Endoscopic Screening for Gastric Cancer

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2006;4: Endoscopic Screening for Gastric Cancer YOCK YOUNG DAN,* J. B. Y. SO, and KHAY GUAN YEOH*, *Department of Gastroenterology and Hepatology and Department of Surgery, National University Hospital, Singapore; and Department of Medicine, National University of Singapore, Singapore Background & Aims: Population endoscopic screening for gastric cancer is generally deemed not to be costeffective except in Japan, where its prevalence is very high. However, in the absence of screening, patients present with advanced disease, and prognosis is poor. We conducted a cost utility analysis to determine whether endoscopic screening for stomach cancer in intermediate-risk population would be cost-effective and to better define the high-risk groups in the population who would benefit from such strategy. Methods: Costeffectiveness analysis was performed by using a Markov Model. Simulation was performed on Singapore (intermediate-risk) population and various high-risk subgroups. Comparison was made between 2-yearly endoscopic mass screening program versus no screening. Data sources were extracted from relevant studies published from identified via systematic PUBMED search. Main outcome measures were deaths caused by stomach cancer averted, cost per life saved, and incremental cost-effectiveness ratio expressed as cost per quality-adjusted life year (QALY) saved. Results: Screening of high-risk group of Chinese men (age-standardized rate, 25.9/100,000) from years old is highly cost-effective, with cost benefit of United States $26,836 per QALY. Screening this cohort of 199,000 subjects prevents 743 stomach cancer deaths and saves 8234 absolute life years. Cost of averting 1 cancer death is United States $247,600. Cost-effectiveness was most sensitive to incidence of stomach cancer and cost of screening endoscopy. Conclusions: Screening of stomach cancer in moderate to high-risk population subgroups is cost-effective. Targeted screening strategies for stomach cancer should be explored. G astric cancer is the second most common cancer incosts accrued by the individual as well as by society. the world and is responsible for some 600,000 Singapore is a multicultural society with the highest risk of 1 gastric cancer seen predominantly in Chinese men from 50 deaths worldwide annually, constituting significant disease burden, especially in Asia. Japan has been conducting population screening for gastric cancer since the 1960s, and remarkable improvement in survival rates from gastric cancer has been achieved as a result of early 2 4 detection and consequently higher cure rates. Screening for gastric cancer in asymptomatic population has been shown to increase the detection of early gastric cancer. 5 Overall 5-year survival is significantly better for tumors detected by screening compared with 6 those detected by open access clinic. Outside Japan, prognosis of stomach cancer is poor because patients tend to present with advanced disease. Screening for stomach cancer by endoscopy is not practiced outside Japan because screening is deemed not to be cost-effective. Yet, population subgroups with increased risk of stomach cancer have been well-documented. These include subpopulations in countries of intermediate epidemiologic risks such as East Asia and Europe, 7 those with family history of stomach cancer, atrophic gastritis, 8,9 Helicobacter pylori (HP) infection, 10 intestinal 6 metaplasia, and gastric dysplasia. The role of screening or surveillance endoscopy in these intermediate-risk populations has never been studied by cost-effectiveness analysis. We sought to determine whether primary screening endoscopy would be a cost-effective way to screen for gastric cancer in an intermediate-risk population. Population subgroups in which cost-effectiveness can be maximized are further defined. The optimal frequency of screening and its ramifications are also examined. Materials and Methods Model Structure Cost utility analysis 11,12 was performed by using a Markov Model. 13 The primary decision analysis model compared mass screening for gastric cancer by 2-yearly endoscopy versus the current strategy of no screening endoscopy and investigating only when there are alarm symptoms. All analyses were performed from societal perspective to incorporate all Abbreviations used in this paper: ASR, age-standardized rate; EGD, esophagogastroduodenoscopy; HP, Helicobacter pylori; ICER, incremental cost-effectiveness ratio; GNI, gross national income; QALY, quality-adjusted life years; USD, United States dollars; WHO, World Health Organization by the American Gastroenterological Association Institute /06/$32.00 doi: /j.cgh

2 710 DAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 6 Figure 1. Epidemiology of gastric cancer in Singapore (ASR, 21.6/ 100,000). (A) Comparison of incidence of gastric cancer by race and sex. The Chinese male population has the highest incidence at 25.7/ 100,000, 3-fold higher than the Indians or Malays. (B) Incidence of gastric cancer by age group and sex. Incidence rises exponentially from 50 years of age. The highest risk population would be the Chinese male population after 50 years of age. lation after screening stops. At each cycle, patients are channeled into one of 3 Markov states: healthy, gastric cancer by various stage, and dead. Patients who survived 5 years after 17 treatment are presumed to have been cured of gastric cancer. We conducted the analysis by assuming screening esophagogastroduodenoscopy (EGD) with biopsy of suspicious lesions 18 will have 84% sensitivity and 100% specificityand by assuming full compliance and adherence with regular screening and appropriate treatment. In determining the cost-effectiveness (willingness-to-pay) threshold for our model, we followed the principle set forth by the Commission on Macroeconomics and Health convened by the World Health Organization (WHO), which recommended pegging the cost-effectiveness threshold to the country s economic spending power, represented by indices such as gross 19 national income per capita (GNI). We thus adopted the conventional United States threshold of United States dollars (USD) 50,000/quality-adjusted life years (QALY) but rationalized it to the Singapore context by adjusting it to USD28,000 on the basis of the ratio of Singapore s 2003 GNI per capita 20 (USD21,230) to that of USA (USD37,610). Screening and Treatment Variables Incidence and prevalence rates. There are no prospective randomized controlled studies looking at efficacy of screening endoscopy. As such, variables were obtained from years and older (age-standardized ratio [ASR], 25.9/ best available data by synthesis from various sources Table ( 1). 100,000) 14 (Figure 1) Screening this cohort would potentially Age and gender-specific incidence of stomach cancer for the reap the highest benefit by detecting cancers in an earlier population were taken from the latest available publication of 14 curable stage. Modeling was thus performed on this basethe Singapore Cancer Registry for Adjustment population of Singapore Chinese men in the target age groupwas made for 2.3% annual decrease in incidence of gastric of years old. Cost-effectiveness was then compared withcancer. 21 strategies that screen various population subgroups in the same Age and gender-specific life expectancies were obtained 22 age category: whole Singapore population (ASR, 16/100,000), from Singapore Health Statistics The distribution of men (ASR, 21/100,000), women (ASR, 11/100,000), and Chi-gastrinese men with HP (estimated ASR, 42.8/100,000). the unscreened population were obtained from local stud- cancer by TNM staging and survival data by stage in In the screened pathway, the selected subpopulation is puties. 23,24 For the screened population, stage distribution of through 2-yearly screening by primary endoscopy. Suspicious gastric cancer diagnosed by screening was projected from the 2 lesions will be biopsied. Positive cases of gastric cancer are staged largest mass screening studies using endoscopy on asymptom- population in Japan, identified by systematic review of the by endoscopic ultrasound and computed tomography imaging ofatic 15 the abdomen. All stage 1 3 tumors are assumed to be treated by literature with PUBMED search for all relevant studies reported from Stage-specific survival was -as 16 5,25 curative resection. In addition, patients with stage 3 tumors will be given chemoradiotherapy with 5-fluorouracil/leucovorin. For sumed to be similar to the gastric cancer subjects in the stage 4 cancers, supportive care with or without palliative chemotherapy will be the primary treatment. Efficacy of HP screening and eradication as well as incidence unscreened population. The unscreened pathway assumes no screening for the pop-oulation at large. Stomach cancers are diagnosed only whenextrapolated from data on the basis of HP epidemiology and identifiable premalignant lesions at first endoscopy were patients present with symptoms. Positive cases of stomachgastric cancer studies 9,26,27 (see supplemental algorithm online cancer are treated with the same treatment protocol. at Screening was conducted from years of age, and Cost. We calculated costs of screening and treatment modeling was performed for the time horizon from 2003 until on the basis of current data obtained from actual hospital death of all subjects (up to 99 years old when 99% of finance records for 2003 in Singapore Table ( 1, and see supplemental algorithm online at All costs subjects would have died). Incidence of gastric cancer is assumed to be similar between the screened and unscreenedaccrued from time of screening until death were calculated and 22 groups. Distribution of gastric cancer staging in the screenedadjusted for inflation (0.5%/year), discounted 3% annually, group is assumed to become similar to the unscreened popu- and expressed in 2003 USD. In situations in which there were

3 June 2006 GASTRIC CANCER SCREENING 711 Table 1. Input Variables and Source Base case Reference source Incidence of stomach cancer Sex-specific ASR (applied as age-specific rate in model for each age) 25.9/100, (see supplementary data) Annual decrease in incidence 2.3%/year 14, 21 Stage of stomach cancer at diagnosis Stage 1:2:3:4 Unscreened population 7%:17%:33%:43% 23, 24 Screened population 85%:4%:8%:3% 5, 25 5-Year survival data (by stage) Stage 1:2:3:4, 90%:70%:40%:0% 23, 24 Screening Age at screening (y) 50 Frequency of screening 2-year Cost (USD) Screening endoscopy 150 Hospital Stages 1 and 2 treatment 4717 Case mix Stage 3 treatment 8480 Records 2003 Stage 4 treatment 1050 Utilities Stages I and II (surgery) , 30 Stage III (chemoradiotherapy) 0.4 Stage IV (palliative care) (average of chemoradiotherapy, surgery, stent) 0.5 Others Annual discount 3% NOTE: Variables used for base case analysis (Chinese men) were obtained from best available data synthesized from various sources including national database and relevant studies identified from systematic PUBMED search for years various treatment options, cost calculation was based on the more conservative option, biasing the analysis against screening. Estimation was also made for additional costs incurred by subjects joining the screening program such as transport, morbidity from complications of endoscopy, as well as follow-up and counseling. In addition, we also factored in logistic cost of establishing a national screening infrastructure that would include quality management, education, and training, as well as information provision by using costs derived from estimates reported in colorectal cancer screening program in 28 the United Kingdom. Quality of life utilities scores were taken from Quality-of- 29 Life Repository for gastric cancer. To increase resolution of utility scores between various gastric cancer stage treatments and where no published references were available, we adjusted the utility scores by mapping EORTC QLQ-C30 quality of life score variance reported for different gastric cancer treatments. 30 Both cost and effects indicators were discounted annually by 3%. Statistics and Outcomes Analysis was performed by using computer software Data 4.0 by Treeage Software Inc, Williamstown, MA. Comparison was made between various strategies of screening (endoscopic screening of specific subpopulation and not the rest) and the reference strategy of non-screening for the whole population. Outcome measures were deaths caused by stomach cancer averted, cost per life saved, cost per life year saved, and incremental cost-effectiveness ratios (ICERs) expressed as USD/quality-adjusted life years (QALY), as well as net health benefit calculated as (Effectiveness Cost/Willingness-to- Pay). The most cost-effective strategy was defined as that with the highest net health benefit compared with the reference strategy of no screening. Sensitivity Analysis Both 1-way and 2-way sensitivity analyses were then applied to base case population of Chinese men to identify the factors that had the greatest impact on cost-effectiveness and also to determine the impact of assumptions and estimations for all variables used. Ranges for sensitivity analysis, including distribution of cancer staging diagnosed during screening, were empirically derived on the basis of range of variables reported in all relevant studies identified in the literature review. Range for cost was extended to cover estimated cost in a developed country (USA) and a developing country (India). Worse case and best case scenarios for clusters of variables were also computed. Results Base Case Analysis Screening of Singapore Chinese male population (ASR, 25.9/100,000) from years old for gastric cancer in 2003 would yield an ICER of USD26,836/ QALY, which would be cost-effective with the average GNI per capita (USD28,000) as cost-effectiveness threshold (Table 2). This would involve a cohort of 198,823 subjects and would incur an incremental cost of

4 712 DAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 6 Table 2. Results of Base Case Analysis Population of Chinese men years old 198,823 Total incremental cost of screening USD183,876,000 Total no. of scopes needed per year 92,216 No. of cancer deaths averted 743 No. needed to screen to prevent 1 cancer death 267 No. of scopes needed to prevent 1 cancer death 2482 Cost of saving 1 cancer death USD247,600 Cost per absolute life year saved USD22,346 Cost per QALY saved USD26,836 NOTE: Screening 198,823 people will result in incremental cost of USD184 million but will result in aversion of 743 deaths due to gastric cancer. This is cost-effective with an ICER of USD 26,836, which is lower than the threshold of USD28,000. USD184 million during a period of 20 years of screening or an average cost of USD9.2 million a year. This expenditure would avert 743 cancer deaths and loss of 8234 absolute life years or 6856 QALY. The number of subjects needed to screen to save 1 cancer death is 267, and number of endoscopic examinations needed per year to prevent 1 cancer death is Sensitivity Analysis Cost-effectiveness was most sensitive to the incidence of gastric cancer, cost of screening endoscopy, and the distribution of cancer stage at screening Table ( 3). Figure 2. Sensitivity analysis on incidence of gastric cancer (ASR and cost of screening [USD]). When the incidence of gastric cancer is low and cost of screening endoscopy is high, such as in the US, screening is not cost-effective. In contrast, when incidence is high like in Japan, it is highly cost-effective, even if screening cost is high. Incidence of gastric cancer. The cost of screening endoscopy in Singapore is USD150. By fixing this cost and varying the incidence of gastric cancer, the ICER ranged widely (Figure 2).. If the population ASR was 1 10/100,000 (equivalent to American men), the ICER was USD64,800/QALY. In contrast, if the population ASR was as high as 70/100,000 (equivalent to ASR for Japanese men), 1 ICER dropped to USD6900/QALY, and screening would be extremely cost-effective. Table 3. Sensitivity Analysis (Screening Chinese Male Population) Variable Range for sensitivity analysis ICER (USD/QALY) (best case scenario) ICER (USD/QALY) (worst case scenario) ASR 10 70/100, ,835 Cost of screening USD ,643 66,382 Stage of stomach cancer at diagnosis Stage 1:2:3:4 Unscreened population 5%:10%:35%:50% 15%:22%:30%:33% 17,990 40,517 Screened population 40%:30%:20%:10% 90%:6%:3.5%:0.5% 20,298 44,854 5-Year survival by stage Stage 1:2:3:4, 70%:50%:30%:0% 99%:90%:60%:0% 21,532 32,510 Cost of treatment (USD) Stage 1: surgery ,000 20,518 27,606 Stage 2: surgery ,000 Stage 3: surgery with chemoradiotherapy ,000 Stage 4: palliative chemotherapy ,800 Starting age of screening (y) ,571 35,125 Frequency of screening ,698 41,872 Sensitivity of diagnosis 70% 95% 24,517 33,618 Specificity of diagnosis 95% 100% 26,319 27,982 Utility scores ,519 27,518 Discount (both cost and benefits) 0% 5% 17,899 27,512 Discount for benefits (discount for cost kept constant at 3%) 0 3% 16,319 21,995 NOTE: Sensitivity analysis was performed to cover the full range of variables reported in the literature. Where there are no data, the widest reasonable range was used to validate the estimations. Incidence of gastric cancer and the cost of screening had the biggest impact on cost-effectiveness.

5 June 2006 GASTRIC CANCER SCREENING 713 Table 4. Comparing Cost-Effectiveness of Various Screening Strategies With No Screening Screening strategy (vs no screening) ASR/100,000 Subpopulation size No. EGDs Total cost Increment needed/y (million) at cost No. cancer deaths averted Life years/ QALY saved ICER (cf to no screening) Inc net health benefit Unscreened , EGD screening, total population , , ,273/ 45, ,200 EGD screening, women only , , / 63, EGD screening, men only , , / 38, EGD screening, Chinese men only ,823 92, / 26, Chinese men, HP screen/eradication, EGD screening only for HP ,823 (79,529) 36, / , Chinese men, HP screen/eradication, EGD screening only for HP with premalignant lesions on first EGD ,823 (35,788) 16, / , NOTE: Selecting subpopulation with higher incidence of gastric cancer makes screening more cost-effective. The strategy of screening Chinese men for HP/eradication and selective EGD screening for those with premalignant lesions yields the lowest ICER compared with no screening. However, the number of cancer deaths averted is only 182. Setting the willingness-to-pay ratio at GNI per capita of USD28,000, the most cost-effective strategy is 2-yearly endoscopic screening of Chinese men as reflected by the highest incremental net health benefit (0.015). Cost. If the cost of endoscopy is increased to USD300 (estimated cost of endoscopy in United States) (Figure 2), the ICER increases to USD45,188/QALY, and screening becomes significantly less cost-effective. Conversely, if cost of endoscopy decreases to USD75, then even screening low-risk population such as Singapore women (ASR, 11.1/100,000) would be potentially cost-effective with ICER of USD25,987/QALY. Age. Cost-effectiveness was also sensitive to the age at which screening starts. The most cost-effective age for starting screening is 65 years (ICER, USD21,333/ QALY). The youngest age from which screening is costeffective is 50 years and is accounted for by the exponential rise in incidence of gastric cancer from 50 years of age. Beyond 70 years of age, screening is no longer cost-effective because of the limited number of life years that can be saved before reaching end of expected life span. Stage of gastric cancer at diagnosis. The distribution of gastric cancer by staging has a moderate impact on cost-effectiveness. With estimates from Japanese data 2,4,5 in screened population, the reference analysis was based on distribution ratio of stages 1:2:3:4 85%:4%: 8%:3%. In the worst case scenario, by using distribution ratio of stages 1:2:3:4 40%:30%:20%:10%, the ICER increased to USD44,854/QALY. Frequency of screening. We also examined the optimal frequency of screening in terms of cost-effectiveness. With an annual screening program, the distribution of cancer staging in the screened population is projected to improve to stages 1:2:3:4 87%:6%:6%: 1%, giving an ICER of USD41,872/QALY. With a 3-year cycle with cancer stage distribution estimated at 60%:15%:15%:10%, the ICER is USD20,435/ QALY. Although it is more cost-effective (ICER of 2-year compared with 3-year screening USD29,985/QALY), only 575 lives will be saved. The survival expectancy of stomach cancer, life expectancy, cost of cancer treatment, and utility score variance only had a small impact on cost-effectiveness. Optimizing Cost-Effectiveness Screening the whole Singapore population or only the male population from years old would have resulted in an ICER of USD45,982 and USD38,435 per QALY, respectively, both of which are higher than the cost-effectiveness threshold (USD28,000) Table ( 4). If only Chinese men with known positive HP serology 26 (estimated ASR, 42.8/100,000) were screened, this would have been extremely cost-effective with an ICER of USD17,455/QALY, provided the HP status of the population was known. Strategies Incorporating Cancer Prevention by Helicobacter pylori Screening and Eradication Recent data from the HP eradication trial in China have not borne out actual cancer incidence reduction, 27 suggesting instead that eradication might be too late for those who have developed premalignant lesions.

6 714 DAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 6 Figure 3. Comparing different strategies of screening. The ICER between different strategies is presented. Although HP screening/eradication followed by selective endoscopy has the lowest cost, they are offset by lower number of cancer deaths averted. Strategy of screening for the whole population or the male population is extendedly dominated (more expensive yet less effective compared with screening Chinese men). We thus modeled for comparison a strategy combining HP screening and eradication for all Chinese men and selective 2-yearly endoscopy screening for those who are HP-positive. Such a strategy would yield an ICER of USD21,800/QALY over the non-screening strategy. With further limiting endoscopic screening to only HPpositive serology subjects with identifiable premalignant lesions at first endoscopy, the ICER over non-screening is USD19,900. Although both these 2 strategies would incur lower costs because less endoscopy is needed, the number of cancer deaths averted was lower. Comparing the net health benefit between these strategies, 2-yearly endoscopic screening of all Chinese men was still the most cost-effective strategy for the defined willingnessto-pay threshold (Table 4, Figure 3). Discussion The pathogenesis of gastric cancer described by the model of Correa 31 of dysplasia, carcinogenesis, early gastric cancer formation, and subsequent invasion to become advanced cancer is well-established. There is reasonable evidence that early detection of gastric cancer does translate into better survival rates. In our model, we showed that screening for gastric cancer by 2-yearly endoscopy in population with moderate risk of gastric cancer (ASR, 25.9/100,000) is potentially cost-effective with an ICER of USD26,836/QALY. Endoscopy is a safe procedure with extremely low rate 32,33 of complication (mortality, 1/3300 1/40,000). It is also more sensitive and specific than barium examination 5 in diagnosing gastric cancer, with positive predic - tive value of 98% 99.8% 34 and false-negative rate -re ported as 0.5%. 18 Given the small caliber of the endoscope and the simplicity of examination, endoscopic screening should be acceptable to most subjects at higher risk of developing gastric cancer. The frequency of screening has previously been proposed at 1 5 years, with the optimal interval being less 2 35 than 3 years. The median doubling time of early gastric cancer is estimated to be 2 3 years 36 ; thus most cases of gastric cancer missed at first endoscopy and subsequently picked up on the second screen within 3 years were still early gastric cancer amenable to curative surgery. 18 This would theoretically increase the sensitiv - ity of screening up to 90% after 2 consecutive EGDs. In our retrospective study on actual gastric cancer cases, endoscopy within prior 24 months would have detected 37 almost 90% of gastric cancers. As such, although the 3-year screening is more cost-effective, we preferred a 2-yearly screening strategy. In estimating variables for the analysis, we used all available data to justify our estimation and performed sensitivity analysis on as wide a range of variables as possible. Estimations were deliberately conservative, erring on the side of bias against screening. For example, we have assumed all subjects with stage 1 gastric cancer undergo surgical resection. In reality, some of the early gastric cancers can be treated by endoscopic mucosal resection and would result in lower cost and higher quality of life. Nevertheless, as in all cost-effectiveness estimates, there are several limitations that need to be considered. Compliance and adherence to screening program might undermine actual cost-effectiveness of a screening program, but they do not detract from its cost-effective potential. There is inherent potential for lead bias in any cancer screening study. Early gastric cancer has been shown to progress to advanced gastric cancer with time and leads to death if untreated, although this might take 38,39 several years to occur. By screening up to 70 years of age, we believe this would limit unnecessary treatment in the elderly. We avoided using overall survival after diagnosis in our model but, rather, calculated yearly survival on the basis of age, stage of cancer, and year from diagnosis to minimize the effect of lead bias. Because no randomized controlled trials on efficacy of gastric cancer screening are available, we used the best available data from retrospective trials, risking errors caused by selection, length-time, and lead-time bias. To mitigate some of these biases, we deliberately ignored the stage-specific survival advantage that was reported in the trials and assumed that survival for each stage was similar for those screened and not screened. This, in fact, might bias against screening because within stage 1 3 disease, early detection might actually result in

7 June 2006 GASTRIC CANCER SCREENING 715 higher chances for cure and real improved stage-specific survival. We also did not consider additional cost or implications that might arise from the endoscopy such as diagnosis and treatment for other incidental findings such as reflux esophagitis, Barrett s esophagus, gastric polyps, erosions, or gastritis, or the tempting prospect of screening for HP by urease test at endoscopy. Utility scores were mapped from various studies because there was no one single study that reported utilities for all the Markov states studied. In projecting the model to different countries, we assumed that utility scores will remain similar, although this is unlikely to be so between different cultures and ethnicity. We did, however, run sensitivity analysis on a wide range of utility scores. How does a country decide whether screening for stomach cancer might be cost-effective? Although our model is simulated on Singapore subpopulation, we have adopted WHO s principle in rationalizing the cost-effectiveness threshold to the country s economic indices such as average per capita income for deciding cost-effectiveness of healthcare program. This better reflects the economic status of a country and its likely willingness to spend on healthcare and is thus of greater relevance. In our model, screening the base population incurs incremental cost of USD184 million, which is similar in magnitude to the cost of other programs such as renal dialysis and breast cancer screening. Many countries including Korea, China, Scotland, Russia, South America, and even USA have subpopulations whose gastric cancer incidences are comparable, if 7 not higher, than Singapore s. With their own specific incidence of gastric cancer, life expectancy, cost of screening endoscopy, and threshold for adopting cost-effective health programs, the estimated cost-effectiveness for screening for various subpopulations can be determined from Figure 2. For example, in a country in which the healthcare system is willing to pay for programs with ICER threshold USD25,000 and the cost of endoscopy screening is USD150, it can be predicted that screening populations with gastric cancer ASR 23.6/100,000 would be cost-effective. Similarly, if the cost of endoscopy is USD300, then only populations in whom the incidence of gastric cancer is high (ASR, 40/100,000) will be cost-effective. It should be noted that the cost of endoscopy in Singapore (USD150) is higher than in most countries in Asia. As the data for other risk factors such as atrophic gastritis, 9 dysplasia, and intestinal metaplasia become clearer, the high-risk population, including those with positive family history of gastric cancer and presence of HP infection, will be better defined. It is likely that a customized screening program that offers more frequent surveillance of high-risk groups and vice versa would further optimize cost-effectiveness. In conclusion, screening for gastric cancer by endoscopy can potentially be cost-effective and is most sensitive to incidence of gastric cancer in the population, cost of screening endoscopy, and relative distribution by stage of gastric cancer diagnosed through screening. The optimal screening strategy of selecting subpopulations for screening can be defined by cost-effectiveness modeling. Prospective population screening trials should be initiated to confirm this lifesaving and cost-saving potential. Supplementary Data Note: To access the supplementary algorithm accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at References 1. Pisani P, Parkin DM, Bray F, et al. Estimates of the worldwide mortality from 25 cancers in Int J Cancer 1999;83: Fukao A, Tsubono Y, Tsuji I, et al. The evaluation of screening for gastric cancer in Miyagi Prefecture, Japan: a population-based case-control study. Int J Cancer 1995;60: Hisamichi S. Screening for gastric cancer. World J Surg 1989;13: Hisamichi S, Sugawara N, Fukao A. Effectiveness of gastric mass screening in Japan. Cancer Detect Prev 1988;11: Kubota H, Kotoh T, Masunaga R, et al. Impact of screening survey of gastric cancer on clinicopathological features and survival: retrospective study at a single institution. Surgery 2000;128: Whiting JL, Sigurdsson A, Rowlands DC, et al. The long term results of endoscopic surveillance of premalignant gastric lesions. Gut 2002;50: Ferlay J, Bray F, Pisani P, et al. Globocan 2000: cancer incidence, mortality and prevalence worldwide. Volume IARC Cancer- Base no 5 (version 1.0) ed. Lyon, France: IARC Press, Walker IR, Strickland RG, Ungar B, et al. Simple atrophic gastritis and gastric carcinoma. Gut 1971;12: Watabe H, Mitsushima T, Yamaji Y, et al. Predicting the development of gastric cancer from combining Helicobacter pylori antibodies and serum pepsinogen status: a prospective endoscopic cohort study. Gut 2005;54: IARC working group on the evaluation of carcinogenic risks to humans. Schistosomes, liver flukes, and Helicobacter pylori: views and expert opinions of an IARC working group on the evaluation of carcinogenic risks to humans, IARC monographs on the evaluation of carcinogenic risks. Lyon, France: International Agency for Research on Cancer, 1994: Weinstein MC, Siegel JE, Gold MR, et al. Recommendations of the panel on cost-effectiveness in health and medicine. JAMA 1996;276: Drummond MF, Jefferson TO. Guidelines for authors and peer reviewers of economic submissions to the BMJ: the BMJ economic evaluation working party. BMJ 1996;313: Sonnenberg FA, Beck JR. Markov models in medical decision making: a practical guide. Med Decis Making 1993;13:

8 716 DAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No Chia HL, Shanugaratnam K. Cancer incidence in Singapore Singapore: Singapore Cancer Registry, American Joint Committee on Cancer. AJCC cancer staging manual. 6th ed. New York: Springer, 2002: Macdonald JS, Smalley SR, Benedetti J, et al. Chemoradiotherapy after surgery compared with surgery alone for adenocarcinoma of the stomach or gastroesophageal junction. N Engl J Med 2001;345: Dupont JB Jr, Lee JR, Burton GR, et al. Adenocarcinoma of the stomach: review of 1,497 cases. Cancer 1978;41: Hosokawa O, Tsuda S, Kidani E, et al. Diagnosis of gastric cancer up to three years after negative upper gastrointestinal endoscopy. Endoscopy 1998;30: Commission on Macroeconomics and Health. Macroeconomics and health: investing in health for economic development report of the Commission on Macroeconomics and Health executive summary. Geneva: World Health Organization, World Development Indicators database. Volume Washington, DC: the World Bank Group, Coleman MP, Esteve J, Damiecki P, et al. Trends in cancer incidence and mortality. Lyon, France: IARC Science Publications, Department of Statistics. Yearbook of statistics, Singapore Singapore: Department of Statistics, Ti TK. Pattern and surgical treatment of gastric cancer in Singapore. Br J Surg 1993;80: Koong HN, Chan HS, Nambiar R, et al. Gastric cancers in Singapore: poor prognosis arising from late presentation. Aust N Z J Surg 1996;66: Matsukuma A, Furusawa M, Tomoda H, et al. A clinicopathological study of asymptomatic gastric cancer. Br J Cancer 1996;74: Fock KM. Helicobacter pylori infection: current status in Singapore. Ann Acad Med Singapore 1997;26: Wong BC, Lam SK, Wong WM, et al. Helicobacter pylori eradication to prevent gastric cancer in a high-risk region of China: a randomized controlled trial. JAMA 2004;291: Robert G, Brown J, Garvican L. Cost of quality management and information provision for screening: colorectal cancer screening. J Med Screen 2000;7: Glimelius B, Hoffman K, Graf W, et al. Cost-effectiveness of palliative chemotherapy in advanced gastrointestinal cancer. Ann Oncol 1995;6: Blazeby JM, Conroy T, Bottomley A, et al. Clinical and psychometric validation of a questionnaire module, the EORTC QLQ-STO 22, to assess quality of life in patients with gastric cancer. Eur J Cancer 2004;40: Correa P. A human model of gastric carcinogenesis. Cancer Res 1988;48: Hart R, Classen M. Complications of diagnostic gastrointestinal endoscopy. Endoscopy 1990;22: Silvis SE, Nebel O, Rogers G, et al. Endoscopic complications: results of the 1974 American Society for Gastrointestinal Endoscopy survey. JAMA 1976;235: Dekker W, Tytgat GN. Diagnostic accuracy of fiberendoscopy in the detection of upper intestinal malignancy: a follow-up analysis. Gastroenterology 1977;73: Kim YS, Park HA, Kim BS, et al. Efficacy of screening for gastric cancer in a Korean adult population: a case-control study. J Korean Med Sci 2000;15: Fujita S. Biology of early gastric carcinoma. Pathol Res Pract 1978;163: Rajnakova AHK, Tun M, Yeoh KG. What is the appropriate endoscopic surveillance interval for patients with helicobacter pyloriassociated gastritis and intestinal metaplasia in a country with moderate risk of gastric cancer? (abstract). Gastroenterology 2004;126(Suppl 2):A Yamazaki H, Oshima A, Murakami R, et al. A long-term follow-up study of patients with gastric cancer detected by mass screening. Cancer 1989;63: Tsukuma H, Oshima A, Narahara H, et al. Natural history of early gastric cancer: a non-concurrent, long term, follow up study. Gut 2000;47: Address requests for reprints to: Khay Guan Yeoh, MD, Department of Gastroenterology, National University Hospital, 5, Lower Kent Ridge Road, Singapore mdcykg@nus.edu.sg; fax: Supported by SCS grant GN-15, awarded by the Singapore Cancer Syndicate, Agency for Science, Technology and Research, Singapore (to K.G.Y.).