Time Trends in Cost of Caring forPeople With Gastroesophageal Reflux DiseaseBernard S. Bloom, Ph.D., Ravi Jayadevappa, Ph.D., Peter Wahl, B.A., and John Cacciamanni, M.D.Division of Geriatrics, Department of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania

OBJECTIVE: Increasing acceptance of the many permutationsof gastroesophageal reflux disease (GERD) has led to di-verse study of the disease and its effects. The goal of thisstudy was to estimate medical care costs attributable to adefined GERD population over time.

METHODS: A retrospective cohort control design was used.All participants were identified from the database of a man-aged care organization serving 300,000 people in the north-eastern United States. The index population (n5 600) wasdefined as anyone who obtained medical services during1997 or 1998, for any International Classification of Dis-eases (ninth revision, Clinical Modification) codes sugges-tive of GERD, and/or anyone who received at least oneprescription and one refill for antisecretory or GERD med-ications during at least two 3-month periods in 1997 or1998. A matched cohort (n5 600) without any diagnosis ofGERD was randomly selected as a control group. Bothpopulations were observed restrospectively from January 1,1990 through December 31, 1998.

RESULTS: The cost of treating GERD averaged around $510per year, about 15% of all medical costs for those withGERD. Treating people with GERD was about 2-fold morecostly than treating those without GERD, a marginal cost of$1500 to $2000 per annum.

CONCLUSION: Although GERD is a low-cost disease totreat, the cost of treating people with GERD is subtantiallygreater than that for a comparable population withoutGERD. Two explanations may account for the large differ-ence of costs between the study populations. First, theGERD group may be sicker than the control group. Diseaseseverity variables and diagnoses associated with GERDwere more commonly diagnosed in the GERD group. Sec-ond, an additional disease that is not treated appropriatelyincreases the cost of treatment geometrically for all diseases.(Am J Gastroenterol 2001;96(suppl):S64–S69. © 2001 byAm. Coll. of Gastroenterology)

INTRODUCTION

Effective treatments are widely available for gastroesopha-geal reflux disease (GERD), but for those with frequentand/or persistent symptoms, correct diagnosis and aggres-sive treatment are required to control symptoms and prevent

recurrence. There has been much discussion recently aboutthe cost of prescription pharmaceuticals commonly used totreat GERD, in part because disease definition has not beenentirely agreed upon. Currently, more money is expendedon these medications than on any other single medication(IMS Health, 1999), even though the vast majority of pro-fessional care is provided in ambulatory settings (1–4).Whether relatively low cost pharmaceutical and ambulatorycare substitutes for high cost hospital care and diagnosticand therapeutic procedures in GERD has not been examinedrecently. Research results published in the 1980s on multi-ple upper GI diseases found that histamine-2 receptor an-tagonists reduced hospitalizations, operations, work loss,and disability (5–9).

Recent estimates of costs of medical care for GERDfound similar results across comparable countries (1–4).Direct medical costs of treating GERD were about $500 peryear, and GERD medications accounted for between onehalf and two thirds of expenditures. The indirect and non-medical costs of GERD, like reduced quality of life, timeaway from normal activities, and reduced productivity, haveonly begun to be measured. These costs may prove to bemuch greater than direct medical costs, and appropriatetreatment may have its most important effects in these areas.If so, patients and employers may be bearing these costsoutside the medical care system.

The goal of this study was to estimate the costs over timeof caring for people with GERD, and differentiate GERD-related and non–GERD related medical care costs. But un-resolved issues remain with GERD, making it difficult tocompare study results or draw conclusions about the state ofknowledge—for example, no agreed upon definition ofGERD and no accurate, accepted, and uniform diagnostictest.

STUDY DESIGN

PopulationA retrospective cohort control study was designed. Thepopulation was selected from a managed care organization(MCO) serving about 300,000 people in a mainly rural areain the northeastern United States.

This MCO is the largest provider in its region, withprimary through tertiary care available. Enrolled popula-

THE AMERICAN JOURNAL OF GASTROENTEROLOGY Vol. 96, No. 8, Suppl., 2001© 2001 by Am. Coll. of Gastroenterology ISSN 0002-9270/01/$20.00Published by Elsevier Science Inc. PII S0002-9270(01)02587-4

tions obtain nearly all of their care within the system. TheMCO owns its hospitals and most ambulatory centers. It hasmultiple forms of delivery including a staff model healthmaintenance organization, provider networks, and indepen-dent practitioners. Physicians are paid by salary, capitation,and/or fees for service. Additionally, the MCO uses multiplepayment methods to all other providers and varies paymentsfor services based on type of insurance plan held by theenrollee, contractual arrangements with providers, andMCO-specific cost accounting derivations.

All participants were identified from the MCO patientservice database. They were defined as anyone who ob-tained medical care during 1997 or 1998 for any upper GIdiagnosis (based on International Classification of Diseases[ICD; ninth revision, Clinical Modification] codes) sugges-tive of GERD, and/or anyone who received at least oneprescription plus one refill for antisecretory or GERD med-ications during at least two 3-month periods in 1997 or 1998(Table 1). This identified population was observed as farback as its members were eligible in the MCO computerizeddatabase, or to January 1, 1990, and through December 31,1998. Participants may have been initially diagnosed duringthe enrollment period or at any time up to 25 yr previously.Use of antibiotics at the same time as any prescription orrefill for a study medication and/or any peptic ulcer diag-nosis ruled out participation. Using a unique patient identi-fier number, we collected data on all services used for alldiagnoses for each identified participant.

The random sampling method found approximately21,000 people with GERD meeting study entry criteria and260,000 who never had a diagnosis of or treatment forGERD during the 9 study years. Both populations werematched by age, gender, length of time in the MCO, andprescription benefits insurance plan. A randomly selectedsample of 600 from each group was chosen from the eligiblematched GERD and control populations. All persons weretracked retrospectively for up to 9 calendar years.

Resource Use and CostAll services provided were attributed to GERD or all othernon-GERD diagnoses based on recorded diagnostic servicecodes. When multiple diagnoses were listed, GERD or re-lated upper GI diagnosis had to be the first listed diagnosisfor the service to be attributed to GERD. However, allambulatory prescriptions for any antisecretory or acid in-hibitor medication, upper GI endoscopy, pH monitoring,and other tests commonly used for GERD were attributed toGERD in the absence of any other upper GI diagnosis likepeptic ulcer. The strict definition of services attributable toGERD and the desire not to become entangled in jointproduct allocation issues mean that study results are con-servative estimates of costs of GERD care. These are bal-anced somewhat by broad inclusion diagnoses.

Data collected included patient characteristics (age, gen-der, date of original diagnosis of GERD, date of originalinsurance plan enrollment, health insurance coverage), di-agnostic and therapeutic procedures performed (e.g., GIendoscopy and 24-h pH monitoring), all inpatient hospital-izations for any diagnosis, physician and ambulatory clinicvisits, laboratory and X-ray testing, and ambulatory phar-maceuticals. Costs are defined as actual payments for spe-cific services by the MCO. Costs per service were attributedto each service for each study patient from actual paymentsfor that patient.

Data AnalysisAnnual payments during the study period were aggregatedfor each patient and added across all patients by diagnosiscategory (GERD and non-GERD). Mean, median, and mea-sures of distribution were calculated for study populations.Significance tests were performed for patient characteristics,services used, and costs. Physician characteristics were notavailable. We compared byx2 statistics the differences inproportions between the two study populations for severityof illness adjusters and diagnoses associated with GERD.

Two cost estimates were calculated. First, average cost ofGERD treatment as part of total medical care for personswith GERD was calculated by attributing services sepa-rately to GERD and all other (non-GERD) care. Second wasthe marginal cost of care due to GERD (i.e., the differencein total costs of all medical care between GERD and non-GERD populations). The added cost burden (marginal cost)was assumed due to GERD and its synergistic effects withmultiple diagnoses.

Next, longitudinal (1990–1998) and cross-sectional datawere combined to form a pooled time series data set. Thegeneralized least squares technique was used to develop aregression model. The dependent variable was total cost,and the independent variables were age, gender, diseaseseverity variables, and diagnoses associated with GERD.The model was used for both GERD and non-GERD pop-ulations, and the parameter estimate and intercepts werecompared.

There is growing evidence about the association and

Table 1. Criteria for Entry to the Study

ICD diagnoses for study entryDyspepsia (536.8)Nausea and vomiting (787.1)Heartburn (787.0)Dysphagia (787.2)

Medications required for study entryProton pump inhibitorsPromotilitySucralfateHistamine-2 receptor antagonistsMetoclopromide

Exclusion diagnosesGastric ulcer (ICD 531)Ulcer site unspecified (ICD 533)Duodenal ulcer (ICD 532)Gastrojejunal ulcer (ICD 534)

S65AJG – August, Suppl., 2001 Time Trends in Cost of Caring for People With GERD

perhaps overlap of GERD and other diagnoses. We con-trolled for these potential biases in the pooled analysisthrough identification of common illnesses associated withGERD, including asthma (ICD 493), unexplained cough(ICD 782.2), hiatal hernia (ICD 553.3), dysphagia (ICD787.0 and ICD 787.2), unexplained epigastric pain (ICD789.0), and unexplained chest pain (ICD 786.59). We alsotested for differences of overall severity of illness betweenGERD and non-GERD populations by measuring differ-ences of prevalence and resource use for diabetes mellitis(ICD 250), angina pectoris (ICD 413), myocardial infarction(ICD 410), and heart failure (ICD 429).

RESULTS

Patient CharacteristicsThere were no statistically significant differences by any ofthe selection variables—age, gender, enrollment date, andtype of health insurance plan—between GERD and non-GERD populations (Table 2). Mean and median ages weresimilar at 43 yr for men and women in both cohorts. About25% of each population had no pharmaceutical benefitsplan. Participant accrual into the study by year had no effecton age and gender distribution.

The rates of diagnoses used as severity of illness indica-tors were low in both populations. However, they weresignificantly higher in the GERD population (Table 3). TheGERD population also had significantly more diagnoses

commonly associated with GERD than the non-GERDgroup, except unexplained cough. Here, too, the rates weregenerally low in both study groups, except for dysphagia inthe GERD population and epigastric pain in both groups.However, total numbers of diagnoses in both GERD andnon-GERD populations were similar.

Annual CostsThe mean cost of treating GERD across the 9 study yearswas $510 per year, with inconsistent increases and largeannual variations (Table 4). GERD costs were generallyabout 15% of total annual medical care costs for the GERDpopulation, ranging from a low of 13.7% in 1993 to a highof 24.7% in 1992. Medical costs of treating people withGERD were substantially higher than costs of care forpeople without GERD (i.e.,marginal costs of GERD). Theyvaried from a low of 20.9% higher in 1992 to as much as2.7-fold higher in 1998.

Costs of non-GERD care in the GERD population and allcare in the control population increased dramatically in

Table 2. Population Distribution by Age and Gender

Age

GERD Population Non-GERD Population

% Female % Male % Female % Male

,20 yr 8.4 7.1 9.0 8.020–29 yr 6.1 4.7 4.4 3.730–39 yr 6.7 7.9 8.0 7.740–49 yr 12.1 11.1 10.6 11.750–59 yr 8.4 9.3 8.5 9.060–69 yr 5.2 4.0 5.9 5.2$70 yr 4.2 4.6 3.8 4.4Total 51.3 48.7 50.3 49.7N 304 289 300 297

Table 3. Prevalence of Diagnoses Used as Severity of Illness Adjusters and Diagnoses Associated With GERD

DiagnosisPercent of GERD

(Population n5 591)Percent of Non-GERD(Population n5 589) p

Severity of illness adjustersDiabetes mellitus 5.3 2.9 0.04Myocardial infarction 4.9 2.7 0.05Angina pectoris 5.9 3.2 0.03Heart failure 5.3 2.9 0.04

Diagnoses associated with GERDAsthma 8.6 3.6 0.001Hiatal hernia 8.1 1.8 0.001Unexplained cough 1.0 0.7 0.5Unexplained chest pain 5.8 2.4 0.003Dysphagia 30.7 1.6 0.001Epigastric pain 45.0 16.1 0.001

Table 4. Mean Annual Medical Costs for GERD and Non-GERD Populations

Year

GERD Population Non-GERDPopulationGERD Non-GERD Total

1990 352 2,290 2,291 1,559SD 194 2,289 2,288 1,763

1991 404 2,571 2,608 1,368SD 532 5,670 5,680 3,594

1992 654 2,506 2,654 2,098SD 999 4,636 4,806 5,461

1993 595 4,194 4,344 2,438SD 560 12,112 12,162 5,922

1994 505 3,319 3,466 1,718SD 519 7,889 7,932 3,942

1995 553 2,784 2,955 1,627SD 520 5,725 5,747 4,432

1996 595 4,247 4,403 1,869SD 730 9,531 9,512 5,406

1997 508 6,441 6,710 4,143SD 676 14,270 14,310 17,607

1998 434 5,744 5,794 2,238SD 546 10,290 11,885 6,678

S66 Bloom et al. AJG – Vol. 96, No. 8, Suppl., 2001

1997. This was due to a sharp increase in the numbers ofnon-GERD diagnostic and therapeutic procedures in bothGERD and non-GERD populations during 1997. There wasno consistent pattern in distribution of procedures done, bydiagnoses and type (e.g., cardiovascular, orthopedic).Among GERD patients there was no detectable increase ofGERD or GI procedures in 1997. There was also no changein the low rate of operations related to GERD, even withincreasing popularity of laparoscopic fundoplication andincreasing incidence of carcinoma of the esophagus. Therewas no change in MCO policies that could affect rates ofdiagnostic and therapeutic procedures either. Thus, wecould find no explanation for this increase, and assumed itwas a random event.

Incremental Costs of GERDThe difference of total medical care costs between GERDand non-GERD populations can be assumed attributable tothe effects of GERD,ceteris paribus.In this study, theGERD population consistently had higher medical coststhan the non-GERD population because of higher non-GERD costs, diagnoses thought not to have any basis intheir GERD (Table 5). The cost differences due to signifi-cantly higher rates of severity of illness of diagnosis adjust-ers associated with GERD had an additive cost effect. How-ever, these costs were similar in both groups; it was thehigher prevalence of these diagnoses in the GERD popula-tion, not cost per episode, that contributed to increased totalcosts.

Costs Over TimeTime since original GERD diagnosis had little effect oncosts per year (Table 6). The marginal cost based on year ofinitial GERD diagnosis was inconsistently variable, thussupporting the observation that time was not an importantdeterminant of GERD care cost. Costs of care for the GERDpopulation declined after the first year after diagnosis. Theyremained level until the sixth year after diagnosis, when aslow and steady increase began. Among the non-GERDpopulation, costs of all care remained essentially flat exceptfor the seventh study year, when costs more than doubled.Again, this is the single year of the unexplained high rate ofdiagnostic and therapeutic procedures.

Mean annual cost was 2- to 3-fold greater than median.Approximately 2-fold annual SDs about the mean werecommon, but always for non-GERD care. A few outliersevery year with high cost non-GERD care skewed costs ofcare for both GERD and non-GERD populations. Nearlyalways this was due to a few (two or three) very expensivehospitalizations, none of which were related to GERD;payments were in the $80,000–200,000 range.

Distribution of CostsINPATIENT COSTS. GERD was rarely a primary or sec-ondary reason for hospitalization (Table 7). During the 9study years, there were only 38 hospitalizations for whichGERD was the primary reason. Mean annual hospitalizationcosts for GERD were consistently low, with the exception of1992 and 1997. Overall hospitalization costs were verysimilar: means of $11,207 for the GERD population and$11,230 for the non-GERD cohort. However, mean hospi-talization costs for non-GERD diagnoses in the GERD pop-ulation were 10–35% higher than for the control populationwhen people were hospitalized. Again, 1997 stands out: thecontrol population had inpatient care that was 31% morecostly than care for the GERD population. Hospital costsaccounted consistently for approximately 40% of total costsin GERD patients and 45% in the non-GERD cohort.

AMBULATORY COSTS. Ambulatory costs (mainly phy-sician and clinic visits and laboratory and X-ray services,

Table 5. Marginal (Incremental) Cost of Caring for the GERDPopulation

YearGERD

PopulationNon-GERDPopulation

Marginal Costof GERD

1990 2291 1559 7321991 2608 1368 12401992 2654 2098 5561993 4344 2438 19061994 3466 1718 17481995 2955 1627 13281996 4403 1869 25341997 6710 4143 25671998 5794 2238 3556

Table 6. Mean Annual Medical Costs for GERD and Non-GERD Populations Since Year of First GERD Diagnosis

Year

GERD PopulationNon-GERDPopulation

MarginalCost ofGERDGERD Non-GERD Total

1990 432 5529 5435 2368 30671991 504 3899 3717 2007 17101992 554 4633 4378 2367 20111993 490 4514 4174 1674 25001994 528 3386 3081 1564 15171995 560 3881 3547 1794 17531996 475 4493 4103 4634 25311997 460 4544 4032 1920 21121998 231 2433 4153 822 3331

Table 7. Mean Annual Inpatient Costs for GERD and Non-GERD Populations

Year

GERD PopulationNon-GERDPopulation

MarginalCost ofGERDGERD Non-GERD Total

1990 NA 775 775 533 2421991 NA 1286 1286 506 7801992 39 791 830 969 21391993 13 1823 1836 1118 7181994 5 1415 1420 688 7271995 7 1177 1184 774 4101996 8 1794 1805 905 9001997 14 3118 3132 1876 12561998 13 2261 2274 876 1398

NA 5 not available.

S67AJG – August, Suppl., 2001 Time Trends in Cost of Caring for People With GERD

but excluding prescription pharmaceuticals) were consistentover time in the non-GERD population (Table 8). Amongthe GERD group, stable costs from 1990 to 1994 werefollowed by a sharp increase beginning in 1996 that led toa doubling of outpatient costs by 1998. Across study years,ambulatory costs were about 50% of total costs in both theGERD and control populations.

AMBULATORY PRESCRIPTION PHARMACEUTICALCOSTS. The cost of ambulatory prescription pharmaceuti-cals accounted for about 12% of total costs in the GERDgroup and 9% in the non-GERD group (Table 9). However,pharmaceuticals were about 40% of GERD costs in theGERD group.

Ambulatory medication costs increased slowly for thenon-GERD population during the study period, from about$150 per year to about $220 per year, about one fourth toone third the cost of all pharmaceuticals. In the GERDpopulation, both GERD and non-GERD pharmaceutical ex-penditures increased inconsistently but in parallel over time,from about $300 to about $800 per year.

Pooled Time Series AnalysisThe pooled (combined longitudinal and cross-sectional) re-gression analysis found significantly higher costs for GERDand non-GERD populations for diagnoses used as severityof illness adjusters and diagnoses commonly associated with

GERD (R2 5 0.522). Total cost in GERD and non-GERDcohorts was explained by the independent variables of se-verity of illness adjusters (diabetesp 5 0.0003, anginap 50.0002, myocardial infarctionp , 0.0001, and heart failurep 5 0.0001) and symptoms commonly associated withGERD (dysphagiap 5 0.002, hiatal herniap 5 0.001,asthmap 5 0.0002, unexplained coughp 5 0.001, unex-plained chest painp 5 0.0002, and unexplained epigastricpain p 5 0.0002). The y intercept (GERD5 $44,485;non-GERD5 $23,485) indicates the significant differencein mean costs of care for the two populations.

DISCUSSION

The costs of GERD treatment were relatively low—in thelate 1990s about $500 per year, but rising and falling un-evenly each year. About 40% of GERD costs were forprescription pharmaceuticals. GERD represented about 10–15% of total medical costs for people with the condition,with non-GERD care the large majority of costs. GERDhospital costs were very small indeed. Our results werecomparable to those of other recently published studiesmeasuring direct medical costs (1–3) as well as estimatesfrom other countries (4).

The cost of treating GERD, however, is very differentfrom the cost of caring for people with GERD. People withGERD incurred approximately 2-fold higher overall medi-cal care costs than a comparable population without GERD.The main cost driver in the GERD population was theirnon-GERD diagnoses and treatments.

There are two likely explanations for cost differencesbetween populations with and without GERD. First, al-though both populations were randomly chosen from theirrespective large universes after stratification by age, gender,length of time in the MCO, and health insurance coverage,the GERD population selected may have been sicker, asdefined by expenditures for diabetes mellitis, angina pecto-ris, myocardial infarction, and heart failure. The pooledregression analysis found that disease severity variables, allwith low prevalence rates, were significantly related tohigher total costs in the GERD population. Thus, there maybe causality between GERD and non-GERD diseases, al-though the direction is unknown.

A second explanation for cost differences, one which wethink more likely, is that sick people are sick, and that as thenumber of diseases a person has increases arithmetically thecosts of caring for all diseases increases geometrically (10).It may be that GERD, like other chronic diseases, addsdisproportionately to the entire burden of illness (11). Thishas been found for low risk disease like GERD and high riskdisease like Alzheimer’s dementia (11, 12). It may be alsothat inappropriately treated GERD leads to emergence orrecurrence of symptoms of non-GERD conditions. This mayoccur in diseases associated with GERD, like asthma, ordiseases of unknown association, like diabetes. We also donot know if the reverse is true. Does cost decline for persons

Table 8. Mean Annual Ambulatory Costs for GERD and Non-GERD Populations

Year

GERD PopulationNon-GERDPopulation

MarginalCost ofGERDGERD Non-GERD Total

1990 12 1505 1517 1027 4901991 37 1283 1320 860 4601992 43 1536 1579 1127 4521993 56 2143 2190 1339 8511994 45 1679 1724 1025 6991995 31 1470 1501 817 6841996 43 2307 2350 929 14211997 124 3097 3162 1248 19141998 160 3237 3367 1352 2015

Table 9. Mean Annual Pharmaceutical Costs for GERD andNon-GERD Populations

Year

GERD PopulationNon-GERDPopulation

MarginalCost ofGERDGERD Non-GERD Total

1990 NA NA NA NA NA1991 NA NA NA NA NA1992 65 188 294 149 1451993 100 238 373 154 2191994 100 257 438 169 2691995 181 399 528 227 3011996 190 510 825 251 5741997 221 479 867 239 6281998 194 446 724 215 509

NA 5 not available.

S68 Bloom et al. AJG – Vol. 96, No. 8, Suppl., 2001

with multiple diagnoses if any single disease is treatedappropriately (13)? Does overall cost decline geometricallyif all diseases are appropriately treated?

LimitationsThe first limitation is that we were unable to attribute coststo the various forms of GERD, low risk and cost likeheartburn to higher risk and cost like erosive esophagitis andBarrett’s esophagus. Often the same diagnostic workup andtreatments were used, and coding was rarely specific.

Second, there were no data on patient and family out-of-pocket costs. Also, there were no data on over the counteruse of GERD medications—for example, histamine-2 re-ceptor antagonists for dyspepsia arising from use of nonste-roidal anti-inflammatory drugs.

Next, we did not attribute costs across multiple diagnosesfor the same ambulatory visit or hospitalization. We dealtwith the issue of joint product by attributing cost only toGERD if the first-listed diagnosis for the visit or hospital-ization was related to GERD. This may underestimate truecosts of GERD. To the contrary, we attributed all GERDmedication use and GI diagnostic tests without specificdiagnosis codes to GERD costs, which will likely overesti-mate costs of GERD care.

Fourth, a potential bias was that persons with no orlimited pharmaceutical benefits, about 25% of the enrolledpopulation, were excluded from the study. Last, only peoplewho come to the MCO for care could be included in thestudy; those who self-treated were excluded.

In conclusion, people with GERD are costly to care for,relative to people who do not have GERD. Surprisingly, thisis due overwhelmingly to the costs of treating non-GERDdiagnoses. It appears that symptoms experienced by peoplewith GERD affect these other conditions as well. In essence,poor health signaled by greater expenditures for all care is afunction of the synergy among all diagnoses. GERD adds itsown nonlinear effects to this total disease burden. Thephenomenon of arithmetic increase of diagnoses associatedwith geometric increase in costs of care appears with manydiseases (10, 11). However, we do not know if appropriateGERD treatment will return overall medical service use to alevel at or near that of people without GERD. Finally, thisstudy shows clearly that a diagnosis, particularly that of a

chronic disease, cannot be examined in isolation, from ei-ther a clinical or an economic perspective.

Reprint requests and correspondence:B. Bloom, Ph.D., Divi-sion of Geriatrics, Department of Medicine, University of Penn-sylvania, 3615 Chestnut Street, Philadelphia, PA 19104-2676.

Received Oct. 6, 2000; accepted Dec. 26, 2000.

REFERENCES

1. Lair TJ, Anderson DR, Crawley JA. Utilisation and expendi-tures in the treatment of patients with gastro-intestinal refluxdisease. Pharmacoeconomics 1998;15:325–36.

2. Keyser MS, Crawley JA, Goldberg GA, Shaw MM. Gastro-esophageal reflux disease in a managed care setting: Chargesby place and type of service. J Managed Care Pharmacy1998;3:336–43.

3. Levin TR, Schmittdiel JA, Kunz K, et al. Costs of acid-relateddisorders to a health maintenance organization. Am J Med1997;103:520–8.

4. Stålhammar N-O, Carlsson J, Mu¨ller-Lissner S, et al. Costeffectiveness of omeprazole and ranitidine in intermittanttreatment of symptomatic gastro-oesphageal reflux disease.Pharmacoeconomics 1999;16:483–97.

5. Bloom BS, Jacobs J. Cost effects of restricting cost-effectivetherapy. Med Care 1985;23:872–80.

6. Bloom BS, Fox NA, Jacobs J. Patterns of care and expendi-tures by California Medicaid for peptic ulcer, and other acid-related diseases. J Clin Gastroenterol 1989;11:615–20.

7. Bloom BS, Fendrick AM, Ramsey SD. Changes in pepticulcer, and gastritis/duodenitis in Great Britain, 1970-1985.J Clin Gastroenterol 1989;12:100–8.

8. Bloom BS, Gessner U. Long-term technology assessment.Mortality, hospitalization, and work-loss due to peptic ulcer,and gastritis/duodenitis in Federal Republic of Germany. Int JTechnol Assess Health Care 1989;5:215–26.

9. Bloom BS. Cross-national changes of effects of peptic ulcerdisease. Ann Intern Med 1991;114:558–62.

10. Bloom BS, Tibi-Levy Y, Harari A, Fendrick AM. Directmedical costs of unstable angina pectoris in a defined popu-lation. J Managed Care Pharmacy 1999;5:39–44.

11. Taylor DH, Sloan FA. How much do persons with Alzhei-mer’s disease cost Medicare? J Am Geriatr Soc 2000;48:639–46.

12. Weiner M, Powe NR, Weller WE, et al. Alzheiner’s diseaseunder managed care: Implications from Medicare utilizationand expenditure patterns. J Am Geriatr Soc 1998;46:762–70.

13. Lanes SF, Birmann BM, Walker AM, et al. Characterisation ofasthma management in the Fallon Community Health Planfrom 1988 to 1991. Pharmacoeconomics 1996;10:378–85.

S69AJG – August, Suppl., 2001 Time Trends in Cost of Caring for People With GERD