 3-month followup. J Am Coll Surg. 1999;188:368–376. Abstract | Full Text | Full-Text PDF (267 KB) | CrossRef

27. 27 Viljakka M, Nevalainen J, Isolauri J. Lifetime costs of surgical versus medical treatment of severe gastro-oesophageal reflux disease in Finland. Scand J Gastoenterol. 1997;32:766–772.

28. 28 Van Den Boom G, Go PMMYH, Hameeteman W, Dallemagne B. Cost effectiveness of medical versus surgical treatment in patients with severe or refractory gastroesophgeal reflux disease in The Netherlands. Scand J Gastoenterol. 1996;31:1–9.

29. 29 Hinder R, Filipi C, Wetscher G, Neary P, DeMeester T, Perdikis G. Laparoscopic Nissen fundoplication is an effective treatment for gastroesophageal reflux disease. Ann Surg. 1994;220:472–483. MEDLINE

30. 30 Collett D, Cadiere G. Conversions and complications of laparoscopic treatment of gastroesophageal reflux disease: formation for the Development of Laparoscopic Surgery for Gastroesophageal Reflux Disease Group. Am J Surg. 1995;169:622–626. Abstract | Full-Text PDF (579 KB) | CrossRef

31. 31 Medina L, Veintimilla R, Williams M, Fenoglio M. Laparoscopic fundoplication. J Laparoendosc Surg. 1996;6:219–226. MEDLINE

32. 32 Cadiere G, Himpens J, Rajan A, Muls V, Lemper JC, Bruyns J, et al. Laparoscopic Nissen fundoplication: laparoscopic dissection technique and results. Hepatogastroenterology. 1997;44:4–10. MEDLINE

33. 33 Fingerhut A, Etienne J, Millat B, Comandella M. Laparoscopic antireflux surgery—technique and results. Therapeutische Umschau. 1997;54:492–499. MEDLINE

34. 34 Landreneau R, Wiechmann R, Hazelrigg S, Santucci TS, Boley TM, Magee MJ, et al. Success of laparoscopic fundoplication for gastroesophageal reflux disease. Ann Thorac Surg. 1998;66:1886–1893. MEDLINE | CrossRef

35. 35 Peters JH, DeMeester TR, Crookes P, Oberg S, de Vos Shoop M, Hagen JA, et al. The treatment of gastroesophageal reflux disease with laparoscopic Nissen fundoplication: prospective evaluation of 100 patients with “typical” symptoms. Ann Surg. 1998;228:40–50. MEDLINE | CrossRef

36. 36 Ritter M, Peters J, DeMeester T, Crookes PF, Mason RJ, Green L, et al. Outcome after laparoscopic fundoplication is not dependent on a structurally defective lower esophageal sphincter. J Gastrointest Surg. 1998;2:567–572. CrossRef

37. 37 Hunter J, Smith C, Branum G, Waring JP, Trus TL, Cornwell M, et al. Laparoscopic fundoplication failures: patterns of failure and response to fundoplication revision. Ann Surg. 1999;230:595–599. MEDLINE | CrossRef

38. 38 Rantanen TK, Salo JA, Sipponen JT. Fatal and life-threatening complications in antireflux surgery: analysis of 5502 operations. Br J Surg. 1999;86:1573–1577. MEDLINE | CrossRef

39. 39 Soper N, Dunnegan D. Anatomic fundoplication failure after laparoscopic antireflux surgery. Ann Surg. 1999;229:669–675. MEDLINE | CrossRef

40. 40 Bais JE, Bartelsman JFWM, Bonjer HJ, Cuesta MA, Go PM, Klinkenberg-Knol EC, et al. Laparoscopic or conventional Nissen fundoplication for gastro-oesophageal reflux disease: randomised clinical trial. Lancet. 2000;355:170–174. Abstract | Full Text | Full-Text PDF (80 KB) | CrossRef

41. 41 Rattner DW, Brooks DC. Patient satisfaction following laparoscopic and open antireflux surgery. Arch Surg. 1995;130:289–294. MEDLINE

42. 42 Dallemagne B, Weerts J, Jehaes C, Markiewicz S. Causes of failures of laparoscopic antireflux operations. Surg Endosc. 1996;10:305–310. MEDLINE | CrossRef

43. 43 Stein H, Feussner H, Siewert J. Failure of antireflux surgery: causes and management strategies. Am J Surg. 1996;171:36–40. Abstract | Full-Text PDF (682 KB) | CrossRef

44. 44 Curet MJ, Josloff RK, Schoeb O, Zucker KA. Laparoscopic reoperation for failed antireflux procedures. Arch Surg. 1999;134:559–563. MEDLINE | CrossRef

45. 45 Floch N, Hinder R, Klingler P, Branton SA, Seelig MH, Bammer T, et al. Is laparoscopic reoperation for failed antireflux surgery feasible?. Arch Surg. 1999;134:733–737. MEDLINE | CrossRef

46. 46 Horgan S, Phol D, Bogetti D, Eubanks R, Pellegrini C. Failed antireflux surgery: what have we learned from reoperation?. Arch Surg. 1999;134:809–817. MEDLINE | CrossRef

47. 47 Pointner R, Bammer T, Then P, Kamolz T. Laparoscopic refundoplications after failed antireflux surgery. Am J Surg. 1999;178:541–544. Abstract | Full Text | Full-Text PDF (146 KB) | CrossRef

48. 48 Bass E, Steinberg E, Pitt H. Cost-effectiveness of extracorporeal shock-wave lithotripsy versus cholecystectomy for symptomatic gallstones. Gastoenterol. 1991;101:189–199.

49. 49 Wong J, Koff B, Rine F. Cost-effectiveness of interferon-alpha-2b treatment for hepatitis B e antigen-positive chornic hepatitis B. Ann Intern Med. 1995;122:664–675. MEDLINE

50. 50 Kellett J, Clarke J. Comparison of “accelerated” tissue plasminogen activator with streptokinase for treatment of suspected acute myocardial infarction. Med Decis Making. 1995;15:297–310. CrossRef

51. 51 Sarasin F, Eckman M. Management and prevention of thromboembolic events in patients with cancer-related hypercoagulable states: a risky business. J Gen Intern Med. 1993;8:476–486. MEDLINE | CrossRef

52. 52 Danese M, Powe N, Sawin C. Screening for mild thyroid failure at the periodic health examination. JAMA. 1996;276:285–292. MEDLINE

White River Junction, Vt, and Lebanon, NH

From the Department of Surgery and VA Outcomes Group, VA Medical Center, White River Junction, Vt; and the Departments of Surgery and Gastroenterology, Dartmouth-Hitchcock Medical Center, Lebanon, NH

☆ Dr Finlayson and Dr Birkmeyer are supported by Career Development Awards from the VA Health Services Research and Development program.

☆☆ The views expressed herein do not necessarily represent those of the Department of Veterans Affairs or the United States government.

★ Reprint requests: Dr Jean Y. Liu, VA Surgery (112), Department of Veterans Affairs Hospital, White River Junction, VT 05009.

★★ 0039-6060/2003/$30.00 + 0

PII: S0039-6060(02)21628-1

doi:10.1067/msy.2003.122

2 of 28