- Email: [email protected]
Cost-effectiveness of screening a population with chronic gastroesophageal reflux
Cost-effectiveness of screening a population with chronic gastroesophageal reflux Paul J. Nietert, PhD, Marc D. Silverstein, MD, Mahesh S. Mokhashi, MD, Christopher Y. Kim, MD, Tammy F. Glenn, FNP, Vicki A. Marsi, MBA, Robert H. Hawes, MD, Michael B. Wallace, MD, MPH Charleston, South Carolina
Background: Persons with chronic esophageal reflux are at increased risk for the development of Barrett’s esophagus and adenocarcinoma. Recently developed ultrathin endoscopes are less expensive and better tolerated than standard endoscopes, they can be used without sedation, and are sensitive and specific for Barrett’s esophagus. The cost-effectiveness of one-time screening strategies were evaluated for 50-year-old patients with chronic reflux: no screening, standard endoscopy, and screening by an ultrathin endoscope. Methods: Markov models were created to simulate the clinical course for patients with chronic reflux. Costs and quality-adjusted life-years were estimated from cancer registry data, published medical data, and expert opinion. Results: Under baseline assumptions, no screening resulted in average costs of $11,785 per person and 19.3226 quality-adjusted life-years. Ultrathin endoscopy screening resulted in costs of $12,119 per person and 19.3326 quality-adjusted life-years, yielding a marginal cost-effectiveness ratio of $55,764 per quality-adjusted life-year. Using standard endoscopy yielded costs of $12,332 with only slightly greater effectiveness, yielding a marginal cost-effectiveness ratio of $709,260 when compared with ultrathin endoscopy and $86,833 compared with no screening. Results were most sensitive to variation in the incidence of cancer in the population with Barrett’s esophagus. Conclusions: Screening for Barrett’s esophagus with ultrathin endoscopy is more cost-effective than standard endoscopy, and both strategies appear to improve quality-adjusted life-years among patients with chronic reflux at costs that are similar to those of other accepted preventive measures. (Gastrointest Endosc 2003;57:311-8.)
The incidence of adenocarcinoma of the esophagus is increasing in the United States.1 Almost all of these cancers occur in patients with Barrett’s esophagus (BE),2 a condition in which metaplastic columnar epithelium replaces the normal squamous epithelium of the esophagus. The presence of BE and chronic GERD are the most significant identifiable risk factors for the development of adenocarcinoma of the esophagus. Patients with chronic GERD are 7 to 44 times more likely to develop adenocarcinoma of the esophagus than patients without reflux.3 Surveillance of patients with known BE is associated with earlier detection of adenocarcinoma Received June 11, 2002. For revision August 27, 2002. Accepted October 30, 2002. Current affiliations: Center for Health Care Research, Medical University of South Carolina, Department of Medicine, Division of General Internal Medicine, Medical University of South Carolina, Digestive Diseases Center, Division of Gastroenterology and Hepatology, Medical University of South Carolina, and Department of Veterans Affairs Medical Center, Charleston, South Carolina. Funded, in part, by a grant from the U.S. Office of Naval Research (No. N00014-99-1-0784) to the Medical University of South Carolina. Dr. Wallace was funded by the American Digestive Health Foundation (TAP Pharmaceutical Outcomes Research Award). Reprint requests: Paul J. Nietert, PhD, Center for Health Care Research, 135 Cannon St., Suite 403, P.O. Box 250837, Charleston, SC 29425. 0016-5107/2003/$30.00 + 0 doi:10.1067/mge.2003.101 VOLUME 57, NO. 3, 2003
and improved survival.4 However, autopsy series suggest less than 5% of patients with BE are currently under surveillance.5 Screening and surveillance of BE is widely practiced and recommended by the professional societies concerned with digestive diseases,6 but no prospective efficacy data are available. Furthermore, decision analysis studies have primarily addressed surveillance strategies after BE has been diagnosed.7,8 The vast majority of patients with BE and adenocarcinoma have never been diagnosed, and the majority of those presenting with adenocarcinoma have not been under surveillance.5,9 If surveillance is to be effective, a program of widespread screening is necessary to identify at risk cases. Major limitations of widespread screening are cost and patient tolerance of upper endoscopy. Endoscopic examination with ultrathin endoscopes as small as 3.1 mm in diameter has recently been developed and has been shown to be well tolerated and potentially less expensive than standard endoscopy (SE).10-16 The purpose of this study was to evaluate the costeffectiveness of ultrathin, unsedated endoscopy (UTE) compared with standard endoscopic screening and no screening for the detection and surveillance of BE. BE was defined in this study as any length of intestinal metaplasia within the tubular esophagus. GASTROINTESTINAL ENDOSCOPY
311
P Nietert, M Silverstein, M Mokhashi, et al.
Chronic gastroesophageal reflux: cost-effectiveness of population screening
Figure 1. Cost-effectiveness model comparing 3 strategies of screening for Barrett’s esophagus. The “M” symbolizes the beginning of a Markov model. GERD, Gastroesophageal reflux disease; UTE, ultrathin endoscopy; SE, standard endoscopy; LGD, low-grade dysplasia; HGD, high-grade dysplasia. PATIENTS AND METHODS The cost-effectiveness (CE) analysis was designed to determine the average lifetime-costs and average lifetime quality-adjusted life-years (QALYs) associated with each of 3 strategies: no screening, screening with UTE, and screening with SE. In these analyses, all patients are assumed to have GERD, defined for the purposes of the study as heartburn and/or acid reflux at least once per week. For all strategies, patients identified as having BE with no or low-grade dysplasia are assumed to be in a surveillance program, undergoing SE every other year. The strategies of screening with UTE and of screening with SE were compared against a strategy of no screening. In addition, the UTE and SE strategies were compared against one another. The analysis was performed with statistical software (Data 3.5.6, Treeage Software, Williamstown, Mass.) and the overall decision analysis model is shown in Figure 1. In both the UTE and SE strategies, 5% of patients are assumed to refuse the procedure, and for these patients costs and effectiveness are assumed to be the same as those in the no screening arm. Markov models The clinical course of disease for 50-year-old patients with chronic GERD was simulated by using Markov models, whereby the annual changes in disease states for patients were simulated from transition probabilities based only on the last known disease state of the patient, independent of history. Estimates of the annual transitions among the mutually exclusive disease states used data from a variety of sources. Key variables and sources used in these models are listed in Table 1. This type of study is exempt from Institutional Review Board review because no human subjects were involved. 312
GASTROINTESTINAL ENDOSCOPY
The “no screening” strategy: This strategy is meant to depict the traditional course of treatment for patients with GERD. Of the patients in this strategy, 3% are assumed to have undiagnosed BE, whereas the rest are considered to be in a clinical state free of BE and any cancer. Patients with BE in the model progress into symptomatic states, defined as having dysphagia and/or upper GI bleeding, at the rate of approximately 1.5% per year, a rate based on our experience that corresponds to about 1 in 6 patients who eventually develop symptoms within 10 years. Patients with symptoms undergo SE, assumed to be the reference standard, and both symptomatic and asymptomatic patients with BE may progress to esophageal adenocarcinoma. Patients free of BE remain free of BE, although they progress to cancer at substantially lower rates than BE patients. All patients identified with high-grade dysplasia (HGD) or with local or regional cancer undergo esophagectomy. Median life expectancy data regarding increased life expectancy for patients undergoing esophagectomy were obtained from our own analyses of the Surveillance, Epidemiology, and End Results (SEER) registry public use data17 (see Table 1). Patients identified with metastatic disease receive palliative care alone. All patients in noncancerous disease states receive pharmacologic treatment for GERD. Half of patients taking medication receive histamine-2 receptor antagonists, and the rest proton-pump inhibitors. The UTE strategy. Of patients screened by UTE, a certain proportion will be identified as having a columnarlined esophagus. The accuracy of this classification is a direct function of the population prevalence and the diagnostic test characteristics (sensitivity and specificity) of UTE (see Table 1). These patients then undergo further diagnostic testing with SE to confirm the presence of BE. As with the no screening strategy, patients ultimately identified with HGD or with local or regional cancer receive esophagectomy, and those identified with metastatic disease have palliative care alone. Patients who are not found to have BE or cancer receive only pharmacologic treatment. The SE strategy. In this strategy, all patients with chronic GERD are screened once with SE, which is assumed to be the reference standard for the diagnosis of BE. As with the other strategies, patients identified with HGD or with local or regional cancer undergo esophagectomy, and patients free of BE and cancer receive only pharmacologic treatment. Quality of life adjustment In this study the primary effectiveness outcome was quality-adjusted life-years (QALYs). Provenzale et al.8 reported the median utility for esophagectomy to be 0.97 (interquartile range 0.83-1.0), and this value was incorporated in the Markov states for anyone undergoing esophagectomy, with sensitivity analysis being performed on this measure. In the absence of direct reports of quality of life from patients with esophageal cancer, estimates of quality of life (QOL) utilities were based on published data obtained from the National Health Interview Survey. Using these national data, Gold et al.18 (1998) reported a QOL score of 0.47 for VOLUME 57, NO. 3, 2003
Chronic gastroesophageal reflux: cost-effectiveness of population screening
P Nietert, M Silverstein, M Mokhashi, et al.
Table 1. Key variables used in the Markov models Variable Prevalence of: GERD in general population Undiagnosed BE in GERD population Relative risk of developing EA given GERD Median life expectancy for EA Local disease Local disease with esophagectomy Regional disease Regional disease with esophagectomy Metastatic disease Incidence of EA among patients with BE
Relative risk of developing EA among patients with HGD compared with those with no dysplasia or LGD Annual rate of patients with GERD developing symptoms requiring SE Probability of having: LGD if patient has BE HGD if patient has BE Annual probability of: Developing LGD if patient has no dysplasia Developing HGD if patient has no dysplasia Developing HGD if patient has LGD Regressing to no dysplasia from LGD Stage distribution of EA: In unscreened population (local: regional: metastatic) In screened population (local: regional: metastatic) Sensitivity of UTE Specificity of UTE Probability of death from SE Cost of: Administration of UTE Administration of SE without biopsy Administration of SE with biopsy H2 receptor antagonist Proton pump inhibitor CT scan Esophagectomy Palliative care for metastatic disease Case after esophagectomy QOL utility for: Esophagectomy Local and regional EA Metastatic EA
Value
Range used in sensitivity analyses
Reference
19.8% 3.0% 9.68*
5%-25% 0%-15% 5.25-16.54
Locke, 199741 Cameron, 199742 Lagergren, 19993
1.37 y 3.68 y 0.93 y 1.13 y 0.37 y 1/227 life years
1.0-1.7 y 2.8-4.6 y 0.7-1.2 y 0.8-1.4 y 0.3-0.5 y 1/100-1/400 life years
28.0
13.0-63.0
SEER, 200017 SEER, 200017 SEER, 200017 SEER, 200017 SEER, 200017 Drewitz, 199735 O’Connor, 199943 Katz, 199844 Cameron, 198530 Reid, 200045
1.5%
0.5%-2.5%
Expert opinion
0.0735 0.00735
0.055-0.092 0.006-0.009
O’Connor, 199943 O’Connor, 199943
0.045 0.010 0.095 0.75
0.034-0.056 0.008-0.013 0.071-0.119 0.563-0.938
O’Connor, 199943 O’Connor, 199943 O’Connor, 199943 Expert opinion
40%: 30%: 30%
0.95 0.95 0 per 1000
20%: 30%: 50% 60%: 30%: 10%† 50%: 30%: 20% 90%: 10%: 0%† 0.75-1.00 0.75-1.00 0-1 per 1000
Mokhashi, 200029 Mokhashi, 200029 Expert opinion
$97 $346 $507 $74/y $727/y $544 $29,415 $40,798 $1236/y
$50-$300 $200-$600 $350-$750 $0-$1306 $50-$1568 $200-$800 $15,000-$40,000 $15,000-$50,000 $500-$2000
MUSC cost data MUSC cost data MUSC cost data Red Book, 199920 Red Book, 199920 MUSC cost data Provenzale, 19998 Provenzale, 19998 Provenzale, 19998
0.97 0.47 0.47
0.75-1.00 0.25-0.75 0.25-0.75
Provenzale, 19998 Gold, 199818 Gold, 199818
70%: 30%: 0%
SEER, 200017 Streitz, 19934
BE, Barrett’s esophagus; EA, esophageal adenocarcinoma; LGD, low-grade dysplasia; HGD, high-grade dysplasia; UTE, batterypowered endoscope; MUSC, Medical University of South Carolina; QOL, quality of life. *This relative risk is actually a weighted average of 3 separate relative risks as reported by Lagergren et al. (1999)3 associated with 3 categories of reflux frequency. †For sensitivity analyses involving stage distribution, the percent of patients diagnosed with regional disease was held constant, whereas the percent of patients with local and metastatic disease was allowed to vary.
patients with GI cancer. This value was incorporated in the Markov states of patients with late-stage cancer. Perspective and costs This analysis was performed from the prospective of a third-party payer. Where possible, national cost estimates VOLUME 57, NO. 3, 2003
or estimates used in similar decision models in other publications (e.g., Provenzale et al.8, 1999) were used in the model. Annual costs of pharmacologic treatment of patients with GERD were determined by obtaining a published list of average retail drug prices in the United States19 and subtracting a $250 annual deductible as well GASTROINTESTINAL ENDOSCOPY
313
P Nietert, M Silverstein, M Mokhashi, et al.
Chronic gastroesophageal reflux: cost-effectiveness of population screening
Table 2. Breakdown of UTE and SE cost estimates
Technical fee ($) Professional fee ($) Pathology technical fee ($) Pathology professional fee ($) Total ($1999) Total ($2001)
UTE
SE without biopsy
SE with biopsy
74 15 0 0 89 97
183 135 0 0 318 346
228 135 36 66 465 506
as a $22 per month co-pay for brand-name medications and $7 per month for generic medications. The 1999 Drug Topics Red Book of average wholesale drug prices was used to determine the upper limit of drug costs used in the sensitivity analyses.20 Direct and indirect costs associated with administration of the UTE and SE (with and without biopsy) were determined from the Medical University of South Carolina financial cost accounting data. Technical UTE costs were based on the estimated per-procedure cost of the instrument depreciated over its expected lifetime, and all costs related to cleaning of the instrument (disposable supplies, personnel). Because ultrathin endoscopes have no accessory channel or automatic air/water channels, they can be disinfected by soaking in a standard endoscopy cleaning solution and do not require automated washers, thus further reducing the cost of service. A breakdown of the cost estimates used in the analysis is shown in Table 2. Provider costs including durable and disposable equipment were calculated from actual costs to perform the procedures. Third party reimbursements were based on South Carolina Medicare and commercial health maintenance organizations (Blue Cross) for esophagoscopy (UTE) and upper endoscopy with biopsy (standard endoscopy). Cost of esophagectomy, postesophagectomy health care, and palliative care for metastatic disease and health state utilities were obtained from published data.8,21 All costs were adjusted to 2001 dollars by using the healthcare component of the Consumer Price Index. Time discount rate Future healthcare costs and life-years were adjusted based on the assumption that people value a given amount of money higher in the present than in the future. Although discount rates vary within published data, recent recommendations for cost effectiveness analysis made by The Panel on Cost-Effectiveness in Health and Medicine include a base case discount rate of 3%,22 which has been adopted in this study. Model calibration Because Provenzale et al.8 published the results of a study that included cost and effectiveness measures for screening of patients with established BE, results from a simplified version of our model were compared with their data. After adjusting our cost estimates to 1995 dollars and calibrating several parameters in our model, similar results were obtained for both cost and effectiveness for 314
GASTROINTESTINAL ENDOSCOPY
annual screening of patients with BE. Variables modified by the calibration process included risk of death during esophagectomy and postesophagectomy mortality. Sensitivity analyses and model simulations Because of uncertainty in the assumptions made in our model, sensitivity analyses were performed on the key variables in Table 1. This method helps identify variables that have the greatest impact on the QALYs associated with each screening strategy. For example, examining how an extremely low annual cost (i.e., $50) associated with proton pump inhibitors (PPIs) impacted the model helps provide an understanding of the results in the context of generic PPIs becoming available. In addition to the sensitivity analyses, for each screening strategy clinical outcomes for hypothetical cohorts of 100,000 patients were determined by using Monte Carlo simulations. These simulations use the given information (e.g., probabilities, costs) to predict the future clinical course for the hypothetical patients. With each strategy, it was thus possible to determine how many patients developed cancer and the stage distribution of those who develop cancer.
RESULTS Baseline scenario The outcomes of modeling the 3 screening strategies are shown in Table 3. Under a strategy of no screening, the baseline assumptions used in the model indicate that 882 cases of esophageal adenocarcinoma would be expected over the lifetimes of 100,000 patients 50 years of age with GERD, with 266 cases having distant disease. Under the UTE strategy, only 431 cases of esophageal adenocarcinoma would be expected, with 26 cases of distant disease, and under the strategy of screening with SE, 396 cases of esophageal cancer would be expected including 13 with distant disease. Screening with UTE would provide an average gain in life expectancy of 0.126 months compared with a strategy of no screening, whereas SE would provide a gain in life expectancy of 0.130 months. Under the no-screening strategy, total costs associated with treatment for 100,000 patients with chronic GERD would be approximately $1.18 billion. The average discounted life expectancy for an average 50-year-old patient with GERD would be 19.327 years. Screening with UTE would increase the total cost estimate to $1.21 billion and incur a net gain of 0.0060 discounted QALYs, yielding a marginal CE ratio of $55,764 per QALY. Screening with SE would cost $1.23 billion and provide a net gain of 0.0063 discounted QALYs when compared with the no-screening strategy, yielding a marginal CE ratio of $86,883 per QALY. However, compared with the UTE strategy, screening with SE provides VOLUME 57, NO. 3, 2003
Chronic gastroesophageal reflux: cost-effectiveness of population screening
P Nietert, M Silverstein, M Mokhashi, et al.
Table 3. Outcome of screening strategies to prevent esophageal adenocarcinoma in hypothetical cohorts of 100,000 persons 50 years of age who were followed until death Outcome measure Total adenocarcinomas prevented Distant cancers prevented Total number of resections performed Average increase in life expectancy QALYs saved* Costs* Prescription drugs for GERD UTE SE Surgery Palliative care Other Total cost of care*
No screening
Standard endoscopy
0 0 234 0 mo 0 $1,153,200,000 $0 $1,600,000 $14,700,000 $6,500,000 $2,500,000 $1,178,500,000
Ultrathin endoscopy
486 253 963 0.130 mo 630
(97.9%) (0.0%) (0.1%) (1.2%) (0.6%) (0.2%) (100.0%)
$1,157,400,000 $0 $38,700,000 $25,200,000 $3,200,000 $8,700,000 $1,233,200,000
(93.9%) (0.0%) (3.1%) (2.0%) (0.3%) (0.7%) (100.0%)
451 240 926 0.126 mo 600 $1,157,200,000 $8,800,000 $9,500,000 $24,700,000 $3,400,000 $8,400,000 $1,211,900,000
(95.5%) (0.7%) (0.8%) (2.0%) (0.3%) (0.7%) (100.0%)
QALY, Quality-adjusted life-year; SE, standard endoscopy; UTE, ultrathin battery-powered endoscopy. *Future quality-adjusted life-years saved and future costs were discounted at an annual rate of 3%.
only 0.0003 additional QALYs, resulting in a marginal CE ratio of $709,260 per QALY. The cost effectiveness comparisons are summarized in Table 4.
Table 4. Summary of the comparison of the average cost and effectiveness associated with each strategy
Strategy
Average cost per person*
Average effectiveness per person*
Marginal cost/effectiveness ratio
No screening UTE SE
$11,785 $12,119 $12,332
19.3266 QALYs 19.3326 QALYs 19.3329 QALYs
— $55,764/QALY† $709,260/QALY‡ $86,883/QALY†
Sensitivity analyses Figure 2 demonstrates how the marginal CE ratios decrease with increasing incidence rates of esophageal adenocarcinoma among patients with BE. If the incidence is 1 per 100 patient-years, as has been reported for patients with long-segment BE,23 both the UTE and SE strategies would provide significant gains in QALYs at relatively low marginal costs. Compared with no screening, the marginal CE ratio for the UTE strategy was $15,265 per QALY, whereas the marginal CE ratio for the SE strategy compared with no screening was $24,617. The incremental CE ratio comparing SE with UTE in this case was $211,197. However, if the incidence of esophageal adenocarcinoma among patients with BE is 1 per 400 patient-years, then both the UTE and SE screening strategies actually result in decreases in QALYs, primarily because of complications and decreased QOL associated with esophagectomy. Thus, in this instance, the no-screening strategy would dominate (provide greater effectiveness and lower costs than) the UTE and SE screening strategies. When the cost of administering the UTE strategy was allowed to vary between $50 and $300, the marginal CE ratio comparing the UTE and no screening strategies increased somewhat ($48,861 to $86,474 per QALY); however, if the cost of the UTE administration exceeded $330, the SE screening strategy became the most cost-effective strategy. Thus, the cost of the UTE must be minimized for this strategy to remain cost-effective. As the cost of administering the SE was allowed to increase relative to the cost of VOLUME 57, NO. 3, 2003
*Future quality-adjusted life-years saved and future costs were discounted at an annual rate of 3%. †Compared with the no-screening strategy. ‡Compared with the UTE strategy.
administering the UTE, the incremental CE ratio comparing these strategies increased as well. Another key variable that influenced the marginal CE ratios was the QOL utility chosen for those receiving esophagectomy. Table 3 shows that many more esophagectomies occur under the UTE and SE strategies than under a no-screening strategy because of the identification of patients with high-grade dysplasia. Although the QOL utility for esophagectomy does not affect the costs associated with any of these strategies, it has a direct impact on the expected QALYs. The baseline assumption was that esophagectomy only slightly diminished QOL for an individual, with the utility of 0.97 being obtained from published results of a time trade-off study of patients having received esophagectomy. If this value is allowed to decrease to 0.90, the CE ratios for the UTE and SE strategies are both relatively high (respectively, $146,963/QALY and $233,915/QALY) compared with the no-screening strategy. If this value decreases below 0.86, then the UTE and SE strategies are both dominated by the no-screening strategy. GASTROINTESTINAL ENDOSCOPY
315
P Nietert, M Silverstein, M Mokhashi, et al.
Chronic gastroesophageal reflux: cost-effectiveness of population screening
This was due to the fact that our model assumed that all patients were treated with drugs, even those in the no screening arm. So as the drug costs varied, the total costs for each strategy changed similarly across the 3 arms of the model. Last, because the discount rate used in the model was allowed to vary from 0% to 10%, the CE ratio for UTE compared with no screening did increase from $46,300 per QALY to $113,579 per QALY. DISCUSSION
Figure 2. Sensitivity analysis of the incidence of esophageal adenocarcinoma among patients with Barrett’s esophagus, comparing UTE and SE with the no-screening strategy. For incidence rates of 1 case per 400 and 1 case per 350 patient-years, the marginal CE ratios are undefined because the no-screening strategy dominates both the UTE and the SE strategy.
As the probability associated with death during SE was increased to 1 per 1000, the no screening and UTE screening strategies were both still preferable to the SE screening strategy. In this case, the marginal CE ratio comparing the UTE strategy with no screening was $138,435 per QALY saved. Although the prevalence of undiagnosed BE in the cohort did impact the overall numbers of cancers and esophageal resections, its influence on the marginal CE ratio for screening with the UTE was minimal. For example, when the prevalence of undiagnosed BE was assumed to be 10%, almost 3 times as many cancers were prevented (UTE, 1257; SE, 1329) and many more resections were necessary (UTE, 3086; SE, 3209; no screening, 778). The marginal CE ratio comparing the SE strategy with the UTE strategy was reduced substantially, from $709,260/QALY to $238,782/QALY; however, the marginal CE ratio for the UTE strategy was only somewhat diminished from baseline (from $55,764/QALY to $43,353/QALY). Other sensitivity analyses yielded some notable results. For example, the probability that a patient would agree to undergo either UTE or SE did not impact the marginal CE ratios. This was due to the fact that as the percentage of patients agreeing to undergo the procedures varied, the costs and QALYs incurred by those patients changed in direct proportion to one another. Only if more than 90% of patients refused UTE would UTE no longer be considered a cost-effective strategy. Another set of parameters that had no measurable impact on the CE ratios was the cost of pharmacotherapy for GERD. 316
GASTROINTESTINAL ENDOSCOPY
This study finds that screening patients with chronic GERD for BE is cost-effective and comparable with other commonly accepted medical procedures such as hemodialysis,24 screening for colorectal cancer,25 and mammography.7,8,26 Screening with UTE, which generally does not require sedation, is more cost effective than SE with sedation, despite the loss of some optical quality and the inability to obtain biopsy specimens. Endoscopic screening of patients with GERD for BE and surveillance of patients with known BE is widely practiced and is recommended by the American College of Gastroenterology.6 There is currently no prospective randomized controlled trial of screening or surveillance of patients with BE. It is estimated that such a trial would require at least 5000 patients and more than 10 years of follow-up7 to demonstrate efficacy. In the absence of such data, multiple decision analysis studies have suggested that surveillance of patients with known BE is cost effective. These studies have focused on the surveillance of patients with known BE, not on the screening for BE itself. It has been estimated from autopsy studies that there are 20 patients with undetected BE for every one patient known to have BE.5 Without broader detection of BE, surveillance strategies will have only minimal impact. Decision analysis studies suggest that screening for BE with standard endoscopic techniques among patients with chronic GERD may be cost effective.27 However, it is estimated that more than 1400 patients with reflux would need to be followed for 1 year to identify 1 case of cancer. This analysis incorporates new advances in endoscopic technology (UTE), the availability of nonphysician endoscopists to perform screening, and current, lower estimates of the incidence of carcinoma in patients with BE. Thin-caliber endoscopes (38 mm) allow for unsedated, and thus less expensive endoscopy with high levels of patient tolerance,11,12,16,28 and are highly accurate for the detection of BE.29 These ultrathin endoscopes can be passed transnasally with only topical anesthesia by trained allied health professionals and do not VOLUME 57, NO. 3, 2003
Chronic gastroesophageal reflux: cost-effectiveness of population screening
require expensive external equipment such as video monitors and electronic light, or air and water sources, thus allowing community-based screening. There are several limitations to our analysis. Using UTE for widespread screening of BE assumes that the subsequent surveillance of BE will be beneficial. As discussed above, the benefit of BE surveillance is only indirectly suggested by retrospective studies and decision analyses. These models are highly dependent on their assumptions, especially the incidence of cancer development in patients with BE. There is considerable variation in the estimates of cancer incidence in patients with BE, ranging from 1/52 to 1/441 patient-years.23,30-35 Our screening model assumed a rate of 1/227 patientyears, identical to the model of Provenzale et al.,8 which demonstrated surveillance to be cost effective. The surveillance component of our model also had a cost-effectiveness comparable to that of the model of Provenzale et al.8 If more recent estimates suggesting a higher incidence rate (1/100 patient years)34,35 are accurate, this would make screening and surveillance even more cost effective. Our model is also based on one-time screening at age 50 years; incident cases of BE developing after age 50 would not be captured. However, Cameron and Lomboy36 have shown that the prevalence of BE plateaus in the sixth decade of life with only a 1% increase per year in subsequent decades. Increasing the starting age for screening to 60 years would potentially capture more cases of BE but would miss malignancies arising between age 50 and 60 years. Our model also made several simplifying assumptions; for example, patients with nondysplastic Barrett’s or low-grade dysplasia (LGD) were followed with surveillance every other year. Recommendations and practice patterns for the surveillance of patients with LGD vary considerably, and the short-term incidence of progression from LGD to HGD or carcinoma is poorly understood. Last, the present study evaluates direct costs from the prospective of a third party payer. Many other indirect costs are also relevant such as time away from work, and (for SE) time expended by a partner to escort patients to and from the endoscopy procedure. Many other indirect benefits of early detection also exist including more lifeyears in which income can be earned and time spent with family. Although all of these are important, and would likely further favor screening with UTE, they are beyond the scope of this analysis. Several questions remain regarding the application of UTE for general screening of patients for BE. The optimal trade-offs of endoscope size, accuracy, and patient tolerance are unknown. Smaller-caliber endoscopes (3-8 mm) are generally better tolerated VOLUME 57, NO. 3, 2003
P Nietert, M Silverstein, M Mokhashi, et al.
than standard endoscopes (9-13 mm), but may have lower optical quality. Endoscopic procurement of biopsy specimens to confirm BE can currently be performed with endoscopes larger than 5 mm. Thus patients screened with endoscopes less than 5 mm in diameter who have characteristic endoscopic findings of BE (salmon-colored mucosa in the distal esophagus), must still undergo conventional endoscopy to obtain specimens. Our model has incorporated the additional cost and risk of converting to SE whenever BE is suspected. Technologic advances such as smaller video charged-coupled devices (CCDs), brush cytology, and optical methods of tissue analysis37,38 may overcome these obstacles. Who will perform population wide screening remains to be answered. Allied health professionals can perform flexible sigmoidoscopy with accuracy rates similar to those of gastroenterologists39,40 and can likely be trained to perform upper endoscopy. In conclusion, the present decision analysis model demonstrates that the use of unsedated UTE may be a cost-effective method for population-based screening of patients with chronic GERD for the presence of BE. Further study is needed to determine the optimal types of endoscopes for screening and patient acceptance of screening in a broad population. REFERENCES 1. Blot WJ, Devesa SS, Fraumeni JF Jr. Continuing climb in rates of esophageal adenocarcinoma: an update. JAMA 1993;270:1320. 2. Antonioli D. The esophagus. In: Henson D, Alobores-Saavdera J, editors. The pathology of incipient neoplasia. Philadelphia: WB Saunders; 1993. p. 64-84. 3. Lagergren J, Bergstrom R, Lindgren A, Nyren O. Symptomatic gastroesophageal reflux as a risk factor for esophageal adenocarcinoma. N Engl J Med 1999;340:825-31. 4. Streitz JM Jr, Andrews CW Jr, Ellis FH Jr. Endoscopic surveillance of Barrett’s esophagus. Does it help? J Thorac Cardiovasc Surg 1993;105:383-8. 5. Cameron AJ, Zinsmeister AR, Ballard DJ, Carney JA. Prevalence of columnar-lined (Barrett’s) esophagus. Comparison of population-based clinical and autopsy findings. Gastroenterology 1990;99:918-22. 6. Sampliner RE. Practice guidelines on the diagnosis, surveillance, and therapy of Barrett’s esophagus. The Practice Parameters Committee of the American College of Gastroenterology. Am J Gastroenterol 1998;93:1028-32. 7. Provenzale D, Kemp JA, Arora S, Wong JB. A guide for surveillance of patients with Barrett’s esophagus. Am J Gastroenterol 1994;89:670-80. 8. Provenzale D, Schmitt C, Wong JB. Barrett’s esophagus: a new look at surveillance based on emerging estimates of cancer risk. Am J Gastroenterol 1999;94:2043-53. 9. Corley DA, Levin TR, Habel LA, Weiss NS, Buffler PA. Surveillance and survival in Barrett’s adenocarcinomas: a population-based study. Gastroenterology 2002;122:633-40. 10. De Gregorio BT, Poorman JC, Katon RM. Peroral ultrathin endoscopy in adult patients. Gastrointest Endosc 1997;45: 303-6. GASTROINTESTINAL ENDOSCOPY
317
P Nietert, M Silverstein, M Mokhashi, et al.
Chronic gastroesophageal reflux: cost-effectiveness of population screening
11. Zaman A, Hapke R, Sahagun G, Katon RM. Unsedated peroral endoscopy with a video ultrathin endoscope: patient acceptance, tolerance, and diagnostic accuracy. Am J Gastroenterol 1998;93:1260-3. 12. Zaman A, Hahn M, Hapke R, Knigge K, Fennerty MB, Katon RM. A randomized trial of peroral versus transnasal unsedated endoscopy using an ultrathin videoendoscope. Gastrointest Endosc 1999;49:279-84. 13. Sorbi D, Gostout CJ, Henry J, Lindor KD. Unsedated smallcaliber esophagogastroduodenoscopy (EGD) versus conventional EGD: a comparative study. Gastroenterology 1999;117: 1301-7. 14. Dumortier J, Ponchon T, Scoazec JY, Moulinier B, Zarka F, Paliard P, et al. Prospective evaluation of transnasal esophagogastroduodenoscopy: feasibility and study on performance and tolerance. Gastrointest Endosc 1999;49:285-91. 15. Craig A, Hanlon J, Dent J, Schoeman M. A comparison of transnasal and transoral endoscopy with small-diameter endoscopes in unsedated patients. Gastrointest Endosc 1999;49:292-6. 16. Dean R, Dua K, Massey B, Berger W, Hogan WJ, Shaker R. A comparative study of unsedated transnasal esophagogastroduodenoscopy and conventional EGD. Gastrointest Endosc 1996;44:422-4. 17. SEER (Surveillance, Epidemiology, and End Results) Cancer Incidence Public-Use Database, 1973-1997: August 1999 Submission. Volume 2000. Bethesda (MD): National Cancer Institute; 2000. 18. Gold MR, Franks P, McCoy KI, Fryback DG. Toward consistency in cost-utility analyses: using national measures to create condition-specific values. Med Care 1998;36:778-92. 19. Nissen DE. Mosby’s drug consult. St. Louis: Mosby; 2002. 20. Drug Topics Red Book. Oradell (NJ): Medical Economics Co; 1999. 21. Schmitt C, Provenzale D, Brazer S, Davis D, Phillips B, Wolfe W. Quality of life in esophageal disorders [abstract]. Gastroenterology 1994;106:A26. 22. Gold M, Siegel J, Russell L, Weinstein M. Cost-effectiveness in health and medicine. Oxford: Oxford University Press; 1996. 23. Williamson WA, Ellis FH, Jr., Gibb SP, Shahian DM, Aretz HT, Heatley GJ, et al. Barrett’s esophagus. Prevalence and incidence of adenocarcinoma. Arch Intern Med 1991;151: 2212-6. 24. Lysaght MJ. Maintenance dialysis population dynamics: current trends and long-term implications. J Am Soc Nephrol 2002;13(Suppl 1):S37-40. 25. Winawer SJ, Fletcher RH, Miller L, Godlee F, Stolar MH, Mulrow CD, et al. Colorectal cancer screening: clinical guidelines and rationale. Gastroenterology 1997;112:594-642. 26. Wright TA, Gray MR, Morris AI, Gilmore IT, Ellis A, Smart HL, et al. Cost effectiveness of detecting Barrett’s cancer. Gut 1996;39:574-9. 27. Soni A, Sampliner RE, Sonnenberg A. Screening for highgrade dysplasia in gastroesophageal reflux disease: is it costeffective? Am J Gastroenterol 2000;95:2086-93. 28. Shaker R. Unsedated trans-nasal pharyngoesophagogastroduodenoscopy (T-EGD): technique. Gastrointest Endosc 1994;40:346-8. 29. Mokhashi M, Glenn T, Jost C, Walace MB, Kim CY, Palesch
318
GASTROINTESTINAL ENDOSCOPY
30.
31.
32.
33. 34.
35.
36.
37.
38.
39.
40. 41.
42.
43.
44.
45.
YY, et al. Can a 3.1 mm stand-alone battery powered esophagoscope screen the esophagus for esophagitis and Barrett’s? A prospective blinded comparison with a standard videoendoscope [abstract]. Gastrointest Endosc 2000; 51:AB274. Cameron AJ, Ott BJ, Payne WS. The incidence of adenocarcinoma in columnar-lined (Barrett’s) esophagus. N Engl J Med 1985;313:857-9. Robertson CS, Mayberry JF, Nicholson DA, James PD, Atkinson M. Value of endoscopic surveillance in the detection of neoplastic change in Barrett’s oesophagus. Br J Surg 1988;75:760-3. Hameeteman W, Tytgat GN, Houthoff HJ, van den Tweel JG. Barrett’s esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989;96:1249-56. Ovaska J, Miettinen M, Kivilaakso E. Adenocarcinoma arising in Barrett’s esophagus. Dig Dis Sci 1989;34:1336-9. Sharma P, Morales TG, Bhattacharyya A, Garewal HS, Sampliner RE. Dysplasia in short-segment Barrett’s esophagus: a prospective 3-year follow-up. Am J Gastroenterol 1997;92:2012-6. Drewitz DJ, Sampliner RE, Garewal HS. The incidence of adenocarcinoma in Barrett’s esophagus: a prospective study of 170 patients followed 4.8 years. Am J Gastroenterol 1997;92:212-5. Cameron AJ, Lomboy CT. Barrett’s esophagus: age, prevalence, and extent of columnar epithelium. Gastroenterology 1992;103:1241-5. Wallace M, Van Dam J. Enhanced gastrointestinal diagnosis: light-scattering spectroscopy and optical coherence tomography. Gastrointest Endosc Clin N Am 2000;10:71-80. Wallace MB, Perelman LT, Backman V, Crawford JM, Fitzmaurice M, Seiler M, et al. Endoscopic detection of dysplasia in patients with Barrett’s esophagus using light scattering spectroscopy. Gastroenterology 2000;119:677-82. Wallace MB, Kemp JA, Meyer F, Horton K, Reffel A, Christiansen CL, et al. Screening for colorectal cancer with flexible sigmoidoscopy by nonphysician endoscopists. Am J Med 1999;107:214-8. Maule WF. Screening for colorectal cancer by nurse endoscopists. N Engl J Med 1994;330:183-7. Locke GR 3rd, Talley NJ, Fett SL, Zinsmeister AR, Melton LJ 3rd. Prevalence and clinical spectrum of gastroesophageal reflux: a population-based study in Olmsted County, Minnesota. Gastroenterology 1997;112:1448-56. Cameron AJ. Epidemiology of columnar-lined esophagus and adenocarcinoma. Gastroenterol Clin North Am 1997;26:48794. O’Connor JB, Falk GW, Richter JE. The incidence of adenocarcinoma and dysplasia in Barrett’s esophagus: report on the Cleveland Clinic Barrett’s Esophagus Registry. Am J Gastroenterol 1999;94:2037-42. Katz D, Rothstein R, Schned A, Dunn J, Seaver K, Antonioli D. The development of dysplasia and adenocarcinoma during endoscopic surveillance of Barrett’s esophagus. Am J Gastroenterol 1998;93:536-41. Reid BJ, Levine DS, Longton G, Blount PL, Rabinovitch PS. Predictors of progression to cancer in Barrett’s esophagus: baseline histology and flow cytometry identify low- and highrisk patient subsets. Am J Gastroenterol 2000;95:1669-76.
VOLUME 57, NO. 3, 2003
Background: Persons with chronic esophageal reflux are at increased risk for the development of Barrett’s esophagus and adenocarcinoma. Recently developed ultrathin endoscopes are less expensive and better tolerated than standard endoscopes, they can be used without sedation, and are sensitive and specific for Barrett’s esophagus. The cost-effectiveness of one-time screening strategies were evaluated for 50-year-old patients with chronic reflux: no screening, standard endoscopy, and screening by an ultrathin endoscope. Methods: Markov models were created to simulate the clinical course for patients with chronic reflux. Costs and quality-adjusted life-years were estimated from cancer registry data, published medical data, and expert opinion. Results: Under baseline assumptions, no screening resulted in average costs of $11,785 per person and 19.3226 quality-adjusted life-years. Ultrathin endoscopy screening resulted in costs of $12,119 per person and 19.3326 quality-adjusted life-years, yielding a marginal cost-effectiveness ratio of $55,764 per quality-adjusted life-year. Using standard endoscopy yielded costs of $12,332 with only slightly greater effectiveness, yielding a marginal cost-effectiveness ratio of $709,260 when compared with ultrathin endoscopy and $86,833 compared with no screening. Results were most sensitive to variation in the incidence of cancer in the population with Barrett’s esophagus. Conclusions: Screening for Barrett’s esophagus with ultrathin endoscopy is more cost-effective than standard endoscopy, and both strategies appear to improve quality-adjusted life-years among patients with chronic reflux at costs that are similar to those of other accepted preventive measures. (Gastrointest Endosc 2003;57:311-8.)
The incidence of adenocarcinoma of the esophagus is increasing in the United States.1 Almost all of these cancers occur in patients with Barrett’s esophagus (BE),2 a condition in which metaplastic columnar epithelium replaces the normal squamous epithelium of the esophagus. The presence of BE and chronic GERD are the most significant identifiable risk factors for the development of adenocarcinoma of the esophagus. Patients with chronic GERD are 7 to 44 times more likely to develop adenocarcinoma of the esophagus than patients without reflux.3 Surveillance of patients with known BE is associated with earlier detection of adenocarcinoma Received June 11, 2002. For revision August 27, 2002. Accepted October 30, 2002. Current affiliations: Center for Health Care Research, Medical University of South Carolina, Department of Medicine, Division of General Internal Medicine, Medical University of South Carolina, Digestive Diseases Center, Division of Gastroenterology and Hepatology, Medical University of South Carolina, and Department of Veterans Affairs Medical Center, Charleston, South Carolina. Funded, in part, by a grant from the U.S. Office of Naval Research (No. N00014-99-1-0784) to the Medical University of South Carolina. Dr. Wallace was funded by the American Digestive Health Foundation (TAP Pharmaceutical Outcomes Research Award). Reprint requests: Paul J. Nietert, PhD, Center for Health Care Research, 135 Cannon St., Suite 403, P.O. Box 250837, Charleston, SC 29425. 0016-5107/2003/$30.00 + 0 doi:10.1067/mge.2003.101 VOLUME 57, NO. 3, 2003
and improved survival.4 However, autopsy series suggest less than 5% of patients with BE are currently under surveillance.5 Screening and surveillance of BE is widely practiced and recommended by the professional societies concerned with digestive diseases,6 but no prospective efficacy data are available. Furthermore, decision analysis studies have primarily addressed surveillance strategies after BE has been diagnosed.7,8 The vast majority of patients with BE and adenocarcinoma have never been diagnosed, and the majority of those presenting with adenocarcinoma have not been under surveillance.5,9 If surveillance is to be effective, a program of widespread screening is necessary to identify at risk cases. Major limitations of widespread screening are cost and patient tolerance of upper endoscopy. Endoscopic examination with ultrathin endoscopes as small as 3.1 mm in diameter has recently been developed and has been shown to be well tolerated and potentially less expensive than standard endoscopy (SE).10-16 The purpose of this study was to evaluate the costeffectiveness of ultrathin, unsedated endoscopy (UTE) compared with standard endoscopic screening and no screening for the detection and surveillance of BE. BE was defined in this study as any length of intestinal metaplasia within the tubular esophagus. GASTROINTESTINAL ENDOSCOPY
311
P Nietert, M Silverstein, M Mokhashi, et al.
Chronic gastroesophageal reflux: cost-effectiveness of population screening
Figure 1. Cost-effectiveness model comparing 3 strategies of screening for Barrett’s esophagus. The “M” symbolizes the beginning of a Markov model. GERD, Gastroesophageal reflux disease; UTE, ultrathin endoscopy; SE, standard endoscopy; LGD, low-grade dysplasia; HGD, high-grade dysplasia. PATIENTS AND METHODS The cost-effectiveness (CE) analysis was designed to determine the average lifetime-costs and average lifetime quality-adjusted life-years (QALYs) associated with each of 3 strategies: no screening, screening with UTE, and screening with SE. In these analyses, all patients are assumed to have GERD, defined for the purposes of the study as heartburn and/or acid reflux at least once per week. For all strategies, patients identified as having BE with no or low-grade dysplasia are assumed to be in a surveillance program, undergoing SE every other year. The strategies of screening with UTE and of screening with SE were compared against a strategy of no screening. In addition, the UTE and SE strategies were compared against one another. The analysis was performed with statistical software (Data 3.5.6, Treeage Software, Williamstown, Mass.) and the overall decision analysis model is shown in Figure 1. In both the UTE and SE strategies, 5% of patients are assumed to refuse the procedure, and for these patients costs and effectiveness are assumed to be the same as those in the no screening arm. Markov models The clinical course of disease for 50-year-old patients with chronic GERD was simulated by using Markov models, whereby the annual changes in disease states for patients were simulated from transition probabilities based only on the last known disease state of the patient, independent of history. Estimates of the annual transitions among the mutually exclusive disease states used data from a variety of sources. Key variables and sources used in these models are listed in Table 1. This type of study is exempt from Institutional Review Board review because no human subjects were involved. 312
GASTROINTESTINAL ENDOSCOPY
The “no screening” strategy: This strategy is meant to depict the traditional course of treatment for patients with GERD. Of the patients in this strategy, 3% are assumed to have undiagnosed BE, whereas the rest are considered to be in a clinical state free of BE and any cancer. Patients with BE in the model progress into symptomatic states, defined as having dysphagia and/or upper GI bleeding, at the rate of approximately 1.5% per year, a rate based on our experience that corresponds to about 1 in 6 patients who eventually develop symptoms within 10 years. Patients with symptoms undergo SE, assumed to be the reference standard, and both symptomatic and asymptomatic patients with BE may progress to esophageal adenocarcinoma. Patients free of BE remain free of BE, although they progress to cancer at substantially lower rates than BE patients. All patients identified with high-grade dysplasia (HGD) or with local or regional cancer undergo esophagectomy. Median life expectancy data regarding increased life expectancy for patients undergoing esophagectomy were obtained from our own analyses of the Surveillance, Epidemiology, and End Results (SEER) registry public use data17 (see Table 1). Patients identified with metastatic disease receive palliative care alone. All patients in noncancerous disease states receive pharmacologic treatment for GERD. Half of patients taking medication receive histamine-2 receptor antagonists, and the rest proton-pump inhibitors. The UTE strategy. Of patients screened by UTE, a certain proportion will be identified as having a columnarlined esophagus. The accuracy of this classification is a direct function of the population prevalence and the diagnostic test characteristics (sensitivity and specificity) of UTE (see Table 1). These patients then undergo further diagnostic testing with SE to confirm the presence of BE. As with the no screening strategy, patients ultimately identified with HGD or with local or regional cancer receive esophagectomy, and those identified with metastatic disease have palliative care alone. Patients who are not found to have BE or cancer receive only pharmacologic treatment. The SE strategy. In this strategy, all patients with chronic GERD are screened once with SE, which is assumed to be the reference standard for the diagnosis of BE. As with the other strategies, patients identified with HGD or with local or regional cancer undergo esophagectomy, and patients free of BE and cancer receive only pharmacologic treatment. Quality of life adjustment In this study the primary effectiveness outcome was quality-adjusted life-years (QALYs). Provenzale et al.8 reported the median utility for esophagectomy to be 0.97 (interquartile range 0.83-1.0), and this value was incorporated in the Markov states for anyone undergoing esophagectomy, with sensitivity analysis being performed on this measure. In the absence of direct reports of quality of life from patients with esophageal cancer, estimates of quality of life (QOL) utilities were based on published data obtained from the National Health Interview Survey. Using these national data, Gold et al.18 (1998) reported a QOL score of 0.47 for VOLUME 57, NO. 3, 2003
Chronic gastroesophageal reflux: cost-effectiveness of population screening
P Nietert, M Silverstein, M Mokhashi, et al.
Table 1. Key variables used in the Markov models Variable Prevalence of: GERD in general population Undiagnosed BE in GERD population Relative risk of developing EA given GERD Median life expectancy for EA Local disease Local disease with esophagectomy Regional disease Regional disease with esophagectomy Metastatic disease Incidence of EA among patients with BE
Relative risk of developing EA among patients with HGD compared with those with no dysplasia or LGD Annual rate of patients with GERD developing symptoms requiring SE Probability of having: LGD if patient has BE HGD if patient has BE Annual probability of: Developing LGD if patient has no dysplasia Developing HGD if patient has no dysplasia Developing HGD if patient has LGD Regressing to no dysplasia from LGD Stage distribution of EA: In unscreened population (local: regional: metastatic) In screened population (local: regional: metastatic) Sensitivity of UTE Specificity of UTE Probability of death from SE Cost of: Administration of UTE Administration of SE without biopsy Administration of SE with biopsy H2 receptor antagonist Proton pump inhibitor CT scan Esophagectomy Palliative care for metastatic disease Case after esophagectomy QOL utility for: Esophagectomy Local and regional EA Metastatic EA
Value
Range used in sensitivity analyses
Reference
19.8% 3.0% 9.68*
5%-25% 0%-15% 5.25-16.54
Locke, 199741 Cameron, 199742 Lagergren, 19993
1.37 y 3.68 y 0.93 y 1.13 y 0.37 y 1/227 life years
1.0-1.7 y 2.8-4.6 y 0.7-1.2 y 0.8-1.4 y 0.3-0.5 y 1/100-1/400 life years
28.0
13.0-63.0
SEER, 200017 SEER, 200017 SEER, 200017 SEER, 200017 SEER, 200017 Drewitz, 199735 O’Connor, 199943 Katz, 199844 Cameron, 198530 Reid, 200045
1.5%
0.5%-2.5%
Expert opinion
0.0735 0.00735
0.055-0.092 0.006-0.009
O’Connor, 199943 O’Connor, 199943
0.045 0.010 0.095 0.75
0.034-0.056 0.008-0.013 0.071-0.119 0.563-0.938
O’Connor, 199943 O’Connor, 199943 O’Connor, 199943 Expert opinion
40%: 30%: 30%
0.95 0.95 0 per 1000
20%: 30%: 50% 60%: 30%: 10%† 50%: 30%: 20% 90%: 10%: 0%† 0.75-1.00 0.75-1.00 0-1 per 1000
Mokhashi, 200029 Mokhashi, 200029 Expert opinion
$97 $346 $507 $74/y $727/y $544 $29,415 $40,798 $1236/y
$50-$300 $200-$600 $350-$750 $0-$1306 $50-$1568 $200-$800 $15,000-$40,000 $15,000-$50,000 $500-$2000
MUSC cost data MUSC cost data MUSC cost data Red Book, 199920 Red Book, 199920 MUSC cost data Provenzale, 19998 Provenzale, 19998 Provenzale, 19998
0.97 0.47 0.47
0.75-1.00 0.25-0.75 0.25-0.75
Provenzale, 19998 Gold, 199818 Gold, 199818
70%: 30%: 0%
SEER, 200017 Streitz, 19934
BE, Barrett’s esophagus; EA, esophageal adenocarcinoma; LGD, low-grade dysplasia; HGD, high-grade dysplasia; UTE, batterypowered endoscope; MUSC, Medical University of South Carolina; QOL, quality of life. *This relative risk is actually a weighted average of 3 separate relative risks as reported by Lagergren et al. (1999)3 associated with 3 categories of reflux frequency. †For sensitivity analyses involving stage distribution, the percent of patients diagnosed with regional disease was held constant, whereas the percent of patients with local and metastatic disease was allowed to vary.
patients with GI cancer. This value was incorporated in the Markov states of patients with late-stage cancer. Perspective and costs This analysis was performed from the prospective of a third-party payer. Where possible, national cost estimates VOLUME 57, NO. 3, 2003
or estimates used in similar decision models in other publications (e.g., Provenzale et al.8, 1999) were used in the model. Annual costs of pharmacologic treatment of patients with GERD were determined by obtaining a published list of average retail drug prices in the United States19 and subtracting a $250 annual deductible as well GASTROINTESTINAL ENDOSCOPY
313
P Nietert, M Silverstein, M Mokhashi, et al.
Chronic gastroesophageal reflux: cost-effectiveness of population screening
Table 2. Breakdown of UTE and SE cost estimates
Technical fee ($) Professional fee ($) Pathology technical fee ($) Pathology professional fee ($) Total ($1999) Total ($2001)
UTE
SE without biopsy
SE with biopsy
74 15 0 0 89 97
183 135 0 0 318 346
228 135 36 66 465 506
as a $22 per month co-pay for brand-name medications and $7 per month for generic medications. The 1999 Drug Topics Red Book of average wholesale drug prices was used to determine the upper limit of drug costs used in the sensitivity analyses.20 Direct and indirect costs associated with administration of the UTE and SE (with and without biopsy) were determined from the Medical University of South Carolina financial cost accounting data. Technical UTE costs were based on the estimated per-procedure cost of the instrument depreciated over its expected lifetime, and all costs related to cleaning of the instrument (disposable supplies, personnel). Because ultrathin endoscopes have no accessory channel or automatic air/water channels, they can be disinfected by soaking in a standard endoscopy cleaning solution and do not require automated washers, thus further reducing the cost of service. A breakdown of the cost estimates used in the analysis is shown in Table 2. Provider costs including durable and disposable equipment were calculated from actual costs to perform the procedures. Third party reimbursements were based on South Carolina Medicare and commercial health maintenance organizations (Blue Cross) for esophagoscopy (UTE) and upper endoscopy with biopsy (standard endoscopy). Cost of esophagectomy, postesophagectomy health care, and palliative care for metastatic disease and health state utilities were obtained from published data.8,21 All costs were adjusted to 2001 dollars by using the healthcare component of the Consumer Price Index. Time discount rate Future healthcare costs and life-years were adjusted based on the assumption that people value a given amount of money higher in the present than in the future. Although discount rates vary within published data, recent recommendations for cost effectiveness analysis made by The Panel on Cost-Effectiveness in Health and Medicine include a base case discount rate of 3%,22 which has been adopted in this study. Model calibration Because Provenzale et al.8 published the results of a study that included cost and effectiveness measures for screening of patients with established BE, results from a simplified version of our model were compared with their data. After adjusting our cost estimates to 1995 dollars and calibrating several parameters in our model, similar results were obtained for both cost and effectiveness for 314
GASTROINTESTINAL ENDOSCOPY
annual screening of patients with BE. Variables modified by the calibration process included risk of death during esophagectomy and postesophagectomy mortality. Sensitivity analyses and model simulations Because of uncertainty in the assumptions made in our model, sensitivity analyses were performed on the key variables in Table 1. This method helps identify variables that have the greatest impact on the QALYs associated with each screening strategy. For example, examining how an extremely low annual cost (i.e., $50) associated with proton pump inhibitors (PPIs) impacted the model helps provide an understanding of the results in the context of generic PPIs becoming available. In addition to the sensitivity analyses, for each screening strategy clinical outcomes for hypothetical cohorts of 100,000 patients were determined by using Monte Carlo simulations. These simulations use the given information (e.g., probabilities, costs) to predict the future clinical course for the hypothetical patients. With each strategy, it was thus possible to determine how many patients developed cancer and the stage distribution of those who develop cancer.
RESULTS Baseline scenario The outcomes of modeling the 3 screening strategies are shown in Table 3. Under a strategy of no screening, the baseline assumptions used in the model indicate that 882 cases of esophageal adenocarcinoma would be expected over the lifetimes of 100,000 patients 50 years of age with GERD, with 266 cases having distant disease. Under the UTE strategy, only 431 cases of esophageal adenocarcinoma would be expected, with 26 cases of distant disease, and under the strategy of screening with SE, 396 cases of esophageal cancer would be expected including 13 with distant disease. Screening with UTE would provide an average gain in life expectancy of 0.126 months compared with a strategy of no screening, whereas SE would provide a gain in life expectancy of 0.130 months. Under the no-screening strategy, total costs associated with treatment for 100,000 patients with chronic GERD would be approximately $1.18 billion. The average discounted life expectancy for an average 50-year-old patient with GERD would be 19.327 years. Screening with UTE would increase the total cost estimate to $1.21 billion and incur a net gain of 0.0060 discounted QALYs, yielding a marginal CE ratio of $55,764 per QALY. Screening with SE would cost $1.23 billion and provide a net gain of 0.0063 discounted QALYs when compared with the no-screening strategy, yielding a marginal CE ratio of $86,883 per QALY. However, compared with the UTE strategy, screening with SE provides VOLUME 57, NO. 3, 2003
Chronic gastroesophageal reflux: cost-effectiveness of population screening
P Nietert, M Silverstein, M Mokhashi, et al.
Table 3. Outcome of screening strategies to prevent esophageal adenocarcinoma in hypothetical cohorts of 100,000 persons 50 years of age who were followed until death Outcome measure Total adenocarcinomas prevented Distant cancers prevented Total number of resections performed Average increase in life expectancy QALYs saved* Costs* Prescription drugs for GERD UTE SE Surgery Palliative care Other Total cost of care*
No screening
Standard endoscopy
0 0 234 0 mo 0 $1,153,200,000 $0 $1,600,000 $14,700,000 $6,500,000 $2,500,000 $1,178,500,000
Ultrathin endoscopy
486 253 963 0.130 mo 630
(97.9%) (0.0%) (0.1%) (1.2%) (0.6%) (0.2%) (100.0%)
$1,157,400,000 $0 $38,700,000 $25,200,000 $3,200,000 $8,700,000 $1,233,200,000
(93.9%) (0.0%) (3.1%) (2.0%) (0.3%) (0.7%) (100.0%)
451 240 926 0.126 mo 600 $1,157,200,000 $8,800,000 $9,500,000 $24,700,000 $3,400,000 $8,400,000 $1,211,900,000
(95.5%) (0.7%) (0.8%) (2.0%) (0.3%) (0.7%) (100.0%)
QALY, Quality-adjusted life-year; SE, standard endoscopy; UTE, ultrathin battery-powered endoscopy. *Future quality-adjusted life-years saved and future costs were discounted at an annual rate of 3%.
only 0.0003 additional QALYs, resulting in a marginal CE ratio of $709,260 per QALY. The cost effectiveness comparisons are summarized in Table 4.
Table 4. Summary of the comparison of the average cost and effectiveness associated with each strategy
Strategy
Average cost per person*
Average effectiveness per person*
Marginal cost/effectiveness ratio
No screening UTE SE
$11,785 $12,119 $12,332
19.3266 QALYs 19.3326 QALYs 19.3329 QALYs
— $55,764/QALY† $709,260/QALY‡ $86,883/QALY†
Sensitivity analyses Figure 2 demonstrates how the marginal CE ratios decrease with increasing incidence rates of esophageal adenocarcinoma among patients with BE. If the incidence is 1 per 100 patient-years, as has been reported for patients with long-segment BE,23 both the UTE and SE strategies would provide significant gains in QALYs at relatively low marginal costs. Compared with no screening, the marginal CE ratio for the UTE strategy was $15,265 per QALY, whereas the marginal CE ratio for the SE strategy compared with no screening was $24,617. The incremental CE ratio comparing SE with UTE in this case was $211,197. However, if the incidence of esophageal adenocarcinoma among patients with BE is 1 per 400 patient-years, then both the UTE and SE screening strategies actually result in decreases in QALYs, primarily because of complications and decreased QOL associated with esophagectomy. Thus, in this instance, the no-screening strategy would dominate (provide greater effectiveness and lower costs than) the UTE and SE screening strategies. When the cost of administering the UTE strategy was allowed to vary between $50 and $300, the marginal CE ratio comparing the UTE and no screening strategies increased somewhat ($48,861 to $86,474 per QALY); however, if the cost of the UTE administration exceeded $330, the SE screening strategy became the most cost-effective strategy. Thus, the cost of the UTE must be minimized for this strategy to remain cost-effective. As the cost of administering the SE was allowed to increase relative to the cost of VOLUME 57, NO. 3, 2003
*Future quality-adjusted life-years saved and future costs were discounted at an annual rate of 3%. †Compared with the no-screening strategy. ‡Compared with the UTE strategy.
administering the UTE, the incremental CE ratio comparing these strategies increased as well. Another key variable that influenced the marginal CE ratios was the QOL utility chosen for those receiving esophagectomy. Table 3 shows that many more esophagectomies occur under the UTE and SE strategies than under a no-screening strategy because of the identification of patients with high-grade dysplasia. Although the QOL utility for esophagectomy does not affect the costs associated with any of these strategies, it has a direct impact on the expected QALYs. The baseline assumption was that esophagectomy only slightly diminished QOL for an individual, with the utility of 0.97 being obtained from published results of a time trade-off study of patients having received esophagectomy. If this value is allowed to decrease to 0.90, the CE ratios for the UTE and SE strategies are both relatively high (respectively, $146,963/QALY and $233,915/QALY) compared with the no-screening strategy. If this value decreases below 0.86, then the UTE and SE strategies are both dominated by the no-screening strategy. GASTROINTESTINAL ENDOSCOPY
315
P Nietert, M Silverstein, M Mokhashi, et al.
Chronic gastroesophageal reflux: cost-effectiveness of population screening
This was due to the fact that our model assumed that all patients were treated with drugs, even those in the no screening arm. So as the drug costs varied, the total costs for each strategy changed similarly across the 3 arms of the model. Last, because the discount rate used in the model was allowed to vary from 0% to 10%, the CE ratio for UTE compared with no screening did increase from $46,300 per QALY to $113,579 per QALY. DISCUSSION
Figure 2. Sensitivity analysis of the incidence of esophageal adenocarcinoma among patients with Barrett’s esophagus, comparing UTE and SE with the no-screening strategy. For incidence rates of 1 case per 400 and 1 case per 350 patient-years, the marginal CE ratios are undefined because the no-screening strategy dominates both the UTE and the SE strategy.
As the probability associated with death during SE was increased to 1 per 1000, the no screening and UTE screening strategies were both still preferable to the SE screening strategy. In this case, the marginal CE ratio comparing the UTE strategy with no screening was $138,435 per QALY saved. Although the prevalence of undiagnosed BE in the cohort did impact the overall numbers of cancers and esophageal resections, its influence on the marginal CE ratio for screening with the UTE was minimal. For example, when the prevalence of undiagnosed BE was assumed to be 10%, almost 3 times as many cancers were prevented (UTE, 1257; SE, 1329) and many more resections were necessary (UTE, 3086; SE, 3209; no screening, 778). The marginal CE ratio comparing the SE strategy with the UTE strategy was reduced substantially, from $709,260/QALY to $238,782/QALY; however, the marginal CE ratio for the UTE strategy was only somewhat diminished from baseline (from $55,764/QALY to $43,353/QALY). Other sensitivity analyses yielded some notable results. For example, the probability that a patient would agree to undergo either UTE or SE did not impact the marginal CE ratios. This was due to the fact that as the percentage of patients agreeing to undergo the procedures varied, the costs and QALYs incurred by those patients changed in direct proportion to one another. Only if more than 90% of patients refused UTE would UTE no longer be considered a cost-effective strategy. Another set of parameters that had no measurable impact on the CE ratios was the cost of pharmacotherapy for GERD. 316
GASTROINTESTINAL ENDOSCOPY
This study finds that screening patients with chronic GERD for BE is cost-effective and comparable with other commonly accepted medical procedures such as hemodialysis,24 screening for colorectal cancer,25 and mammography.7,8,26 Screening with UTE, which generally does not require sedation, is more cost effective than SE with sedation, despite the loss of some optical quality and the inability to obtain biopsy specimens. Endoscopic screening of patients with GERD for BE and surveillance of patients with known BE is widely practiced and is recommended by the American College of Gastroenterology.6 There is currently no prospective randomized controlled trial of screening or surveillance of patients with BE. It is estimated that such a trial would require at least 5000 patients and more than 10 years of follow-up7 to demonstrate efficacy. In the absence of such data, multiple decision analysis studies have suggested that surveillance of patients with known BE is cost effective. These studies have focused on the surveillance of patients with known BE, not on the screening for BE itself. It has been estimated from autopsy studies that there are 20 patients with undetected BE for every one patient known to have BE.5 Without broader detection of BE, surveillance strategies will have only minimal impact. Decision analysis studies suggest that screening for BE with standard endoscopic techniques among patients with chronic GERD may be cost effective.27 However, it is estimated that more than 1400 patients with reflux would need to be followed for 1 year to identify 1 case of cancer. This analysis incorporates new advances in endoscopic technology (UTE), the availability of nonphysician endoscopists to perform screening, and current, lower estimates of the incidence of carcinoma in patients with BE. Thin-caliber endoscopes (38 mm) allow for unsedated, and thus less expensive endoscopy with high levels of patient tolerance,11,12,16,28 and are highly accurate for the detection of BE.29 These ultrathin endoscopes can be passed transnasally with only topical anesthesia by trained allied health professionals and do not VOLUME 57, NO. 3, 2003
Chronic gastroesophageal reflux: cost-effectiveness of population screening
require expensive external equipment such as video monitors and electronic light, or air and water sources, thus allowing community-based screening. There are several limitations to our analysis. Using UTE for widespread screening of BE assumes that the subsequent surveillance of BE will be beneficial. As discussed above, the benefit of BE surveillance is only indirectly suggested by retrospective studies and decision analyses. These models are highly dependent on their assumptions, especially the incidence of cancer development in patients with BE. There is considerable variation in the estimates of cancer incidence in patients with BE, ranging from 1/52 to 1/441 patient-years.23,30-35 Our screening model assumed a rate of 1/227 patientyears, identical to the model of Provenzale et al.,8 which demonstrated surveillance to be cost effective. The surveillance component of our model also had a cost-effectiveness comparable to that of the model of Provenzale et al.8 If more recent estimates suggesting a higher incidence rate (1/100 patient years)34,35 are accurate, this would make screening and surveillance even more cost effective. Our model is also based on one-time screening at age 50 years; incident cases of BE developing after age 50 would not be captured. However, Cameron and Lomboy36 have shown that the prevalence of BE plateaus in the sixth decade of life with only a 1% increase per year in subsequent decades. Increasing the starting age for screening to 60 years would potentially capture more cases of BE but would miss malignancies arising between age 50 and 60 years. Our model also made several simplifying assumptions; for example, patients with nondysplastic Barrett’s or low-grade dysplasia (LGD) were followed with surveillance every other year. Recommendations and practice patterns for the surveillance of patients with LGD vary considerably, and the short-term incidence of progression from LGD to HGD or carcinoma is poorly understood. Last, the present study evaluates direct costs from the prospective of a third party payer. Many other indirect costs are also relevant such as time away from work, and (for SE) time expended by a partner to escort patients to and from the endoscopy procedure. Many other indirect benefits of early detection also exist including more lifeyears in which income can be earned and time spent with family. Although all of these are important, and would likely further favor screening with UTE, they are beyond the scope of this analysis. Several questions remain regarding the application of UTE for general screening of patients for BE. The optimal trade-offs of endoscope size, accuracy, and patient tolerance are unknown. Smaller-caliber endoscopes (3-8 mm) are generally better tolerated VOLUME 57, NO. 3, 2003
P Nietert, M Silverstein, M Mokhashi, et al.
than standard endoscopes (9-13 mm), but may have lower optical quality. Endoscopic procurement of biopsy specimens to confirm BE can currently be performed with endoscopes larger than 5 mm. Thus patients screened with endoscopes less than 5 mm in diameter who have characteristic endoscopic findings of BE (salmon-colored mucosa in the distal esophagus), must still undergo conventional endoscopy to obtain specimens. Our model has incorporated the additional cost and risk of converting to SE whenever BE is suspected. Technologic advances such as smaller video charged-coupled devices (CCDs), brush cytology, and optical methods of tissue analysis37,38 may overcome these obstacles. Who will perform population wide screening remains to be answered. Allied health professionals can perform flexible sigmoidoscopy with accuracy rates similar to those of gastroenterologists39,40 and can likely be trained to perform upper endoscopy. In conclusion, the present decision analysis model demonstrates that the use of unsedated UTE may be a cost-effective method for population-based screening of patients with chronic GERD for the presence of BE. Further study is needed to determine the optimal types of endoscopes for screening and patient acceptance of screening in a broad population. REFERENCES 1. Blot WJ, Devesa SS, Fraumeni JF Jr. Continuing climb in rates of esophageal adenocarcinoma: an update. JAMA 1993;270:1320. 2. Antonioli D. The esophagus. In: Henson D, Alobores-Saavdera J, editors. The pathology of incipient neoplasia. Philadelphia: WB Saunders; 1993. p. 64-84. 3. Lagergren J, Bergstrom R, Lindgren A, Nyren O. Symptomatic gastroesophageal reflux as a risk factor for esophageal adenocarcinoma. N Engl J Med 1999;340:825-31. 4. Streitz JM Jr, Andrews CW Jr, Ellis FH Jr. Endoscopic surveillance of Barrett’s esophagus. Does it help? J Thorac Cardiovasc Surg 1993;105:383-8. 5. Cameron AJ, Zinsmeister AR, Ballard DJ, Carney JA. Prevalence of columnar-lined (Barrett’s) esophagus. Comparison of population-based clinical and autopsy findings. Gastroenterology 1990;99:918-22. 6. Sampliner RE. Practice guidelines on the diagnosis, surveillance, and therapy of Barrett’s esophagus. The Practice Parameters Committee of the American College of Gastroenterology. Am J Gastroenterol 1998;93:1028-32. 7. Provenzale D, Kemp JA, Arora S, Wong JB. A guide for surveillance of patients with Barrett’s esophagus. Am J Gastroenterol 1994;89:670-80. 8. Provenzale D, Schmitt C, Wong JB. Barrett’s esophagus: a new look at surveillance based on emerging estimates of cancer risk. Am J Gastroenterol 1999;94:2043-53. 9. Corley DA, Levin TR, Habel LA, Weiss NS, Buffler PA. Surveillance and survival in Barrett’s adenocarcinomas: a population-based study. Gastroenterology 2002;122:633-40. 10. De Gregorio BT, Poorman JC, Katon RM. Peroral ultrathin endoscopy in adult patients. Gastrointest Endosc 1997;45: 303-6. GASTROINTESTINAL ENDOSCOPY
317
P Nietert, M Silverstein, M Mokhashi, et al.
Chronic gastroesophageal reflux: cost-effectiveness of population screening
11. Zaman A, Hapke R, Sahagun G, Katon RM. Unsedated peroral endoscopy with a video ultrathin endoscope: patient acceptance, tolerance, and diagnostic accuracy. Am J Gastroenterol 1998;93:1260-3. 12. Zaman A, Hahn M, Hapke R, Knigge K, Fennerty MB, Katon RM. A randomized trial of peroral versus transnasal unsedated endoscopy using an ultrathin videoendoscope. Gastrointest Endosc 1999;49:279-84. 13. Sorbi D, Gostout CJ, Henry J, Lindor KD. Unsedated smallcaliber esophagogastroduodenoscopy (EGD) versus conventional EGD: a comparative study. Gastroenterology 1999;117: 1301-7. 14. Dumortier J, Ponchon T, Scoazec JY, Moulinier B, Zarka F, Paliard P, et al. Prospective evaluation of transnasal esophagogastroduodenoscopy: feasibility and study on performance and tolerance. Gastrointest Endosc 1999;49:285-91. 15. Craig A, Hanlon J, Dent J, Schoeman M. A comparison of transnasal and transoral endoscopy with small-diameter endoscopes in unsedated patients. Gastrointest Endosc 1999;49:292-6. 16. Dean R, Dua K, Massey B, Berger W, Hogan WJ, Shaker R. A comparative study of unsedated transnasal esophagogastroduodenoscopy and conventional EGD. Gastrointest Endosc 1996;44:422-4. 17. SEER (Surveillance, Epidemiology, and End Results) Cancer Incidence Public-Use Database, 1973-1997: August 1999 Submission. Volume 2000. Bethesda (MD): National Cancer Institute; 2000. 18. Gold MR, Franks P, McCoy KI, Fryback DG. Toward consistency in cost-utility analyses: using national measures to create condition-specific values. Med Care 1998;36:778-92. 19. Nissen DE. Mosby’s drug consult. St. Louis: Mosby; 2002. 20. Drug Topics Red Book. Oradell (NJ): Medical Economics Co; 1999. 21. Schmitt C, Provenzale D, Brazer S, Davis D, Phillips B, Wolfe W. Quality of life in esophageal disorders [abstract]. Gastroenterology 1994;106:A26. 22. Gold M, Siegel J, Russell L, Weinstein M. Cost-effectiveness in health and medicine. Oxford: Oxford University Press; 1996. 23. Williamson WA, Ellis FH, Jr., Gibb SP, Shahian DM, Aretz HT, Heatley GJ, et al. Barrett’s esophagus. Prevalence and incidence of adenocarcinoma. Arch Intern Med 1991;151: 2212-6. 24. Lysaght MJ. Maintenance dialysis population dynamics: current trends and long-term implications. J Am Soc Nephrol 2002;13(Suppl 1):S37-40. 25. Winawer SJ, Fletcher RH, Miller L, Godlee F, Stolar MH, Mulrow CD, et al. Colorectal cancer screening: clinical guidelines and rationale. Gastroenterology 1997;112:594-642. 26. Wright TA, Gray MR, Morris AI, Gilmore IT, Ellis A, Smart HL, et al. Cost effectiveness of detecting Barrett’s cancer. Gut 1996;39:574-9. 27. Soni A, Sampliner RE, Sonnenberg A. Screening for highgrade dysplasia in gastroesophageal reflux disease: is it costeffective? Am J Gastroenterol 2000;95:2086-93. 28. Shaker R. Unsedated trans-nasal pharyngoesophagogastroduodenoscopy (T-EGD): technique. Gastrointest Endosc 1994;40:346-8. 29. Mokhashi M, Glenn T, Jost C, Walace MB, Kim CY, Palesch
318
GASTROINTESTINAL ENDOSCOPY
30.
31.
32.
33. 34.
35.
36.
37.
38.
39.
40. 41.
42.
43.
44.
45.
YY, et al. Can a 3.1 mm stand-alone battery powered esophagoscope screen the esophagus for esophagitis and Barrett’s? A prospective blinded comparison with a standard videoendoscope [abstract]. Gastrointest Endosc 2000; 51:AB274. Cameron AJ, Ott BJ, Payne WS. The incidence of adenocarcinoma in columnar-lined (Barrett’s) esophagus. N Engl J Med 1985;313:857-9. Robertson CS, Mayberry JF, Nicholson DA, James PD, Atkinson M. Value of endoscopic surveillance in the detection of neoplastic change in Barrett’s oesophagus. Br J Surg 1988;75:760-3. Hameeteman W, Tytgat GN, Houthoff HJ, van den Tweel JG. Barrett’s esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989;96:1249-56. Ovaska J, Miettinen M, Kivilaakso E. Adenocarcinoma arising in Barrett’s esophagus. Dig Dis Sci 1989;34:1336-9. Sharma P, Morales TG, Bhattacharyya A, Garewal HS, Sampliner RE. Dysplasia in short-segment Barrett’s esophagus: a prospective 3-year follow-up. Am J Gastroenterol 1997;92:2012-6. Drewitz DJ, Sampliner RE, Garewal HS. The incidence of adenocarcinoma in Barrett’s esophagus: a prospective study of 170 patients followed 4.8 years. Am J Gastroenterol 1997;92:212-5. Cameron AJ, Lomboy CT. Barrett’s esophagus: age, prevalence, and extent of columnar epithelium. Gastroenterology 1992;103:1241-5. Wallace M, Van Dam J. Enhanced gastrointestinal diagnosis: light-scattering spectroscopy and optical coherence tomography. Gastrointest Endosc Clin N Am 2000;10:71-80. Wallace MB, Perelman LT, Backman V, Crawford JM, Fitzmaurice M, Seiler M, et al. Endoscopic detection of dysplasia in patients with Barrett’s esophagus using light scattering spectroscopy. Gastroenterology 2000;119:677-82. Wallace MB, Kemp JA, Meyer F, Horton K, Reffel A, Christiansen CL, et al. Screening for colorectal cancer with flexible sigmoidoscopy by nonphysician endoscopists. Am J Med 1999;107:214-8. Maule WF. Screening for colorectal cancer by nurse endoscopists. N Engl J Med 1994;330:183-7. Locke GR 3rd, Talley NJ, Fett SL, Zinsmeister AR, Melton LJ 3rd. Prevalence and clinical spectrum of gastroesophageal reflux: a population-based study in Olmsted County, Minnesota. Gastroenterology 1997;112:1448-56. Cameron AJ. Epidemiology of columnar-lined esophagus and adenocarcinoma. Gastroenterol Clin North Am 1997;26:48794. O’Connor JB, Falk GW, Richter JE. The incidence of adenocarcinoma and dysplasia in Barrett’s esophagus: report on the Cleveland Clinic Barrett’s Esophagus Registry. Am J Gastroenterol 1999;94:2037-42. Katz D, Rothstein R, Schned A, Dunn J, Seaver K, Antonioli D. The development of dysplasia and adenocarcinoma during endoscopic surveillance of Barrett’s esophagus. Am J Gastroenterol 1998;93:536-41. Reid BJ, Levine DS, Longton G, Blount PL, Rabinovitch PS. Predictors of progression to cancer in Barrett’s esophagus: baseline histology and flow cytometry identify low- and highrisk patient subsets. Am J Gastroenterol 2000;95:1669-76.
VOLUME 57, NO. 3, 2003