The BARI Protocol BypassAngioplasty … TheBARIProtocol Protocol for the BypassAngioplasty Revascularization Investigation In this supplement we present the design of the Bypass Angioplasty

V-1

The BARI Protocol

Protocol for the Bypass AngioplastyRevascularization Investigation

In this supplement we present the design of the Bypass Angioplasty RevascularizationInvestigation (BARI). The BARI team of investigators, data coordinators, staff, and supportcommittees are committed to providing the highest quality data to scientifically test theproposed hypotheses. Considerable controversy exists about the extension of percutaneoustransluminal coronary angioplasty (PTCA) into the arena of therapy for multivessel coronaryartery disease. Although coronary artery bypass surgery has been thoroughly compared withmedical therapy in randomized trials, studies of PTCA to date have been observational innature. The presumption that the results of trials of coronary bypass surgery can be applied tothe use of PTCA is not established; this creates a dilemma that requires resolution in the mostobjective and scientific manner. In the Wangensteens's book (-The Rise of Surgery, 1978), thereis an excellent discussion of the controversy that surrounded the work of Semmellweiss. In oneof the earliest efforts to bring science to the bedside and to influence medical practice, heinvestigated childbirth outcome in terms of whether the physician's hands were washed beforedelivery. Those times were filled with extreme reluctance on the part of the medicalestablishment to accept the observations of Semmellweiss. As the Wangensteens suggest, "Trial,not debate, is the proper manner in which to resolve a question of this kind." It is in this spiritthat the BARI investigators are conducting the BARI trial. To the best of our ability, we willprovide information that will benefit patients and physicians in making decisions on revascu-larization procedures. (Circulation 1991;84[suppl VI :V-1-V-27)

T he application of invasive therapy to the treat-ment of severe coronary artery disease(CAD) has progressively increased during

the past decade. In the United States in 1988 approx-imately 250,000 patients with CAD were treated withpercutaneous transluminal coronary angioplasty(PTCA), and nearly as many patients received coro-nary artery bypass graft (CABG) surgery (NationalCenter for Health Statistics, 1988; personal commu-nication). These numbers represent dramatic in-creases in the use of invasive procedures comparedwith those of 1980, when 6,000 PTCAs and 137,000CABG operations were performed.' Therefore, cor-onary revascularization is an important component ofhealth care costs in the United States, with currentdirect costs easily exceeding the prior estimate ofover $5 billion for CABG alone in 1984.2At the present time, the choice between PTCA and

CABG for patients with multivessel CAD who needrevascularization and who are suitable for eitherprocedure represents a clinical dilemma because therelative indications for PTCA and CABG in thesepatients are not yet clearly defined.

Before informed therapeutic choices between thetwo procedures can be made, controlled studies areneeded to objectively compare the benefits and risksof an initial strategy of PTCA versus CABG inappropriately selected patients. The Bypass Angio-plasty Revascularization Investigation (BARI) has

been designed to accomplish this through randomassignment of revascularization strategy and system-atic follow-up over 5 years. Although the primaryclinical indication for revascularization varies amongBARI patients (symptom relief or treatment of pro-found ischemia), all patients are judged to be atrelatively high risk for subsequent cardiac events.

In addition to the clinical trial component, allpatients who are eligible but refuse random assign-ment are asked to participate in the BARI Registry.The registry also contains a 5-10% random sample ofthose who are deemed ineligible for random assign-ment because they are considered angiographicallyunsuitable for PTCA and/or CABG.Evidence from this clinical trial will provide a

scientific basis for choosing PTCA or CABG as theinitial revascularization treatment of severe multi-vessel coronary disease.

Specific AimsBARI is a comparative study of PTCA and CABG,

the two most prevalent revascularization methodsused to treat advanced CAD. The study focuses onthe treatment of patients who have multivessel dis-ease and severe angina or ischemia, those who re-quire revascularization, and those who are suitablefor either procedure. The primary aim of BARI is totest the hypothesis that an initial strategy of PTCA inthese eligible patients compared with CABG does

V-2 Supplement V Circulation Vol 84, No 6 December 1991

not compromise clinical outcome during a 5-yearfollow-up period.

Aims of the Randomized Clinical TrialBecause CABG is of established benefit for pa-

tients with severely symptomatic multivessel disease,the use of an alternate treatment strategy, regardlessof its potential efficacy, must not impose a greaterrisk of mortality than CABG. For this reason, and toprovide a reliable end point for calculation of samplesize, mortality is the primary end point of the trial.The sample size was selected to enable BARI to ruleout with high probability that the 5-year mortalityrate with PTCA exceeds the 5-year mortality ratewith CABG by more than 2.5%. (For a discussion ofsample size calculations, see Appendix 3.) Althoughmortality is essential in assessing the safety of thePTCA strategy in patients with multivessel CAD,other end points of clinical outcome are of criticalimportance, particularly if there is no difference inmortality between the two treatment strategies. Be-cause the acceptable difference in mortality is small,a large sample size is required. This large sample sizealso provides sufficient power to examine treatmentdifferences in rates of myocardial infarction (MI),repeat revascularization, and recurrent severe anginaor ischemia. Distribution of exercise capacity, ven-tricular function, and need for medication will alsobe compared by treatment assignment at selectedfollow-up points.

In addition, BARI will provide much-needed an-

swers to questions concerning the economic and psy-chosocial aspects of myocardial revascularization. Theeconomic and quality-of-life consequences of PTCAand CABG strategies over a 5-year follow-up periodwill be compared. BARI will provide estimates ofinitial and continuing indirect and direct costs of thetwo procedures and provide measures on quality oflife. These critical data are collected in-depth at seven

participating BARI centers in an ancillary study ofeconomics and quality of life (SEQOL). This study isfunded by the Robert Wood Johnson Foundation,having had initial funding from Advanced CatheterSystems. In addition to SEQOL, data on the numberof significant cardiac hospitalizations, employmentstatus, and limitations of activities are collected for allBARI patients. (Also see resource use and quality oflife [p V-8].)Although BARI is designed for the overall com-

parison of PTCA versus CABG as the initial strategy,it will also provide comparative data for predeter-mined subgroups, which are defined by the variousclinical and angiographic presentations. Clinical sub-groups of special interest are those with unstableangina, stable Canadian Cardiovascular Society Clas-sification functional class III or IV angina, or class I

or II angina in the presence of either documentedischemia or recent Q wave MI. Angiographic sub-

groups will be defined by the number of significantlystenosed vessels, the number and proportion of myo-cardial territories with a jeopardized coronary sup-ply, the complexity of the lesion and vessel anatomy,and the degree of left ventricular function.

Aims of the Bypass Angioplasty RevascularizationInvestigation RegistryThe BARI Registry will include eligible patients

who refuse random assignment as well as a 5-10%

sample of patients who are excluded from the trialbased on angiographic criteria. The group of eligiblepatients who refused random assignment will lenditself to the investigation of selection factors involvedin the choice of PTCA versus CABG. In addition, wewill be able to compare the results of treatmentselected by choice with that selected by randomallocation. The Registry patients who are ineligiblefor random assignment because of angiographic cri-teria will be used to assess how angiographic exclu-sion practices differ across sites, which treatmentsuch patients actually receive, and the long-termoutcome with the given treatment.

Patient SelectionInclusion Criteria

Patients included in BARI must be representativeof those who have multivessel coronary disease and aretreated for severe angina or myocardial ischemia incurrent clinical practice. Specifically, eligible patientsmust meet the following criteria: clinically severeangina or objective evidence of ischemia that requiresthe need for a revascularization procedure, angio-graphically documented multivessel coronary dis-ease, suitability for both PTCA and CABG, andinformed consent for random assignment.

Patient Screening for Exclusion CriteriaThe population that is considered for BARI con-

sists of those patients who undergo diagnostic coro-nary arteriography in a BARI institution and may

include patients with off-site angiograms.Exclusions from screening. Patients are first evalu-

ated for entry on the screening log. A patient isconsidered eligible for screening and is placed on thescreening log, allowing the patient to be trackedthrough the remainder of the screening system if thepatient has none of the following exclusions: absenceof significant coronary disease, primary congenitalheart disease, primary valvular heart disease, primarymyocardial heart disease (including patients with aventricular aneurysm, which requires surgery), priorPTCA or CABG, single-vessel CAD, and/or an age.80 years.Clinical and major angiographic exclusions. Patients

placed on the screening log are evaluated for clinicaland major angiographic exclusions. If none of thefollowing exclusions are present, the patient is con-sidered clinically eligible: age <17 years, geographi-cally inaccessible or unable to return for follow-up,

Definitions for terms printed in bold italics may be found in theGlossary; see Appendix 2.

The BARI Protocol V-3

No CADSgle VD

Prior ProcedureCongenita Heart DiseaseValvular Heat DiePrmary Myocardial DiseaseAge280 years

Left Main DiseaseInsudfcent SymptomsEmergency Revasc.Other

FIGURE 1. Flowchart showingpatient screening mechanism.A full list of exclusion criteria appears within the text of theProtocol (seepp V-2-V-3, Patient Selection). *Other screeninglog exclusions include noncardiac illness expected to limitsurvival, age <17, geographic inaccessibility, clinical contrain-dications to PTCA or CABG, inability to understand or

cooperate with protocol, enrolled in competing study, techni-cally unsatisfactory angiogram, extensive ascending aorticcalcification, primary coronary spasm, concomitant majorsurgery required, and pregnancy.

insufficient angina or objective evidence of isch-emia, unstable angina or acute MI, which requiresemergency revascularization, left main stenosis>50% or of a character that precludes angioplasty,noncardiac illness that is expected to limit survival,extensive ascending aortic calcification, primarycoronary spasm, inability to understand or cooper-ate with protocol requirements, coronary angio-gram that is technically unsatisfactory, suspected or

known pregnancy, enrollment in a competing clini-cal trial, contraindication to CABG or PTCA be-cause of a coexisting clinical condition, and/orconcomitant major surgery that is required (e.g.,aortic and/or mitral valve surgery, carotid endarter-ectomy, and/or resection of left ventricular or ab-dominal aortic aneurysm).

Final angiographic exclusion. Clinically eligible pa-tients are evaluated for final angiographic eligibility.The surgeon and angioplasty operator assess the pa-tient's suitability for each procedure according to their

technical expertise and considerations of patient safety.On the basis of angiographic findings, patients areexcluded at this point if the patient is judged to beunsuitable for PTCA and/or CABG. (The criteria forthis decision are described on pp V-3-V-6, PTCA andCABG guidelines, respectively.)Because the BARI investigators recognize the

limitations of trying to prospectively describe all thefeatures that define suitability for PTCA and CABG,the final evaluation for eligibility involves the subjec-tive judgment of both the surgeon and the angio-plasty operator. This results in a certain degree ofdiversity across centers, an additional strength of thetrial that allows the results to be applicable to thebroad group of patients considered for angioplasty.Baseline data collected on all patients will allowBARI to characterize in detail angiographic findingsof patients considered clinically eligible for the trial.The population excluded at the final angiographiclevel will be studied by using a 5-10% randomsample that is selected for inclusion in the registry.Informed consent. Eligible patients who sign an

informed consent for random assignment are enteredin the clinical trial. Those who refuse random assign-ment but sign an informed consent for follow-up areentered in the registry. Those who refuse follow-upare recorded, but no further data are collected.

Prototype consent forms for randomly assignedand registry patients are shown in Appendix 8. Localinstitutional review boards may prefer a modificationof the consent forms to provide additional informa-tion. Any local change in the consent forms mustmeet Public Health Service requirements.Random assignment of patients. The Coordinating

Center (CC) prepared the sequence of random as-signment of patients to the treatment groups beforetheir enrollment. Random assignment was stratifiedby clinical site, and within each clinical stratum,blocks of varying length were used. The sequence ofrandom assignment was verified and incorporatedinto the BARI computer system for each clinical site.Details of the design and implementation of randomassignment are presented in Appendix 4.

Participation in other trials. Randomly assignedBARI patients may not participate in any otherclinical trial while they are participating in BARI.However, registry patients are free to participate inother studies. BARI clinical sites may participate inother studies for which BARI-eligible patients arealso eligible, but they must continue to meet theircommitment to-randomly assign an adequate numberof BARI patients.

Angioplasty GuidelinesCriteria for Acceptable Candidacy to PercutaneousTransluminal Coronary AngioplastyTo be suitable for PTCA, a patient must have

anatomic characteristics associated with a reasonableprobability of successful balloon dilatation. The ar-teries that are patent but significantly narrowed


should be able to sustain prolonged occlusion with-out the development of cardiogenic shock. Althoughthe potential for complete revascularization is not arequirement for entry of patients in BARI, theremust be a reasonable probability for successful reliefof the major stenoses presumed to be contributing toactive myocardial ischemia. These criteria are judgedby a BARI-certified angioplasty operator before ran-dom assignment. Reasons for excluding patients aredocumented on the Angiographic Exclusion formand include the following: PTCA of (a) vessel(s)responsible for ischemia is unlikely to be successfulbecause of excessive tortuosity of vessels proximal tothe lesion, excessive angulation within the lesion,excessive lesion length, total chronic occlusion, orinability to dilate because of excessive calcification;and PTCA would be excessively dangerous becauseabrupt closure is likely and would result in cardio-genic shock, or a major side branch cannot beadequately protected.

Guidelines for Strategy in Percutaneous TransluminalCoronary AngioplastyOnce a patient has been randomly assigned to

PTCA, the procedure must be performed within 2weeks. Clinical information derived from the medicalhistory, physical examination, ECG, and noninvasivestress testing, coupled with the results of coronaryand left ventricular cine angiography, determine thestrategy to be used for each patient assigned toPTCA. The aim of the procedure is to maximize theeffectiveness of PTCA in relieving ischemia andminimizing the risk of procedure-related untowardevents. A lesion may be targeted for PTCA if all ofthe following conditions are met: a stenosis repre-sents a 50% or greater diameter reduction by calipermeasurement, the normal vessel diameter adjacent tothe site of stenosis is > 1.5 mm, and the vesselsupplies a sizable region of viable myocardium. Non-significant lesions, lesions located distally or in smallarteries, and lesions in arteries that supply areas ofinfarction are not routinely dilated. For each patienta hierarchy of lesion priority is set in such a way thatPTCA is attempted first in lesions that are most likelyto be responsible for the patient's ischemia. Beforerevascularization of a patient in the study, the clinicalimportance and suitability of each lesion for PTCA iscategorized, and a treatment plan is specified by aBARI-certified operator. These data will be used todefine patient subsets for analysis of PTCA outcomeand to assess operator performance.

Guidelines for the Percutaneous TransluminalCoronary Angioplasty ProcedureFor randomly assigned patients, all PTCAs should

be performed by a certified BARI angioplasty oper-ator, and the initial PTCA must be performed by acertified BARI operator. An experienced catheter-ization laboratory staff should assist, and backupcardiac surgical support must be immediately avail-able. Preprocedure medication should include aspi-

rin (unless contraindicated) and other medicationdeemed appropriate for the clinical status of thepatient. Each patient should be fully heparinizedduring the procedure.PTCA should be performed in a cardiac catheter-

ization laboratory that is capable of providing high-quality video images with immediate replay (includ-ing 4- or 5-in. image-intensifier modes), biplaneimaging or rapidly available orthogonal single-planeimages, compound angulated projections, hemody-namic monitoring, and high-quality film processing.A full range of commercially available guiding anddilating catheters and guide wires should be avail-able. PTCA may be performed by either the brachialor femoral approach.Each procedure begins with a coronary cine angio-

gram of the vessels to be dilated. At least two scoutprojections of each vessel are obtained. For eachtargeted lesion, PTCA is attempted with the goal ofachieving <50% residual stenosis and normal TIMI(grade 3) distal flow*.2 A PTCA procedure is consid-ered completed when the patient is removed fromthe cath lab table.

Considerations of patient safety or logistics mayrequire that the initial PTCA procedure be per-formed over more than one session. If this approachis used the decision to do so must be made by the endof the first procedure, and all subsequent proceduresshould be performed within 2 weeks after the first.After PTCA cine angiography of treated arteriesshould be repeated in the same projections as ini-tially used. Additional projections may be acquired asneeded.Heparin should be continued for 24-48 hours after

the procedure in patients whose PTCA was per-formed for total occlusion or was associated withlesion dissection, thrombosis, transient occlusion, orin whom distal embolization was observed. Oralcalcium antagonist therapy should be continued for 4weeks, and a regimen of one aspirin tablet per dayshould be indefinitely continued.

In the case of abrupt closure, every effort should bemade to reestablish patency and flow and avoid MI.This effort may include emergency CABG or the useof new technology devices.

ComplicationsPatients are monitored for adverse events through-

out the hospitalization. These complications andtheir definitions are listed in Appendix 5. The needfor additional revascularization procedures, includ-ing emergency CABG and repeat PTCA for abruptreclosure, are also recorded. Each lesion that issubjected to PTCA is assessed for the occurrence ofdissection and acute closure.

Guidelines for Repeat Percutaneous TransluminalCoronary Angioplasty

Patients who have an initially successful PTCAmay undergo repeat PTCA when anatomic and clin-

*TIMI flow criteria are given in the Glossary; see Appendix 2.


ical circumstances are judged suitable. This mayoccur under two circumstances: restenosis of a suc-cessfully dilated coronary artery that is associatedwith clinical manifestations, either by recurrent an-gina or significant ischemia documented by objectivemeasures (see p V-8); and recurrent ischemia forwhich repeat angiography indicates the developmentof new significant CAD that is responsible for theischemia and amenable to PTCA.When PTCA is repeated, the procedure should be

performed by a BARI operator according to protocolguidelines. Indications for repeat PTCA will be moni-tored carefully, particularly for patients participating ina study of the 1-year follow-up angiography. RepeatPTCA must be performed in accordance with BARIindications for such procedures. Deviations are consid-ered protocol violations and require defense and justi-fication by the responsible BARI investigators.

Patients that are randomly assigned to CABG whohave recurrent ischemia that is associated with by-pass conduit or native artery stenosis or occlusion forwhich PTCA is deemed desirable should have PTCAperformed by a certified BARI operator in accor-dance with the BARI protocol.

Guidelines for Subsequent Coronary Artery BypassGraft Surgery

It is possible that patients who are randomlyassigned to PTCA will have CABG before, during, orafter their initial PTCA procedure. To ensure theappropriate use of CABG in such patients, thefollowing guidelines are recommended. 1) Once thepatient is randomly assigned, PTCA must be per-formed within 2 weeks to minimize the time duringwhich pre-PTCA crossover could occur. 2) Duringthe initial PTCA procedure, patients who experienceclosure of an artery that was previously patent butnarrowed may require emergency CABG for relief ofischemia or infarction if they are refractory to repeatangioplasty or medical therapy. The decision to pro-ceed with emergency CABG under this circumstanceshould be strictly based on the need to provideappropriate patient care. 3) After an initially success-ful PTCA, indications for subsequent elective CABGare either recurrence or persistence of disablingsymptoms that are accompanied by evidence of myo-cardial ischemia resulting from inadequate or unsat-isfactory PTCA and anatomy that is judged to beunsuitable for repeat PTCA. 4) Subsequent CABGmay be necessary for the recurrence of symptoms andischemia after a period of symptomatic relief withevidence of restenosis of a previously successfullydilated coronary artery, as described in the previoussection. Patients who require revascularization afterinitial PTCA should first be considered candidatesfor repeat PTCA. PTCA may be repeated more thanonce if there is a repetitive recurrence of symptomsor severe ischemia as defined. The decision to pro-ceed with CABG in such patients should be based onthe presence of angina or ischemia of sufficientseverity to warrant surgery in the presence of evi-

dence that the initial or repeat PTCA has beenunsuccessful. If PTCA would be particularly difficultor associated with an increased risk of untowardevents, CABG may be considered without repeatPTCA. 5) Subsequent CABG may be necessary forthe recurrence of symptoms or ischemia resultingfrom the development of new CAD. Considerationshould be given first to repeat PTCA. If repeat PTCAis judged to be inappropriate or not feasible, thenCABG should be considered if the symptoms aredisabling despite optimal medical therapy or if severeischemia is documented.

Angioplasty Operator CertificationCriteria for certification include participation as an

independent operator in more than 300 electivePTCA procedures, of which at least 100 were multi-vessel disease cases; demonstration of a success rateper lesion of 85% or greater for subtotal lesionsamong the last 100 cases; overall incidence perpatient of PTCA-related acute myocardial infarctionor emergency CABG of 5% or less; and an overallmortality rate of 2% or less for elective PTCApatients.To complete certification requirements, each BARI

angioplasty operator submits to the Central Radio-graphic Laboratory (CRL) preprocedure and postpro-cedure films of five consecutive PTCA proceduresperformed on patients with multivessel disease. TheCentral Radiographic Laboratory evaluates the filmsand determines the quality of the procedures, request-ing additional films as needed. In-depth informationabout the CRL may be found in Appendix F. Thecurrent BARI certified angioplasty operators at eachparticipating clinical site are listed in Appendix 8.

Classification of Outcome of PercutaneousTransluminal Coronary AngioplastyTo define what constitutes a successful PTCA

procedure is complex. One approach will be to assesslesion improvement. The following are requirementsfor complete lesion improvement: TIMI grade 3 flow,luminal diameter reduced by .20%, and residualstenosis of <50% diameter narrowing. If partial, thelesion that is subjected to PTCA has all of thefollowing features: TIMI grade 3 flow and luminaldiameter reduced by >20% but residual stenosis ofnot <50% diameter narrowing. If there is no lesionimprovement, neither of the above definitions for thisimprovement has been met.

Patients who undergo a PTCA procedure in whicheach diseased vessel is not dilated but all targetedvessels are improved is classified as "incompletelyrevascularized by intent." Patients in whom PTCAresults are partially satisfactory (that is, not all tar-geted vessels are improved) are classified as "incom-pletely revascularized but not by intent."

Lesion ClassificationIn the analysis of PTCA outcome, it will be impor-

tant to describe the characteristics of the lesions that


are targeted for treatment. The lesion classificationsystem developed for use in BARI is describedbelow.

1) BARI class A-A lesion is considered to beclass A if it exhibits all of the following characteristicsand has no class B or C characteristics: discretenesswith critical narrowing of <10 mm in length, vesseldiameter adjacent to the site of stenosis of > 1.5 mm,lesion accessible and not excessively tortuous, subto-tal occlusion, and concentricity with smooth borders.

2) BARI class B -A lesion is considered to be classB if it exhibits at least one of the following charac-teristics but none of those of class C: discretenesswith critical narrowing 10-20 mm in length, recent(within 3 months) total occlusion, moderate vesseltortuosity proximal to the lesion, irregular borders,ostial location, significant calcification, lesion in bi-furcation, moderate vessel angulation within lesion,thrombus, ulceration, and/or eccentricity.

3) BARI class C-A lesion is considered to be classC if it exhibits any of the following characteristics:excessive vessel tortuosity proximal to the lesion orexcessive vessel angulation at its site, chronic (>3months) total occlusion (TIMI grade 0) or an un-known period of total occlusion, critical narrowing of>20 mm in length, and/or inability to protect majorside branches.

New Technology DevicesThe New Technology Committee monitors the

development of new techniques and devices such asstents, cardiopulmonary support system, atherec-tomy, and laser and recommends if and under whatcircumstances they can be used in BARI. No newdevices, neither those of Investigative Device Exemp-tion (IDE) nor those of new technology that arefederally approved, are to be used as an initialstrategy in randomly assigned patients. The devicescan be used in clinical situations such as abruptclosure with hemodynamic compromise in which,based on local experience and judgment, the tech-nique is in the best interest of the patient. Thisrestriction does not apply to registry patients. Aftercompletion of the initial single or planned stagedPTCA procedure, new technology devices may beused if additional coronary interventions are re-quired.

Surgical GuidelinesCriteria for Acceptable Candidacy for Surgery

Before random assignment, a BARI-certified sur-geon must deem the patient suitable for CABG.Specifically, patients must have the following charac-teristics: target vessels of an adequate size for inser-tion of a bypass graft (i.e., luminal diameter of >1mm in all arteries to be bypassed); satisfactory distalrunoff; no severe diffuse atherosclerotic involvementof distal coronary arteries including the absence ofmultiple discrete severe lesions throughout thecourse of the artery to be bypassed; absence of

extreme aortic calcification; and disease severeenough to warrant surgery.Although the patient must meet the entry angio-

graphic criteria for multivessel disease, the bypass ofadditional arteries with only 50-60% luminal diam-eter narrowing by visual assessment in a patientrandomly assigned to CABG is permissible.

Guidelines for Operative ManagementOnce the patient is randomly assigned to CABG,

the procedure must be performed within 2 weeks.Rigid control of all aspects of the management of thepatient during and after surgery is not possible, butthe surgery form documents the techniques andmethods used in the surgical management of thesepatients. Anesthetic techniques are not standardized.Cold potassium cardioplegia (either crystalloid orblood) is the protection of choice; however, coldischemic arrest may be routinely preferred by somesurgeons or in special situations, and this decision isleft to the individual surgeon's judgment.The internal mammary artery should be used for

revascularization of the left anterior descending coro-nary artery whenever feasible. The choice of conduitfor revascularization of other arteries depends on theexperience and judgment of the surgeon. Details of thecannulation technique, methods of myocardial preser-vation, grafts used, aortic cross clamping, duration ofcardiopulmonary bypass, perioperative medications,and patient status are recorded.

Preoperative and postoperative use of antiplateletdrugs is recommended. For elective procedures 100mg dipyridamole four times daily for the 2 dayspreceding CABG may be used, with daily aspirinintake after operation when deemed reasonable. Inthe absence of evidence that long-term dipyridamoletherapy is essential after CABG, aspirin not exceed-ing 325 mg/day will be acceptable unless there arecontraindications to the use of antiplatelet drugs.

ComplicationsPatients are monitored for adverse events through-

out their hospitalization. These complications andtheir definitions are listed in Appendix 5.

Guidelines for Repeat Coronary Artery Bypass GraftSurgery or Subsequent Percutaneous TransluminalCoronary AngioplastyRepeat CABG should be considered on either the

recurrence of significant clinical manifestations thatsuggest a need for further invasive therapy to relievemyocardial ischemia or on discovery of evidence ofprofound ischemia with exercise testing (see InclusionCriteria, p V-2). Final decisions regarding the appro-priateness of CABG or subsequent PTCA in patientswho have had prior CABG are based on angiographicevidence of graft narrowing or progressive and severeatherosclerosis in ungrafted vessels to a narrowing ofthe luminal diameter of .50% by caliper measure-ment. Patients who have severe lesions in ungraftedcoronary arteries or bypass grafts may be considered for

Radom Initial RevasculariAMgnt Completed

-_ 2-ek_~WidowRequied

BY

0 2wks 4wks l4wks

-_ Wmndow-for

Iniialaui.cvisit


Tebphon Tebphone TdephoneConat Contact' C tact

6rnos 1 yr 2yrs 3yrs 4yrs_

ainic vl

5yrs

FIGURE 2. Follow-up chart forrandomly assigned patients. Patientfollow-up is described more fully onp V-7. Follow-up for regitry pa-tients occurs at identical intervalsexcept that the 4-14-week contact isomitted, and all follow-up is doneby telephone contact.

ainiCVWjt b Clinic Viit

a) Telephone Contact + ECG requested from physicnb) Cinhc Vsit: Resing ECG, Exercse Test, Upid Levels

PTCA if recurrent symptoms or signs of profoundischemia are thought to be related to lesions in vesselsthat are amenable to PTCA.

Surgeon CertificationEach surgeon who performs CABG on patients

who are randomly assigned in the BARI trial willrequire initial certification that is based on the fol-lowing criteria: practice as an attending staff surgeonfor 3 years or more, majority of practice devoted tocoronary artery surgery, most recent 100 consecutiveprimary, elective, isolated CABG operations with amortality rate of no more than 2% (death within 30days of procedure) and an MI rate of no more than4% (new Q waves within 30 days of procedure),performance as principal surgeon of 100 or moreCABGs with internal mammary artery grafts, and theprincipal BARI cardiac surgeon at the participatingBARI Clinical Unit is satisfied that the judgment,technical performance, results, and care after theprocedure meet current standards of the institution'sDepartment of Surgery.

Surgeon ParticipationBARI surgeons are an integral part of the study.

They serve as representatives on all working commit-tees and share responsibility with the cardiologistsfor adherence to protocol and policy decisions. Thesurgeons who are currently participating in the BARItrial are listed in Appendix 8.

Patient Follow-up and End Point AscertainmentEffective follow-up depends on the relationship

developed by the BARI staff with referring physi-cians and patients. All BARI centers make an effortto actively involve referring physicians in the study.This helps to ensure that therapy guidelines arefollowed and that subsequent revascularization is

performed at BARI institutions according to theguidelines.

Patients in BARI are followed for a minimum of 5years from the time of entry into the randomly assignedtrial or registry. During this follow-up period ninemajor end points are ascertained: these are mortality,MI, angina/chest pain, myocardial ischemia, subse-quent revascularization, resource use, quality of life,angiographic characteristics at 5 years, and left ventric-ular function at 5 years.

Scheduled Follow-upScheduled follow-up points are at 4-14 weeks

(randomly assigned patients only); at 6 months; andat 1, 2, 3, 4, and 5 years after study entry. Randomlyassigned patients alternate between clinic visits (4-14weeks and 1, 3, and 5 years) and telephone contact (6months and 2 and 4 years), and registry patients havetelephone contacts only (see Figure 2). During theclinic visit, randomly assigned patients undergo ECGand exercise treadmill tests, and blood is drawn forfasting-state serum lipid levels. When the follow-up isby telephone, an ECG is requested from the primaryphysician. Data collected for the randomly assignedand registry patients are identical with the exceptionof the early follow-up, lipid levels, and exercisetreadmill tests.

Information that is collected at each contact in-cludes symptomatic status, health behavior (diet,exercise, etc.), and quality of life. If a follow-upcontact is not possible within the specified timewindow, the evaluation takes place thereafter as soonas possible. During each scheduled follow-up, eventsthat require data collection to document end pointsare identified.

Ascertainment ofEnd PointsAt each follow-up contact all hospitalizations that

occurred since the last contact are identified. A

a a


Hospital Course form is required for the followingconditions: cardiac hospitalization for 3 days orlonger, cardiac procedure (coronary angiography,PTCA, CABG), hospitalization of any length forcardiac arrest, and hospitalization of any length for acondition thought to be a complication of a revascu-larization procedure.Data from the Follow-up and Hospital Course

forms are used to identify study end points. For eachend point additional data are collected as outlinedbelow.

1) Mortality-when a patient is identified as de-ceased, the CC must be notified within 3 days bymeans of a Notification of Major Event form. Inaddition, the following information must be sent tothe CC: a mortality data form, a narrative reportfrom the Principal Investigator that describes theevents surrounding the death and that is signed bythe BARI surgeon and/or PTCA operator involvedwith the case, a death certificate, a coroner's report(if autopsy is done), a cath lab report (if the patientdied within 30 days of PTCA), a surgical report (ifthe patient died within 30 days of CABG), any ECGsperformed within 24 hours of death, and any cardiacenzymes drawn within 24 hours of death. This infor-mation is used by the Morbidity and Mortality Classi-fication Committee (MMCC) to categorize each death.

2) MI-when MI is suspected, all ECGs that arerecorded within 3 days after the event are sent to theCentral ECG Laboratory (in-depth informationabout the Central Electrocardiographic Laboratory[CEL] may be found in Appendix 7), and a SuspectedMI report form is required. These data are used todetermine whether the event was an MI. When dataare missing or inconsistent, the MMCC reviews allavailable data to determine whether an MI occurred.MIs that do not result in hospitalization will beidentified through the yearly scheduled ECGs.

3) Angina -angina, which is classified as stable orunstable, is assessed at scheduled follow-up pointsand during hospitalization. Stable angina is classifiedaccording to the Canadian Cardiovascular SocietyClassification (CCSC). Unstable angina is classifiedaccording to the presence or absence of acceleration(increased frequency, severity, duration), pain thatlasts 20 minutes or longer and is associated withnegative cardiac enzymes, pain that continues within14 days after infarction, transient ECG changes, andpain at rest.

4) Myocardial ischemia -exercise treadmill testingwill be used to assess ECG evidence of myocardialischemia in randomly assigned patients only. Anexercise ECG is performed at 4-14 weeks and 1, 3,and 5 years. Additional exercise ECGs are requiredwhen myocardial ischemia is a reason for subsequentrevascularization. All exercise ECGs are evaluatedby the CEL for a myocardial ischemic response.

5) Subsequent revascularization-all revascular-ization procedures subsequent to the initial revascu-larization require data collection forms, includingRepeat Revascularization, PTCA or CABG Proce-

dure, and Hospital Course forms, and preprocedureand postprocedure ECGs. Angiography in asymptom-atic patients is discouraged except for the scheduledangiogram at 5 years and the 1-year angiogram inpatients enrolled in the 1-year angiographic study.No patient should undergo repeat revascularizationon the basis of anatomic findings alone. To be eligiblefor angiography during follow-up, a BARI patientmust have clinically severe angina, ischemia, or in-farction as described by at least one of the followingcriteria: unstable angina, CCSC III or IV angina,CCSC I or II angina and an ejection fraction of<50%, MI (Q wave or non-Q wave) after the initialprocedure, and/or severe ischemia on noninvasive test-ing after initial revascularization. After angiography,the decision to perform a repeat revascularizationprocedure is left to the discretion of the investigatorand the referring physician; however, the guidelinesfor subsequent revascularization (given in detail onpp V-3-V-7) must be followed in randomly assignedpatients, and revascularization should be performedby a BARI surgeon or angioplasty operator.

6) Resource use-resource use is calculated byusing the mean number of significant cardiac hospi-talizations. SEQOL is collecting bills from initial andfollow-up hospitalizations as well as data concerningfollow-up outpatient visits, tests, procedures, andmedications. Data are also being collected concern-ing the wider economic impact of the two procedures,particularly on employment. These data will be ana-lyzed to compare the profile of long-term costs andbenefits of PTCA and CABG.

7) Quality of life -at baseline and at each follow-uppoint, quality of life is assessed by using a brief ques-tionnaire. Data on employment, self-care, social life,home life, sex life, and leisure activities are included.The SEQOL ancillary study collects data on an array ofquality-of-life measures at baseline and during follow-up, including functional status, emotional health,health concerns, and social functioning.

8) Angiographic assessment- the extent of revas-cularization will be angiographically assessed at the5-year follow-up. The following criteria will be eval-uated: the condition of CABG revascularization con-duits (stenosis, disease, patency); condition of dilatedsegments; revascularization status (e.g., global leftventricular coronary perfusion and regional perfu-sion in relation to baseline stenoses of a >50%reduction in luminal diameter); progression of pre-existing lesions or development of new lesions innative vessels that affects revascularization status;and adequacy of revascularization, as indicated byintegration of regional coronary distribution withregional wall motion.

9) Left ventricular function -left ventricular ejec-tion fraction will be angiographically obtained atbaseline and at 5 years. In the event that a contrastleft ventriculogram is not obtained or is technicallyinadequate, a radionuclide angiogram will be re-quired as a surrogate measurement.


Concomitant Therapy GuidelinesIntroduction

Patients who are enrolled in BARI must receivecareful attention to concomitant medical therapy,regardless of assignment to PTCA or CABG. It isimportant to maintain comparability of concomitanttherapy for the two treatment strategies to avoid bias inthe comparison between the strategies. Although mostpatients are followed primarily by their local physicians,the BARI investigator must establish an appropriaterelationship with the patient and referring physician toensure careful attention to concomitant medical ther-apy according to the guidelines outlined below.

Risk Factor ModificationRisk factor modification will be initiated for all

patients after enrollment in the study. Because allrandomly assigned patients will be seen at the 4-14week follow-up, it is the responsibility of the BARIinvestigators to initiate risk factor modification if ithas not already been undertaken at the time ofrevascularization.Behavior modification. Instruction for behavior

modification in the areas of smoking, exercise, anddiet is initiated for all patients after enrollment in thestudy and is reinforced at each follow-up clinic visitand telephone contact.

Treatment of specific medical problems. Hyperten-sion-treatment may include common medicationsthat are used after PTCA and CABG (calciumchannel blockers, /3-blockers) as a first-line therapyin an attempt to simplify medical therapy. Angioten-sin converting enzyme (ACE) inhibitors, which are

used as a first-line antihypertensive therapy in clini-cal practice, may also be used as indicated. Thecriterion for treatment is a resting systolic bloodpressure of >160 mm Hg or a diastolic blood pres-sure of >90 mm Hg on two measurements. The goalof therapy is to maintain a blood pressure of < 140/90mm Hg. Diabetes - this will be treated with diet, oralhypoglycemic agents, and insulin, as clinically indi-cated. When control is poor, consultation with a

diabetologist should be obtained. Lipid abnormali-ties -total cholesterol, HDL, LDL, and triglyceridesare obtained at baseline, 4-14 weeks, and annuallythereafter. Baseline diet histories are encouragedthough not recorded for BARI. Patients are in-structed in an appropriate diet program based on

individual patient profile. Diet therapy remains an

integral part of the approach to lipid disorders andmust be maintained regardless of drug regimens.Every reasonable effort should be made to achievethe following target lipid levels: total cholesterol lessthan 180 mg/dl, HDL cholesterol greater than 45mg/dl, LDL cholesterol less than 110 mg/dl, triglyc-erides less than 120 mg/dl, and a total cholesterol/HDL ratio of less than 4.5.

Patients who do not sufficiently respond to diet andexercise programs should begin lipid-lowering drugtherapy, following the algorithm proposed by the Na-

tional Cholesterol Education Program.8 Surgery or arecent MI may temporarily lower the serum cholesterolvalue, and this must be taken into account whenevaluating lipid values.

Additional Concomitant TherapyAll patients should be indefinitely maintained on

antiplatelet therapy with one aspirin per day unlessthe drug is contraindicated or not well tolerated. Inthe latter case, alternative therapy with other anti-platelet drugs may be considered. Recurrent myocar-dial ischemia should be treated with nitrates, calciumchannel blockers, and /3-blockers according to usualclinical practice.

Data AnalysisIntroductionBARI compares the safety and efficacy of the PTCA

and CABG treatment strategies rather than the treat-ments themselves. All primary analyses will be con-ducted by intention to treat regardless of what treat-ment or treatments patients received. Comparisons oftreatment strategy will be made not only for the entirerandomly assigned trial population but also withinrelevant subgroups.

End PointsTo compare end points, 5-year incidence rates and

confidence limits will be calculated for the differencein rates between the initial CABG and initial PTCAstrategies. The primary study end point is mortality at5 years. Other major end points include MI, angina/chest pain, myocardial ischemic response, subse-quent revascularization, resource use, quality of life,and measures of angiographic characteristics and leftventricular function at 5 years.

Composite End PointsTo simultaneously analyze several outcomes, com-

posite end points will be created. For example, angina-free and event-free survival are best viewed in a hier-archical manner, which has been used to expressoutcome in the National Heart, Lung, and BloodInstitute PTCA Registry. When assessing long-termsymptomatic relief, subsequent procedures and phar-macological intervention must be taken into account.This will be done in a cross-sectional analysis. Whengiven sufficient intervention, it is expected that mostpatients will end up either asymptomatic or mildlysymptomatic at follow-up. The comparison of the twotreatments will be based on the number and types ofprocedures and the amount of medication that isrequired to achieve asymptomatic status. The distribu-tion of the most severe form of intervention is expectedto be inherently different between the two treatmentstrategies. The CABG group undergoes a more exten-sive and invasive technique initially, whereas the PTCAstrategy group is expected to have a greater number ofsubsequent procedures during follow-up.


SubgroupsAs much as possible, we will limit formal analyses to

those subgroups that have been specified a priori. Itshould be noted that because of the smaller samplesizes, these subgroup analyses will have limited power.Multivariate models that include treatment by covari-ate interaction will be used to see whether the treat-ments differentially affect outcome in various clinicalsubgroups. Important baseline factors that define sub-groups include the following: symptomatic status (an-gina/ischemia pattern), left ventricular function, extentof myocardial ischemia, and angiographic risk (e.g., thepresence of a class III lesion or the amount of myocar-dium at risk). Some subgroups might also be created onthe basis of exploratory statistical analyses. Analysis ofa multitude of such subgroups creates a multiple-comparison problem that will be acknowledged.

Methods ofAnalysisMethods of analysis can be classified by whether

data are collected cross sectionally (e.g., 5-year ven-triculography), longitudinally (e.g., angina status ateach follow-up point), or the time to event (e.g., timefrom baseline to death).Analysis of baseline data. Comparability of the two

treatment groups is expected because of the processof random assignment; nevertheless, distributions ofbaseline characteristics within the two groups will beexamined. The following set of key baseline charac-teristics will be defined: known risk factors for car-

diac mortality (e.g., history of congestive heart fail-ure, previous MI, history of diabetes, and history ofhypertension); usual coronary risk factors (lipid lev-els, smoking status, and blood pressure levels); an-

giographic risk factors (severity of coronary arterydisease, impaired left ventricular function); durationand nature of symptoms, functional capacity, medi-cation use, and results of baseline ECG; and demo-graphic factors (age, sex, race, education, and socio-economic status).

Suitable techniques will be used to test for compara-bility of the distribution of these variables between thetwo treatment groups. Results of these analyses will beused to describe the study population and to determineany imbalance for which adjustment must be made inthe final analysis of treatment comparisons.

Time-to-event analyses. Standard life table tech-niques will be used to analyze mortality and MI.All-cause mortality and cardiac mortality will beseparately assessed. Cumulative event rates for sur-vival and MI-free survival will be compared betweenthe two treatment groups.

Cross-sectional and longitudinal analyses. Some im-portant BARI end points, such as the status of anginaand myocardial ischemic response, will be measuredand analyzed cross sectionally.Some patients originally present with angina and

others are asymptomatic with ischemic manifestationat entry. Patients with angina can be analyzed for therecurrence of angina, whereas ischemic patients can

be separately analyzed for ischemia at follow-up.This analysis (in effect two subgroup analyses) woulddescribe the relative efficacy of the two treatmentsfor relief of the specific symptoms that the originalstrategy was intended to relieve. Alternatively, thesubgroups can be combined and analyzed for therecurrence of any symptom.

Observations at multiple time points will be han-dled by various approaches. One method will simplybe to assess the outcome at final follow-up. Anotherapproach is to average the severity of response overmultiple observations during follow-up. Profile analy-sis, generalized linear models, and survival modelsthat allow inclusion of time-dependent covariates(e.g., events, medications) may be appropriate toexamine differences between treatments over time inan exploratory manner.

Further cross-sectional analyses include the follow-ing examples. Exercise testing- exercise test results willbe compared at each time point by treatment. Theassociation of exercise test results and subsequentrevascularization procedures will be examined. Returnto work- employment status before and after treat-ment will be cross tabulated. The interpretation of suchdata will take into account the fact that a large portionof the BARI population will be men and women nearretirement age who will not remain in the active laborforce, regardless of the success of the procedure. Costanalysis- a major ancillary study will be devoted to thethorough evaluation of cost data generated from BARIby seven clinical centers. Analysis of cost data in themain study will be limited to the comparison of thenumber of significant cardiac hospitalizations for eachof the two initial strategies. Follow-up angiographicanalyses-the angiographic follow-up data, collected at1 year at selected centers and at 5 years at all centers-will provide information in the following areas: graftpatency, patency of dilated segments, and progression!regression of the atherosclerotic process in both treatedand untreated segments. The distribution of changes inthese areas between baseline and follow-up angiogra-phy will be compared by treatment group. A myocardialjeopardy index, which is computed by the CRL fromthe angiography readings, will allow overall assessmentof revascularization results.

Interim Data AnalysisEnd point data are not revealed to clinical inves-

tigators during the course of the study but arepresented at semiannual meetings of the Safety andData Monitoring Board (SDMB). First and foremost,mortality and MI rates by treatment group are pre-sented. These rates are expected to be relatively lowand similar in the two treatment groups. Proceduralcomplications are reported for each treatment group,as are rates of subsequent procedures during follow-up. Should one of the treatment groups experiencesignificantly higher rates of mortality or MI, theSDMB would consider either early termination of thetrial or modification of the protocol to exclude pa-tients in certain subgroups.


Registry AnalysesEligible but not randomly assigned patients. It is as-

sumed that the BARI clinical trial results would holdtrue in a larger population of patients whose diseasecharacteristics are similar to those of the BARI pa-tients. To test this assumption, outcomes from therandomly assigned trial will also be studied in BARI-eligible patients who do not consent to random assign-ment. If after adjustments for differences between therandomly assigned and the eligible but not randomlyassigned groups the long-term outcome by treatment issimilar, then the trial results can be generalized topatients who meet BARI criteria.

In spite of careful statistical adjustments to maxi-mize comparability, the investigation of end pointsamong eligible but not randomly assigned patientswill only have the status of a good observational studyand will not be suitable for providing primary esti-mates for treatment comparisons. In particular, inthe registry population treatment is recommendedbased on a variety of selection factors, many of whichare subtle and cannot be measured. An additionalproblem'to be solved in analyses of such data will bethe definition and time of treatment assignment.As is typical for observational studies, registry

analyses will also explore and generate hypothesesregarding the relation of various patient characteris-tics to outcome (predictors of events). The charac-teristics to be considered will be the key baselinevariables of the randomly assigned study. Periproce-dural outcomes such as mortality, myocardial infarc-tion, and other complications will serve as end pointsin these analyses; procedure-related complicationsand outcomes will in turn also serve as risk factors forlong-term events.

Statistical' methods such as linear regression, logis-tic regression, and Cox regression will be used toobtain adjusted estimates of the effect of prognosticvariables on outcome. As in the trial, testing forqualitative interaction between treatments and riskmay be appropriate.

Patients clinically but not angiographically eligible. A5-10% random sample of patients who meet clinicalcriteria for the randomly assigned trial but are ex-cluded on angiographic grounds will be used to assessthe appropriateness of angiographic exclusion forBARI.

In this sample baseline characteristics as well asin-hospital outcome and major follow-up events willbe analyzed by the given treatment. Results of theseanalyses will lend insight into the prognostic impor-tance of those vessel/lesion characteristics that arenot otherwise present in BARI-eligible patients.

Determination of time 0. For the randomly assignedpatients, the clear choice for time 0 is the time ofrandom assignment. For registry patients, the time ofassignment of the BARI identification number isused. Random assignment determines the treatmentgroup in the clinical trial, but a clear-cut treatmentassignment does not apply to registry patients. Here,

treatment can be defined by the management re-ceived within 2 weeks of BARI ID assignment, oralternatively within 3 months. The 2-week rule isbased on the requirement that the assigned treat-ment be delivered within 2 weeks in the trial.

Quality Control/Reproducibility of MeasurementsTo establish the reproducibility of key measure-

ments in BARI, replicate blinded assessments will beperformed on randomly selecited samples. Measuresin this category include angiographic baseline char-acteristics, measure of PTCA outcome, occurrence ofMI, cause of death, and functional test outcome.(Details for some of these analyses are described inthe Central Laboratory sections located in Appen-dixes 6 and 7).

Reliability will be assessed by appropriate statisti-cal techniques that are applicable to the replicationof measurement design.

Exploratory Methods ofEnd Point AnalysisCompleteness of revascularization. With respect to

CABG, it has been claimed that the more completethe revascularization the better the results. Thisclinically appealing statement has never been ade-quately tested because no one would randomly assignpatients to complete versus incomplete revasculariza-tion. The BARI trial will indirectly test this hypoth-esis because the PTCA strategy does not necessarilyaim for or achieve the completeness of revasculariza-tion that CABG does.Percutaneous transluminal coronary angioplasty as a

delay for coronary artery bypass graft surgery. An addi-tional goal of BARI is to learn the rate and timing ofthe crossover from PTCA to CABG. A PTCA patientis a crossover case if he or she undergoes CABG atany time after random assignment either before orafter PTCA. The percentage of patients who crossover, as well as the length of delay to CABG, will beanalyzed.Repeat revascularization. The attribution of events

in the presence of subsequent revascularization by atreatment other than the one randomly assigned is ofparticular concern in BARI. Although the literaturehas described many avenues of analysis in suchsituations, none of these analyses can be interpretedin a classical statistical sense. However, to assess theimpact of crossover two ancillary methods of analysisare proposed, with the clear understanding that theresults obtained must be interpreted in the samemanner as any other observational study results.The first exploratory method is to censor follow-up

at the time of the different method of revasculariza-tion. This approach looks at' patients only during thecontinuation of their originally assigned treatmentand does not count events after initiation of the othertreatment. This method of analysis has less power todetect differences between the treatment groupsbecause many patients may be lost because of asubsequent procedure. More importantly, thismethod could also lead to biased results if subse-


quent procedures occur frequently and in a highlyselective manner, removing individuals who are ei-ther at more or less risk than others for the end pointof interest. This possibility must be examined beforesuch analysis, and great care should be exercised indrawing inferences.A second auxiliary method is the transitional life

table analysis, in which patients are considered to bemembers of their original treatment group until thedifferent revascularization procedures and members ofthe other group thereafter. This approach allows allfollow-up data to be used in the analysis, although theremay again be bias because patients who subsequentlyreceive the other treatment may be different from thosewho remain with their treatment assignment.4

These two methods have been useful in previousstudies.4

Organizational StructureIntroduction

Investigators in the BARI study collaboratethrough an organizational structure designed tomaintain the continuity of study operations and tofacilitate effective communication and cooperationamong components.

Operational ComponentsClinical centers. Fourteen primary clinical centers

with one or more hospitals are participating in BARI.All are located at major medical centers in theUnited States and Canada. The clinical centers areresponsible for the screening and recruitment ofeligible patients, performance of PTCA and CABG,coordination of patient care and follow-up, and col-lection of all clinical information and test data thatare required by the BARI protocol. Individual clini-cal sites may also consider interaction with otherinstitutions to enhance recruitment; such relation-ships consist of the following specific categories.

Coinvestigational sites. Each one of these functionsas a satellite of an original BARI clinical center anduses resources of the parent site. At these sites thesame protocol activities as at the associated parentsite are performed; however, computer and datamanagement resources are shared with the parentsite. Investigators at these sites become voting mem-bers of the Steering Committee if requirements forprotocol and performance criteria (certification ofsurgeons, angioplasty operators, and coordinators)and minimum recruitment levels (3 patients/mo) areachieved.European parallel study. The Institute for Clinical

and Experimental Medicine in Prague, Czechoslova-kia, participates in the BARI study through a UnitedStates-Czechoslovakian scientific agreement. Ran-dom assignment in this parallel study began in Feb-ruary of 1989. All data that are collected will beanalyzed as an ancillary study to BARI.

Coordinating center. The CC at the University ofPittsburgh has primary responsibility for the BARI

study design, data collection and management, andanalysis of BARI results. The CC staff maintain theprotocol and operations manual, design and supportthe data entry and management system, and imple-ment certification and procedures of random assign-ment. CC staff are responsible for preparing anddistributing regular BARI progress reports and min-utes, preparing reports for the SDMB, monitoringend point results, preparing data analyses, and en-suring the quality and accuracy of data collection. CCinvestigators will ultimately play a key role in thepreparation of reports for publication.

Central radiographic laboratory. The BARI CRL islocated at Stanford University, Palo Alto, Calif. Thelaboratory is responsible for the receipt, review, andanalysis of all BARI radiographic test data, includingthe assessment of PTCA performance and success,on all randomly assigned patients from each of theBARI clinical sites and for transmission of the resultsto the CC in a timely fashion. In addition, thislaboratory is responsible for monitoring the quality,completeness, and timeliness of BARI radiographicprocedures that are performed at the clinical sites.The CRL will also be responsible for reviewing theangiograms of the patients who are excluded fromBARI eligibility based on angiography criteria andwho are subsequently selected for the registry. (Thefunctions of the CRL are discussed further in Ap-pendix 6.)

Central ECG laboratory. The BARI CEL is locatedat St. Louis University. It is responsible for qualitycontrol and interpretation of all resting and exerciseECGs required by the BARI protocol. (The func-tions of the CEL are discussed further in Appendix7.)

Coordinating Center for the Study of Economics andQuality of Life. The SEQOL coordinating center iscurrently located at Stanford University, which movedfrom Duke University in January of 1990, and isresponsible for the coordination and quality control ofdata collection at the seven BARI clinical sites that areparticipating in the SEQOL study (Boston University,Cleveland Clinic Foundation, Duke University, MayoClinic, St. Louis University, University of Alabama, andUniversity of Michigan). The ancillary study activitiesare supported by a grant from the Robert WoodJohnson Foundation.

Administrative ComponentsStudy Chair. The Study Chair, which is appointed

by the National Heart, Lung, and Blood Institutedirector, has major responsibility for the scientificdirection of the BARI trial. The Study Chair hasbeen appointed to serve for the duration of the studyunless other arrangements are made by mutualagreement between the Chair and the NationalHeart, Lung, and Blood Institute director. In theevent that the Study Chair is vacated because ofdeath, resignation, or inability to serve because ofserious illness, the National Heart, Lung, and BloodInstitute director will appoint a new Chair.


National Heart, Lung, and Blood Institute ProgramOffice. The National Heart, Lung, and Blood InstituteProgram Office is in the Cardiac Diseases Branch ofthe Division of Heart and Vascular Diseases and isresponsible for the overall direction and monitoring ofBARI. The Program Office participates in the generalorganizational and scientific guidance of the study andmonitors the study progress for the institute. Statisticalguidance is provided by the National Heart, Lung, andBlood Institute.

Steering Committee. The Steering Committee is com-posed of the Study Chair and the Principal Investiga-tors from each BARI Clinical Site, the Central Labo-ratories, the Coordinating Center, and the NationalHeart, Lung, and Blood Institute Program Office. Thiscommittee provides the scientific direction for the studyand periodically meets to assess progress. The SteeringCommittee developed the BARI protocol and is re-

sponsible for its execution.The following technical subcommittees have been

appointed: Angiography, PTCA, CABG, Design andAnalysis, End Points, Forms, Entry Criteria, Con-comitant Therapy, Economic, and New Technology.These subcommittees are charged with overseeingspecific areas of the BARI protocol.

Operations Committee. The Operations Committeeis responsible for the daily conduct of BARI as an

extension of the Steering Committee. Weekly confer-ence calls are conducted to ensure that importantissues are appropriately addressed and problemsresolved. It is composed of the Study Chair, thePrincipal Investigator of the Coordinating Center,representatives from the National Heart, Lung, andBlood Institute Program Office, Principal Investiga-tors of the Central Laboratories, and other BARIparticipants on an ad hoc basis as required.

Policy and Publication Committee. The Policy andPublication Committee (PPC) has the responsibilityof approving ancillary studies and analyses of theBARI data base beyond the primary baseline andoutcome papers. Principal Investigators as well as

associate investigators at all BARI clinical and cen-

tral sites are encouraged to develop protocols andpublish research papers with BARI study data. ThePPC will ensure that these protocols are methodolog-ically sound and that the publication of high-qualitymanuscripts progresses efficiently.

Independent ComponentsSafety and Data Monitoring Board. The SDMB is an

external review committee appointed by the NationalHeart, Lung, and Blood Institute that has the respon-sibility of protecting the scientific integrity of the BARItrial. The Board reviews interim trial results and ad-vises the National Heart, Lung, and Blood Institute onall policy matters. The SDMB includes senior scientists,cardiologists, surgeons, a biostatistician, and an ethicist.The Study Chair, the Principal Investigator from theCoordinating Center, and representatives from theNational Heart, Lung, and Blood Institute ProgramOffice also participate as nonvoting members.

The group meets semiannually; in addition, atmonthly intervals the CC provides data to the Chairof the SDMB to ensure early identification of poten-tial adverse outcomes of therapy during the recruit-ment phase.

Morbidity and Mortality Classification Committee.The MMCC is an external National Heart, Lung, andBlood Institute-appointed group responsible for thereview and classification of MI and death among

study patients. Classification is based on case sum-

maries provided by the CC. The Committee serves

independent of BARI and clinical and central labo-ratory investigators.

Data Management/ComputingOverviewThe CC has implemented a distributed data entry

system for BARI. Each clinical site has been pro-

vided with an IBM PS/28 (a registered trademark ofInternational Business Machines Corp.) microcom-puter system that is equipped with the hardware andsoftware necessary for data collection and manage-

ment at the clinical site. Data are transferred weeklyfrom each site to the CC, where they are maintainedin an aggregate database on the CC's Digital Equip-ment Corporation VAX 6310 computer system (Dig-ital Equipment Corporation, Maynard, Mass.).

Distributed Data Entry and ManagementThe BARI distributed data entry system consists of

six phases. At the site level, the forms* are enteredinto the microcomputer and intraform edits per-

formed. Data are then transmitted to the data centerby means of a telecommunications link and uploadedinto data sets on the VAX computer. At this pointinterform edits are done, and then finally the data are

appended to the aggregate BARI data base. Thesephases are discussed in greater detail below and are

also described in the Data Management Manual.Form entry. The Epidemiology Data Center at the

University of Pittsburgh has designed and imple-mented a screen-oriented data entry program for theIBM microcomputer called the PoP data entry sys-tem. In this system the pages of study forms appearon the microcomputer screen in much the same wayas they appear on paper. Data are transcribed frompaper forms to the corresponding data fields that are

displayed on the microcomputer. During data entryeach field is interactively subjected to data typeverification and range checks. Data are further veri-fied by using a double entry system. All data must beentered twice before subsequent system processingcan occur. Each discrepancy that requires resolutionduring this process is recorded in an audit file thatcontains a full account of all changes made to thestudy data records, the date of each change, and theID number of the operator who makes the changes.

*Data Entry Form sets may be ordered at a cost of $50.00 fromthe BARI Data Coordinating Center, University of Pittsburgh, 127Parran Hall/130 DeSoto Street, Pittsburgh, PA 15261.


Intraformt edits. After entry and verification, eachdata record undergoes defined intraform logic andconsistency checks. A report that lists the nature ofeach error is generated for resolution by the clinicalsite staff. Corrections to the data are then madethrough the PoP update module.Data transmission. Data records that are free of

errors are flagged for transmission to the CC. Anautomated data file transfer system retrieves datarecords from each clinical center once a week bymeans of modem and telephone lines. This occurs atnight while the centers' microcomputers are not inuse. The CC host computer dials the telephonenumber of the clinical center's modem. As a securitymeasure, the center's modem, on recognition of theincoming call, disconnects the line and then redialsthe CC to proceed with file transfer. The entiretransaction is recorded in an audit file that is re-viewed the next morning. Unsuccessful transfers areidentified and resolved during that day.Data that are contained in files transmitted to the

CC can no longer be accessed by the clinical center.Any subsequent changes to the data must be madethrough the submission of data base update requests,which is discussed below.Load data. At the CC data records are transferred

to the VAX 6310 computer and loaded into System1032 Data Base Management System (DBMS) datasets. (System 1032 is a product of CompuServe DataTechnologies, Boston.)

Interform edits. At the VAX level each data set isprocessed through extensive interform edits to en-sure consistency and validity across all forms. Errorreports are generated and sent to the clinical centersfor resolution. Data changes that are required toresolve errors at this level are made by submitting adata base update record by means of the PoP system.These records are retrieved during the weekly datafile transfer and are loaded into the VAX system,where a program accesses the data base and per-forms the requested change. This system generatesan audit trail of all changes made to the BARI database at the VAX level.Appending data. The data transmitted weekly to the

CC from the clinical sites are loaded into the aggregatedata base on the VAX. An append audit trail isgenerated that shows all records loaded into S1032 datasets and any records that are rejected. All data analysesoriginate from this final, clean, aggregate data base.

Form InventoryPoP provides an automatic form inventory system.

Information from each entered form is automaticallyextracted from the data record and recorded in aseparate inventory data base. The inventory data baseis used to monitor data management progress at thesite level. In addition, this data base generates sched-uling reports that inform the clinical staff of patientswho are due for a protocol visit during the next month.

Coordinating Center Statistical ComputingThe aggregate study data base is maintained in the

System 1032 DBMS. Files for analysis that use statisti-cal programs such as SAS, SPSS, BMDP, and MINITAB (SASInstitute Inc., Cary, N.C.; SPSS Inc., Chicago, Ill.;BMDP Statistical Software, Los Angeles; and Minitab,Inc., State College, Pa; respectively) are easily gener-ated. The 1032 DBMS includes a powerful program-ming language that allows new variables to be createdfrom existing information. All study analyses are com-pleted on the VAX computer by using standard statis-tical packages whenever possible.

Data SecurityClinical site microcomputer. Each clinical site is

responsible for confidentiality by appropriately re-stricting physical access to the BARI microcomputer,which is placed in a locked room after working hours.

Data backup is forced daily and can also beperformed more often if desired. The clinical sitedata manager is responsible for placing backup dis-kettes in a safe storage area in a location remote fromthe microcomputer. Damage to the hard disk resultsin the loss of no more than 1 day's work. Allmicrocomputer drives are backed up once eachmonth.To gain access to the PoP data entry system, users

must go through a log-in procedure. Each user has aunique identification name. Audit trail files associateall PoP data system activities with the user's identi-fication name. The clinical site data manager isresponsible for maintaining a confidential list ofsystem users and passwords.

Treatment assignment file. The treatment assign-ment file on each site computer contains the treat-ment assignment schedule for that site. When apatient is randomly assigned, the PoP module selectsthe next treatment assignment in the list and associ-ates it with the randomly assigned patient's name andBARI identification name number. The BARI iden-tification and the treatment assigned are then storedin a random assignment file.

Because of its highly sensitive nature, the randomassignment file is encrypted. This ensures that thenext available random assignment cannot be deter-mined by any method other than that of the PoPrandom assignment module. The random assignmentfile is protected from accidental deletion and cannotbe modified.

In an effort to prevent improper assignment torandomly assigned treatment, the PoP random as-signment module requires that patient data thatpertain to exclusion criteria be entered. Treatmentwill be assigned only after data corresponding toappropriate criteria have been entered. Once a pa-tient receives a treatment assignment, there is no wayof removing the patient from the random assignmentfile.

Coordinating Center VAX Computer. The CC en-sures patient confidentiality by using alphanumeric


identification names to identify forms. This identifi-cation name is easily linked to the patient name atthe site level only.

Study data files at the CC are protected againstinadvertent change or access by nonproject person-

nel. The VAX computer system provides protectionat the individual file level by means of passwords andfile protection codes that are controlled by the proj-ect manager. File protection codes can permit certainusers to have "read only" access to data without theability to make changes.Backup of VAX files is performed daily. In addi-

tion, a special tape backup is made weekly for theentire VAX computer system. These tapes are storedin a room and location that are separate from thedaily backup medium. Therefore, a computer systemfailure will at worst result in the loss of 1 day's work.In a worst-case disaster (fire, explosion, etc.), no

more than 1 week's work would be lost.

Appendix 1: AcronymsAAP, Angiographic Assessment Program; ACE,

angiotensin converting enzyme; BARI, Bypass An-gioplasty Revascularization Investigation; BMDP, Bio-Medical Data Program; CABG, coronary artery by-pass graft surgery; CAD, coronary artery disease;CASS, Coronary Artery Surgery Study; CC, Coordi-nating Center; CCSC, Canadian Cardiovascular So-ciety classification; CEL, Central Electrocardio-graphic Laboratory; CRL, Central RadiographicLaboratory; DBMS, Data Base Management System;ECG, electrocardiogram; EF, ejection fraction; E1TT,exercise treadmill testing; IMA, internal mammaryartery; LAO, left anterior oblique; LV, left ventricle;MC, Minnesota Code; MI, myocardial infarction;MMCC, Morbidity and Mortality Classification Com-mittee; NHLBI, National Heart, Lung, and BloodInstitute; NIH, National Institutes of Health; PI,Principal Investigator; PTCA, percutaneous translu-minal coronary angioplasty; RAO, right anterioroblique; SAS, Statistical Analysis System; SDMB,Safety and Data Monitoring Board; SEQOL, Studyof Economics and Quality of Life; spss, StatisticalPackage for the Social Sciences; TIMI, Thrombolysisin Myocardial Infarction.

Appendix 2: Bypass Angioplasty RevascularizationInvestigation Glossary of Definitions

Canadian Cardiovascular Society ClassificationofAngina

Class I: Ordinary physical activity does not cause

angina, such as walking and climbing stairs. Anginaoccurs with strenuous or rapid prolonged exertion atwork or recreation.

Class II: Slight limitation of ordinary activity, suchas walking or climbing stairs rapidly; walking uphill;walking or stair climbing after meals; the limitation ofordinary activity in cold or windy weather, whileunder emotional stress, or during the few hours afterawakening; walking more than two blocks on a level

surface; or climbing more than one flight of ordinarystairs at a normal pace and under normal conditions.

Class III: Marked limitation of ordinary physicalactivity, such as walking one to two blocks on a levelsurface or climbing one flight of stairs under normalconditions and at a normal pace; comfortable at rest.

Class IV: Inability to carry on any physical activ-ity without discomfort. Anginal syndrome may bepresent at rest.

Clinically Severe Angina or IschemiaClinically severe angina or ischemia involves any of

the following conditions: 1) non-Q wave myocardialinfarction stabilized 4 hours to 6 weeks; 2) unstableangina stabilized 4 hours to 6 weeks; 3) stable CCSCIII or IV angina; 4) stable CCSC I or II angina andone or more of the following: a) severe ischemia onnoninvasive testing, b) Q wave MI (stabilized for 24hours, revascularization clinically indicated, and abil-ity to randomly assign patients within 30 days afterMI), c) resting ejection fraction <50%, d) a history ofstable class III or IV angina and current intensivemedical therapy (with exercise testing waived); 5) noangina but severe ischemia on noninvasive testing andprior Q wave MI; 6) no angina within 6 weeks ofstudy entry but a history of prior angina, with currentsevere ischemia on noninvasive testing.

Definite Q Wave Myocardial InfarctionA definite Q wave MI involves the two-step wors-

ening of Minnesota code Q waves, as determined bythe CEL.

Multivessel Coronary DiseaseThe BARI definition of multivessel coronary dis-

ease requires that two or more myocardial territories(anterior, lateral, inferior/posterior) be jeopardized.Therefore, the following definition could also betermed multiterritory disease. To be eligible forBARI a patient must have a 50% or greater diameterreduction by caliper measurement in arteries thatsupply at least two of the three major myocardialterritories (Table 1, Figure 3). These arteries must be>1.5 mm in diameter adjacent to the site of thestenoses. Assignment of coronary segments to aspecific myocardial territory (anterior, lateral, inferi-or/posterior) includes consideration of coronary ar-tery dominance (Table 1).

TABLE 1. Coronary Segments*

Myocardialterritory Right dominant Left dominant Balanced

Anterior 12-17, 29 12-17, 29 12-17, 29Lateral 18-22, 28 18-22, 28 18-22, 28

Inferior/posterior 1-9 23-27 1-5, 23-26

*See Figure 3.


10

/23

27-_- 26-.- 25-

FIGURE 3. Coronary artery map. Right coronary artery: 1, proximal; 2, middle; 3, distal; 4, posterior descending; 5, rightposteroatrioventricular; 6, first posterolateral; 7, second posterolateral; 8, third posterolateral; 9, inferior septal artery; 10, acutemarginal artery. Left coronary artery: 11, left main; 12, proximal left anterior descending; 13, middle left anterior descending; 14,distal left anterior descending; 15, first diagonal; 15a, first diagonal branch; 16, second diagonal; 16a, second diagonal branch; 17,anterior septals; 18, proximal circumflex; 19, middle circumflex; 19a, distal circumflex; 20, 21, 22, first, second, and third obtusemarginal; 20a, 21a, 22a, first, second, and third obtuse marginal branches; 23, left atrioventricular; 24, 25, 26, first, second, andthird posterolaterals; 27, left posterior descending; 28, ramus; 28a, ramus branch; 29, third diagonal; 29a, third diagonal branch.

Myocardial Ischemic Response (End Point Definition)A myocardial ischemic response is indicated by any

of the following exercise-induced ECG responses: 1)horizontal or downsloping ST segment depression.1 mm 80 msec after the J-point; 2) a slow upslopingST segment depressed .1.5 mm 80 msec after theJ-point; and/or 3) ST segment elevation .1 mm 80msec after the J-point in a non-Q wave lead, each ascompared with the rest tracing.

New Technology DevicesNew technology devices include those currently

holding Investigational Device Exemption status andany one deemed "new technology" by the NewTechnology Committee. Some examples are stents,atherectomy, and laser catheters. (See pp V-3-V-6,Angioplasty Guidelines.)

Severe Ischemia on Noninvasive Testing (EntryCriteria Definition)

Severe ischemia on noninvasive testing is definedas the results of testing as listed herein: 1) Exercisetreadmill testing: exercise-limiting definite angina,with final exercise stage less than Bruce stage 3; andexercise-induced severe ST segment response, with afinal exercise stage less than Bruce stage 3. In addi-tion, patients who do exercise in Bruce stage 3 mayalso be considered for random assignment if theischemic changes occur within the first 6 minutes andthe patient is judged to be a candidate for revascu-larization. 2) Exercise 201T1: multiple reversible de-fects, increased lung uptake and a single reversible or

fixed defect, and one fixed defect and one reversibledefect remote from the fixed defect. 3) Persantine20`T1: multiple reversible defects and one reversibleand one fixed defect. 4) Exercise radionuclide ven-triculography: resting ejection fraction >0.50 andexercise ejection fraction <0.50, with a decline inejection fraction of at least 5% in the absence of leftbundle branch block or significant arrhythmia; andresting ejection fraction .0.50, with a severe STsegment response and work load <450 kpm.Severe ST Segment ResponseA severe ST segment response may be indicated by

any of the following exercise-induced ECG re-sponses: 1) horizontal or downsloping ST segmentdepression >1.5 mm 80 msec after the J-point; 2)slow upsloping ST segment depression .2.0 mm 80msec after the J-point; and/or 3) ST segment eleva-tion >1 mm 80 msec after the J-point in a non-Qwave lead, each as compared with the rest tracing.Suspected Myocardial InfarctionMI is suspected if any episode of chest pain lasts 20

minutes or longer or if there is the occurrence of newQ waves in the ECG or if there is elevation of cardiacenzymes, as well as any other reason to suspect anMI.

Flow Classification (Thrombolysis in MyocardialInfarction): A Randomized Clinical Trial ofTissue- Type Plasminogen Activator (t-PA) in PatientsWith Acute Myocardial InfarctionFour specific grades of coronary flow rate, as

developed in the TIMI study, are given below.


Grade 3 flow. Antegrade flow into the terminalcoronary artery segment distal to a stenosis occurs aspromptly as anterograde flow into a terminal coro-nary artery segment proximal to the stenosis. Theclearance of contrast material from the distal coro-nary artery terminal segment is as rapid as clearancefrom an uninvolved, more proximally located coro-nary artery segment.

Grade 2 flow. Contrast material opacifies the ter-minal coronary artery segment distal to the stenosis.However, the entry of contrast material into theterminal coronary artery segment is perceptiblyslower than it is into a more proximally locatedterminal coronary artery segment. Alternatively, theclearance of contrast material from a terminal coro-nary artery segment distal to a stenosis is noticeablyslower than that from a comparable segment that isnot preceded by a significant stenosis.

Grades 1 and 0 flow. There is no obvious flow ofcontrast material beyond the stenosis in question. Asmall amount of contrast material may penetrate theocclusion (grade 1); however, the contrast materialfails to fully opacify the coronary artery beyond theocclusion.Flow of contrast material not assessable. In this

situation there may be competitive flow to a terminalcoronary artery segment. For instance, total occlu-sion in the proximal right coronary artery may beassociated with both anterograde (ipsilateral) andretrograde (contralateral) collateral flow to the distalright coronary artery bed. There may be relativelyfaint visualization of the distal right coronary bed bythe anterograde collateral flow and also by retro-grade collateral flow. In this situation it is not possi-ble to accurately assess the flow rates of contrastmaterial in the terminal right coronary artery bedbecause of the competitive nature of flow. It may alsobe impossible to judge relative flow in a proximaltotal occlusion in which collateral flow to the termi-nal coronary segments is poor. If the terminal coro-nary segments in this situation are poorly visualized,the correct judgment concerning perfusion may beunclear, and this choice should be indicated.

Appendix 3: Sample SizeMortality was used as the basis for selecting the

sample size for the trial. For PTCA (a procedure lessinvasive than CABG) to be recommended as aninitial revascularization strategy, it is of primaryimportance to know that the long-term results withPTCA are not appreciably worse than with CABG interms of mortality. With this in mind, the basis of theBARI sample size calculation was to choose a samplesize large enough so that, with a high probability, theupper confidence limit for the absolute differencebetween PTCA and CABG mortality rates would notexceed a specified value.5 This specified value repre-sents the maximum acceptable absolute differencebetween the CABG and PTCA event rates. In BARIthe specified value was taken to be 2.5%, in light ofthe estimated 5% mortality rate discussed below. In

other words, the sample size was chosen in such away that if the two strategies were in fact identical interms of mortality, the study would have a highprobability of ruling out excess mortality of 2.5% ormore in the PTCA group relative to the CABGgroup.Review of the data from the National Heart, Lung,

and Blood Institute PTCA Registry and the CASSTrial and Registry indicates that for patients withmultivessel CAD, a 5-year mortality rate of 5% aftereither therapy was a reasonable estimate. For thepurpose of calculating sample size, we assume this tobe the true common mortality for both PTCA andCABG patients in BARI. Under this assumption, asample size of 2,400 patients (1,200 assigned to eachtreatment) yields a probability of 88% that the upper95% confidence limit for the true difference betweenthe 5-year mortality rates of PTCA and CABG willnot exceed 2.5%. Note that if the upper confidencelimit of the true difference is 2.5%, then the observeddifference must be appreciably less than 2.5%. Forexample, if the observed 5-year mortality rates were5.9% for PTCA and 5% for CABG, the observeddifference would be 0.9% and the upper 95% confi-dence limit for the true difference would be 2.4%.

Appendix 4: Random AssignmentStratificationThe sequence of random assignment is stratified by

clinical center. This allows for approximately equalallocation of the two treatments within each center aswell as preservation of the overall balance betweenthe two treatment groups.Because of the large number of patients antici-

pated (1,200 for each treatment), the probability ofserious imbalance in the distribution of risk factors isunlikely. In case of such an occurrence, adjustmentcan be accomplished during the analysis. This can bedone without a significant loss of efficiency, and it ispreferable to the administratively more complex pro-cess of stratifying during random assignment. There-fore, the sequence of random assignment is stratifiedby clinical center only, and other clinically importantfactors will be adjusted for (if necessary) in analysesafter random assignment.

Design ofRandom AssignmentThe BARI clinical trial requires a scheme of

random assignment that will provide a balance in thetotal number of patients assigned to each treatmentgroup both early and late in the trial. Early balance isnecessary to maximize the power to detect differ-ences for safety-monitoring purposes and to adjustfor any early learning curve effects in the trial. Toachieve this balance, a blocked scheme of randomassignment has been implemented, with generationof a separate sequence for each clinical center asdescribed below.

Block lengths were selected at random. After thisnumber was chosen, a permutation of this number of


treatment assignments, with half being PTCA andhalf CABG, were selected at random from all possi-ble permutations. This permutation then becamepart of the assignment sequence. The process wasrepeated until the assignment sequence was of alength of 500 or more. (Each center is expected torecruit an average of 175 patients for a trial size of2,400 patients, but assignment schemes are longer toallow some clinics to recruit more patients in caseother clinics fall behind the projected recruitmentrates or have to stop recruitment.)

Treatment AssignmentsBy using the blocked random assignment scheme

indicated above, the CC has generated the treatmentassignment schemes for all centers. The file with therandom assignment sequence for each center is en-coded and stored on the same microcomputer that isused for data entry. Because a high level of security isessential for this file, it is encrypted on each micro-computer to ensure that there is no unauthorizedaccess to these blinded data. Security issues arediscussed further on p V-13 (Data Management).When a patient is to be randomly assigned, the

clinic coordinator uses the software developed forrandom assignment. The coordinator is promptedwith a list of questions that reflect BARI trial inclu-sion and exclusion criteria. Answers to these ques-tions must match the entry criteria (i.e., BARI eligi-bility must be verified) before the program willindicate that the patient is indeed eligible and pro-vide the patient's treatment assignment. The CCclosely monitors the use of the random assignmentprogram.A contract is maintained at all centers to guarantee

service within 4 hours of notification of computerbreakdown. The CC is available to assign treatments bytelephone in the event that computer repair is delayed.Copies of each random assignment sequence are storedat the CC in secured computer files.

Appendix 5: ComplicationsAbrupt Closure (Regardless of Previous Dilatation)

There is clinical evidence of one or more coronaryarteries abruptly closing.Arterial Embolus of Extremity or Permanent Lossof PulseAcute occlusion of a main or distal arterial trunk

supply in a limb assessed by the lack of detectabledistal arterial pulsations (by pulsation or Dopplerexamination) that had previously been observable.The loss of pulse may or may not be associated withischemia of the affected limb.

Cardiogenic ShockHypotension (systolic blood pressure of less than

80 mm Hg) that is associated with reduced urineoutput, decreased mental acuity, and compensatoryvasoconstriction.

Cardiac TamponadeHemodynamic compromise, with hypotension and

low cardiac output secondary to increased pericardialpressure.

Cerebrovascular Accident (CVA)A focal neurological deficit that is still at least

partially evident more than 24 hours after its onset.

Chest Tubes for More Than 5 Days After Surgery(Self-Explanatory)

ComaProfound depression in the level of conscious-

ness, reflected by the loss of contact with theenvironment and loss of spontaneous movement.Brain stem activity (respiration and response todeep pain) may or may not be preserved.

Congestive Heart Failure (CHF)Manifested by one or more of the following fea-

tures: dyspnea on exertion, edema, fatigue, orthop-nea, and/or paroxysmal nocturnal dyspnea. Otherfindings that support the clinical diagnosis includethe presence of an S3 gallop, elevated jugular venouspressure, and radiographic evidence of pulmonarycongestion. Verification by a physician's statement inthe medical record is required.

Deep Wound InfectionPositive culture from the mediastinal tissues be-

neath the sternum.

DementiaBroad-based loss in higher intellectual function

(including cognitive, perceptual, calculational, and/orrecall functions) that is evident to family membersand close associates or demonstrated on serial func-tional testing.

HemorrhageSevere bleeding sufficient to require transfusion of

packed red blood cells.

Hypersensitivity ReactionAn allergic reaction to iodine-containing radio-

graphic contrast media or protamine, which ismarked by the development of urticaria, wheezing,prolonged hypotension, or laryngospasm.

HypotensionReduction in systolic blood pressure to <90 mm Hg

or reduction by .30 mm Hg compared with the base-line value, which persists for more than 1 minute andrequires treatment.


Nonfatal Cardiac Arrest That Requires CPRor Countershock (Self-Explanatory)

Postthoracotomy SyndromeClinical syndrome of pleuro/pericarditis mani-

fested by chest pain and fever.

Pulmonary Edema (Cardiac)Acute congestive heart failure resulting in the

accumulation of pulmonary interstitial and alveolarfluid, manifested by profound dyspnea, orthopnea,rales, and evidence of pulmonary congestion on chestX-ray.

Pulmonary EmbolusAcute occlusion of one or more branches of the

pulmonary arteries with thrombotic material, usuallyoriginating from the iliofemoral or pelvic veins, mani-fested by the abrupt onset of pleuritic chest pain,worsened gas exchange, increased pulmonary arterypressure, or frank hemodynamic collapse.

Renal FailureProgressively deteriorating renal function requir-

ing dialysis.

Reoperation for BleedingReoperation performed to remedy excessive bleed-

ing after initial surgery.

Respiratory Failure That Includes NoncardiacPulmonary Edema and Adult Respiratory DistressSyndrome (ARDS)

Inability of the patient to maintain adequate gas

exchange during spontaneous ventilation, even withthe assistance of supplemental oxygen. In cases

where a patient is receiving mechanical ventilatoryassistance after surgery, respiratory failure shall beconsidered the inability to wean the patient frommechanical ventilation within 48 hours of completionof the surgical procedure.

In noncardiac pulmonary edema the same clinicaland radiographic appearances that are in cardiacpulmonary edema are present. However, the pulmo-nary capillary wedge pressure is 15 mm Hg.ARDS is the increased interstitial and alveolar lung

water resulting from lung injury from any of a variety ofcauses in the absence of a cardiac etiology.

Superficial Wound InfectionPositive culture from the surgical or PTCA wound

site.

Transient Ischemic Attack (TL4)A focal neurological defect (usually correspond-

ing to a single vascular territory) that spontaneouslyresolves so that no residual evidence exists within24 hours.

Wound DehiscenceThe splitting or bursting open of a procedural

wound.

Appendix 6: Angiography and Functions of theCentral Radiographic Laboratory

Purpose of the Central Radiographic LaboratoryThe CRL is responsible for processing all BARI

data that relate to angiograms. Through the develop-ment and implementation of the AAP, the CRLensures that angiographic data regarding coronaryanatomy and lesion morphology are collected in astandard format across all sites. Angiographic datafor all BARI patients are entered into the AAP ateach site. In addition, the CRL performs centralangiographic readings for all randomly assigned pa-tients. The central readings performed by the CRLwill be used in the primary analysis of BARI.

Angiographic Procedural RequirementsNitroglycerin is usually administered immediately

before coronary arteriography. Notations of cathetersize, nitroglycerin administration, and projectionsused are documented on a Procedure Report Form.Multiple standard views are obtained by using caudalRAO and cranial LAO projections of the left coro-nary artery and paired LAO/RAO projections of theright coronary artery when possible. Six-inch or 7-in.image-intensifier modes are used for standard views.Before PTCA two views are obtained of each lesion,and these views are replicated after PTCA. A leftventricular angiogram in which a 300 RAO view isused is required, and a cranially angulated steepLAO view is also obtained when possible.

Required AngiogramsAll angiograms that are performed on BARI patients

must be entered into the AAP at the clinical site.Angiograms are assigned to the following categories.

Baseline. The angiogram that is used to documentangiographic eligibility for BARI is considered thebaseline angiogram. This angiogram must be enteredinto the AAP within 5 working days of randomassignment.

Before and after initial PTCA. For patients that arerandomly assigned to PTCA, data must be enteredwithin 5 working days of completion of the procedure.

Intercurrent angiograms. All angiograms that areperformed during follow-up, whether they are part ofa subsequent revascularization procedure or not, areconsidered intercurrent angiograms. These must beentered within 5 working days of the angiogram forrandomly assigned patients.

Angiographic Data Entry, Management,and InterpretationThe acquisition of angiographic data is performed

by using the AAP, a microcomputer and softwaresystem that was developed specifically for use inBARI. The AAP provides a uniform format for the


entry and storage of coronary and left ventricularangiographic data at the clinical sites and providestools (calipers, tables, reports) for clinicians andangiographers at each institution. Electronic calipersare used to measure catheter diameter (as a calibra-tion), lesion diameter, and the diameter of normalreference segments of the vessel proximal and distalto the lesion. The microcomputer system facilitatesthe electronic transfer of data from the clinical sitesto the CRL. The same AAP system is used at theCRL for angiographic interpretation and data analy-sis. The microcomputer hardware was provided tothe original 14 clinical centers and to the CRL by agift from Hewlett-Packard Co., Palo Alto, Calif.

All baseline, PTCA, and intercurrent films forrandomly assigned patients are sent to the CRL bycommercial delivery service as soon as possible afterpatient discharge. Films are read at the CRL andreturned within 20 working days whenever possible.Site angiographic readings that entered into the AAPare transmitted to the CRL by way of telephoneconnections on a weekly basis.At the CRL evaluation of baseline and intercur-

rent angiograms is done by a senior angiographer,who reviews coronary vessel distribution, branching,and myocardial territory supplied. Lesions are as-sessed for location, severity (calipers), and morphol-ogy. LV angiography is evaluated by using quantita-tive methods for the measurement of ejectionfraction and regional wall motion. These evaluationsare compared with the initial site reading, and ifthere are no major differences the CRL reading isfinal. In the case of major differences, the film is readby a second CRL reader and adjudication is per-formed. Evaluation of PTCA films is performed by asenior angioplasty operator, who reviews the studiesfor lesion severity and morphology before and afterPTCA. The same adjudication procedures that wereoutlined above are performed during this evaluation.

Quality ControlAngiograms. The CRL reviews all randomly as-

signed studies for image quality and protocol compli-ance. Feedback is provided to the site, the angiogra-pher, and the Operations Committee.PTCA procedures. Films that document PTCA pro-

cedures are reviewed to assess PTCA outcome. Filmsare graded with respect to protocol compliance aswell as angiographic findings.CRL angiographic readings. The CC randomly des-

ignates angiographic studies to be reread at the CRL.This rereading is performed in a blind manner tocompare the initial CRL reading with the rereadingof the same angiogram to determine the amount ofvariability in the CRL interpretation of specific an-giographic variables.

Appendix 7: Electrocardiography and Functions ofthe Central Electrocardiographic Laboratory

Purpose of the Central ElectrocardiographicLaboratory (CEL)The CEL analyzes all rest and exercise ECGs that

are collected in BARI. Rest ECGs are evaluated for

evidence of QRS, ST, and T wave items by using thestandard Minnesota code and supplemental criteria.7Exercise ECGs are evaluated for ischemia based onthe BARI definitions for severe ischemia on nonin-vasive testing (entry criteria definition) and myocar-dial ischemic response (end point definition).

Resting ECG Acquisition by Using MarquetteMAC-12 Units

Through an educational grant from MarquetteElectronics (Milwaukee, Wis.), the CEL has pro-vided Marquette MAC-12 units to the original 14clinical centers that did not have Marquette equip-ment. Baseline, preprocedure, postprocedure, andscheduled follow-up ECGs are acquired on Mar-quette equipment. This standard recording appara-tus facilitates ECG transmission to the CEL and thesystem provides a repository for digitized ECGs,allowing easy retrieval for subsequent coding.

ECG CodingRest ECGs are evaluated by using the Minnesota

Code. To more accurately describe the patient popula-tion in BARI, additional data are collected as part of anextension of the Minnesota Code. The actual height ordepth of the ST segment shift is recorded, and codeshave been added to describe the depth and extent of Twave inversion in the anterior precordial leads. Finally,Q wave codes have been added to identify characteris-tics that, along with ST segment and T wave changes,are associated with a high probability of myocardialinfarction.ECGs are received and logged in at the CEL. When

multiple ECGs are sent to document the postproce-dure period or evaluate a suspected myocardial infarc-tion, the ECGs that have the best representation of Qwaves and ST segment changes are chosen for analysis.ECGs are first evaluated for technical quality, includingthe presence of missing leads, excess artifacts, excessbaseline wandering, switched leads, and invalid calibra-tion. ECGs are logged into an inventory system, and acoding clerk measures Q wave depth and width, R waveamplitude, ST segment displacement, and T waveamplitude by using an eight-power magnified loupecalibrated in increments of 0.1 mm. These measure-ments are verified by the coding supervisor and thenentered into a MicroVAX computer. Logical edits arethen performed to verify the codes, and any discrepan-cies are resolved. The codes are then reviewed by aphysician, who has the option to make changes to thedata. All digitized ECGs are transmitted to DalhousieUniversity, where electronic determination of the Min-nesota Code is performed. Any discrepancies betweenthe Dalhousie and CEL readings are adjudicated, withthe CEL making the final code determination. The finaldata are used to create summary files that are trans-mitted weekly to the CC.

Quality Control ofECG CodingCoding clerks and physician electrocardiographers

must pass a comprehensive coding test before the


actual coding of project tracings. Within the CEL,ECGs are periodically recirculated to detect codingdeficiencies. A staff member who is identified ashaving inadequate performance undergoes remedialtraining until performance improves. Kappa statisticsare calculated biannually for the physicians andlaboratory supervisors.

In addition, the CC is recirculating a blinded sampleof 200 baseline ECGs for Q wave interpretation. A25% sample of ECGs that indicate Q waves duringfollow-up and an additional 0.5% of ECGs without Qwaves are also being recirculated. This serves to docu-ment the reliability of the central diagnosis of thismajor BARI end point. Interreader variability of binarydata will be assessed by using the Kappa statistics.

Exercise Test AcquisitionExercise tests are performed by using a motor-

driven treadmill and following the Bruce protocol8with no or at least one half of a warmup stage ifnecessary. A physician is in attendance throughoutthe test, and the patient and ECG tracings arecarefully observed. All patients undergo a pretestevaluation, during which ongoing drug therapy andindications for stopping the exercise are ascertained.The Mason-Likar 12-lead torso ECG is the lead

set used. Four specific tracings are required: uprightimmediately before exercise; peak exercise; immedi-ate postexercise; recovery period (3-5 minutes afterexercise); or an ECG showing maximum postexercisechange.

Exercise Test DigitizationExercise ECGs are first examined for the presence of

excess artifacts, baseline wandering, and missing leadsor ECGs. Each test is reviewed by the laboratorysupervisor or associate to determine the presence ofexclusion codes such as left bundle branch block, leftventricular hypertrophy, etc. The maximum depth ofST segment depression, maximum height of ST eleva-tion in a non-Q wave lead, maximum number ofabnormal leads, and maximum height of ST elevationin a Q wave lead are calculated. A Q wave lead is onein which the Q wave width exceeds 0.03 seconds. This isdetermined from the resting ECG that undergoesMinnesota Coding. These exercise ECG data are trans-mitted weekly to the CC.

Quality Control of Exercise ECG DataA 2% sample of scheduled and unscheduled exer-

cise ECGs are submitted for reproducibility studies.The sample is enriched with 25% of all exercise teststhat indicate a myocardial ischemic response. Varia-bles to be analyzed for reproducibility include themaximum number of abnormalities in any leadgroup, the total number of abnormal leads, and themaximum depth of ST segment depression.

Serial ComparisonFor both rest and exercise ECGs, serial comparisons

are performed by using the most recent rest ECG or

exercise test and the ECG or test that preceded it. BothECGs or sets of ECGs are available to the laboratorysupervisor and physician at the time of overreading.The standard Minnesota and supplemental codes areused to determine significant change in Q, ST, and Twave items.

Appendix 8Form A

Consent for Participation in the Research StudyEntitled Bypass Angioplasty RevascularizationInvestigation (BARI), a Randomized Trial ComparingCoronary Bypass Surgery to Coronary AngioplastyStudy Investigators:Office of Human Research:

General Description and Purpose of ResearchYou are invited to participate in a trial sponsored

by the National Heart, Lung, and Blood Institutewhich involves comparison of two treatments forpatients with severe myocardial ischemia and coro-nary artery disease. You have coronary artery dis-ease involving at least two of the major coronaryarteries that supply blood to your heart muscle.Your doctors have determined that you are acandidate for either percutaneous transluminal cor-onary angioplasty (PTCA), a procedure in which aballoon catheter will be passed across the narrowingin your coronary artery to open the narrowing, orcoronary artery bypass graft (CABG) surgery, anoperation in which the narrowing in the coronaryartery will.be bypassed by using a vein from your legor an artery from your chest. One of the potentialcomplications of using the balloon catheter (PTCA)is that the narrowing may return at the site ofdilation. This renarrowing has been found in asmany as 30% of patients undergoing this procedureat various medical centers. Narrowing generallyoccurs within the first 6 months, and a secondballoon angioplasty procedure usually can be per-formed. When the narrowing does not recur withinthe initial 6 months, it is not likely to occur in thesubsequent 3-5 years.

I understand that the risks of CABG and PTCAare considered to be similar and that it is possible fordeath to occur in approximately 1-2% of patientsundergoing these procedures. The likelihood of sus-taining a heart attack (myocardial infarction) duringeither of these procedures is approximately 5%.During the study, 2,400 patients will be assigned bychance to either form of therapy during a period of 2years in 14 hospitals.

If I agree to participate in the randomized clinicalresearch study, one of these two revascularizationprocedures will be chosen randomly (by chance) asthe method to correct the narrowings of my coronaryarteries. If I am selected to receive CABG, thisprocedure will be carried out within 2 weeks after mycoronary angiogram was made, at a time that is


mutually agreeable to me and the physicians caringfor me.

If I am selected to receive PTCA, I agree toundergo a repeat heart catheterization and to haveballoon angioplasty performed on one or more of thenarrowings in my coronary arteries. The PTCA pro-cedure is successfully performed in approximately80% of patients. Should this procedure fail becauseof complications resulting from the procedure, Iunderstand that it may be necessary to proceedimmediately with CABG.

After the treatment, whether it is PTCA or CABG,I will be carefully followed for 5 years and will beseen, or contacted by telephone, throughout myhospital stay, at 4-12 weeks after hospital discharge,at 6 and 12 months after the procedure, and everyyear thereafter. An exercise test will be performed at4-14 weeks and at 1, 3, and 5 years after theprocedure to determine maximal exercise capacity;this test may provide diagnostic information concern-ing my state of health. I understand that certaincomplications may occur during or after an exercisetest, such as abnormal blood pressure response,dizziness, irregular heart beat, or, very rarely, a heartattack or death. However, the risk of a complicationduring exercise testing is infrequent, and the risk isfurther decreased by the presence of appropriatemedical facilities and an experienced medical team.At 5 years, I will be admitted to the hospital for a

repeat coronary angiogram. Coronary angiographydefines the presence, location, and extent of coronarydisease and associated heart muscle function. Thetest will assess the results of the PTCA or CABG.Complications from a coronary angiogram are infre-quent (less than 1%) and include a blood clot wherethe catheter is introduced and the possibility of astroke, heart attack, or death (less than 5%). The riskis no greater than that incurred during the initialcoronary angiography.

Risks and BenefitsI understand that there are possible risks to me if I

agree to participate in this study. The risks are those ofthe PTCA procedure, CABG, exercise tests, and coro-nary angiography as stated above. I understand that mydoctor has recommended that I have a revasculariza-tion procedure (PTCA or CABG) and that the testprocedures described above are routinely used in theassessment and follow-up of patients with coronaryartery disease. With participation in this study there isa risk that I may receive the less effective of thetreatments being studied. However, the potential ben-efits include the possibility that I may receive the moreeffective therapy. The best medical knowledge at thistime does not permit a scientific recommendation tome as to which option is better.

Voluntary Consent, Right to WithdrawIf I agree to participate in this study, I understand

that my care will in no way be compromised. Myparticipation in this study is voluntary. My refusal to

participate will involve no penalty or compromise inmy medical care or loss of benefits to which I amotherwise entitled. I may discontinue participation atany time without penalty or loss of the benefits towhich I am entitled.

Alternative TreatmentsPossible alternatives to my participation in this

study include medical therapy with standard ap-proved drug regimens, coronary angioplasty, or cor-onary bypass surgery. Each of these three options isstandard in the management of patients with coro-nary artery disease.

Investigational Sponsorship and Cost ConsiderationsThe physician investigators listed on this form;

research personnel associated with this study at; and the National Heart, Lung, and

Blood Institute (the sponsor of the study) mayinspect and copy my medical records relating to thisstudy. The results of the study will be reported to acoordinating center selected for data analysis. Con-fidentiality of my medical record will be maintainedby the use of a numerical or alphabetical code. Inthe event of any publications regarding this study,my identity will not be disclosed. I will be informedof any significant new findings developed during thecourse of the research which may relate to mywillingness to continue in the study. These findingsmay also be published in the medical literature.The choice of PTCA or CABG may affect the

number of days I will be hospitalized. In general,CABG requires a longer hospitalization than PTCAdoes. However, PTCA may require repeat admis-sions in the initial year after the procedure becauseof recurrent narrowings, which may require a secondangioplasty procedure. One of the purposes of thisstudy is to determine whether there are differences inthe total number of days of hospitalization, over thelong term, depending on the initial revascularizationprocedure chosen. My physician has recommendedthat I would best be treated by having PTCA orCABG. Therefore, the costs of the PTCA and of theCABG, the direct hospital expenses, and the directphysician fees that are considered clinically indicatedfor my care will not be paid for by the sponsor of thisstudy. The use of exercise studies with or without aradionuclide is considered routine clinical care in thefollow-up of patients with coronary artery diseaseand will not be paid for by the sponsor. The cost ofthe coronary angiogram performed 5 years afterCABG will be paid for by the sponsor of the study ifthis test is not required as part of usual care.

(Medical Center Title), in fulfilling its public re-sponsibility, accepts professional liability and respon-sibility for physical injury if it is caused by negligenceof the Center and its employees or agents. No personshall have any authority, orally or in writing, tochange the terms of the foregoing. Any questions thatI have concerning the research study or my partici-


pation in it, before or after my consent, will beanswered by

In the event I believe that I have suffered anyinjury as a result of participation in the researchproject, I am to contact , who will be ableto refer me to an individual who will review thematter with me, identify other resources that may beavailable to me, and provide information concerningadditional inquiries. If I am a woman, I am notpregnant, to the best of my knowledge. I am notparticipating in any other medical research study. Ihave read the above statement and have been able toexpress concerns which have been satisfactorily re-sponded to by the investigator.

I believe I understand the purpose of this study, aswell as the potential benefits and risks that areinvolved. I hereby give my informed and free consentto be a participant in this study.

Patient's Signature Date

DateWitness's Signature

I certify that I have explained to the above indi-vidual the nature, purpose, potential benefits, andpossible risks associated with participating in thisresearch study. I have answered any questions thathave been raised and have witnessed the abovesignature. These elements of informed consent con-form to the assurance given by Depart-ment of Health and Human Services, to protect therights of human subjects. I provided the patient acopy of this signed consent document.

Investigator's Signature Date

Appendix 8Form B

Patient Consent to Participate in the BypassAngioplasty Revascularization Investigation Registry

Principal Investigator

to participate in this Registry. If you decide toparticipate, you will be interviewed each year for thenext 5 years or more by mail or telephone. Shouldyou move from your present address, a private agencymay be used to determine your new home or place ofwork. These interviews will take approximately one-half hour of your time. The questions asked will beabout your symptoms of angina or other heart prob-lems. You will also be asked whether or not you haveundergone any hospitalizations because of your heartproblems or whether you have undergone a repeatPTCA or CABG or have been hospitalized for treat-ment of a heart attack. The questions are not con-sidered to have psychological stress. No standardtreatment will be required by your treating physiciansnor will any standard treatment be withheld fromyou. You will receive no payment or other compen-sation for participating in this Registry. There may beno immediate benefit to you, but the combinedinformation from this Registry may help your physi-cian and other physicians in the future to betterunderstand the effects of the treatment you havereceived on ischemic heart disease and its course.

I understand that, upon my request, a BARIresearch investigator will answer questions about thestudy. I may refuse to participate or discontinueparticipating in the study at any time without penaltyor loss of benefits available to me as a patient at thismedical center.

I understand that no commitment is made to pro-vide complimentary medical care or compensation forany adverse results of my participation in this study.Further information concerning institutional policiesin this regard or information about the conduct of thisstudy or the rights of research subjects may be ob-tained from

Confidentiality of information concerning partici-pants will be maintained. Names of participants ormaterial identifying participants will not be releasedwithout written permission, except as such release isrequired by law. Medical records related to this studymay be made available to the Food and Drug Ad-ministration, as provided for in federal regulations.

Date Signature of Participant

Approved by Institutional Review Board

Date_Date Signature of BARI Investigator

Obtaining Consent

Some patients at the Medical Centerundergoing percutaneous transluminal coronary an-gioplasty (PTCA) or coronary artery bypass graft(CABG) surgery are being enrolled in a Registry ofscientific data as part of a national collaborativeclinical study comparing the outcome of these proce-dures. The purpose of this Registry is to providescientific information about the merits of PTCAcompared with CABG, another procedure for restor-ing the flow of blood to heart muscle. You are invited

Appendix 9BARI

List of ParticipantsClinical Centers

University of AlabamaPrincipal Investigator:PTCA Operators:

Surgeons:

William J. Rogers, MDWilliam A. Baxley, MDLarry S. Dean, MDGary S. Roubin, MDJames K. Kirklin, MD


Associate Investigators:

Coordinators:

Data Manager:Staff:

Former Participants:

John W. Kirklin, MDAlbert Pacifico, MDGeorge L. Zorn, MDEdgar Charles, PhDThomas D. Paine, MDLarry E. Maske, RNTerri E. Morgan, RNLeah C. Carr (SEQOL)John A. Trobaugh, RAKaren W. AndersonFredericka HarrisThomas Bulle, MDJ. Bradley Cavender, MDPaul J. Garrahy, MD

Brown University: Rhode Island HospitalPrincipal Investigator:PTCA Operators:

Surgeon:Associate Investigators:

Coordinators:


David 0. WilliamsThomas M. Drew, MDDavid 0. WilliamsArun K. Singh, MDGeorge N. Cooper, MDBarry L. Sharaf, MDMark Macedo, RNJanice L. Wheeler, RNJohn Moran, MDEdward S. Thomas, MDHarvey White, MD

Bellevue Hospital (Satellite to Brown)Principal Investigator:PTCA Operators:

Surgeons:

Associate Investigator:Coordinator:

Boston UniversityPrincipal Investigator:PTCA Operators:

Surgeons:


Coordinators:


Cleveland Clinic FoundationPrincipal Investigator:PTCA Operators:

Surgeons:


Coordinator:Data Manager:Staff:

Frederick Feit, MDMichael J. Attubato, MDFrederick Feit, MDStephen B. Colvin, MDAubrey C. Galloway, MDPeter F. Pasternack, MDSonja Shapiro, RN

David P. Faxon, MDJohn E. Brush, MDDavid P. Faxon, MDGary R. Garber, MDAlice K. Jacobs, MDNicholas A. Ruocco, MDJames D. Fonger, MDRichard S. Shemin, MDMichael A. Bettmann, MDJesse W. Currier, MDJames A. Rothendler, MDThomas J. Ryan, MDDonald A. Weiner, MDBeth R. Hankin, RNMary E. Mazur, RNRoger M. Mills, MDGaetano Paone, MD

Patrick L. Whitlow, MDIrving Franco, MDPatrick L. Whitlow, MDDelos Cosgrove, MDFloyd Loop, MDBruce Lytle, MDRobert Stewart, MDPaul C. Taylor, MDAlexios P. Dimas, MDWilliam Proudfit, MDBenjamin Robalino, MDEric Topol, MDKevin Vaska, MDGeorge Williams, PhDAmy Rogers, RNSharon SenickKathy Comella, RNMarsha Lowrie, RNJoyce Tedrick, RN


Duke UniversityPrincipal Investigator:PTCA Operators:

Surgeons:


Coordinators:

Data Manager:Staff:Former Participants:

John Frierson, MDJay Hollman, MD

Robert M. Califf, MDRobert P. Bauman, MDVictor S. Behar, MDYihong Kong, MDMitchell W. Krucoff, MDKenneth G. Morris, MDRobert H. Peter, MDHarry R. Phillips, MDRichard Stack, MDJames E. Tcheng, MDRobert H. Jones, MDH. Newland Oldham, MDPeter Van Trigt, MDThomas Bashore, MDDonald F. Fortin, MDDavid J. Frid, MDKerry Lee, PhDMichael J. Miller, MDE. Magnus Ohman, MDDavid B. Pryor, MDAlan N. Tenaglia, MDMary Ann Sellers, RNEllen T. Hampton, RNTerri DanielsLeonard SantoroSandra G. Burks, RNStephanie Caminiti, RNPeter J. Quigley, MDJ. Scott Rankin, MDJoan Richard, RN

Harvard University: Beth Israel HospitalPrincipal Investigator: Donald S. Baim, MDCoprincipal Investigator: Robert Safian, MDPTCA Operators: Julian Aroesty, MD

Donald S. Baim, MDDaniel Diver, MDBeverly Lorell, MD

Surgeons: Robert Johnson, MDRobert Thurer, MDRonald Weintraub, MD

Coordinator: Mary CunnionFormer Participants: Raymond McKay, MD

Tia DeFeo-Fraulini, MSCarolyn McCabe

Maine Medical Center (Satellite to Harvard)Principal Investigator:PTCA Operators:

Surgeons:

Associate Investigator:Coordinators:

University of MassachusettsPrincipal Investigator:PTCA Operator:Surgeons:

Mirle A. Kellett Jr., MDWarren D. Alpern, MDRichard A. Anderson, MDD. Joshua Cutler, MDMirle A. Kellett Jr., MDPaul W. Sweeney, MDDesmond J. Donegan, MDSaul Katz, MDRobert S. Kramer, MDChris A. Lutes, MDJeremy R. Morton, MDEdward R. Nowicki, MDJoan F. Tryzelaar, MDRichard L. White, MDCostas T. Lambrew, MDPamela Birmingham, RNJane Conner Kane, RNNancy Tooker, RN

Bonnie H. Weiner, MDBonnie H. Weiner, MDDani Bitran, MDJohn Moran, MDOkike N. Okike, MDThomas Pezzella, MD


Coordinators:

Data Manager:Former Participants:

Thomas J. VanderSalm, MDMarie Borbone, RNPaul Wanta, RNTheresa WisnewskiJoseph Benotti, MDJames Dalen, MDJohn Gaca, MDJeffrey Leppo, MDMichael Pasque, MDMarilyn Shay, RN


Coordinator:Data Manager:Staff:

Yves Leclerc, MDConrad Pelletier, MDAndr6 Arseneault, MDGilles Hudon, MDJacques Lesp6rance, MDDavid D. Waters, MDJohanne Trudel, RNClaudette FailleHuguette FlageolLucette Whittom, RN

Mayo ClinicPrincipal Investigator:PTCA Operators:

Surgeons:


Coordinator:Data Managers:


Medical College of VirginiaPrincipal Investigator:PTCA Operators:

Surgeons:

Associate Investigator:Coordinators:


University of MichiganPrincipal Investigator:PTCA Operators:

Surgeons:

Coordinator:Staff:


Montreal Heart InstitutePrincipal Investigator:PTCA Operators:

Surgeons:

Toronto Hospital (Satellite to Montreal)Michael Mock, MDJohn Bresnahan, MDDavid Holmes, MDGuy S. Reeder, MDCharles Mullany, MDThomas A. Orszulak, MDHartzell Schaff, MDPeter B. Berger, MDBernard Gersh, MDRaymond Gibbons, MDStephen L. Kopecky, MDFred Nobrega, MDRobert S. Schwartz, MDHugh C. Smith, MDSylvia Matheson, RNLisa KellyMary PetersonLou Ann Pierre, RNDennis Bresnahan, MDRonald Vlietstra, MD

Michael J. Cowley, MDMichael J. Cowley, MDGermano DiSciascio, MDGeorge Vetrovec, MDAlbert Guerraty, MDDavid Salter, MDAndrew Wechsler, MDChauncey W. Crandall, MDDavid Debottis, RNAnn Maziarz, RNKim Kelley, RNR. R. Lower, MDAmar Nath, MDBonnie Sechrist, RNS. Szentpetery, MD

Bertram Pitt, MDEric Bates, MDStephen Ellis, MDLinda Lee, MDEric J. Topol, MDJoseph Walton, MDSteven Bolling, MDMichael Deeb, MDMarvin Kirsh, MDKaren Burek, RNLinda Belzowski, RNMarkus Schwaiger, MDHui-lee ShuDiane Scarpace, RNMack Stirling, MDPeter ThomasmaEric J. Topol, MDSteve Werns, MD

Martial G. Bourassa, MDRaoul Bonan, MDGilles Cdt6, MDJacques Cr6peau, MDPierre DeGuise, MDYves Castonguay, MD

Principal Investigator:PTCA Operators:

Surgeons:


Coordinator:Staff:

New York Medical CollegePrincipal Investigator:PTCA Operator:Surgeons:

Coordinator:Data Manager:Former Participants:

St. Louis UniversityPrincipal Investigator:PTCA Operator:Surgeons:

Associate Investigator:Coordinator:Staff:


Jewish Hospital (Satellite toPrincipal Investigator:Coprincipal Investigator:PTCA Operators:

Surgeons:


Coordinators:

Staff:

Leonard Schwartz, MDHarold Aldridge, MDLeonard Schwartz, MDDavid Uden, MDTirone David, MDChris Feindel, MDBernard Goldman, MDIrving Lipton, MDLynda Mickleborough, MDRichard Weisel, MDDavid Almond, MDMichael McLoughlin, MDLeon Zelovitsky, MDLinda Ganassin, RNKaren Mackie, RN

Michael V. Herman, MDMelvin B. Weiss, MDRichard Moggio, MDRichard Pooley, MDGeorge Reed, MDMohan Sarabu, MDDoris Efstathakis, RN, BSNYonina Sait, PAPeter Praeger, MDKathleen Ryman, MDEric Somberg, MDJonathan H. Stein, MD

Bernard R. Chaitman, MDMorton J. Kern, MDGeorge Kaiser, MDVallee Willman, MDRobert Wiens, MDKatherine Galan, RNJane Fehl, LPNBarbara PooleHendrick Barner, MDUbeydullah Deligonul, MDMichel G. Vandormael, MD

St. Louis)Ronald J. Krone, MDNicholas Kouchoukos, MDRonald J. Krone, MDAli Salimi, MDNicholas Kouchoukos, MDThomas H. Wareing, MDPatricia Cole, MDRobert Kleiger, MDSandor Kovacs, MDMichael Rich, MDAnil Shah, MDMary Caruso, RNJane Humphrey, RNLisa Spinner, RNCynthia BensonGail EisenkramerThelma JonesEthel KellyJean MooreRose UmsteadJuanita Weaver


BARI/Parallel StudyInstitute of Clinical and Experimental MedicinePrague, CzechoslovakiaPrincipal Investigator:PTCA Operators:

Surgeons:


Coordinator:Data Manager:

Georgetown UniversityPrincipal Investigator:PTCA Operator:Surgeons:


Coordinator:

Vladimir Stanek, MDAlfred Belin, MDJosef Kovac, MDPavel Firt, MDJan Pirk, MDVladimir Kocandrle, MDMichael Zelhzko, MDRuzena Jandova, MDErhard Tchernoster, Ing.

Former Site

Kenneth Kent, MDKenneth Kent, MDNevin M. Katz, MDRobert Wallace, MDLarry Elliott, MDCurtis Green, MDJames Lavelle, MDCharles Rackely, MDBeverly Shriver, RN

Central Electrocardiographic Laboratory

St. Louis University Medical CenterPrincipal Investigator: Bernard R. Chaitman, MDAssociate Investigators: Robert D. Wiens, MD

Preben Bjerregaard, MDLeonardo Maitas, MDBonpei Takase, MDLiwa T. Younis, MD, PhD

Coordinators: Leslee Shaw, MAKaren Stocke

Programmers: Li-Yung Lui, MSKim Russell

Staff: Steve CannonTeresa CuttsDebbie KarglWillie MathisMark Muehl, MABarbara PooleLaura SchaeferArt Stelken, MAJesse Williams

Former Participants: Brian BilgereMary Ellen HauesenJeanine Sabatino

Coordinating Center

University of Pittsburgh, Pittsburgh, PennsylvaniaPrincipal Investigators:

Coordinator:Assistant Coordinator:Statisticians:

Data Coordinator:Data Managers:

System Programmers:

System Support:

Consultants:


PiCardia

Division of HeNational Heart

NationalBeti

Program Administrator:Coinvestigators:


rogrnac DiDart a

t, Lu]Ins

hesda

Katherine M. Detre, MD,DrPH

Sheryl F. Kelsey, PhDKim Sutton-Tyrrell, RN, DrPHBarbara L. NaydeckHuiman X. Barnhart, MAJoyce F. Killinger, MSJoseph E. MelvinAllan D. Rosen, MSAnn Rodewald Steenkiste, MSSharon Weber CrowRegina M. HardisonMeg Cooper, MEdGail Harger, MSISDave W. BurryJeffrey P. MartinEmil A. MaurerWilliam P. Amoroso, MPHLarry Kamons, MSISTimothy E. Kuntz, MSISNancy H. Remaley, MSISJoel Greenhouse, PhDRobert L. Hardesty, MDBarry F. Uretsky, MDLynette M. AndersonDonald Borrebach, MSMary Ann Carr, MSRichard Holubkov, MSDiane F. HurshJennifer A. MetzlerAngela SpadaroJohn Wilson, PhD

am Officeiseases Branchand Vascular Diseasesng, and Blood Institute;titutes of Healtha, MarylandGeorge Sopko, MD, MPHMichael Horan, MDRachel Solomon, M.H.S.David Zucker, PhDMichael J. Domanski, MDLouis Offen, MDThomas Robertson, MDJoel Verter, PhD

Central Radiographic Laboratory

Stanford University Medical CenterPrincipal Investigators:


Coordinator:Staff:

Former Participant:

Edwin L. Alderman, MDMichael Stadius, MDByron Brown, PhDWilliam Sanders, MSEELewis Wexler, MDBrooke Hollak, RNTerri BeamGao Shao-Zhou, MDRobert Moore, MD

kncillary Study

SEQOL: Study of Economics and Quality of LifeStanford University School of MedicinePrincipal Investigator: Mark Hlatky, M.D.Coordinator: Kathryn Cavanaugh, MPHFormer Participant: N. Clapp-Channing, MPH, RN

Safety and Data Monitoring BoardChair: J. David Bristow, MD, Oregon

Health Sciences UniversityMembers: James Childress, PhD,

University of VirginiaTimothy Gardner, MD, JohnsHopkins Hospital

Herman Gold, MD,Massachusetts GeneralHospital

J. Ward Kennedy, MD,University of Washington

Genell Knatterud, PhD,Maryland Medical ResearchInstitute

Robert Roberts, MD, BaylorCollege of Medicine

John Waldhausen, MD,Hershey Medical Center

Carl White, MD, University ofMinnesota Hospital


Morbidity and Mortality Classification CommitteeChair: Ronald Prineas, MD, PhD,

University of MiamiMembers: Charles Fisch, MD, Krannert

In,titute of CardiologyLeoni Greene, MD,Harborview Hospital

Robert B. Karp, MD,University of ChicagoMedical Center

Spencer B. King III, MD,Emory University Hospital

Jay Mason, MD, University ofUtah Medical Center

Jack L. Titus, MD, PhD,United Hospital

Office of Study ChairMayo Clinic FoundationRochester, Minnesota

Robert Frye, MDProfessor of MedicineChairman of Internal Medicine

References1. Feinlieb M, Havlik RJ, Zillim RF, Pokras R, McCarthy E,

Moien M: Coronary heart disease and related procedures:

National Hospital Discharge Survey Data. Circulation 1989;79(suppl I):I-13-1-18

2. Harrison DC: Cost containment in mediums: Why cardiology?Am J Cardiol 1985;56:1OC-15C

3. The Expert Panel: Report of the National Cholesterol Educa-tion Program Expert Panel on Detection, Evaluation, andTreatment of High Blood Cholesterol in Adults. Arch InternMed 1988;148:36-69

4. Peduzzi P, Detre KD, Whittes J, Holford T: The intent-to-treatanalysis and the problem of crossovers: An example from theVA Coronary Bypass Surgery Study. J Thorac Cardiovasc Surg1991;101:481-487

5. Makuch R, Simon R: Sample size requirements for evaluatinga conservative therapy. Cancer Treat Reports 1978;62:1037-1040

6. Grizzle J-E: A note on stratifying versus complete randomassignment in clinical trials. Controlled Clin Trials 1982;3:365-368

7. Prineas RJ, Crow RS, Blackburn H: The Minnesota Codemanual of electrocardiographic findings. Boston, John WrightPGS Inc, 1982

8. Bruce ZA: Exercise testing of patients with coronary arterydisease. Ann Clin Res 1971;3:323-332

KEY WORDS: * PTCA * multivessel CAD * CABGrevascularization

Documents

The BARI Protocol BypassAngioplasty … TheBARIProtocol Protocol for the BypassAngioplasty Revascularization Investigation In this supplement we present the design of the Bypass Angioplasty