Predictors of Outcome in a Multicenter Port-AccessValve RegistryDonald D. Glower, MD, Lawrence C. Siegel, MD, Karen J. Frischmeyer, DVM,Aubrey C. Galloway, MD, Greg H. Ribakove, MD, Eugene A. Grossi, MD,Newell B. Robinson, MD, William H. Ryan, MD, and Stephen B. Colvin, MDDuke University Medical Center, Durham, North Carolina, Heartport, Inc, Redwood City, California, New York UniversityMedical Center, New York, New York, St. Francis Hospital, Roslyn, New York, and Presbyterian Hospital, Dallas, Texas

Background. The aim of this study was to examine thepredictors of outcome in patients undergoing isolatedvalve operation using port-access techniques.

Methods. Logistic regression analysis was performedin a prospective, multi-institutional registry of patientsundergoing isolated aortic valve replacement (AVR, n 5252), mitral repair (MVP, n 5 491), or mitral replacement(MVR, n 5 568) using port-access techniques from 1997 to1999.

Results. Endoaortic balloon occlusion was used in 2%(AVR), 93% (MVP), and 90% (MVR) of cases. Conversionto full sternotomy occurred in 3.8% of all cases. For allpatients, early mortality was 50 of 1,311 (3.8%) and onsetof new atrial fibrillation occurred in 140 of 1,311 (11%)

patients. The determinants of 30-day mortality were redo,age, and MVR or AVR. The determinants of reoperationfor bleeding were age, reoperation, and MVR. Age was apredictor for stroke, and age and low or medium volumecenter were predictors of new atrial fibrillation.

Conclusions. Excellent short-term results can be ob-tained using port-access techniques in isolated mitral oraortic valve operations. Patient outcome is not related toinstitutional case volume, and the primary determinantsof outcome after port-access valve procedures are gener-ally patient-related factors.

(Ann Thorac Surg 2000;70:1054–9)© 2000 by The Society of Thoracic Surgeons

Port-access was introduced as a technique to performvalvular operations using cardiopulmonary bypass,

and combinations of a small right thoracotomy and/orendovascular aortic occlusion [1]. Since the introductionof this technique, several series of mitral valve [2–11] oraortic valve [12, 13] operations have been reported, gen-erally representing the initial experiences of single insti-tutions. Although the results of port-access for mitral oraortic valve operation have been good, controversy per-sists as to whether the results of port-access differ fromthose of median sternotomy [5, 13, 14] and whetherfactors such as institutional volume or learning curvesinfluence the results of port-access procedures [15]. Fur-thermore, few results from aortic valve replacementusing port-access have been reported [12, 13]. To addressthese issues, a multi-institutional registry of port-accessprocedures (the Port-Access International Registry[PAIR]) was initiated in 1997 and terminated in 1999. Thisreport contains the results of this PAIR registry forisolated aortic or mitral valve operation.

Material and Methods

The PAIR database was used to conduct a detailedanalysis of patient demographics, comorbid conditions,

valvular disease, operative techniques, and clinical out-comes for consecutive patients undergoing isolated aorticor mitral valve surgery. The PAIR is a prospective obser-vational, consecutive cohort registry that was initiated inJune 1997, managed by an independent contract researchorganization, and sponsored by Heartport, Inc (RedwoodCity, CA) [8]. All institutions trained in port-access min-imally invasive cardiac surgery were eligible to partici-pate. In all, 104 institutions elected to participate at oneof three levels: (1) a brief case report form (CRF) within-hospital follow-up; (2) brief CRF with 30-day follow-up; or (3) comprehensive CRF with 30-day follow-up (seeAppendix). Patient exclusion criteria included agegreater than 85 years; need for emergent operation;illness limiting life expectancy to less than 2 years; andany significant femoral, iliac, or aortic arterial disease

The brief CRF collected core patient data (includinggender, height, weight, age, hospitalization and proce-dure dates, operative and reoperative status, and proce-dure) and perioperative complications in the followingcategories: reoperation, pulmonary, vascular, neurologic,infection, cardiac, multisystem failure, renal failure, othermajor morbidity, and death. In addition to the fields onthe brief form, the comprehensive CRF collected data onpatients’ history and comorbid conditions, preoperative

Presented at the Sixth Annual Cardiothoracic Techniques and Technol-ogies Meeting 2000, Fort Lauderdale, FL, Jan 27–29, 2000.

Address reprint requests to Dr Glower, Duke University Medical Center,Box 3851, Durham, NC 27710; e-mail: glowe001@mc.duke.edu.

Doctor Lawrence C. Siegel and Karen J. Frischmeyerare employees of Heartport Inc, Redwood City, CA

© 2000 by The Society of Thoracic Surgeons 0003-4975/00/$20.00Published by Elsevier Science Inc PII S0003-4975(00)01748-3

status, operative technique, transfusions, and postoper-ative care. Perioperative complication fields were identi-cal on all forms. The data incorporated a coding system toensure that the patients and surgeons were not identifiedand institutional identities were maintained as confiden-tial. Missing or inconsistent data were queried. Onlypatients with complete forms were included in an anal-ysis data set. In all, 3,586 forms were submitted and 3,436forms were included in the analysis data set. The analysisdata set contained 1,311 isolated mitral or aortic valvesurgery patients. A comprehensive report form wasavailable for 738 patients, and the brief report form wasavailable for remaining patients. Unless otherwise stated,all data refer to all 1,311 patients.

Patient demographic and operative data were summa-rized as mean plus or minus standard deviation, median,and 25th to 75th percentiles, or prevalence, as appropri-ate. Perioperative morbidity and mortality incidence and95% confidence interval [16] were analyzed on an inten-tion-to-treat basis. Independent predictors of mortality(in-hospital or 30-day), stroke, reoperation for bleeding,and new-onset atrial fibrillation were determined usingmultivariable logistic regression analysis. The effect ofinstitutional case volume was examined by dividing thereporting institutions into three groups based on the totalvolume of port-access cases reported in this series. Thus,three institutions had high volume ($ 75 cases), 11institutions had medium volume (20 to 74 cases), and 90institutions had low volume (1 to 19 cases). In addition toinstitutional case volume, other categorical independentvariables that were examined were age (# 65 years, . 65years), previous cardiac operation (yes, no), heart failure(New York Heart Association classes I to III, New YorkHeart Association class IV), and procedure (aortic valvereplacement, mitral valve repair, mitral valve replacement).

Results

Port-access cardiac surgery techniques have been re-ported extensively [1–11] and will not be described here.A total of 1,311 isolated port-access aortic valve (AVR)surgery, mitral valve repair (MVP), and mitral valvereplacement (MVR) cases reported between July 1997and August 1999 were analyzed. This report includescases from 104 institutions [8], with a distribution ofaortic or mitral cases contributed per institution (Fig 1).Of these 1,311 cases, 252 (19%) were AVR, 491 (37%) were

MVP, and 568 (43%) were MVR. Follow-up was at hospi-tal discharge in 23% of cases and at 30 days in 77%.

Demographic characteristics of patients who underwent port-access valve surgery are reported in Table 1.There was a high proportion of reoperation patients formitral valve replacement and significant preoperativecomorbidity in all three patient groups (Table 2). Of theMVR cases reported on the comprehensive form, 32%were reoperation patients. Aortic valve disease in AVRpatients was characterized as stenotic in 74% of patientsand as insufficient (rating of 3 or 4 out of 4) in 42% ofpatients. The most frequently indicated aortic valve dis-ease etiologies were calcific (46%) and myxomatous de-generative (26%). Mitral valve disease in MVP patientswas predominantly insufficiency (rating 3 or 4, 75%)whereas mitral valve disease in MVR patients was ste-notic (44%) or insufficient (rating 3 or 4, 76%). The mostfrequently reported valve disease etiologies for MVPwere myxomatous degeneration (58%) and mitral valveprolapse (22%), whereas those for mitral valve diseaseetiologies in MVR patients were rheumatic (41%) andmyxomatous degeneration (24%).

Aortic valve procedures were performed predomi-nantly by means of a partial sternotomy (40%) [17] or

Fig 1. Distribution of cases contributed by 104 institutions. Volumerepresents aortic and mitral cases combined.

Table 1. Demographic Characteristics of Patients WhoUnderwent Port-Access Isolated Mitral or Aortic Valve Surgery

AVR(n 5 252)

MVP(n 5 491)

MVR(n 5 568)

Age (y) 65 6 15 57 6 14 60 6 14Gender (% male) 63 66 39Weight (kg) 79 6 18 76 6 15 72 6 16BSA (m2) 1.9 6 0.2 1.9 6 0.2 1.8 6 0.2Elective (%) 98 99 98Reoperation (%) 12 8 23

Data are given as mean 6 SD.

AVR 5 aortic valve repair or replacement; BSA 5 body surfacearea; MVP 5 mitral valve repair; MVR 5 mitral valve replace-ment.

Table 2. Preoperative Status of Patients Who UnderwentPort-Access Isolated Aortic or Mitral Valve Surgery

AVR(n 5 199)

MVP(n 5 307)

MVR(n 5 232)

Diabetes (%) 11.1 5.9 9.1Renal failure (%) 3.0 2.3 4.3Hypertension (%) 54 39 38Pulmonary hypertension (%) 11 25 39COPD (%) 8.0 6.5 15PVD (%) 4.5 1.3 3.9Myocardial infarction (%) 8.0 8.2 6.0Atrial fibrillation (%) 8.5 19 46LVEF , 30% (%) 1.4 3.7 2.7NYHA class III–IV (%) 57 53 75

COPD 5 chronic obstructive pulmonary disease; LVEF 5 left ven-tricular ejection fraction; NYHA 5 New York Heart Association;PVD 5 peripheral vascular disease; other abbreviations as in Table 1.

1055Ann Thorac Surg MINIMALLY INVASIVE GLOWER ET AL2000;70:1054–9 PORT-ACCESS VALVE REGISTRY: PREDICTORS OF OUTCOME

right anterior thoracotomy (47%) [12, 13, 18, 19]. Mitralvalve operations were performed through a right anterior(87%) or lateral (10%) thoracotomy [3, 5–7]. Medianlength of chest incisions for AVR, MVP, and MVR pro-cedures were 8 cm, 6 cm, and 6 cm, respectively. Median(25th to 75th percentile) skin-to-skin operative times forAVR, MVP, and MVR were 3.6 (range, 2.9 to 4.3), 4.7 (4.0to 5.5), and 5.0 (4.0 to 5.8) hours, respectively. Mediancardiopulmonary bypass times (25th to 75th percentile)for AVR, MVP, and MVR were 103 (range, 80–123), 127(110–158), and 137 (108–180) minutes, and median aorticocclusion times were 73 (range, 61–90), 92 (75–115), and 89(70–116) minutes, respectively. Endoaortic balloon occlu-sion was used in 2% of AVR, 93% of MVP, and 90% ofMVR cases. In all, 86% of mitral valve repair procedureswere performed with an annuloplasty ring (67% Baxter,9% Medtronic, 8% St Jude). Of the mitral valve replace-ments, 19% were completed with a bioprosthetic valve(54% Baxter, 38% Medtronic) and 77% with a mechanicalvalve (90% St Jude, 4% ATS).

A total of 1,261 cases (96%) were successfully com-pleted as port-access, which was defined as use of theEndoCPB or EndoDirect System (Heartport, Inc, Red-

wood City, CA) without conversion to a total mediansternotomy. Multiple reasons for conversion to mediansternotomy were captured on the comprehensive formfor 17 patients and included patient anatomy (4 patients),vascular injury (6), poor visualization/exposure (4), in-ability to place system catheter (1), inadequate cardiopul-monary bypass (1), calcific aorta on palpation (1), andother (4 patients).

Intraoperative or postoperative transfusion therapyincluding homologous red blood cells (RBC), platelets,fresh frozen plasma, or cryoprecipitate, was administeredto 57% of AVR, 58% of MVP, and 66% of MVR patients. Atotal of 109 AVR, 172 MVP, and 150 MVR patientsreceived RBC and the median (25th to 75th percentile)number of units administered were 3 (range, 2 to 5), 2 (2to 5), and 3.5 (2 to 6) units, respectively. Median (25th to75th percentile) postoperative ventilation times for AVR,MVP, and MVR were 6 (range, 4 to 12), 8 (5 to 13), and 7(5 to 15) hours, respectively. Intensive care unit stays forAVR, MVP, and MVR were 21 (range, 18 to 36), 22 (17 to28), and 23 (19 to 48) hours, respectively. In all, 79% ofAVR patients, 81% of MVP patients, and 67% of MVRpatients were walking within 48 hours of surgery. Pa-tients walked within 24 hours of surgery in the case of48% of AVR, 53% of MVP, and 37% of MVR procedures.

In the United States, the median (25th to 75th percen-tile) postoperative length of stay for AVR, MVP, andMVR were 6 (range, 5 to 10), 5 (4 to 7), and 7 (5 to 11) days,respectively. Most patients were discharged home with-out services (82% AVR, 83% MVP, 74% MVR). Rates ofreadmission within 30 days were low in all groups at 6.0%for AVR patients, 8.4% for MVP patients, and 7.5% forMVR patients. Major perioperative morbidity and mor-tality rates are shown in Table 3. Operative mortality was11/252 (4.4%) for AVR, 8/491 (1.6%) for MVP, and 31/568(5.5%) for MVR. Causes of death were generally unre-lated to the port-access approach (Table 4).

Multivariable analysis showed that the major predic-tors of death were reoperation, older age, MVR, and AVR(Table 5). Older age was the only predictor of stroke.Older age, MVR, and reoperation were all predictors forreoperation for bleeding (Table 5). The predictors for new

Table 3. Perioperative Morbidity for Patients WhoUnderwent Port-Access Isolated Aortic or Mitral Valve Surgery

AVR(n 5 252)

MVP(n 5 491)

MVR(n 5 568)

Stroke, unresolved (%) 2.4 2.6 2.8Myocardial infarction (%) 0 0.4 0.2Failed valve repair (%) — 2.0 —Paravalvular leak (%) 0 0.6 0.5Reop for first procedure (%) 1.2 2.0 0.9Reop for bleeding (%) 4.4 2.6 6.2Renal failure (%) 2.8 2.1 3.3Multisystem failure (%) 1.2 0.6 2.6Deep vein thrombosis (%) 0 0 0.5Aortic dissection (%) 0 0.2 0.4New-onset AF (%) 12.7 10.4 10.0Pleural effusion (%) 4.0 4.5 4.0

AF 5 atrial fibrillation; Reop 5 reoperation; other abbreviationsas in Table 1.

Table 4. Causes of Death in Patients Who Underwent Port-Access Isolated Aortic or Mitral Valve Surgery

AVR(n 5 252)

MVP(n 5 491)

MVR(n 5 568)

All deaths (%) 4.4 1.6 5.5Cardiac (%) 1.6 0.2 2.3Infection (%) 0 0.2 0Multisystem failure (%) 0.8 0.4 1.4Neurologic (%) 0.8 0.4 0.2Pulmonary (%) 0.8 0.2 0.2Renal failure (%) 0 0 0.4Vascular (%) 0 0 0.5Other (%) 0.4 0.2 0.5

AVR 5 aortic valve replacement; MVP 5 mitral valve repair; MVR5 mitral valve replacement.

Table 5. Predictors of Selected Outcomes by LogisticRegression Analysis After Port-Access Mitral or Aortic ValveOperation

Outcome Predictor OR (95% CI)

Death Reoperation 2.6 (1.4–4.7)Age 1.7 (0.9–3.0)MVR 2.8 (1.2–6.2)AVR 2.5 (1.0–6.3)

Stroke Age 2.1 (1.04–4.1)Reoperation for

bleedingAge 2.3 (1.3–3.9)MVR 1.8 (1.1–3.1)Reoperation 1.7 (0.9–3.1)

New atrial fibrillation Age 1.7 (1.2–2.4)High-volume center 0.6 (0.4–0.8)

AVR 5 aortic valve replacement; CI 5 confidence interval; MVR5 mitral valve replacement; OR 5 odds ratio.

1056 MINIMALLY INVASIVE GLOWER ET AL Ann Thorac SurgPORT-ACCESS VALVE REGISTRY: PREDICTORS OF OUTCOME 2000;70:1054–9

onset atrial fibrillation were older age and not being ahigh-volume center (Table 5). Institutional case volumedid not significantly affect mortality, stroke, or reopera-tion for bleeding.

Comment

The purposes of this study were to examine prospectivelythe results of port-access aortic or mitral valve surgery ata number of institutions, and to examine the determi-nants of short-term outcome. Like any new techniquethat is evolving rapidly, port-access has a learning curve[9, 15]. One might therefore expect patient outcome to beinfluenced by the volume of port-access cases performedby the operating team. These issues have only beenpartially addressed for port-access by one study, whichfound that procedure time decreased with experience butat different rates at different institutions [15]. To dateseveral reports have described results from port-accessmitral or aortic valve procedures. Mohr and colleagues[7] reported the first large series of port-access mitralprocedures (N 5 51) and reported higher mortality andmorbidity rates in port-access mitral operations (Table 6)than those seen here, in an initial experience. Othershave reported more recent results that were similar tothose reported here (Table 6). Thus, results from port-access mitral operation have been stable or have im-proved despite expansion of the patient pool. At the sametime, the port-access technique has evolved to includegreater use of indirect video to visualize the mitral valve[9], as well as selective aortic (instead of femoral) cannu-lation, with improved stability of the endoclamp and withfewer groin complications [20].

This report examines the influence of institutional casevolume upon patient outcome after port-access mitral ofaortic valve operation. Institutional case volume did notaffect mortality, stroke, or reoperation for bleeding. Thereason that high institutional volume decreased the like-lihood of new-onset atrial fibrillation is unclear but may,for example, involve the routine use of antiarrhythmicagents in mitral valve patients at higher volume institu-tions. Similar patient outcomes from a variety of institu-tions suggest that port-access can be mastered with arelatively short learning curve in a variety of institutionalcircumstances and in a variety of patient populations.Nonetheless, the results reported here tended not to

include the initial 10 to 15 cases (initial learning curve) atthe participating institutions. Thus, this study cannotexclude the possibility that results might not be similaramong institutions in the initial learning curve (first 15cases) [15].

The mortality and morbidity in this series of port-access procedures are similar to those reported by meansof a sternotomy [21]. Although the patient populationis not obviously different from those described in ster-notomy series, few studies have directly compared theresults of sternotomy to those of port-access in a con-trolled fashion [5, 13]. In retrospective analyses, fasterreturn to normal activity, lower incidence of sepsis orwound complications, less fresh plasma transfusion, andshorter hospital stay have been reported for port-accessrelative to sternotomy [5, 13].

In addition, the multivariable predictors of adverseoutcome identified for port-access (Table 5) include fac-tors such as age that have previously been reported asrisk factors for death, stroke, or new atrial fibrillation insternotomy patients. Interestingly, the incidence of newonset of atrial fibrillation for port-access in this series(10.0%–12.7%) is somewhat lower than that previouslyreported for sternotomy in mitral valve patients.

This is the largest series of port-access applied to aorticvalve replacement using a variety of techniques (partialsternotomy, right thoracotomy) advocated by Colvin andassociates, Grossi and coworkers, and Cohn and cowork-ers [12, 13, 17]. Results of aortic valve replacement werenot significantly different from those of mitral valveoperation. The potential advantages of port-access foraortic valve replacement include avoidance of sternot-omy and elimination of arterial, venous, and coronarysinus cannulas from the operative field.

In conclusion, port-access is a viable option for isolatedaortic or mitral valve surgery. Results are comparable tothose of sternotomy in terms of early morbidity andmortality, and are primarily determined by patient char-acteristics and not by the case volume of the operatinginstitution. Results will need to be reassessed as port-access techniques and patient selection criteria evolve.Outcomes, cost, and technical aspects with port-accessvalve procedures will also need to be compared, both tostandard sternotomy and to other minimally invasiveapproaches that do not use an endoaortic clamp orpercutaneous retrograde cardioplegia [17, 22, 23].

Table 6. Reported Outcomes From Port-Access Mitral Valve Operations at Individual Institutions

Author, Year [Ref.] Institution PatientsAortic

DissectionsPerioperative

DeathsValve

Failures

Fann 1997 [3] Stanford 10 0 0 1Galloway 1997 [4] NYU 50 0 0 0Mohr 1998 [7] Leipzig 51 2 5 4Glower 1998 [5] Duke 21 0 0 0Gulielmos 1999 [10] Dresden 24 0 0 0Vanerman 1999 [9] Aalst 75 2 2 1Reichenspurner 1999 [11] Munich 50 0 0 1

Total 281 4 (1%) 7 (2%) 7 (2%)

Data are given as absolute numbers of patients.

1057Ann Thorac Surg MINIMALLY INVASIVE GLOWER ET AL2000;70:1054–9 PORT-ACCESS VALVE REGISTRY: PREDICTORS OF OUTCOME

References

1. Pompili MF, Stevens JH, Burdon TA, et al. Port-accesscoronary bypass with cardioplegic arrest: acute and chroniccanine studies. J Thorac Cardiovasc Surg 1996;112:1268–74.

2. Pompili MF, Yakub A, Siegel LC, Stevens JH, Awang Y,Burdon TA. Port-access mitral valve replacement: initialclinical experience [Abstract]. Circulation 1996;94:I-533.

3. Fann JI, Pompili MF, Burdon TA, Stevens JH, St Ghoar FG,Reitz BA. Minimally invasive mitral valve surgery. SeminThorac Cardiovasc Surg 1997;9:320–30.

4. Galloway AC, Ribakove GH, Miller JS, et al. Minimallyinvasive port-access valvular surgery: initial clinical experi-ence. Circulation 1997;I-508.

5. Glower DD, Landolfo KP, Clements F, et al. Mitral valveoperation via port access versus median sternotomy. EurJ Cardiothorac Surg 1998;14:S143–7.

6. Gulielmos V, Wunderlich J, Dangel M, et al. Minimallyinvasive mitral valve surgery—clinical experiences witha port access system. Eur J Cardiothorac Surg 1998;14:S148 –53.

7. Mohr FW, Falk V, Diegeler A, Walther T, van Son JAM,Autschbach R. Minimally invasive port-access mitral valvesurgery. J Thorac Cardiovasc Surg 1998;115:567–76.

8. Galloway AC, Shemin RJ, Glower DD, et al. First report ofthe Port Access International Registry. Ann Thorac Surg1999;67:51–8.

9. Vanerman H, Wellens F, DeGeest R, Degrieck I, VanPraet F.Video-assisted port-access mitral valve surgery: from debutto routine surgery. Will trocar-port-access cardiac surgeryultimately lead to robotic cardiac surgery? Semin ThoracCardiovasc Surg 1999;11:223–34.

10. Gulielmos V, Wagner F, Waetzig B, et al. Clinical experiencewith minimally invasive coronary artery and mitral valvesurgery with the advantage of cardiopulmonary bypass andcardioplegic arrest using the port-access technique. WorldJ Surg 1999;23:480–5.

11. Reichenspurner H, Weltz A, Gulielmos V, Boehm DH,Reichart B. Port-access cardiac surgery using endovascular

cardiopulmonary bypass: theory, practice, and results.J Card Surg 1999;14:275–80.

12. Colvin SB, Galloway AC, Ribakove G, et al. Port-accessmitral valve surgery. J Card Surg 1998;13:286–9.

13. Grossi EA, Galloway AC, Ribakove GH, et al. Minimallyinvasive port access surgery reduces operative morbidity forvalve replacement in the elderly. Heart Surgery Forum 1999;2:212–5.

14. Baldwin JC. Con: minimally invasive port-access mitralvalve surgery [Editorial]. J Thorac Cardiovasc Surg 1998;115:563–4.

15. Pisano GP, Bohmer RMJ, Edmondson AC. Experience versuslearning: evidence from the adoption of minimally invasivecardiac surgery (working paper 00–036). Boston: HarvardBusiness School, 1999.

16. Zar JH. Biostatistical analysis. 3rd ed. Upper Saddle River,NJ: Prentice-Hall, 1996.

17. Cohn LH, Adams DH, Couper GS, Bichell DP. Minimallyinvasive aortic valve replacement. Semin Thorac CardiovascSurg 1997;9:331–6.

18. Bennetti FJ, Mariani MA, Rizzardi JL, Bennetti I. Minimallyinvasive aortic valve replacement. J Thorac Cardiovasc Surg1997;113:806–7.

19. Pau KK, Yakub A, Awang Y. Minimally invasive aortic valvesurgery: pocket AVR [Letter]. J Thorac Cardiovasc Surg.1998;115:255.

20. Glower DD, Komtebedde J, Clements FM, et al. Direct aorticcannulation for port-access mitral or coronary bypass oper-ations. Ann Thorac Surg 1999:1529–31.

21. Jamieson WRE, Edwards FH, Schwartz M, Bero JW, ClarkRE, Grover FL. Risk stratification for cardiac valve replace-ment. National Cardiac Surgery Database. Ann Thorac Surg1999;67:943–51.

22. Chitwood WR Jr, Wixon CL, Elbeery JR, Moran JF, ChapmanWHH, Lust RM. Video-assisted minimally invasive mitralvalve surgery. J Thorac Cardiovasc Surg 1997;114:773–82.

23. Loulmet DF, Carpentier A, Cho PW, et al. Less invasivetechniques for mitral valve surgery. J Thorac CardiovascSurg 1998;115:772–9.

Appendix

Participating Center City State Participating Center City State

Allegheny General HospitalBaptist Medical CenterBaptist Medical CenterBaptist Medical Center MontclairBaptist Memorial HospitalBarnes-Jewish HospitalBoston Medical CenterBrigham and Women’s HospitalBryn Mawr HospitalCedars-Sinai Medical CenterCentral Baptist HospitalChippenham Medical CenterCleveland Clinic FoundationDeborah Heart and Lung CenterDoctors Medical CenterDuke University HospitalEncino-Tarzana Regional

Medical CenterFlorida Hospital Medical CenterFresno Community Hospital

and Medical CenterGeorgetown University HospitalGrant-Riverside Methodist HospitalsHenrico Doctors’ HospitalHermann HospitalInova Fairfax HospitalJohns Hopkins HospitalLatter Day Saints Hospital

PittsburghOklahoma CityLittle RockBirminghamMemphisSt. LouisBostonBostonBryn MawrLos AngelesLexingtonRichmondClevelandBrown MillsModestoDurhamEncino

OrlandoFresno

WashingtonColumbusRichmondHoustonFalls ChurchBaltimoreSalt Lake City

PAOKARALTNMOMAMAPACAKYVAOHNJCANCCA

FLCA

DCOHVATXVAMDUT

St. Francis Hospital Centerand Health Centers

St. John Medical CenterSt. Joseph Medical CenterSt. Luke’s Medical CenterSt. Mary’s HospitalSt. Thomas HospitalStanford University HospitalThe Sanger Clinic, PA and

Carolinas Medical CenterUnion Memorial HospitalUniversity Community HospitalUniversity HospitalUniversity of LouisvilleDepartment of Thoracic and

Cardiovascular Surgery and JewishHospital Heart/Lung Institute,University of Pennsylvania

University of Southern CaliforniaUniversity of Utah,

School of MedicineUniversity of Virginia Medical CenterVeterans Affairs Medical Center

San DiegoWake Medical CenterWalter O. Boswell Memorial HospitalWestchester County Medical CenterWillis Knighton Medical Center

Beech Grove

TulsaTowsonMilwaukeeRichmondNashvilleStanfordCharlotte

BaltimoreTampaAugustaLouisvillePhiladelphia

PasadenaSalt Lake City

CharlottesvilleSan Diego

RaleighSun CityValhallaShreveport

IN

OKMDWIVATNCANC

MDFLGAKYPA

CAUT

VACA

NCAZNYLA

1058 MINIMALLY INVASIVE GLOWER ET AL Ann Thorac SurgPORT-ACCESS VALVE REGISTRY: PREDICTORS OF OUTCOME 2000;70:1054–9

Appendix (continued)

Participating Center City State Participating Center City State

Lehigh Valley HospitalLittle Company of Mary HospitalLos Robles Regional Medical Ctr.Loyola University Medical CenterMassachusetts General HospitalMedical Center HospitalMedical College of Ohio HospitalMission St. Joseph’s Health SystemNew Mexico Heart InstituteNew York University Medical Ctr.Ochsner Foundation HospitalPalo Alto Veterans Affairs

Medical CenterPresbyterian Hospital of DallasRedding Medical CenterResearch Medical CenterRochester General HospitalRush-Presbyterian-St. Luke’s

Medical CenterSaint Francis HospitalSt. Joseph’s Hospital of AtlantaSchumpert Medical CenterSouthern Illinois University

School of MedicineSpectrum HealthSt. Francis Health Care SystemSt. Francis Hospital

AllentownTorranceThousand OaksMaywoodBostonOdessaToledoAshevilleAlbuquerqueNew YorkNew OrleansPalo Alto

DallasReddingKansas CityRochesterChicago

TulsaAtlantaShreveportSpringfield

Grand RapidsHartfordRoslyn

PACACAILMATXOHNCNMNYLACA

TXCAMONYIL

OKGALAIL

MICTNY

Akademiska Sjukhuset UppsalaBergmannshell-UniversitatsklinikumC.H.U. Lyon Louis PradelC.H.U.R. Metz Hospital Bon SecoursClinique ClairvalDresden UniversityFreeman Hospital

Harley Stree Hospital,Columbia Health Group

Her-kreislaufzentrumHopital BroussaisHopital Europeen de Paris-

La RoseraieHospital do MeixoeiroHospital Doce de OctubreHospital Santa CruzKlinikum GrosshadernKrankenhaus Links der WesserLille/Hospital CardiologiqueOnze Lieve Vrouw ClinicPoliclinico San MatteoRoyal Infirmary of EdinburghUniversity of BarcelonaUniversity of FrankfurtEscorts Heart Institute and ResearchNational Heart Institute

UppsalaBochumLyonMetzMarseilleDresdenNewcastle

upon TyneLondon

LeipzigParisAubervilliers

VigoMadridLinda-a-VelhaMunichBremenLilleAalstPaviaEdinburghBarcelonaFrankfurt a.M.New DelhiKuala Lampur

SwedenGermanyFranceFranceFranceGermanyUK

UK

GermanyFranceFrance

SpainSpainPortugalGermanyGermanyFranceBelgiumItalyScotlandSpainGermanyIndiaMalaysia

1059Ann Thorac Surg MINIMALLY INVASIVE GLOWER ET AL2000;70:1054–9 PORT-ACCESS VALVE REGISTRY: PREDICTORS OF OUTCOME