A atioth l R

sh

1.

2.

3.

diagabrothetioMaAGencoisstiotrecopreideevabederevthevicThthimeseaquara

diagnosis recorded on outpatient physician visits since2006, even surpassing abdominal pain.1 This is remark-able considering the vagaries of the diagnosis. Most pa-tieforexaatypash

demethesunoresa rbeinhesowh

liteinitwarbopbaquallemerastagrotiobeemanThofsynwamemeevi

An overarching definition of GERD must encompassesophageal as well as extraesophageal syndromes: syn-

AGA

INSTITU

TE

GASTROENTEROLOGY 2008;135:13921413nts with heartburn do not have esophagitis, even be-e treatment,2 and this disconnect becomes moreggerated after empirical antisecretory therapy or withpical GERD symptoms. Furthermore, although the

Abbreviations used in this paper: AGA, American Gastroenterologi-cal Association; GERD, gastroesophageal reux disease; H2RA, hista-mine2 receptor antagonist; PPI, proton pump inhibitor; USPSTF, USPreventive Services Task Force.merican Gastroenterological Associe Management of Gastroesophagea

Learning ObjectivesAt the end of this activity, the successful learner

ould:

Demonstrate an understanding of the natural historyand manifestations of reflux diseaseEvaluate the role of diagnostic testing such as endos-copy, esophageal manometry, ambulatory pH moni-toring, and impedance-pH monitoring in the manage-ment of patients with gastroesophageal reflux diseaseEvaluate the options for treatment of patients withcomplicated and uncomplicated reflux disease.

The objectives of this technical review were to evaluategnostic and management strategies for patients withstroesophageal reflux disease (GERD). Specifically, 12ad questions were developed by interaction amongauthors, the American Gastroenterological Associa-

n (AGA) Institute, the Clinical Practice and Qualitynagement Committee, and representatives from theA Institute Council. The questions were designed tocapsulate the major management issues leading tonsultations for GERD in clinical practice in 2008. Theue of management of Barretts esophagus was inten-nally excluded, because this will be the focus of a lateratise. However, the indications for performing endos-py were within our purview. For each question, a com-hensive literature search was conducted, pertinent ev-nce reviewed, and the quality of relevant dataluated. The resultant conclusions were based on thest available evidence or, in the absence of quality evi-nce, the expert opinion of the authors of the technicaliew and medical position statement. The strength ofse conclusions was weighed using US Preventive Ser-es Task Force (USPSTF) grades detailed in Table 1.e details of the development methodology used fors and subsequent AGA Institute technical reviews anddical position statements as well as the literaturerch methodology and yield associated with each of theestions in this technical review are available as a sep-te document on the AGA Institute Web site.GERD has been the most common gastrointestinalthogenesis of reflux esophagitis and reflux symptomsare common elements, the two have several indepen-n Institute Technical Review oneflux Disease

nt determinants as well. Finally, with respect to treat-nt, potent inhibition of gastric acid secretion reduceslethality of gastric juice to esophageal epithelial cells

ch that esophagitis will heal, irrespective of whether ort gastroesophageal reflux is reduced or symptoms areolved. Ironically, as refractory esophagitis has becomeare clinical problem, refractory GERD symptoms hascome a substantial one. In the postproton pumpibitor (PPI) world, it is patients with symptoms, notphagitis, who confront the practitioner in the over-elming majority of clinical encounters for GERD.

Diagnosis and Initial Therapy1. What Is an Operational Definition ofGERD? What Is the Distinction BetweenGERD and Episodic Heartburn?Regardless of how many citations are identified by

rature review, there can be no criterion standard def-tion of GERD because the threshold distinction be-een physiologic reflux and reflux disease is ultimatelyitrary. Hence, these questions can only be answered byinion, and presumably the best opinion upon which tose the answers is that of experts (USPSTF grade andality not applicable). Fortuitously, a recent and unpar-eled attempt at gaining consensus in defining GERDanated from a panel of world experts utilizing a 4-it-tion Delphi process that spanned a 2-year period.3 Theted objective of that unique international consensusup was to develop a global definition and classifica-n of GERD, using rigorous methodology, that couldused clinically by primary care physicians and thatbraces the needs of physicians, patients, researchers,d regulatory bodies from different parts of the world.3

e output of the Montreal consensus group was a series50 statements pertaining to the diagnosis of GERDdromes. For each statement, the level of consensuss determined by vote along a 6-point scale of agree-nt ranging from strong agreement to strong disagree-nt and, when applicable, the quality of supportingdence was evaluated. 2008 by the AGA Institute0016-5085/08/$34.00

doi:10.1053/j.gastro.2008.08.044

droplireacococatweuasyn(Ficlakntoest

tiobetgro

drotroepindanclinbutheaftsioconhesubmo

meunbeinsymbunigDrdepopablemiwiThcoherelpatiebysureplidferintpathefulwogroHetieoldtoeffino

Ta

StrA:

stc

B:e[t

C:pecj

D:sfo

InsrEod

QuGo

wa

Faionge

Poooie

NO5 candgrasca

AGA

INSTITUTE

October 2008 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE 1393mes with tissue injury as well as those without. Grap-ng with this dilemma, the Montreal consensus panelched very strong agreement in defining GERD as andition which develops when the reflux of stomachntents causes troublesome symptoms and/or compli-ions.3 Thereafter, the panel specified that symptomsre troublesome if they adversely affected an individ-ls well-being and delineated the broad array of GERDdromes that have been demonstrated or proposedgure 1). Note that the extraesophageal syndromes aressified as of established or proposed association, ac-owledging that while the evidence on hand is sufficientlink these syndromes to reflux, it is insufficient toablish causation.A test of the word troublesome in the Montreal defini-

ble 1. USPSTF Recommendations and Grades

ength of recommendationsThe USPSTF strongly recommends that clinicians provide [theervice] to eligible patients. The USPSTF found good evidencehat [the service] improves important health outcomes andoncludes that benefits substantially outweigh harms.The USPSTF recommends that clinicians provide [this service] toligible patients. The USPSTF found at least fair evidence thatthe service] improves important health outcomes and concludeshat benefits outweigh harms.The USPSTF makes no recommendation for or against routinerovision of [the service]. The USPSTF found at least fairvidence that [the service] can improve health outcomes butoncludes that the balance of benefits and harms is too close toustify a general recommendation.The USPSTF recommends against routinely providing [theervice] to asymptomatic patients. The USPSTF found at leastair evidence that [the service] is ineffective or that harmsutweigh benefits.uff: The USPSTF concludes that the evidence is insufficient toecommend for or against routinely providing [the service].vidence that the [service] is effective is lacking, of poor quality,r conflicting and the balance of benefits and harms cannot beetermined.

ality of evidenceod: Evidence includes consistent results from well-designed,ell-conducted studies in representative populations that directlyssess effects on health outcomes.r: Evidence is sufficient to determine effects on healthutcomes, but the strength of the evidence is limited by theumber, quality, or consistency of the individual studies,eneralizability to routine practice, or indirect nature of thevidence on health outcomes.r: Evidence is insufficient to assess the effects on healthutcomes because of limited number or power of studies,mportant flaws in their design or conduct, gaps in the chain ofvidence, or lack of information on important health outcomes.

TE. The USPSTF grades its recommendations according to one oflassifications (A, B, C, D, Insuff) reflecting the strength of evidencemagnitude of net benefit (benefits minus harms). The USPSTF

des the quality of the overall evidence for a service on a 3-pointle (good, fair, poor).n of GERD comes in attempting to draw a distinctionween GERD and episodic heartburn. The Montrealup believed that for the typical esophageal GERD syn-

admetiome, defining a threshold at which heartburn becomesublesome was useful in planning treatment trials oridemiologic studies but not in clinical practice. In theividual case, symptom frequency, symptom intensity,d psychosocial factors must also be considered. Hence, inical practice, the determination of whether or not heart-rn is troublesome should be made by the patient withoutuse of arbitrary cutoffs for frequency or duration and

er the patient is assured of the benign nature of occa-nal symptomatic heartburn. Presumably, a patient willclude that if heartburn regularly interferes with his or

r normal daily activities, it is troublesome; the morestantial the limitations imposed on his or her life, there troublesome it is.In clinical trials of symptomatic GERD, a thresholdasure of heartburn severity needs to be established foriformity in the study population. However, there hasen little consistency among symptomatic GERD trialshow this was done, with different studies utilizingptom thresholds as low as 2 mild episodes of heart-

rn per week and as high as 5 daytime episodes and 1httime episode per week as minimal entry criteria.4,5

opping the threshold of heartburn severity required tofine symptomatic GERD clearly enlarges the diseasepulation, and variability in the definition makes com-risons of results among trials difficult, if not impossi-. Moving forward with this, the Food and Drug Ad-nistration recently issued guidance on the criteria thatll be required in the future to support labeling claims.6

e catchphrase has become patient-reported out-mes. Patient-reported outcomes measure a patientsalth status in terms of symptoms and quality of life asayed by the patient without the interpretation of thetients responses by a physician. Disease-specific pa-nt-reported outcomes must be developed and acceptedthe Food and Drug Administration before their use inpporting a labeling claim. Acceptance of a patient-orted outcome is predicated on its demonstrated va-ity, reliability, and ability to identify meaningful dif-ences in disease-specific measures of importance in theended treatment population. For the example of atient-reported outcome for symptomatic GERD, bothdefining symptom burden and the minimal meaning-increment of improvement attributable to therapyuld need to be developed through patient focusups and vetted by the Food and Drug Administration.nce, the development of a symptomatic GERD pa-nt-reported outcome will result in defining the thresh-symptom burden required for the diagnosis. Needlesssay, few, if any, existing trials substantiating treatmentcacy in symptomatic GERD would meet the criteriaw mandated by the Food and Drug Administration.In summary, the Montreal definition of GERD was

opted to use as a framework throughout this docu-nt. A distinguishing feature of the Montreal defini-n is that it does not use the term nonerosive reflux

disintesonoinceashetrhetroniga s

gaspfoolatprefooesotiobeingobsizasiesdimHolow

inmopreme

6redaneatbeeverec

Fig andareg by re

AGA

INSTITU

TE

1394 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE GASTROENTEROLOGY Vol. 135, No. 4ease, but rather subdivides esophageal syndromeso symptomatic syndromes and syndromes withphageal injury. Hence, functional heartburn doest fit the Montreal definition of GERD, whereas it isluded under the umbrella of nonerosive reflux dis-e. The distinction between GERD and episodicartburn in the Montreal definition is in the wordoublesome. In the absence of esophageal injury,artburn of insufficient intensity to be perceived asublesome by the patient (after assurance of its be-n nature) does not meet the Montreal definition ofymptomatic esophageal GERD syndrome.

2. What Is the Efficacy of LifestyleModifications for GERD? Which ElementsShould Be Recommended and in WhichCircumstances?There are a multitude of recommendations re-

rding lifestyle modifications as GERD therapy. Broadlyeaking, these fall into 3 categories: (1) avoidance ofds that may precipitate reflux (coffee, alcohol, choco-e, fatty foods), (2) avoidance of acidic foods that maycipitate heartburn (citrus, carbonated drinks, spicyds), and (3) adoption of behaviors that may reducephageal acid exposure (weight loss, smoking cessa-n, raising the head of the bed, and avoiding recum-ncy for 23 hours after meals). Most evidence support-

such recommendations is weak, coming fromservational and uncontrolled studies of small samplee, often with surrogate end points. In some cases, suchwith dietary fat, the evidence is conflicting. Early stud-suggested that a high-fat diet was harmful because it

ure 1. The Montreal definition of GERD. The overarching definition mardless of syndrome(s) present and that those syndromes are causedinished lower esophageal sphincter (LES) pressure.7

wever, a subsequent controlled study comparing a-fat diet with a high-fat diet did not find any change

boreflpaLES pressure or objective reflux parameters by pHnitoring.8 Similarly, the reported reduction in LESssure with smoking has not extrapolated to improve-nt of GERD parameters with cessation of smoking.9

The recommendation to elevate the head of the bed by8 inches in patients with reflux is intuitively based onucing esophageal acid exposure by improving clear-ce. A corollary to this recommendation is to avoiding for the 2- to 3-hour period before going to bed,cause this is the period during which the most refluxnts would be anticipated. There is some merit to thisommendation based on a randomized controlled trialpatients with moderately severe esophagitis (USPSTFde B, quality fair).10 The therapeutic gain from raisinghead of the bed with a 20-cm block for the 6-week

ration of the study was 20%30%. However, the sub-up studied (moderate to severe esophagitis) is partic-rly prone to supine reflux and the applicability of theommendation of elevation of the head of the bed tomajority of patients with GERD experiencing heart-

rn predominantly confined to the postprandial perioddubious.Obesity merits special consideration because it hasen the object of substantial study in recent years. Theregood evidence that GERD is associated with obesity.ecifically, epidemiologic data from the Nurses Healthdy suggest a dose-dependent relationship betweenreasing body mass index and frequent reflux symp-s,11 and a large meta-analysis similarly demonstratedose-response relationship between body mass indexd the risk of reporting symptoms of GERD among

tes that troublesome symptoms and/or complications are presentflux.3ofgrathedugroularecthebuis

beisSpStuinctoma danth men and women.12 Evidence also suggests that acidux measured by pH monitoring is increased in obesetients.13 Proposed mechanisms for the obesity effect

incoftrathelesB,animHeexaweranoraft

ificmawhonthsibmoquconGEtheliferectomhisfrowhingsupfacbuproatthe

totomGEnotintivpealstha

resbyresa nasdiawhGEwhlimemhe

abPPrecincclusym(HreveffizozododasyminthefroTheso(Udethegodotheadtheprotheeveclisutheextproagasuextass

AGA

INSTITUTE

October 2008 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE 1395lude alteration in the pressure dynamics and anatomythe esophagogastric junction14 and an increase innsient LES relaxation and reflux frequency.15 However,contention that losing weight will improve GERD is

s robustly supported by the literature (USPSTF gradequality fair). One observational study of 34 overweightd obese patients found that weight loss resulted inprovement in GERD symptoms,16 and the Nursesalth Study analysis found that reflux symptoms werecerbated or improved over time concomitant withight gain or loss, respectively.11 On the other hand, adomized controlled study did not find any objectivesubjective improvement of GERD in 20 obese patientser a significant weight loss.17

In summary, the problem with advocating lifestyle mod-ations as GERD therapy is that there are simply toony and each is too narrowly applicable to enforce theole set on every patient. The Genval Workshop Reportevidence-based reflux management concluded that

ere is currently a significant overestimation of the pos-ility of patients deriving adequate relief from life styledification.18 Thus, from the vantage point of high-ality evidence, there are insufficient data to suggest asistent benefit of lifestyle changes for all patients withRD (USPSTF grade Insuff). However, it is also clear thatre are subsets of patients who may benefit from specificstyle modifications and it is good practice to make thoseommendations to those patients. A patient with symp-s of nighttime heartburn of sufficient severity to disturbor her sleep despite acid suppressive therapy may benefitm elevation of the head of the bed. Similarly, a patiento consistently experiences troublesome heartburn afterestion of alcohol, coffee, or spicy foods despite acidpressive therapy will benefit from avoidance of thesetors. Finally, if the development of troublesome heart-rn paralleled weight gain, it is perfectly reasonable topose weight loss as an intervention that may prevent, orleast postpone, the need for continuous acid suppressiverapy.

3. How Do Antisecretory Therapies Comparein Efficacy and Under What CircumstancesMight One Be Preferable to Another? WhatIs an Acceptable Upper Limit of EmpiricalTherapy in Patients With Suspected TypicalEsophageal GERD Syndromes BeforePerforming Esophagogastroduodenoscopy?Initiating empirical treatment with a PPI amounts

a pragmatic therapeutic trial; if a patient reports symp-s consistent with GERD and responds to therapy forRD, then he or she must have GERD. However, thetion that this constitutes a clinical test blurs the dis-ction between healing esophagitis and resolving puta-e reflux symptoms, the latter of which are neither

rfectly sensitive nor specific for GERD. Such reasoningo ignores the existence of other diagnostic possibilitiest may benefit from PPI therapy or a possible placebo

shwitheponse. The strategy of empirical therapy is also limitedthe observation that PPI therapy is not as robust atolving GERD symptoms as it is resolving esophagitis;egative response to a PPI trial does not exclude GERDa diagnostic possibility. Furthermore, it dampens thegnostic utility of endoscopy because esophagitis,ich would otherwise have been a robust marker ofRD, will likely be healed with treatment regardless ofether or not symptoms resolve. Nonetheless, theseitations withstanding, the current consensus is thatpirical PPI therapy is appropriate for uncomplicatedartburn.3,19,20

As for the choice of therapeutic agent, there is anundance of data demonstrating the effectiveness ofIs in healing esophagitis and in relieving heartburn. Aent Cochrane review examined 134 treatment trialsluding 36,978 patients with esophagitis and con-ded that PPIs exhibit a better healing effect and fasterptom relief than histamine2 receptor antagonists

2RAs), which are in turn better than placebo.21 Thatiew also concluded that there is no major difference incacy among the currently available PPIs (esomepra-le, lansoprazole, omeprazole, pantoprazole, rabepra-le) and that the gain achieved by doubling the standardse of PPI therapy is modest. Available 6- to 12-monthta also suggest that PPIs are effective for maintainingptom relief and preventing recurrence of esophagitis

patients who respond to an acute course of the samerapy. Table 2 summarizes the major observationsm this voluminous pool of therapeutic data.2231

us, abundant data support treating patients withphageal GERD syndromes with antisecretory drugsSPSTF grade A, quality good) and there is ample evi-nce that, as a drug class, PPIs are more effective inse patients than are H2RAs (USPSTF grade A, qualityod). However, the data supporting the use of PPIs withses higher than the standard are minimal. Similarly,re is no evidence of improved long-term efficacy byding a nocturnal dose of an H2RA to twice-daily PPIrapy (USPSTF grade Insuff). This combination wasposed to suppress nocturnal acid breakthrough onassumption that this was clinically relevant.32 How-r, subsequent data have failed to show any associatednical benefit.3335 There are also no high-quality datapporting the use of metoclopramide as either mono-rapy or adjunctive therapy in esophageal or suspectedraesophageal GERD syndromes, making the toxicityfile of the drug36 ample grounds for recommendinginst its use (USPSTF grade D, quality fair). Finally, datapporting the use of PPIs for treatment of patients withraesophageal GERD syndromes with an establishedociation are weak (USPSTF grade B, quality fair).The data in Table 2 are very robust with respect to the

ort-term healing of esophagitis but much more limitedth respect to defining the parameters for maintenancerapy or the management of patients with an inade-

qucalThanaltsuhakecoexprecsymsyndocipanthaofsomsymtheisdieTh

Ta astr

Eso placcurvor spH2R

o plae-res

Hee

plac-respor spH2Rto pl

Hen

plac-respH2R

0.73to ple-res

Mas

placpy isphagd PPive pa

Exte

no b

w momaticose P

NOAll15NNeac

AGA

INSTITU

TE

1396 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE GASTROENTEROLOGY Vol. 135, No. 4ate symptom response to once-daily PPI therapy, basi-ly the 2 most common issues faced by the clinician.e other notable disconnect between clinical trial datad clinical practice is in the use of PPIs twice daily;hough the pharmacodynamics of the drugs logicallypports twice-daily dosing, the pharmaceutical industrys been reluctant to advocate it, primarily due to mar-ting considerations. Hence, guidance on these issuesmes primarily from expert opinion based on clinicalerience. Expert opinion is essentially unanimous inommending twice-daily dosing of PPIs to improveptom relief in patients with an esophageal GERDdrome with an unsatisfactory response to once-dailysing (USPSTF grade B, quality fair). The general prin-le of maintenance therapy is to titrate the strength oftisecretory therapy up or down to find the lowest doset provides satisfactory control of heartburn. As a rulethumb, 80% symptom relief is a reasonable, albeitewhat arbitrary, target; at that point, persistentptoms are often triggered only by indulgence. Fur-rmore, the significance of these unresolved symptoms

ble 2. Summary of GERD Treatment Data With Inhibitors of G

phagitis healing (all severities) PPIs are superior toPPIs dose-response

standard vs highPPIs are superior toH2RA are superior tH2RAs show no dos

artburn resolution (patients withsophagitis)

PPIs are superior toPPIs show no dose

standard vs highPPIs are superior toH2RAs are superior

artburn resolution (endoscopyegative or uninvestigated patients)

PPIs are superior toPPIs show no dosePPIs are superior to

interval 0.60H2RAs are superiorH2RAs show no dos

intenance of esophagitis healing orymptom control (6-12 months)

PPIs are superior toLow-dose PPI thera

patients with esoLow-dose on-deman

endoscopy-negatraesophageal syndromes (withstablished association)

High-dose PPIs areheartburn29

High-dose PPIs shosubgroup of asth

Standard- or high-dsyndrome with ob

TE. Low and standard doses of PPIs are as follows: esomeprazole 20omeprazole and rabeprazole data were 20 mg. Standard doses of H0 mg, ranitidine 150 mg.T, estimated number of patients needed to treat to demonstrate thish group.only in the lifestyle compromises they impose: limitedt, restricting physical activity, poor sleep, and so on.ey should not be framed as a threat to ones well-being.

someffingAlthough PPIs are more effective overall, H2RAs have are rapid onset of action and will suffice for sometients. If patients need to take a PPI twice daily becausey are experiencing breakthrough symptoms towardend of the day or during the night, the optimaling is 3060 minutes before breakfast and dinner.tients may find on-demand or intermittent shorturses of therapy sufficient. However, antacids are mostective once heartburn is already present; PPIs andRAs are more effective in preventing heartburn. Pa-nts whose heartburn has not adequately responded toice-daily PPI therapy should be considered treatmentlures; in other words, that is a reasonable upper limitempirical therapy.

Circumstances in which one antisecretory drug mightpreferable to another primarily relate to side effects oren the onset of effect is a prime consideration. Thest common side effects of PPIs are headache, diarrhea,nstipation, and abdominal pain, none of which occurre frequently than with placebo, but all of which cancur with any drug in the class and can be confirmed in

ic Acid Secretion

ebo: 83% vs 18% at 8 wk, NNT 1.721

e exhibits a plateau: low vs standard dose, NNT 10 at 4 wk;lit dose, NNT 25 at 4 wk21

As: 84% vs 52%,20 RR 0.5121

cebo: 41% vs 20% at 6 wk, NNT 521

ponse curve (standard vs high or split dose)21

ebo: 56% vs 8% at 4 wk, NNT 2322

onse curve: low vs standard dose, 75% vs 79% at 4 wk;lit dose, 73% vs 76% at 4 wk21

As: 77% vs 48% at 412 wk23

acebo: 56% vs 45% at 12 wk24

ebo: 36.7% vs 9.5%, NNT 3422

onse curve (low vs standard dose)As: 61% vs 40%, NNT 5,25 RR 0.66, 95% confidence26

acebo: RR 0.77, 95% confidence interval 0.600.9926

ponse curve (standard vs high dose): 45.8% vs 44.8% at 8 wk27

ebo for maintaining healing: 93% vs 29%28

sufficient to maintain endoscopic remission in 35%95% ofitis20

I therapy yields acceptable symptom control in 83%92% oftients20

etter than placebo for reflux laryngitis syndrome without frequent

dest benefit in morning peak expiratory flow vs placebo forpatients with nocturnal GERD symptoms30

PIs show some improvement in some patients with reflux coughvely demonstrated GERD: NNT 531

40 mg; lansoprazole 15 mg, 30 mg; pantoprazole 20 mg, 40 mg.(all twice daily): cimetidine 400 mg, famotidine 20 mg, nizatidine

fit; RR, risk ratio, compares the probability of treatment failure inmopathethetimPacoeffH2tietwfaifor

bewhmocomooc

jecti

mg,2RAs

benee patients with a test-retest strategy. Potential sideects should also be considered in wholesale transition-of patients from one PPI to another, as shown by the

28lanorpalowforenpriinttheactsuHoap

wievemodifgraoftagwionactap

critomofenBatomintarebrocoofingthoadqu

plicoAlatesbepreIm

disnaorvounepinvsuingis qmealaintevaorhathopeovcointwianbediaevagiaIfwi

symmaofBaalsanallinvdypatinwhspetheesoualowagefrotisasvalGE

AGA

INSTITUTE

October 2008 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE 1397% of patients who failed the switch from omeprazole tosoprazole because of either therapeutic failure (15%)side effects (13%) in a VA hospital experience of 78tients.37 Switching among alternative PPI drugs or to aer dose can usually circumvent side effects; it is rarea patient to exhibit intolerance to any dose of the

tire drug class. As for the issue of onset of action, thismarily pertains to on-demand therapy. If a patientends to take a drug only in response to symptoms,n it should be a rapidly acting drug. The most rapidlying agents are antacids, the efficacy of which can bestained by combining them with an H2RA or a PPI.wever, the clinical benefits of such combination ther-ies have yet to be demonstrated in clinical trials.Some data suggest differences among the various PPIsth respect to healing of erosive esophagitis.38,39 How-r, absolute differences in efficacy in these studies aredest,21 and subgroup analyses have suggested thatferences in healing rates may be more pronounced inde C and D disease. Given that these medications areen given empirically and that grade C and D esoph-itis will be present in only a small fraction of thoseth GERD symptoms, medication selection predicatedprescription plan coverage, side effects, and onset ofion will likely govern the selection of the initial ther-y for most patients.

4. What Is the Role and Priority of DiagnosticTests (Endoscopy With or Without Biopsy,Esophageal Manometry, Ambulatory pHMonitoring, Combined Impedance-pHMonitoring) in the Evaluation of Patients WithSuspected Esophageal GERD Syndromes?As evident in Figure 1, widely accepted diagnostic

teria for esophageal GERD syndromes are (1) a symp-complex attributable to gastroesophageal reflux and

sufficient severity to compromise quality of life or (2)doscopic findings of erosive esophagitis, stricture, orrretts metaplasia. In the simplest case, when symp-s are typical and the patient responds to therapy

ended to address those symptoms, no diagnostic testsrequisite. Rather, diagnostic testing is invoked in 3ad scenarios: (1) to avert misdiagnosis, (2) to identify

mplications of reflux disease, and (3) in the evaluationempirical treatment failures. With respect to perform-endoscopy to screen for Barretts metaplasia, al-

ugh clearly a mainstream issue, this is specificallydressed in the section on chronic management (seeestion 11 in the following text).The discussion of misdiagnosis and identifying com-cations of reflux disease usually revolves around thencept of alarm features found on clinical evaluation.rm features dictate circumstances in which diagnosticting is indicated as part of the initial evaluation, either

cause they suggest that complications of GERD aresent or because they suggest an alternative diagnosis.portant alternative diagnoses include coronary artery

oftobeeease, gallbladder disease, gastric or esophageal malig-ncy, peptic ulcer disease, and eosinophilic, infectious,caustic esophagitis. Proposed alarm features includemiting, evidence of gastrointestinal blood loss, invol-tary weight loss, dysphagia, anemia, chest pain, origastric mass.20,40 It is not always endoscopy that isoked, but some specific evaluation as mandated by thespected disease entity. High-quality evidence support-the broad utility of alarm features as a diagnostic tooluite limited (USPSTF grade Insuff). However, a recentta-analysis addressed the specific issue of the utility ofrm signs and symptoms in diagnosing upper gastro-estinal malignancy based on 15 published prospectiveluations encompassing 46,161 patients, 8669 with onemore alarm feature, and 150 subsequently found tove gastric or esophageal cancer on endoscopy.40 Al-ugh those investigators concluded that alarm features

rformed poorly as a diagnostic test, they reported theerall pooled sensitivity and specificity to be 67% (95%nfidence interval, 54%83%) and 66% (95% confidenceerval, 55%79%), respectively. Individual alarm featuresth the best performance were weight loss, dysphagia,d epigastric mass on examination. Given those num-rs, and viewing this as a screening test rather than agnostic test, it seems reasonable that patients beingluated for GERD should be queried regarding dyspha-and weight loss and examined for an epigastric mass.judged significant, any of these should be evaluatedth endoscopy (USPSTF grade B, quality fair).Dysphagia merits special consideration as an alarmptom because it can be indicative of a stricture orlignancy. However, dysphagia was also reported by 37%11,945 patients with esophagitis without stricture orrretts esophagus participating in esophagitis clinical tri-, and it resolved in 83% with PPI therapy.41 This discrep-cy led the Montreal consensus group to suggest that notdysphagia, but only troublesome dysphagia, warrantsestigation3 (USPSTF grade B, quality fair). Troublesomesphagia is present when patients need to alter their eatingtterns, when patients have symptoms of solid food get-g impacted, when it exhibits a worsening pattern, oren it does not resolve with PPI therapy. Although notcifically mentioned by the Montreal consensus group,latter circumstance is significantly associated with failedphagitis healing.41 A final caveat in the endoscopic eval-tion of dysphagia is that the endoscopist should have athreshold for obtaining multiple (preferably 5) esoph-al mucosal biopsy specimens or even biopsy specimensm multiple levels to evaluate for eosinophilic esophagi-.42 The increasing recognition of eosinophilic esophagitisa confounding clinical entity has increased the potentialue of biopsies when performing upper endoscopy forRD. The traditional teaching that histologic assessment

mucosa in the setting of GERD is of limited utility duethe poor specificity of histologic findings for GERD hasn tempered by the need to differentiate eosinophilic

esocontha

mtidgshle

setsio(Utherefl

is pthaapattortriusutwunmaturis sB,

noidethepoicadeflurefldisesothefolis tprerelcomtiothasinesonostupHtivsturio

disnouaexctaistumothiorrelpesuconspoforallcortomcauthiis

symphcluesorulthedroinatttioansup(Uwereflinghaimcleesofinfaibetgo

ofby

AGA

INSTITU

TE

1398 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE GASTROENTEROLOGY Vol. 135, No. 4phagitis from GERD. A recent systematic review andsensus conference on eosinophilic esophagitis suggestedt

ucosal pinch biopsy specimens should be obtained from all pa-ents in whom EE (eosinophilic esophagitis) is in the differentialiagnosis. Biopsy specimens should be obtained regardless of theross appearance of the mucosa, and multiple biopsy specimensould be obtained from different esophageal locations along thength of the esophagus.43

Given the high rate of eosinophilic disease in theting of dysphagia without an obvious obstructing le-n, such subjects may benefit from mucosal biopsies44

SPSTF grade B, quality fair). No evidence demonstratesutility of routine esophageal biopsies in the setting ofux symptoms without dysphagia.The other broad scenario under which diagnostic testingerformed is in the evaluation of troublesome symptomst have not responded to empirical twice-daily PPI ther-y. Did therapy fail because of troublesome symptomsributable to reflux that did not resolve with PPI therapybecause the symptoms under consideration are not at-butable to reflux? In current practice, this dilemma isally faced when the patient has already been treated with

ice-daily PPI therapy for a significant period and it islikely that endoscopy will reveal esophagitis. Endoscopyy, however, still demonstrate Barretts metaplasia, stric-e, or an alternative upper gastrointestinal diagnosis, so ittill the most useful initial diagnostic test (USPSTF gradequality fair).In the setting of persistent troublesome symptoms andrmal findings on endoscopy, priority should be given tontifying conditions for which an effective alternativerapy exists. In the case of GERD, the only alternative,tentially more effective, therapy is antireflux surgery. Log-lly then, further evaluation should be targeted to thetection of conditions that are likely to respond to antire-x surgery. High-quality evidence on the efficacy of anti-ux surgery exists only for esophagitis and/or excessivetal esophageal acid exposure determined by ambulatoryphageal pH monitoring in a study obtained when PPIrapy was withheld (see discussion of question 12 in thelowing text). Another requirement for antireflux surgeryhat some peristaltic function be preserved. Although thecise cutoff is uncertain, severe peristaltic dysfunction is aative contraindication for antireflux surgery.45 Certainly,plete absence of peristalsis is an absolute contraindica-

n. Finally, it is important to identify alternative diagnosest may masquerade as GERD: functional heartburn, eo-ophilic esophagitis, subtle cases of achalasia, or distalphageal spasm. Given these priorities, the second diag-stic evaluation should be an esophageal manometrydy, which will serve to localize the LES for subsequentmonitoring, to evaluate peristaltic function preopera-ely, and to diagnose the major motor disorders. Recentdies suggest that high-resolution manometry has supe-r sensitivity in recognizing atypical cases of achalasia and

waabGEtal esophageal spasm compared with conventional ma-metry46,47 (USPSTF grade B, quality fair). The third eval-tion should then be to ascertain whether or not there isessive esophageal acid exposure, data that can be ob-ned with a conventional catheter-type pH monitoringdy, a combined impedance-pH study, or a wireless pHnitoring study (USPSTF grade B, quality fair). Whethers examination should be performed with the patient onoff acid suppressive therapy is debated. The unclearevance of normative data for impedance-pH studiesrformed on PPI therapy makes it difficult to interpreth studies. If normal values are not adjusted, then such an-PPI study could unequivocally demonstrate PPI nonre-nse. That, however, rarely occurs. However, a study per-med with PPIs withheld that demonstrates only margin-y abnormal esophageal acid exposure and poor symptomrelation in a patient being evaluated for persistent symp-s despite twice-daily PPI therapy raises concern aboutsality. Given that the next therapeutic consideration ins clinical setting is often antireflux surgery, further workurgently needed to answer this question.At this point in the diagnostic algorithm, troublesomeptoms of heartburn, chest pain, regurgitation, or dys-

agia persist despite normal findings on endoscopy (in-ding mucosal biopsy in the case of dysphagia), normalphageal acid exposure, and a manometry study thated out a major motor disorder. Current thinking is thatmajor remaining possibilities are a hypersensitivity syn-me or a functional syndrome, the distinction being thatthe case of a hypersensitivity syndrome symptoms areributable to reflux events, whereas in the case of a func-nal syndrome they are not. This is a subtle distinctiond a domain in which there is no high-quality evidenceporting one management approach or anotherSPSTF grade Insuff). Because of its added ability to detectakly and nonacidic reflux, the best tool available to testux-symptom association is impedance-pH monitor-,48,49 but the optimal algorithm to use in data analysiss yet to be validated.50 Conversely, the value of a negativepedance-pH monitoring study on or off therapy is morear. In the absence of endoscopic findings, with normalphageal acid exposure, without significant manometricdings, and with an impedance-pHmonitoring study thatled to show significant symptom association probabilityween reflux events and troublesome symptoms, one hasne as far as currently possible to rule out GERD.

5. What Are the Unique ManagementConsiderations in Patients With SuspectedReflux Chest Pain Syndrome?Reflux chest pain syndrome was accepted as one

the esophageal syndromes without mucosal injurythe Montreal consensus group (Figure 1), and there

s strong agreement that chest pain indistinguish-le from ischemic cardiac pain can be caused byRD.3 Often referred to as noncardiac chest pain, the

epbualeuscowichischapetienoetidia

iscoftheouchtiapaeasesowesumonifiomfouagtilimoch

priemtrionsuplaanthe5.4thetiepirica

peheolograwihates

anrefpre

tatsynliscoestulatheorrangitothFunoesounstrapaglarMode

1Gisawam

latfluinimhafladiaheustesfacplo

AGA

INSTITUTE

October 2008 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE 1399idemiology of this disorder is poorly understood,t population-based studies report community prev-nce rates ranging from 23% to 33%.5153 In a studying the General Practice Research Database thatmpared 13,740 patients with new-onset chest painth 20,000 age- and sex-matched patients withoutest pain, the odds of patients with chest pain havinghemic heart disease was 14.9, having GERD was 3.0,ving peptic ulcer disease was 3.0, and having dys-psia was 2.7.54 Another community sample of pa-nts with chest pain concluded that low socioeco-mic status increased the likelihood of cardiacology, while affluent social strata favored noncar-c etiologies.55

Because the morbidity and mortality associated withhemic heart disease are substantially greater than thatGERD and because of the impressive array of availablerapeutic interventions, this diagnosis must be thor-ghly considered before accepting a diagnosis of refluxest pain syndrome. This important priority substan-lly increases the economic impact of the reflux chestin syndrome.53,54,56 However, once ischemic heart dis-e has been adequately considered, the relative rarity ofphageal motor disorders in this group of patients, asll as results from empirical treatment trials of acidppressive therapy, suggest that GERD may be the nextst likely etiology.5759 Illustrative of the rarity of sig-cant motility disorders in this patient group, a man-etric study of 140 patients with noncardiac chest painnd diffuse esophageal spasm in 2%, nutcracker esoph-

us in 10%, hypertensive LES in 10%, and normal mo-ty in 70% of patients.58 Thus, although esophagealtility disorders are potential etiologies of noncardiacest pain, significant motility disorders are rare.59

With finding support for a diagnosis of GERD the nextority, a common clinical strategy in chest pain is anpirical trial of PPI therapy. Meta-analyses of treatmentals in patients with suspected reflux chest pain (basedobjective findings from endoscopy or pH monitoring)ggest clinical benefit with twice-daily PPI therapy overcebo.57,60 The reported pooled sensitivity, specificity,d diagnostic odds ratio for a short course of PPIrapy are 80%, 74%, and 13.83 (95% confidence interval,834.91), respectively. Extending the duration of PPIrapy beyond 4 weeks was not beneficial in these pa-nts.60 A cost-effectiveness analysis has also found em-ical treatment with PPIs to be superior to other clin-l strategies for this patient group.61

In summary, the first priority in patients with sus-cted reflux chest pain syndrome is to exclude ischemicart disease as a potential etiology. Once a cardiac eti-gy is excluded, there is sufficient evidence (USPSTFde A, quality good) to recommend empirical therapy

th twice-daily PPIs for 4 weeks. If a patient continues tove chest pain despite this course of therapy, diagnosticting with esophageal manometry and pH or imped-

incuaallce-pH monitoring can exclude motility disorders orractory reflux symptoms (see also question 4 in thevious text).

6. What Is the Best Initial Management forPatients With Suspected ExtraesophagealReflux Syndromes (Asthma, Laryngitis,Cough)? What Are the Unique ManagementConsiderations With Each? What Is theAppropriate Dose and Course of AntisecretoryTherapy in Each?The Montreal consensus group divided manifes-

ions of GERD into esophageal and extraesophagealdromes, with the latter divided into those with estab-hed or proposed association (Figure 1).3 Chronicugh, laryngitis, and asthma were accepted to have anablished association with GERD on the basis of pop-tion-based studies confirming an increased risk ofse symptoms among patients with either esophagitisesophageal reflux symptoms,51,62,63 with odds ratiosging from 1.2 to 3.0. However, because cough, laryn-is, and asthma have a multitude of potential etiologieser than GERD, they are clearly nonspecific for GERD.rthermore, the causal relationship of GERD with thesenspecific syndromes in the absence of a concomitantphageal GERD syndrome remains controversial andproven. The only randomized controlled trials demon-ating a significant treatment effect for a GERD ther-y in these syndromes were in patients who had esoph-eal GERD syndromes in addition to either chronicyngitis64 or asthma.30,65,66 Considering these data, thentreal consensus group concluded that existing evi-nce supports

) the existence of an association between these syndromes andERD, 2) the rarity of extraesophageal syndromes occurring inolation without concomitant manifestations of the typical esoph-geal syndrome, 3) that these syndromes are usually multifactorialith GERD as one of the several potential aggravating cofactors,nd 4) that data substantiating a beneficial effect of reflux treat-ents on the extraesophageal syndromes are weak.3

Recognizing the difficulty in establishing a causal re-ionship between extraesophageal syndromes and re-x, substantial investigational effort has been expendedvalidating diagnostic tests. However, clinical predictorsplicating GERD in the extraesophageal syndromesve proven elusive, and the premature adoption ofwed diagnostic criteria has likely resulted in the over-gnosis of extraesophageal GERD syndromes. Thealth care impact of this overdiagnosis is substantial,ually resulting in multiple (often repeated) diagnosticts and expensive unsuccessful therapies. Alternativetors (or cofactors) that are often insufficiently ex-red in these nonspecific extraesophageal syndromes

lude postnasal drip, allergic rhinitis, infections, habit-l throat clearing, tobacco, alcohol, excessive voice use,ergens, exercise, temperature or climate changes, emo-

tiodesoptarim

symdiaenpirThhaodpropHofPPwi99excinanoduGEdapaFutweffithusmfinbeynastonnoefi3 sdagea

GEetiingwiha

mewicliagradaextanGEfecD,

lisCoagantroingPP(PPgosigSuexp

his

Ta Twiceecte

rome

ChrLar r)64

Ast r)30

NO e no

AGA

INSTITU

TE

1400 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE GASTROENTEROLOGY Vol. 135, No. 4nal conflicts, and environmental irritants.67,68 Evi-nce supporting the multifactorial nature of the extrae-hageal syndromes comes from therapeutic trialsgeting GERD in which these nonspecific syndromesproved but were not resolved.64,6971

Given the nonspecific nature of the extraesophagealptoms and the poor sensitivity and specificity ofgnostic tests such as pH monitoring, laryngoscopy, ordoscopy for establishing an etiology of GERD,68 em-ical therapy with PPIs has become common practice.e majority of therapeutic trials of these syndromesve used twice-daily dosing of PPIs for treatment peri-s of 34 months.68,69 The rationale for this unap-ved dosing for unapproved indications comes frommonitoring data demonstrating that the likelihood

normalizing esophageal acid exposure with twice-dailyI therapy in patients with GERD is 93% and in thoseth chronic cough, asthma, or laryngeal symptoms is%,72 the logic then being that lesser dosing does notlude the possibility of a poor response because ofdequate acid suppression. Having said that, there arecontrolled studies investigating the optimal dosage orration of PPI therapy in patients with extraesophagealRD syndromes. The only supportive data for twice-ily PPI dosing are uncontrolled open-label studies oftients with suspected reflux laryngitis71 or asthma.73

rthermore, despite widespread treatment with PPIsice daily, high-quality evidence supporting treatmentcacy in these syndromes is still scant.3,68,69 Most en-siastically supportive studies were uncontrolled withall sample size. Subsequent attempts to confirm thesedings with controlled trials have shown no therapeuticnefit of PPIs over placebo for chronic cough31 or lar-gitis.74 A recent controlled trial in patients withhma suggested slight therapeutic benefit with PPIsly in the subgroup of asthmatic patients with bothcturnal respiratory and GERD symptoms but no ben-t in those without nocturnal GERD symptoms.30 Tableummarizes the evidence-based treatment recommen-tions that can currently be made for these extraesopha-l syndromes.In summary, patients with suspected extraesophagealRD syndromes may have GERD as a contributingology but rarely as the sole cause. However, the increas-

ble 3. Recommendation for or Against PPI Therapy (Daily orRecommendation for Treatment of Patients With Susp

With concomitant esophageal synd

onic cough Yes (USPSTF grade Insuff)yngitis Yes (USPSTF grade B, quality faihma Yes (USPSTF grade B, quality fai

TE. The evidence evaluated pertains to the treatment response of thincrimination of GERD as an etiologic factor alongth the lack of accurate confirmatory diagnostic testsve resulted in widespread overdiagnosis and overtreat-

noesocont of these conditions. Nonetheless, empirical therapyth twice-daily PPIs for 2 months remains a pragmaticnical strategy for subsets of these patients if they haveconcomitant esophageal GERD syndrome (USPSTFde B, quality fair; Table 3). However, once- or twice-ily PPIs (or H2RAs) for acute treatment of potentialraesophageal GERD syndromes, including laryngitisd asthma, in the absence of a concomitant esophagealRD syndrome cannot be supported and appears inef-tive based on presently available data (USPSTF gradequality fair).The role of pH or impedance-pH monitoring in estab-hing these diagnoses is controversial and unproven.nversely, similar to the case with symptomatic esoph-eal syndromes, the value of a negative pH or imped-ce-pH monitoring study is clearer. In the absence ofublesome esophageal symptoms or endoscopic find-s, with a failed 8-week therapeutic trial of twice-dailyI therapy, and with normal esophageal acid exposureI therapy withheld) on 24-hour monitoring, one hasne as far as currently possible to rule out GERD as anificant contributor to these nonspecific syndromes.ch patients should have etiologies other than GERDlored.

Chronic Management7. Does GERD Progress in Severity, SuchThat Symptomatic Patients WithoutEsophagitis Develop Esophagitis and BarrettsMetaplasia, or Are These Distinct DiseaseManifestations That Do Not Exist Along aContinuum? If Patients Do Progress, at WhatRate Does This Occur, and Does It WarrantEndoscopic Monitoring?Two potential paradigms for viewing the natural

tory of GERD exist. In the first, GERD is viewed as agressive disease such that, in the absence of effectiveervention, todays patient with nonerosive disease be-mes tomorrows patient with erosive disease, who thencomes a candidate for the development of Barrettsphagus.75 This spectrum of disease approach hasen contrasted with the view that GERD may be aease with phenotypically discrete categories, such as

Daily) and the Strength of Evidence Supporting Thatd Extraesophageal GERD Syndromes

Without concomitant esophageal syndrome

No (USPSTF grade D, quality fair)31

No (USPSTF grade D, quality fair)29

No (USPSTF grade D, quality fair)30

nspecific extraesophageal symptoms.prointcobeesobedisnerosive disease, erosive esophagitis, and Barrettsphagus.76 In this phenotypically preordained view,nversion from one disease manifestation to another is

disini

latadGEmacothelimpocoshsutheremthejectheto

Ta e

Sch NG

Iso NWE

Mca

NES

We 1

Ma 1

Pac N

Baj 5

Kaw NN

Lab

Sto

Son

NEeso

AGA

INSTITUTE

October 2008 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE 1401tinctly unusual, and subjects generally stay in theirtial category.Given the high prevalence of GERD in Western popu-ions, there is a decided paucity of data with which todress the question of long-term manifestations ofRD and the risk of progression in an evidence-basednner. Add to this the fact that essentially all reportedhorts are treatment cohorts, and our ability to describenatural history of the disease becomes even moreited. Several long-term cohort studies have been re-rted and are described in Table 4, along with reportednversion rates to more severe forms of disease.2,7787 Asown, length of follow-up and disease progression varybstantially among the studies. Furthermore, whetherse data represent true progression or the relapsing anditting nature of a chronic disease is unclear. However,data do permit some generalizations. First, in sub-

ble 4. Cohort Studies of the Natural History of Reflux Diseas

AuthorsCohort size andcomposition

Mean or range ofyears followed up

indlbeck et al, 199277 24 3.4(NERD, 16; EE, 8)

lauri et al, 199778 60 1722Severe GERD

symptomsDougall et al, 1997nd 199879,80

77 34.5

(NERD, 23;EE, 33)

tscher et al, 200181 83 2Mild EE

nabe et al, 200282 105 5.5First diagnosis EE

e et al, 200483 18 10NERD

bouj et al, 200584 34 2.9NERD

anishi, 200685 497 5.0 NERD5.3 Normal(NERD, 47; normal,

450)enz et al, 200686 3894 2

(NERD, 1717;LA A/B, 1512;LA C/D, 278;BE, 387)

ltey et al, 200787 684 3.4(GERD, 515

[103 erosive];BE, 169)

tag et al, 20062 2306 7.6 (120)(NERD, 1313; EE,

957; BEexcluded)

RD, nonerosive reflux disease; EE, erosive esophagitis; BE, Barrephagitis.ts with GERD treated in an uncontrolled fashion,re is some risk of progression from nonerosive diseaseerosive esophagitis. However, the risk of developing a

it ipesevicture, conversion to Barretts metaplasia, or develop-adenocarcinoma appears to be low within the 2- to

-year time frame of these studies. Substantial numberssubjects, especially if treated with antisecretory medi-ions,2 appear to regress to lesser grades of erosivephagitis and even to nonerosive disease. Progressiond regression are not predictable based on symptomration or demographics.88

Patients with GERD often inquire how often theyould undergo endoscopy to monitor the condition ofir mucosa. The role of endoscopy in the managementGERD will be discussed more completely in the fol-ing text, but there are no data demonstrating thattine endoscopy to assess for disease progression in

bjects with erosive or nonerosive reflux disease resultsimproved patient outcomes, and this practice shoulddiscouraged (USPSTF grade D, quality fair). However,

Progression dataRegression data

(treatment variable)

ERD to EE: 25%rade 2 to grade 3: 12.5%

Grade 3 to grade 1: 25%

ERD to esophagitis: 17%orsened esophagitis: 40%sophagitis to BE: 0%

NR

ERD to esophagitis: 24%sophagitis to BE: 11%ymptoms only to positive pH studyand/or esophagitis: 31%

NR

4.5% mild esophagitis to BE NR

0.5% (to more severe EE) 29.5% (EE to nonerosive)

ERD to esophagitis: 89% NR

.9% NERD to BE

ormal to LA A/B: 11.3%ERD to LA A/B: 36.2%

NERD to normal: 10.6%

ERD to LA A/B: 24.9%A A/B to LA C/D: 1.6%ERD to LA C/D: 0.6%ERD to BE: 0.5%A A/B to BE: 1.4%A C/D to BE: 5.8%

LA A/B to NERD: 61.3%,LA C/D to LA A/B: 41.8%LA C/D to NERD: 50.4%

E to BE: 1%onerosive to BE: 0%onerosive to EE: NR

NR

ERD to erosive: 15.6%ERD to stricture: 0.1%rosive to stricture: 1.9%ERD to adenocarcinoma: 0.1%

EE to NERD: 43.7%

sophagus; NR, not reported; LA, Los Angeles classification ofstring20ofcatesoandu

shtheoflowrousuinbe

NLNNLLENN

NNEG

tts es likely that slow disease progression will occur over ariod of decades in a small subset of patients, becauseere esophagitis, Barretts esophagus, and esophageal

addiv

GEtostrsiotimwhnomuvelofthehelenseteqesobeof

wibeerouorena ingtromeesocreinten26relplaantomPPnifisubwithedachsivof

streitdoeso

nogearesthewhPPdebecotharevcoeromasuesojecslisuusbareasynthecoaredorectreapontaiqu

eastotheHocotorlowtosomHeingtomneby

AGA

INSTITU

TE

1402 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE GASTROENTEROLOGY Vol. 135, No. 4enocarcinoma are all more prevalent among older in-iduals.89

In summary, data suggest that while subjects withRD may sometimes progress from nonerosive diseaseerosive esophagitis, demonstrating that it is not aictly categorical disease, the reported rates of progres-n are relatively low over a 20-year period, the longeste frame for which published data exist. In patients inom stricture, Barretts metaplasia, and adenocarci-ma were carefully excluded in the setting of a healedcosa at index endoscopy, the likelihood of these de-oping within a 7-year follow-up period is on the order1.9%, 0.0%, and 0.1%, respectively.2 On the other hand,likelihood of developing Barretts esophagus after the

aling of high-grade esophagitis (or unmasking preva-t disease) is about 6%,86 making endoscopy in theting of Los Angeles grade C or D esophagitis an inad-uate examination to exclude the presence of Barrettsphagus. Most importantly, upper endoscopy has noten shown to diminish the risk of cancer in the settingchronic GERD symptoms (USPSTF grade Insuff).

8. What Maintenance Therapy Is Indicatedfor Patients With the Typical EsophagealReflux Syndrome (With or WithoutEsophagitis)? When and How ShouldAntisecretory Therapy Be Decreased orDiscontinued? What, If Any, Risks AreAssociated With This?The utility of maintenance therapy in patients

th GERD depends on the manifestation of the diseaseing monitored, with the strongest data pertaining tosive esophagitis. Subjects not maintained on contin-us acid suppressive therapy have high rates of recur-ce of erosive disease, with some studies documenting80% recurrence rate within 12 months of discontinu-treatment.90,91 Multiple rigorous randomized con-

lled trials,92,93 summarized in a recent high-qualityta-analysis,94 show that the recurrence of erosivephagitis in subjects with GERD is dramatically de-ased by PPI treatment. The meta-analysis found that10 studies assessing maintenance therapy with main-ance doses of PPI (generally half the healing dose) for52 weeks, 36% of subjects taking PPIs experiencedapse of erosive disease, compared with 75% takingcebo. The relative risk of relapse in PPI users was 0.46,d the number needed to treat was 2.4. Reflux symp-s were also better controlled with maintenance-dose

I therapy than with placebo (44.4% vs 73.4% had sig-cant symptoms). The therapeutic gain was even morestantial when healing-dose PPI therapy was comparedh placebo. Although several of the comparisons showedterogeneity among individual trial results, on balance, theta strongly suggest that, when compared with placebo,

ronic acid suppression with PPIs prevents relapse of ero-e esophagitis for at least 612 months in subjects healedthe condition (USPSTF grade A, quality good). Similarly

tiemetheong are randomized controlled trials between H2RAs andher healing- or maintenance-dose PPIs, with subjects ran-mized to H2RAs up to twice as likely to have recurrentphagitis.95,96

The role of daily maintenance therapy in patients withnerosive disease is less clear. Patients with an esopha-l GERD syndrome without esophagitis who initiallyponded to a PPI randomized to maintenance-dose PPIrapy are less likely to have recurrent symptoms thanen randomized to an H2RA97 or placebo.98 WhetherI dosing needs to be continuous as opposed to on-mand (daily until resolution of symptoms) has alsoen studied, using the willingness of the patient tontinue on-demand therapy and the proportion of dayst the patient self-medicates as outcomes. A systematiciew of 17 such studies (15 of which were randomizedntrolled trials) showed that subjects with either non-sive or uninvestigated GERD did well with on-de-nd regimens.99 For example, Tsai et al randomized 622bjects with nonerosive disease to either on-demandmeprazole 20 mg or daily lansoprazole 15 mg.100 Sub-ts assigned to on-demand therapy were actuallyghtly more likely to be willing to continue therapy thanbjects in the continuous therapy arm (93% vs 88%) anded much less medication (0.3 vs 0.8 doses per day). Onlance, the data suggest that on-demand therapy is asonable strategy in patients with an esophageal GERDdrome without esophagitis where symptom control isprimary objective (USPSTF grade B, quality good). In

ntrast, in those with known erosive esophagitis whohealed with continuous PPI therapy and then ran-

mized to either continuous or on-demand therapy, theurrence rates of erosive disease are high in subjectsated with on-demand compared with continuous ther-y (42% vs 19% at 6 months; P .00001).101 Therefore,-demand therapy cannot be recommended for main-ning healing of erosive esophagitis (USPSTF grade D,ality good).The evidence presented in the previous text makes ity to say that continuous PPI therapy is recommendedmaintain a healed mucosa and that discontinuingrapy will likely result in recurrent heartburn.102,103

wever, there are no high-quality data to suggest thatntinuous antisecretory therapy alters the natural his-y of reflux disease other than to reduce the (already) incidence of peptic stricture.2 There are also no datathe effect that intermittent esophageal erosions ore degree of residual symptomatology is harmful.

nce, the main identifiable risk associated with reduc-or discontinuing PPI therapy is of an increased symp-burden. It follows that the decision regarding the

ed for (and dosage of) maintenance therapy is driventhe impact of those residual symptoms on the pa-

nts quality of life rather than as a disease controlasure. Certainly, the data in Table 4 do not supportcontention that residual GERD symptoms predispose

paesothatheingeveapinsbe

adwiPPducosiounsymwiopgesPPof

geapaemforanorofpaofanthedrodia

tenwiastregareesoquAltberynthaase

thedroesotemstereflbledispa21wioutow

inafttenreqGEtombetom

froiscudewiustalatlowmube

secgafocmigenactrisusoftnizof

AGA

INSTITUTE

October 2008 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE 1403tients to the development of Barretts esophagus orphageal adenocarcinoma. Pragmatically, this meanst many subjects beginning PPI therapy will receive thisrapy chronically, but often intermittently. Interest-ly, an accumulating body of literature now shows thatn when subjects are instructed to take daily PPI ther-y for GERD, compliance plummets within 3 months oftituting therapy, such that the majority of subjectscome noncompliant with daily dosing.104

In summary, chronic PPI therapy will be required forequate symptom control in the majority of subjectsth GERD symptoms severe enough to warrant initialI therapy. While many subjects may tolerate dose re-ction of their PPI and maintain adequate symptomntrol, the likelihood of long-term spontaneous remis-n of disease is low and conservative measures arelikely to suffice on their own. Beyond recurrence ofptoms and/or erosive disease, the risks associated

th cessation of therapy, including the possible devel-ment of Barretts esophagus, are minimal. Data sug-t that on-demand dosing or intermittent courses ofIs is the regimen preferred by most patients regardlessphysician instructions.

9. What Maintenance Therapy Is Indicatedfor Patients With Suspected ExtraesophagealReflux Syndromes (Asthma, Laryngitis,Cough)? When and How Should AntisecretoryTherapy Be Decreased or Discontinued?Owing to the nonspecificity of the extraesopha-

l reflux syndromes for GERD, at least 40%50% oftients will have persistent symptoms after 8 weeks ofpirical PPI therapy. In this group of patients, the needcontinued PPI therapy is predicated on the presence

d severity of concomitant esophageal syndromes withwithout mucosal injury. Although the true prevalenceesophageal mucosal abnormalities in this group oftients is unknown, uncontrolled observational studiessmall sample sizes suggest the presence of esophagitisd Barretts mucosa in 12% and 7%, respectively.105,106 In

absence of concomitant esophageal GERD syn-mes, PPI therapy should be discontinued and othergnostic and/or therapeutic avenues explored.There are no trials showing the effectiveness of main-ance therapy for patients in whom empirical therapyth twice-daily PPI therapy results in improvement ofhma, cough, or laryngitis. Thus, recommendationsarding maintenance therapy in this group of patientsbased on expert opinion extrapolated from the typicalphageal reflux syndrome literature (see discussion ofestion 8 in the previous text) (USPSTF grade Insuff).hough early observations suggested some associationtween reflux-induced laryngeal inflammation and la-

geal cancer,107 critical appraisal of these data suggestst such associations are inconclusive and mostly bi-d.108 Hence, the objective of continued maintenance

wiincenrapy in patients with an extraesophageal reflux syn-me is symptom control and, just as with the typicalphageal syndromes, step-down therapy should be at-pted. The likelihood of symptom recurrence withp-down therapy in patients with an extraesophagealux syndrome is currently unknown. However, a dou--blind placebo-controlled trial addressing the issue ofcontinuing PPI therapy in an unselected group oftients on chronic PPI therapy reported a disappointing% likelihood of remaining PPI free at 1 year in patientsth typical GERD symptoms and 48% in patients with-t typical GERD symptoms, highlighting the tendencyard long-term use in many patients.109

In summary, step-down therapy should be attemptedall patients with extraesophageal reflux syndromeser empirical twice-daily PPI therapy. Continuing main-ance PPI therapy should be predicated on either theuirements of therapy for concomitant esophagealRD syndromes or extraesophageal syndrome symp-response. In both cases, maintenance therapy should

with the lowest PPI dose necessary for adequate symp-relief.

10. What Are the Clinical Consequences ofChronic Potent Acid Inhibition? Do ThesePotential Side Effects Warrant SpecificTesting (eg, Bone Density Studies, CalciumSupplementation, Helicobacter pyloriScreening, and so on)?Most of the mortality from reflux disease stems

m its link with esophageal adenocarcinoma, and thereno high-level evidence that the risk is reduced by anyrrently available GERD therapy.110 Epidemiologic evi-nce does, however, suggest a substantial risk reductionth aspirin or nonsteroidal anti-inflammatory druge.111 Apart from esophageal adenocarcinoma, the mor-ity associated with reflux disease is very low, estimated0.46/100,000 in the year 2000.112 Given the profoundlymortality rate of this disease, therapies for GERD

st be extremely safe to satisfy an acceptable risk-nefit ratio.Because PPIs work by profoundly reducing gastric acidretion, which in turn results in a reactive increase instrin secretion, most consideration of long-term risk isused on unwanted effects of secondary hypergastrine-a, hypochlorhydria, or even achlorhydria. Other, moreeric considerations have to do with drug-drug inter-ions and potential teratogenicity. In general, theseks are slight if even demonstrable, but the widespreade of PPIs coupled with the frequent need for chronic,en open-ended therapy mandate that they be scruti-ed. Table 5 summarizes the available data on the riskslong-term PPI use.113128

The data in Table 5 show no worrisome safety signals

th PPIs. The most convincing data link PPI use with anrease in Clostridium difficile colitis and bacterial gastro-teritis, but in each case the magnitude of risk is slight.

WipoanbeicarosriswhgodapylofofanhaAfalw

ica

Ta

Pot RiH NoA No

F Od

V So

C Ne

I Po

PotI PP

I Ne

G Da

GeS Ba

D Cl

AA

P

AGA

INSTITU

TE

1404 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE GASTROENTEROLOGY Vol. 135, No. 4th respect to the hip fracture issue, there are manytential confounders to the data, but the putative mech-ism would be decreased calcium absorption, which hasen shown with PPI use.129 Regardless, it is good med-l practice to screen and treat the elderly for osteopo-is irrespective of PPI use. To summarize all availablek/benefit data on PPIs, their use is strongly justifieden clinically indicated (USPSTF grade A, qualityod). Conversely, there is inadequate evidence to man-te bone density studies, calcium supplementation, Hori screening, or any other routine precautions becausePPI use (USPSTF grade Insuff). Having said that, manythe potential rare side effects of PPIs are dose related,d PPIs should be used for conditions in which theyve proven efficacy and at the minimal effective dosage.ter all, in the absence of benefit, a risk-benefit ratio isays unacceptable.

11. What Is the Role of Endoscopy in Long-term Management of Patients With GERD,and Under What Circumstances ShouldMucosal Biopsy Specimens Be Obtained When

ble 5. Potential Risks of Long-term PPI Therapy

ential risks of hypochlorhydria (trophic, absorptive)ypergastrinemia-induced carcinoid tumorsccelerated progression of atrophic gastritis/gastric cancer withconcomitant H pylori gastritis115

ormation of gastric fundic gland polyps168

itamin B12 malabsorption

alcium malabsorption

ron malabsorption

ential risks of hypochlorhydria (infectious)ncreased risk of C difficile colitis

ncreased risk of community-acquired pneumonia (presumablyaspiration)astric colonization with bacteria that convert nitrates tocarcinogenic N-nitroso compounds that then reflux117

neric pharmacologic risksafety in pregnancy (omeprazole crosses placenta and ispregnancy safety category C; other PPIs are category B)rug-drug interactions; PPIs metabolized by cytochrome P450and may induce or inhibit drug metabolism (phenytoin,warfarin, and so on)naphylaxiscute interstitial nephritis

ancreatitisEndoscopy Is Performed?Because PPI treatment is usually rendered empir-

lly before testing, the sensitivity of endoscopy as a

deversprgnostic test for GERD is poor. Hence, the principale of endoscopy in patients with suspected GERD is theluation of treatment failures and risk management.st of the morbidity and mortality from reflux diseasems from its link with esophageal adenocarcinoma, andt risk has not been shown to be decreased by anyrrent GERD therapy.110 Otherwise, the mortality asso-ted with reflux disease is very low, estimated at 0.46/0,000 in the year 2000, and mainly attributable tomorrhagic esophagitis (38%), ulcer perforation orphageal rupture (19%), aspiration pneumonia (19%),d complications of antireflux surgery (11%).112 Puttingrisk of adenocarcinoma in perspective, data from the

rveillance Epidemiology and End Results database sug-t that there were about 8000 incident cases of esoph-eal adenocarcinoma in the United States in 2004130

d, depending on interpretation of the data, this diseaserden has increased anywhere from 2- to 6-fold relative20 years prior.131,132

The 5-year survival of patients with esophageal adenocar-oma is very poor, but it is greatly improved by early

sk magnitude/possible consequencet demonstrated in humans113

documentation of an increase in atrophic gastritis and no basisto recommend testing or treatment for H pylori before long-termPPI use116

ds ratio of 2.2 for developing fundic gland polyps within 15years,168 negligible, if any, risk of dysplasiame patients show decreased vitamin B12 levels after years ofacid inhibition,113 case reports (2) of clear deficiency114

sted case-control study of UK patients older than 50 years;adjusted odds ratio of 1.44 (95% confidence interval,1.301.59) of hip fracture with PPI use longer than 1 year122

or response to oral iron supplement absorption in 2 iron-deficient individuals improved after cessation of omeprazole; noclear clinical relevance169

I use is independent risk of C difficile diarrhea in antibioticusers, odds ratio of 2.1 (95% confidence interval, 1.23.5)120

sted case-control analysis, adjusted odds ratio for pneumoniawith PPI use of 1.73 (95% confidence interval, 1.332.25)121

ta on PPI use and increased gastric N-nitrosamine remainuncertain and the risk of cancer is speculative113

sed on 345 accidental exposures compared with 787 controls,no observed increased teratogenecity123

inically significant PPI drug-drug interactions are rare (1/millionprescriptions)124

e case report with lansoprazole125

cases worldwide, partially reversible (one case requiresdialysis, no deaths), estimated risk 1/12,500 patient-years oftherapy126,127

pulation-based case-control study adjusted odds ratio of 3.2(95% confidence interval, 1.47.4128diausevaMostethacucia10heesoantheSugesaganbuto

cin

On64

Potection: 58% for patients with tumors detected in situsus 10% for patients with tumors detected after regionalead.133 The other potential benefit of endoscopy in the

seta mkndeagustrGEtifienfounoicaAmpoentomHobecscresocan

geaesopatutityfintivtooubobaag(2)SwrepseresoidespwhcerAngradewaveiAGthdudosymcanEdy

esograinsuscoforofgrathor

diringcincricoadmebe

GEeffilapofSta31enscodisforthaHosufalTh18tearegsures

andisingdaH2areMowi

AGA

INSTITUTE

October 2008 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE 1405ting of chronic GERD is detection of Barretts esophagus,etaplastic change of the distal esophageal mucosa ac-

owledged to be a premalignant condition. The risk ofveloping esophageal adenocarcinoma in Barretts esoph-s is estimated at 0.5% per year.134 Thus, the proposedategy for controlling the risk of cancer is to screen theRD population for Barretts esophagus, to survey iden-ed individuals for the development of dysplasia and ad-ocarcinoma, and to resect or ablate these lesions whennd. In the hope of controlling the risk of adenocarci-ma, past societal guidelines, including those of the Amer-n Society of Gastrointestinal Endoscopy135 and theerican College of Gastroenterology,136,137 have sup-rted consideration of the use of at least once-in-a-lifetimedoscopy to screen subjects with chronic GERD symp-s for the presence of Barretts esophagus or early cancer.wever, these guidelines are based solely on expert opinionause no direct data exist to substantiate the utility ofeening or surveillance endoscopy to detect Barrettsphagus or to monitor the condition for progression tocer.Evidence supporting the strategy of reducing esopha-l adenocarcinoma mortality by screening for Barrettsphagus is scarce. For such a strategy to work well,tients with Barretts esophagus would need to consti-e the majority of patients at risk for cancer, the sever-of reflux symptoms would have to be predictive ofding Barretts esophagus, and endoscopy would effec-ely identify patients with Barretts esophagus, leadingaltered clinical management that improved the clinicaltcome. To date, none of these conditions have beenrne out in clinical trials. (1) In a Swedish population-sed endoscopy study, the prevalence of Barretts esoph-us (1.6%) was not correlated with reflux symptoms.138

In a nationwide case-control study, more than 40% ofedes destined to develop esophageal adenocarcinomaorted no antecedent reflux symptoms.139 (3) In a Kai-Permanente cohort study, 454 of 589 patients withphageal or gastric cardia adenocarcinoma had nontifiable Barretts metaplasia evident in any pathologyecimen.140 (4) In the same cohort, among 64 patientso had undergone endoscopy before detection of can-, only 38% had Barretts esophagus identified.140 (5)alyses of 2 large Barretts esophagus surveillance pro-ms concluded that, although a small number of inci-nt esophageal adenocarcinomas were detected, theres no improvement in survival attributable to the sur-llance program.141,142 These data were reviewed by anA consensus workshop composed of 18 experts ine field of Barretts esophagus in 2004.143 After con-cting an evidence-based review of the utility of en-scopic screening of subjects with chronic GERDptoms for the detection of Barretts esophagus or

cer, this group strongly rejected the statement that

ndoscopic screening for Barretts esophagus andsplasia has been shown to improve mortality from

analcophageal adenocarcinoma and concluded that thede of evidence in support of this intervention wassufficient to form an opinion (USPSTF grade In-ff). Regarding the corollary statement that Endo-pic screening for BE and dysplasia should be per-med in all adults 50 years of age with 510 yearsheartburn, the supporting evidence was againded only at the level of expert opinion, and againe majority of the group either rejected the statementrejected it with reservation (USPSTF grade Insuff).In summary, despite the ubiquity of the practice, noect evidence supports the use of endoscopy as a screen-test for Barretts esophagus or esophageal adenocar-oma in the setting of chronic GERD. Regarding theteria for obtaining mucosal biopsy specimens in theurse of performing an endoscopy, there is no basis tovocate doing this routinely but, clearly, biopsy speci-ns of any areas suspected of being metaplastic shouldobtained and carefully evaluated for dysplasia.

12. What Are Indications for AntirefluxSurgery, and What Is the Efficacy of ThisTherapy?Few topics in the management of patients with

RD are as controversial as the indications for andcacy of antireflux surgery. With the introduction ofaroscopic Nissen fundoplication in 1991, the numberadult antireflux surgeries performed in the Unitedtes rapidly increased from 11,000 per year in 1985 to,695 in 1999.144 This increase was largely driven by thethusiastic endorsement of the procedure by endo-pic surgeons and surgery departments in a number ofease scenarios, typified by guidelines from the SocietyGastrointestinal and Endoscopic Surgeons claimingt the procedure was curative in 85%93% of cases.145

wever, since 1999, the number of adult antirefluxrgeries performed in the United States has steadilylen; by 2003, it had declined by 30% to 23,998 cases.144

e decline has been greatest among young patients,39 years of age (38%), and at teaching versus non-ching hospitals (36% vs 23%). There is also substantialional variation in the utilization of antireflux surgery,ggesting a lack of consensus among practitioners withpect to the appropriate indications.144

Just as with PPI therapy, evidence of the utility oftireflux surgery depends on the manifestation of theease being monitored, with the strongest data pertain-to erosive esophagitis. However, the utility of older

ta randomizing subjects to either medical care withRAs or Nissen fundoplication by the open approach146

limited because of advances in care in both domains.re recent data comparing laparoscopic fundoplication

th PPI therapy are more germane to current practice

d will be considered here. Illustrative of this, Lundell ethave reported 5- and 7- year results of a randomizedntrolled trial of patients with esophagitis treated with

omyeathevsimthe

synpaturinofcomforthiextthoformocanwhpocenfec

trosopGEoftigsupofwidthiaftaltpathemathaterthesurancouHesymsymhafitssymofabanshobel

gersurgea(Usym

wimothaaccavagerdatomofroovilathalonvanapPPgra

higremshdawiAltthenumaus

gerwhingrisgraintgergosynsusomevebaanpawhgre

AGA

INSTITU

TE

1406 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE GASTROENTEROLOGY Vol. 135, No. 4eprazole 2060 mg/day or antireflux surgery. At 7rs, the 2 treatment arms were similar with respect toincidence of recurrent esophagitis (10.3% omeprazole11.8% antireflux surgery).147 Hence, if the outcome ofportance is maintaining a healed esophageal mucosa,2 therapies appear to be equivalent.

As for other manifestations of the esophageal GERDdromes with esophageal injury, there are no data com-ring the efficacy of PPIs with antireflux surgery in stric-e prevention, and controlled data have shown no changethe prevalence of Barretts esophagus or in the incidenceadenocarcinoma when patients treated surgically werepared with those treatedmedically.110,148 It is importantproviders and patients to understand that the risk ofs cancer to the average subject with reflux symptoms isremely low (1 in 10,000 per patient-year).149 Evenugh the safety profile of antireflux surgery is excellenta surgical procedure, the low risks of morbidity andrtality still dwarf any potential benefit in reduction ofcer risk. Even among subjects with Barretts esophagus,o have a higher risk of cancer than the general GERDpulation, randomized controlled trial data150 and a re-t meta-analysis151 fail to substantiate any protective ef-t of surgery against cancer.The relative efficacy of antireflux surgery to PPIs in con-lling symptomatic esophageal syndromes and extrae-hageal syndromes with an established association withRD is less clear. If the analysis is restricted to the controlheartburn and acid regurgitation as determined by inves-ator interview or questionnaire, studies suggest modesteriority of antireflux surgery to PPI therapy, on the ordera 10% therapeutic gain.147,152 However, the data areely divergent. As many as 30% of subjects who undergos procedure continue to use medical therapy by 5 yearser the procedure and surgical revision is common. Also,hough community-based outcome data are sparse com-red with that generated from specialized referral centers,data suggest that subjects from community-based seriesy have poorer outcomes and lower patient satisfactionn those in specialized, presumably higher-volume cen-s.153,154 With respect to the extraesophageal syndromes,re are no controlled data comparing PPIs with antirefluxgery, but observational studies suggest some benefit oftireflux surgery for highly selected patients with refluxgh syndrome66,155 and reflux asthma syndrome.65,156

nce, if the outcome of importance is controlling eitherptomatic esophageal syndromes or extraesophagealptoms in carefully selected patients, antireflux surgery

s greater efficacy than PPI therapy. However, these bene-must be weighed against the deleterious effect of newptoms consequent from antireflux surgery. Dysphagia

sufficient severity to require esophageal dilation occurs inout 6% of patients treated with antireflux surgery,157,158d both controlled147 and uncontrolled trials159 havewn a significant increase in flatulence, an inability toch, and increased bowel symptoms after antireflux sur-

anappay. Given this balance, the recommendation for antirefluxgery is stronger in the case of the symptomatic esopha-l syndromes, especially with troublesome regurgitationSPSTF grade B, quality fair), than for extraesophagealptoms (USPSTF grade C, quality fair).

As alluded to in the discussion of the risks associatedth chronic PPI therapy (question 10), the otherwise lowrbidity and mortality associated with GERD mandatet GERD therapies must be extremely safe to satisfy aneptable risk-benefit ratio. Table 6 summarizes theilable morbidity and mortality data on antireflux sur-y.153,157,160165 Note that unlike the treatment efficacyta, in which case it is reasonable to restrict the analysiscontrolled studies from highly specialized centers, it isre relevant to assess morbidity and mortality datam the perspective of larger data sets reflective of theerall impact of the intervention on public health. Sim-rly, it is reasonable to compare the data in Table 6 witht in Table 5, detailing what is known of the risks ofg-term PPI therapy. Given that comparison, from thetage point of risk, if antireflux surgery and PPI ther-y were estimated to be equally effective for a patient,I therapy should be strongly recommended (USPSTFde A, quality good).In contrast to the data regarding antireflux surgery,h-quality data on endoluminal antireflux proceduresain sparse. Most available studies were designed to

ow proof of principle in this rapidly evolving area. Tote, no studies compare the efficacy of these devicesth either optimal medical therapy or antireflux surgery.hough the latest of these devices show promise,166,167

dearth of comparative data, as well as the smallmber and relatively short follow-up of subjects treated,ke it premature to frame recommendations for theire in GERD (USPSTF grade Insuff).In summary, the current indications for antireflux sur-y are well circumscribed. Patients with esophagitiso are well maintained on medical therapy have noth-to gain from antireflux surgery and incur significant

k; they should be advised against surgery (USPSTFde A, quality good). Patients with esophagitis who areolerant of PPIs will likely benefit from antireflux sur-y and should be so advised (USPSTF grade A, qualityod). Patients with symptoms of the esophageal GERDdrome poorly controlled by PPIs may benefit fromrgery, especially in the setting of persistent trouble-e regurgitation (USPSTF grade B, quality fair). How-

r, the recommendation for antireflux surgery must belanced with a thorough discussion of potential posttireflux surgery symptoms. Finally, carefully selectedtients with extraesophageal GERD syndromes inom a reflux causality has been established to theatest degree possible (see question 4) may benefit from

tireflux surgery, and it should be recommended withpropriate restraint (USPSTF grade C, quality fair). Theucity of comparative data on endoluminal antireflux

proreflsu(U

paGErecdocluwimapetrosimareesothemotio

Ta

DeP 116

D tion

U 57

U Utiliz

M ters,

LifeP 116

D 4 RF

U Utiliz

M ters,W

ReoD tions

U

WDys

U

M

WS

LNF

AGA

INSTITUTE

October 2008 AMERICAN GASTROENTEROLOGICAL ASSOCIATION INSTITUTE 1407cedures with either optimal medical therapy or anti-ux surgery makes it impossible to define the role ofch techniques in our current treatment algorithmSPSTF grade Insuff).

ConclusionsIn formulating guidelines for the management of

tients with GERD, we found the Montreal definition ofRD3 (Figure 1) to be very useful and structured ourommendations around it. Contemplating the broadmain of the Montreal definition, an immediate con-sion was that much of the management of patientsth GERD in terms of diagnostic tests and diseasenagement is based on uncontrolled trials, clinical ex-rience, or expert opinion rather than randomized con-lled clinical trials; high-quality trials in these areasply do not exist. Randomized controlled clinical trials, however, ubiquitous in the domain of therapy for thephageal GERD syndromes, especially pharmacologic

ble 6. Complications From Antireflux Surgery

athopulation-based, Finland; January 1, 1987, to January 1, 1996; n

anish, population-based; 19972005; n 2465 primary fundoplica

S VA database; October 1, 1990, to January 29, 2001; n 31451

S population-based cohorts, Washington state and US Health Care1997164

eta-analysis of reports published from 1991 to 1995; specialty cen036 mo160

-threatening complicationsopulation-based, Finland; January 1, 1987, to January 1, 1996; n

anish, population-based; 19972005; n 2465 primary FP and 12

S population-based cohorts, Washington state and US Health Care1997164

eta-analysis of reports published from 1991 to 1995; specialty cenisconsin managed care network; community practices, n 87153

perations (redo)anish, population-based; 19972005; n 2465 primary fundoplica

S VA database; October 1, 1990, to January 29, 2001; n 31454.5 yr157

isconsin managed care network; community practices, n 87153

phagia severe enough to require dilationS VA database; October 1, 1990, to January 29, 2001; n 31454.5 yr157

eta-analysis of reports published from 1991 to 1995; specialty cen036 mo160

isconsin managed care network; community practices, n 87153

pecialty center, retrospective review of June 1996 to October 1998

, laparoscopic fundoplication; OF, open fundoplication; RF, redo funrapies for esophagitis. Hence, it is not surprising thatst of the highest-level evidence-based recommenda-ns that can be made pertain to that scenario. However,

Iacute management of patients with esophagitis isely a clinical dilemma in current practice. Hence, weused instead on the clinical quandaries that do con-nt the clinician with great regularity and examined thedence that could be brought to bear on those issues.used the USPSTF grades detailed in Table 1 to ascer-

n whether or not particular practices or therapiesuld be recommended based on published evidence.nceptually, USPSTF grades evaluate a risk-benefit as-sment for diagnostic or therapeutic interventions. Al-ugh relatively few of our conclusions achieved thehest USPSTF grade, a substantial number achieved anequate grade upon which to base pro or con recom-ndations. Our major conclusions follow.

Grade A: Strongly Recommended Based onGood Evidence That It Improves ImportantHealth Outcomes

2 LNF; n 3993 OF163 LNF 0.1%OF 0.2%

and 124 RF162 Primary 0.45%RF 0.81%LNF and OF 0.8%

ation Project, 1992 Washington 0.4