J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8

ª 2 0 1 8 B Y T H E AM E R I C A N C O L L E G E O F C A R D I O L O G Y F O UN DA T I O N

P U B L I S H E D B Y E L S E V I E R

Conscious Sedation VersusGeneral Anesthesia inTranscatheter Aortic Valve ReplacementThe German Aortic Valve Registry

Oliver Husser, MD, PHD,a Buntaro Fujita, MD,b Christian Hengstenberg, MD,a,c,d Christian Frerker, MD,e

Andreas Beckmann, MD,f Helge Möllmann, MD,g Thomas Walther, MD,h Raffi Bekeredjian, MD,i Michael Böhm, MD,j

Costanza Pellegrini, MD,a Sabine Bleiziffer, MD,k,l Rüdiger Lange, MD,k,l Friedrich Mohr, MD,m

Christian W. Hamm, MD,n,o Timm Bauer, MD,n Stephan Ensminger, MD,b on behalf of the GARY Executive Board

ABSTRACT

ISS

Fro

Ge

Oe

Au

Mu

Ge

mu

He

bu

Mu

Ce

Ge

ckh

(Ed

Ge

Th

Ma

OBJECTIVES The aims of this study were to report on the use of local anesthesia or conscious sedation (LACS) and

general anesthesia in transcatheter aortic valve replacement and to analyze the impact on outcome.

BACKGROUND Transcatheter aortic valve replacement can be performed in LACS or general anesthesia. Potential

benefits of LACS, such as faster procedures and shorter hospital stays, need to be balanced with safety.

METHODS A total of 16,543 patients from the German Aortic Valve Registry from 2011 to 2014 were analyzed, and

propensity-matched analyses were performed to correct for potential selection bias.

RESULTS LACS was used in 49% of patients (8,121 of 16,543). In hospital, LACS was associated with lower rates of

low-output syndrome, respiratory failure, delirium, cardiopulmonary resuscitation, and death. There was no difference in

paravalvular leakage (IIþ) between LACS and general anesthesia in the entire population (5% vs. 4.8%; p¼ 0.76) or in the

matched population (3.9% vs. 4.9%, p ¼ 0.13). The risk for prolonged intensive care unit stay ($3 days) was significantly

reducedwith LACS (odds ratio: 0.82; 95%confidence interval [CI]: 0.73 to0.92; p¼0.001). Thirty-daymortalitywas lower

with LACS in the entire population (3.5% vs. 4.9%; hazard ratio [HR]: 0.72; 95% CI: 0.60 to 0.86; p < 0.001) and in the

matched population (2.8% vs. 4.6%; HR: 0.6; 95%CI: 0.45 to 0.8; p<0.001). However, no differences in 1-year mortality

between both groups in the entire population (16.5% vs. 16.9%; HR: 0.93; 95% CI: 0.85 to 1.02; p ¼ 0.140) and in the

propensity-matched population (14.1% vs. 15.5%; HR: 0.90; 95% CI: 0.78 to 1.03; p ¼ 0.130) were observed.

CONCLUSIONS Use of LACS in transcatheter aortic valve replacement is safe, with fewer post-procedural

complications and lower early mortality, suggesting its broad application. (J Am Coll Cardiol Intv 2018;11:567–78)

© 2018 by the American College of Cardiology Foundation.

N 1936-8798/$36.00 https://doi.org/10.1016/j.jcin.2017.12.019

m the aKlinik für Herz-und Kreislauferkrankungen, Deutsches Herzzentrum München, Technical University Munich, Munich,

rmany; bDepartment for Thoracic and Cardiovascular Surgery, Heart and Diabetes Center NRW, Ruhr-University Bochum, Bad

ynhausen, Germany; cDivision of Cardiology, Department of Internal Medicine II, Medical University of Vienna, Vienna,

stria; dDeutsches Zentrum für Herz- und Kreislauf-Forschung e.V. (German Center for Cardiovascular Research), Partner Site

nich Heart Alliance, Munich, Germany; eDepartment of Cardiology, Asklepios Klinik St. Georg, Hamburg, Germany; fDeutsche

sellschaft für Thorax-, Herz- und Gefäßchirurgie, Berlin, Germany; gDepartment of Cardiology, St. Johannes Hospital, Dort-

nd, Germany; hDepartment of Cardiac Surgery, Kerckhoff Heart Center, Bad Nauheim, Germany; iCenter for Internal Medicine,

idelberg University Hospital, Heidelberg, Germany; jKlinik für Innere Medizin III, Universitätsklinikum des Saarlandes, Hom-

rg/Saar, Germany; kKlinik für Herz- und Gefäßchirurgie, Deutsches Herzzentrum München, Technical University Munich,

nich, Germany; lInsure (Institute for Translational Cardiac Surgery), Department of Cardiovascular Surgery, German Heart

nter Munich, Technische Universität München, Munich, Germany; mLeipzig Heart Center, University of Leipzig, Leipzig,

rmany; nDepartment of Medical Clinic I, University of Giessen, Giessen, Germany; and the oDepartment of Cardiology, Ker-

off Heart Center, Bad Nauheim, Germany. This work was supported by unrestricted grants from medical device companies

wards Lifesciences, Medtronic, Symetis, JenaValve Technology, Liva-Nova, St. Jude Medical, and Direct Flow Medical), the

rman Heart Foundation, the German Society of Cardiology, and the German Society of Thoracic and Cardiovascular Surgery.

e authors have reported that they have no relationships relevant to the contents of this paper to disclose.

nuscript received October 23, 2017; revised manuscript received December 14, 2017, accepted December 19, 2017.

ABBR EV I A T I ON S

AND ACRONYMS

ASA = American Society of

Anesthesiologists

CI = confidence interval

GA = general anesthesia

HR = hazard ratio

ICU = intensive care unit

LACS = local anesthesia or

conscious sedation

PVL = paravalvular leakage

PPI = permanent pacemaker

implantation

TAVR = transcatheter aortic

valve replacement

THV = transcatheter heart

valve

Husser et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8

Conscious Sedation Versus General Anesthesia in TAVR M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8

568

T ranscatheter aortic valve replace-ment (TAVR) has revolutionized thetreatment of severe symptomatic

aortic valve stenosis and has recently beenshown to be noninferior to conventional sur-gery in intermediate-risk patients (1). As aconsequence, there is a considerable incre-ment in TAVR procedures, with a projectedannual case number of about 17,000 inEurope (2). Additionally, increasing operatorand heart team experience, refinement oftranscatheter heart valves (THVs), intro-ducers, and delivery systems, and economicconsiderations have fostered interest in asimplification of the TAVR procedure.

In this regard, the possibility to performTAVR only in local anesthesia or conscioussedation (LACS) instead of general anesthesia

(GA) appears appealing. Compared with GA, LACSmay be associated with logistic benefits of shorterprocedure times and shorter intensive care unit (ICU)and in-hospital stays (3–5). However, these potentialbenefits need to be carefully weighed against patientsafety, as increased risk for paravalvular leakage(PVL) and need for permanent pacemaker implanta-tions (PPI) has been observed with LACS (5,6).

SEE PAGE 579

In the absence of randomized controlled data,registries offer an opportunity to investigate thevalue of each anesthesiologic strategy in TAVR.Therefore, we analyzed the use of each anesthesio-logic strategy in a large population from theGARY (German Aortic Valve Registry), reporting andassessing the influence of LACS versus GA on earlyand midterm mortality. Apart from conventionalmultivariate adjustment, the impact of LACS versusGA was analyzed in a propensity-matched subset ofpatients to correct for an inherent selection bias andbaseline differences.

METHODS

THE GARY. The GARY is a nonprofit nationwide reg-istry inaugurated in July 2010 by the German Societyof Cardiology and the German Society of Thoracic andCardiovascular Surgery. The aim of GARY is to collectdata on a real-world and all-comer basis for short- andlong-term outcomes and to provide information oncurrent practices of treatment in patients undergoingthe complete spectrum of interventional and surgicalaortic valve interventions in Germany. The protocol

of GARY has been previously described in detail (7).The responsible societies and the BQS Institute areindependent organizations by virtue of their consti-tution both from legal and scientific points of view.GARY receives financial support in form of unre-stricted grants from medical device companies(Edwards Lifesciences, Medtronic, Symetis, Jena-Valve Technology, Liva-Nova, St. Jude Medical, andDirect Flow Medical), the German Heart Foundation,the German Society of Cardiology, and the GermanSociety of Thoracic and Cardiovascular Surgery, noneof which have access to data or any influence onpublications.

STUDY POPULATION AND ENDPOINTS. All patientsundergoing elective or urgent transfemoral TAVR inGermany from 2011 to 2014 were included in thisanalysis. Not included in the analysis were patientsundergoing surgical aortic valve replacement, TAVRusing nontransfemoral access treatment, or in anemergency or ultima-ratio setting. In total, 16,543patients were analyzed and divided into 2 groupsaccording to the primary anesthesiologic strategyused during the procedure (LACS, n ¼ 8,121; GA,n ¼ 8,422). Baseline parameters, procedural charac-teristics, and in-hospital outcomes were analyzedaccording to LACS versus GA. Endpoints of thepresent study were 30-day and 1-year mortality.Patients were followed during the first year afterTAVR, and survival status was determined throughdirect telephone contact. Thirty-day survival statuswas available for all patients, and 1-year follow-upwas complete for >97% of patients. Event-freepatients were censored at last contact alive.

STATISTICAL ANALYSIS. Continuous variables areexpressed as mean � SD or median (interquartilerange) and were compared using the Student t test orthe Mann-Whitney U test, respectively. Discrete var-iables were compared using the chi-square test.

The association of LACS versus GA with time to30-day and 1-year mortality was assessed using Coxproportional hazard models. Hazard ratios (HRs) withtheir respective 95% confidence intervals (CIs) werecomputed. To test the independent influence ofLACS on both outcomes, conventional multivariateadjustment for variables yielding p values <0.100 inunivariate analyses was performed. Because of alikely influence of a learning curve or team experi-ence on outcome, the year of treatment was includedas a covariate. Additionally, to account for centerexperience, centers were categorized in quintilesaccording to cases performed per year.

FIGURE 1 Proportion of Transcatheter Aortic Valve Replacement Procedures Performed in Local Anesthesia or Conscious Sedation and

General Anesthesia per Year and per Center Experience

(A) Total number of procedures per year and proportion of local anesthesia or conscious sedation (LACS) and general anesthesia (GA).

(B) Proportion of LACS and GA according to the quintiles (Q) of center experience (cases per year).

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8 Husser et al.M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8 Conscious Sedation Versus General Anesthesia in TAVR

569

To correct for potential selection bias in the use ofanesthesiologic strategy, 2 approaches were used.First, a propensity score for the probability to un-dergo TAVR in LACS was calculated using multivar-iate regression analysis including variablesassociated with LACS in univariate analysis. Thispropensity score was included into the multivariatemodels. Second, 2 matched cohorts were created bymeans of propensity matching using R version 3.2.3(R Foundation for Statistical Computing, Vienna,Austria) and the package MatchIt (8). In short, 1-to-1nearest neighbor matching was used to identify 1control patient treated with GA for each patienttreated with LACS. To improve matching quality, thecaliper was set at 0.0001, resulting in 2,624 matchedcases per group. Baseline characteristics showingsignificant association with LACS and with 30-daymortality as well as the type of THV were includedin the matching algorithm. Online Figure 1 summa-rizes the study flow and variables used for propensitymatching.

All analyses were performed in the entire as well asin the propensity-matched population. A 2-sidedp value of <0.05 was considered to indicate statistical

significance for all analyses. SPSS version 19.0 (IBM,Armonk, New York) and R were used for analyses.

RESULTS

TEMPORAL EVOLUTION, PROPORTION OF TAVR IN

LACS, AND INFLUENCE OF CENTER EXPERIENCE.

In total, 16,543 patients were analyzed. The medianage was 81 years, 56% were women, and the medianSociety of Thoracic Surgeons score was 4.8%. Overall,LACS was used as the primary anesthesiologic strat-egy in 49% of cases (8,121 of 16,543). From 2011 to2013, the proportion of procedures performed in LACSdecreased with increasing number of procedures andthen remained stable at 46% (p for trend <0.001)(Figure 1A). There was an inverse relationship in theuse of LACS with center experience, with 66% of pa-tients undergoing TAVR in LACS at high-volumecenters (>358 procedures per year) and 38% in the2 lowest quintiles of center experience (p for trend<0.001) (Figure 1B).

PREDICTORS OF LACS AND PROPENSITY MATCHING.

Table 1 shows the baseline characteristics of the entire

TABLE 1 Baseline Characteristics

Entire Population Matched Population

All Patients(n ¼ 16,543)

LACS(n ¼ 8,121)

GA(n ¼ 8,422) p Value

LACS(n ¼ 2,624)

GA(n ¼ 2,624) p Value

Age (yrs) 81 � 6 81 � 6 81 � 6 <0.001 81 � 6 81 � 5 0.68

Age group (yrs)

<75 11.0 (1,807) 11.8 (951) 10.2 (856) 0.003 8.9 (233) 8.7 (229) 0.54

75–84 58.9 (9,714) 58.7 (4,752) 59.0 (4,962) 61.1 (1,602) 62.5 (1,640)

$85 30.2 (4,978) 29.5 (2,387) 30.8 (2,591) 30.1 (789) 28.8 (755)

Female 56.4 (9,331) 55.5 (4,506) 57.3 (4,825) 0.02 58.6 (1,537) 58.2 (1,527) 0.78

Logistic EuroSCORE (%) 16 (10–26) 17 (10–28) 16 (10–26) 0.11 15 (10–25) 15 (10–24) 0.57

STS score (%) 4.8 (3.3–7.1) 4.7 (3.2–7) 4.9 (3.4–7.2) <0.001 4.6 (3.2–6.6) 4.5 (3.2–6.6) 0.84

STS group

<4% 37.2 (6,130) 38.2 (3,087) 36.2 (3,043) 0.024 40.7 (1,068) 40.7 (1,180) 0.42

4%–8% 43.3 (7,150) 42.9 (3,468) 43.8 (3,682) 43.8 (1,149) 45.0 (1,180)

>8% 19.5 (3,219) 19.0 (1,535) 20.0 (1,684) 15.5 (407) 14.3 (375)

BMI (kg/m2) 27 (24–30) 27 (24–30) 27 (24–30) 0.26 27 (24–30) 27 (24–30) 0.02

BMI group (kg/m2)

<25 36.3 (5,938) 36.8 (2,953) 35.9 (2,985) 0.42 36.2 (938) 33.6 (872) 0.06

25–30 39.2 (6,413) 39.0 (3,131) 39.4 (3,282) 40.3 (1,043) 40.6 (1,054)

>30 24.4 (3,996) 24.2 (1,938) 24.7 (2,058) 23.5 (607) 25.8 (670)

Prior sternotomy 17.7 (2,915) 17.9 (1,451) 17.4 (1,464) 0.39 16.8 (439) 16.3 (426) 0.62

Prior PCI 28.8 (4,771) 28.7 (2,331) 29.0 (2,441) 0.69 28.4 (744) 29.5 (773) 0.38

NYHA functional class $III 85.1 (14,077) 84.8 (6,886) 85.4 (7,191) 0.29 84.1 (2,208) 84.8 (2,226) 0.49

CCS class $3 13.5 (2,230) 6.4 (1,060) 7.1 (1,170) 0.11 13.1 (343) 12.8 (335) 0.74

ASA class $4 13.5 (2,233) 7.8 (636) 19.0 (1,597) <0.001 5.8 (152) 5.8 (151) 0.95

Pacemaker/ICD 12.8 (2,120) 13.7 (1,110) 12.0 (1,010) 0.001 8.3 (218) 7.7 (201) 0.39

Coronary artery disease 54.1 (8,944) 55.6 (4,518) 52.6 (4,426) <0.001 52.6 (1,380) 54.3 (1,425) 0.21

Pulmonary hypertension* 22.9 (3,730) 25.5 (2,037) 20.4 (1,693) <0.001 17.7 (465) 18.7 (491) 0.35

Atrial fibrillation 29.7 (4,905) 29.3 (2,383) 29.9 (2,522) 0.40 26.8 (703) 27.1 (712) 0.78

Diabetes mellitus 13.8 (2,275) 13.8 (1,123) 13.7 (1,152) 0.78 11.6 (305) 12.1 (318) 0.59

Hypertension 87.0 (14,213) 87.7 (7,062) 86.3 (7,151) 0.009 90.6 (2,377) 90.5 (2,375) 0.93

Peripheral artery disease 14.4 (2,383) 14.5 (1,180) 14.3 (1,203) 0.66 12.7 (332) 11.9 (313) 0.42

COPD 12.6 (2,086) 12.2 (992) 13.0 (1,094) 0.13 10.8 (283) 11.2 (295) 0.60

Serum creatinine (>2 mg/dl) 5.2 (859) 5.3 (434) 5.0 (425) 0.39 4.2 (109) 4.7 (123) 0.35

On dialysis 2.7 (453) 2.8 (229) 2.7 (224) 0.53 1.7 (44) 1.6 (42) 0.83

Neurological dysfunction 2.9 (473) 2.8 (229) 2.9 (244) 0.76 2.5 (66) 2.4 (62) 0.72

Mitral regurgitation IIþ 27.7 (4,483) 28.5 (2,266) 26.9 (2,217) 0.02 20.2 (529) 20.4 (534) 0.86

Tricuspid regurgitation IIþ 18.9 (3,022) 19.7 (1,547) 18.2 (1,475) 0.02 12.2 (320) 12.0 (314) 0.80

Mean gradient (mm Hg) 43 (33–53) 42 (32–53) 44 (34–54) <0.001 43 (33–54) 44 (35–54) 0.22

LV ejection fraction <30% 8.7 (1,434) 8.6 (696) 8.8 (738) 0.66 6.8 (178) 6.6 (174) 0.83

Values are mean � SD, % (n), or median (interquartile range). *Defined as pulmonary artery pressure >55 mm Hg.

ASA ¼ American Society of Anesthesiologists; CCS ¼ Canadian Cardiovascular Society; COPD ¼ chronic obstructive pulmonary disease; EuroSCORE ¼ European System forCardiac Operative Risk Evaluation; GA ¼ general anesthesia; ICD ¼ implantable cardioverter-defibrillator; LACS ¼ local anesthesia or conscious sedation; LV ¼ left ventricular;NYHA ¼ New York Heart Association; PCI ¼ percutaneous coronary interventions; STS ¼ Society of Thoracic Surgeons.

Husser et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8

Conscious Sedation Versus General Anesthesia in TAVR M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8

570

patient population and according to LACS and GA.Independent predictors to undergo TAVR in LACSwere younger age, lower Society of Thoracic Surgeonsscore and American Society of Anesthesiologists(ASA) class, presence of a permanent pacemaker,pulmonary hypertension, lower mean transaorticgradient, and year of procedure and increasing centerexperience (see Online Table 1 for complete results ofmultivariate analysis). Propensity matching resultedin 2 cohorts of 2,624 patients per anesthesiologic

strategy. Online Figures 2A and 2B show the distri-bution of the propensity scores before and aftermatching, and differences in baseline characteristicsbetween LACS versus GA disappeared after matching(Table 1).

PROCEDURAL RESULTS AND COMPLICATIONS.

Procedural characteristics and in-hospital complica-tions according to LACS versus GA are depicted inTable 2 for the entire and matched populations. The

TABLE 2 Procedural Characteristics and In-Hospital Outcomes

Entire Population Matched Population

All Patients(n ¼ 16,543)

LACS(n ¼ 8,121)

GA(n ¼ 8,422) p Value

LACS(n ¼ 2,624)

GA(n ¼ 2,624) p Value

Procedural characteristics

Elective setting 84.9 (14,039) 85.0 (6,904) 84.7 (7,135) 0.60 86.1 (2,259) 87.3 (2,290) 0.21

Balloon-expandable THV 51.7 (7,891) 40.4 (3,105) 63.3 (4,786) <0.001 55.3 (1,452) 54.3 (1,425) 0.45

Procedure time (min) 82 (55–100) 65 (50–90) 77 (60–107) <0.001 67 (52–91) 77 (60–107) <0.001

Fluoroscopy time (min) 15 (11–20) 14 (10–20) 15 (11–21) <0.001 14 (10–19) 15 (11–21) <0.001

Contrast agent (ml) 140 (100–200) 140 (105–195) 140 (100–200) 0.16 140 (100–190) 150 (100–220) <0.001

Procedural complications

Procedural success 97.6 (16,152) 97.8 (7,944) 97.5 (8,208) 0.13 97.8 (2,566) 97.8 (2,566) 0.999

Device malposition 2.0 (337) 1.9 (152) 2.2 (185) 0.14 1.5 (40) 2.7 (71) 0.003

Device embolization 0.3 (53) 0.3 (24) 0.3 (29) 0.58 0.2 (5) 0.6 (15) 0.03

Paravalvular leakage IIþ 4.9 (658) 5.0 (318) 4.8 (340) 0.76 3.9 (85) 4.9 (106) 0.13

Conversion sternotomy 0.8 (137) 0.3 (27) 1.3 (110) <0.001 0.3 (8) 1.4 (38) <0.001

Bleeding ($2 U) 18.3 (2,993) 17.3 (1,388) 19.4 (1,605) <0.001 17.0 (442) 20.0 (517) 0.01

Vascular complication 9.3 (1,534) 10.9 (888) 7.7 (646) <0.001 11.1 (290) 8.6 (226) 0.003

New pacemaker* 20.3 (2,929) 22.2 (1,558) 18.5 (1,371) <0.001 20.4 (490) 20.8 (504) 0.71

Post-procedural and in-hospital complications

CPR post 2.8 (455) 2.1 (174) 3.3 (281) <0.001 1.8 (48) 3.5 (91) <0.001

Respiratory failure† 2.3 (381) 1.8 (144) 2.8 (237) <0.001 1.7 (44) 2.7 (72) 0.009

Low-output syndrome‡ 0.9 (144) 0.4 (32) 1.3 (112) <0.001 0.5 (14) 1.3 (33) 0.005

Psychological syndromes§ 3.3 (539) 2.5 (207) 3.9 (332) <0.001 2.5 (65) 4.0 (105) 0.002

Myocardial infarction 0.3 (55) 0.3 (27) 0.3 (28) 0.999 0.3 (9) 0.4 (1) 0.82

Disabling strokek 1.4 (229) 1.5 (121) 1.3 (108) 0.25 1.0 (25) 1.4 (38) 0.10

Dialysis (acute/chronic) 3.8 (632) 4.0 (323) 3.7 (309) 0.30 2.6 (69) 2.9 (76) 0.56

Late mobilization 22.9 (3,457) 22.5 (1,645) 23.4 (1,812) 0.16 22.8 (535) 21.5 (528) 0.30

Days in ICU¶ 2 (1–3) 2 (1–3) 2 (1–3) 0.82 2 (1–3) 2 (1–3) 0.001

>1 day in ICU 62.1 (10,261) 62.4 (5,066) 61.8 (5,195) 0.44 61.8 (1,620) 65.5 (1,715) 0.01

Days in hospital¶ 9 (7–13) 9 (7–13) 9 (7–12) 0.001 9 (7–13) 9 (7–12) 0.11

In-hospital mortality 3.4 (567) 2.8 (226) 4.0 (341) <0.001 2.4 (62) 3.8 (99) 0.003

Values are % (n) or median (interquartile range). *Excluding patients with pacemaker at baseline. †Requiring intubation or tracheotomy. ‡Requiring intra-aortic counter-pulsation or mechanical support. §Requiring medical treatment mobilization on day 3 or later. kDefined as stroke resulting in a modified Rankin Scale score of 2 or more at90 days or an increase in at least one modified Rankin Scale category from an individual’s pre-stroke baseline. ¶Only for patients discharged alive.

CPR ¼ cardiopulmonary resuscitation; ICU ¼ intensive care unit; THV ¼ transcatheter heart valve; other abbreviations as in Table 1.

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8 Husser et al.M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8 Conscious Sedation Versus General Anesthesia in TAVR

571

majority of cases was performed in an elective setting(85%). Balloon-expandable THVs were less frequentlyused with LACS in the entire population (40% vs.63%; p < 0.001), but matching resulted in an equaldistribution (55% vs. 54%; p ¼ 0.450).

Overall, procedural success was achieved in97.6%. Comparing LACS with GA in both the entireand the matched populations, procedural durationand fluoroscopy time were significantly shorter,conversion to sternotomy and bleeding complica-tions were less frequent, and the rate of vascularcomplications was higher. After matching, the rate ofdevice embolization and malposition was lower withLACS, and a difference in need for new pacemakerimplantation observed in the entire population dis-appeared. There was no difference in PVL IIþ be-tween both groups in the entire (5% vs. 4.8%; p ¼0.76) or in the matched population (3.9% vs. 4.9%;p ¼ 0.13).

Post-procedural course was less complicated withLACS compared with GA. This was the case for boththe entire and the matched population with a lowerincidence of low-output syndrome, respiratory fail-ure, post-operative delirium, cardiopulmonaryresuscitation, and in-hospital death.

Duration of ICU and hospital stay was not differentbetween both anesthesiologic strategies in the entirepopulation analysis (Figure 2A). However, in thematched population analysis (Figure 2B), LACS wassignificantly associated with shorter ICU stays (higherproportion of cases with #1 day [38% vs. 34%,p ¼ 0.003] and lower proportion of cases with $4 days[19% vs. 22%; p ¼ 0.001]). The risk for prolonged ICUstay ($3 days) was significantly reduced with LACS(odds ratio: 0.82; 95% CI: 0.73 to 0.92; p ¼ 0.001).

30-DAY AND 1-YEAR MORTALITY. Overall, 30-dayand 1-year mortality was 4.2% (695 of 16,543) and

FIGURE 2 Days in Intensive Care Unit and in Hospital According to Local Anesthesia or Conscious Sedation Versus General Anesthesia

Days in intensive care unit (ICU) and in hospital according to local anesthesia or conscious sedation (LACS) or general anesthesia (GA) in the entire population (A) and in

the matched population (B).

Husser et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8

Conscious Sedation Versus General Anesthesia in TAVR M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8

572

16.7% (2,761 of 16,543), respectively. Baseline char-acteristics associated with both outcomes were usedfor adjustment in the multivariate analysis (OnlineTables 2 and 3). Figures 3A, 3B, and 4 show the ratesand risk for mortality. Thirty-day mortality wassignificantly lower in patients undergoing TAVR inLACS compared with GA in both the entire (3.5% vs.4.9%, p < 0.001) and the matched (2.8% vs. 4.6%;HR: 0.60; 95% CI: 0.45 to 0.80; p < 0.001) analysis.The unadjusted risk for 30-day mortality was signifi-cantly lower with LACS (HR: 0.71; 95% CI: 0.61 to0.83; p < 0.001), an effect that persisted after multi-variate and propensity score adjustment (odds ratio:0.72; 95% CI: 0.60 to 0.86; p < 0.001). The mortalitybenefit observed at 30 days did not translate into asignificant difference in 1-year mortality in the entirepopulation (16.5% vs. 16.9%, p ¼ 0.380; adjusted HR:0.93; 95% CI: 0.85 to 1.02; p ¼ 0.140) or in the

propensity-matched population (14.1% vs. 15.5%; HR:0.90; 95% CI: 0.78 to 1.03; p ¼ 0.130) (Figures 3A,3B, and 4).

SUBGROUP ANALYSES. Subgroup analyses of thedifferential effect of LACS versus GA on 30-day(Figure 5) and 1-year (Figure 6) mortality in theentire and matched populations were performed.When performing TAVR in LACS, a trend towardlower risk for 30-day mortality was observed inpatients of older age, lower ASA class (<4), andpulmonary hypertension. No significant interactionfor the effect of LACS versus GA on 30-day mor-tality was observed in case of conversion to ster-notomy. For 1-year mortality, LACS was associatedwith a significant reduction in women and in pa-tients with pulmonary hypertension (matched pop-ulation only).

FIGURE 3 30 and 1-Year Mortality According to Local Anesthesia or Conscious Sedation Versus General Anesthesia

Rates and unadjusted and adjusted risk for 30-day mortality (A) and 1-year mortality (B) in the entire population and in the matched population. *Adjusted for variables

associated with 30-day or 1-year mortality in the univariate analysis. See Online Table 1 for the results of the multivariate analysis. #Adjusted for the propensity score

(PS) to undergo transcatheter aortic valve replacement in LACS. See Online Table 1 for multivariate predictors of LACS. CI ¼ confidence interval; GA ¼ general

anesthesia; HR ¼ hazard ratio; LACS ¼ local anesthesia or conscious sedation.

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8 Husser et al.M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8 Conscious Sedation Versus General Anesthesia in TAVR

573

DISCUSSION

The use of LACS and GA during TAVR was analyzedin a large national registry. The results can besummarized as follows. First, almost one-half ofTAVR procedures were performed in LACS, primarilyat high-volume centers. Second, LACS was associatedwith faster procedure times, fewer procedural com-plications, and a more favorable post-proceduralcourse. Third, 30-day mortality was lower withLACS, an effect that persisted after multivariateadjustment and propensity matching. This benefitappeared more pronounced in female patients andthose with older age, lower ASA class, and pulmonaryhypertension. Fourth, despite this risk reduction,1-year survival did not differ, except for femalepatients.

AVAILABLE DATA ON LACS VERSUS GA IN TAVR.

The number of TAVR procedures performed in LACSis rising (9). Increasing operator experience, reduc-tion of device profiles, and use of percutaneousarterial closure systems led some groups to exclu-sively use LACS (10). Several studies (3,5,11–20) anda recent meta-analysis (6) investigated the impact ofLACS versus GA in TAVR providing data on more

than 5,000 patients to date. There are, however,several issues that may hamper the relevance andthe scope of the available data. First, the majority ofthese studies were conducted in the early years ofTAVR, and only 3 studies were conducted after 2012(16,19,20). Thus, these reports may rather reflect thelearning curve of the performing teams whenswitching from GA to LACS with increasing experi-ence. Second, sample sizes of these previous studieswere comparatively small, with only 4 studiesincluding more than 250 patients (5,15,17,18). Third,only 3 studies addressed inherent patient selectionbias using adequate statistical methods such aspropensity matching (5,18,19). Therefore, the pre-sent study significantly adds to the body of evi-dence by including more than 16,000 patients up to2014 and using propensity matching to address po-tential selection bias and the influence of a learningcurve.

LACS VERSUS GA: PROS AND CONS. The imple-mentation of LACS indicated an increased experienceof the heart team performing the procedure and isoften driven by potential benefits. These benefitsinclude the possibility for periprocedural neurolog-ical assessment, less vasopressor use (12,19), shorter

FIGURE 4 Survival in the Entire and Propensity-Matched Populations According to Local Anesthesia or Conscious Sedation Versus

General Anesthesia

Kaplan-Meier cumulative survival according to local anesthesia or conscious sedation (LACS) versus general anesthesia (GA) in the entire

population (A) and the matched population (B). TAVR ¼ transcatheter aortic valve replacement.

Husser et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8

Conscious Sedation Versus General Anesthesia in TAVR M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8

574

procedure times (12,14–16,19), and shorter ICU stayswith earlier ambulation. Consistent with this obser-vation, we detected a clear benefit of LACS, withfaster procedures and a less complicated peri- andpost-procedural course, persisting even after pro-pensity matching. Especially encouraging is theobservation of a significantly lower rate of low-outputor psychological syndromes, which was reduced byalmost 50% with LACS in the propensity-matchedpopulation. This overall less complicated post-procedural course may result in a significant reduc-tion in health care expenses, as already suggested bya recent study (16). However, regarding the length ofICU and in-hospital stay, results of previous studieshave not been univocal. Although some observed areduction of the length of stay (3,12–16), othersfailed to show statistically significant differences(11,17–19,21) or even reported longer stays after TAVRin LACS (5,20). The present study shows that after

propensity matching, LACS was associated withshorter ICU stays and a reduced risk for prolongedICU stay. When looking to the overall in-hospital stay,these benefits were not observed. The median lengthof stay was 9 days in the present study. There areseveral factors that need to be taken into consider-ation when interpreting this observation. First, thestudy population includes patients treated between2011 and 2014 and therefore may not entirely becomparable with contemporary patients referred forTAVR. Second, in Germany, if the actual length ofstay falls below the diagnosis-related group–determined average length of stay, this difference isdeducted from the total revenues obtained from acase, offering little stimulus to fall below the averagelength of stay. However, significantly shorter ICUstays with LACS observed in the present study sup-port the notion of possible shorter hospital stays withLACS. In this regard, LACS is not inferior to GA, and

FIGURE 5 Subgroup Analysis of Risk for 30-Day Mortality According to Local Anesthesia or Conscious Sedation Versus General Anesthesia

Odds ratios adjusted for variables associated with 30-day mortality in the univariate analysis and for the propensity score to undergo transcatheter aortic valve

replacement in local anesthesia or conscious sedation (LACS) for the entire population (A) and the matched population (B). See the Online Table for the results of the

multivariate analysis. Values are % (n). ASA ¼ American Society of Anesthesiologists; BMI ¼ body mass index; CCS ¼ Canadian Cardiovascular Society; CI ¼ confidence

interval; COPD ¼ chronic obstructive pulmonary disease; GA ¼ general anesthesia; HR¼ hazard ratio; LVEF ¼ left ventricular ejection fraction; NYHA ¼ New York Heart

Association; STS ¼ Society of Thoracic Surgeons.

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8 Husser et al.M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8 Conscious Sedation Versus General Anesthesia in TAVR

575

future studies focusing on possible economic benefitsof LACS with earlier discharge are warranted. Thesepotential economic and logistic benefits of LACS inTAVR need to be well balanced by a similar safetyprofile compared with GA. Although most studieshave shown similar procedural success rates withboth strategies (5,12,14,16), one showed lower rateswith LACS (15). The possible advantages of GA includeless unrest of the patient and the additional optionfor respiratory hold and for minimal movement dur-ing implantation of the prosthesis. Indeed, unex-pected patient movement (22) may have played a rolein the observation of a higher permanent pacemakerrate due to misplacement of the THV (15,18) or ahigher rate of major vascular complications (18) usingLACS. The present study calls these findings intoquestion. In the univariate analysis we found asignificantly higher rate of PPI in the LACS group,after adjustment of THV type, we found no higher PPIrate in the matched population. Self-expanding THVswere more frequently used in patients undergoingLACS and are a known factor for PPI after TAVR.

Therefore, our data indicate that the anesthesiologicstrategy does not influence PPI rate, and rates of de-vice malposition and embolization were even lowerwith LACS. However, our data confirm higher rates ofvascular complications with LACS, indicating a po-tential influence of patient movement on the safety ofvascular access and closure.

The integration of periprocedural imaging in formof transesophageal echocardiography to early recog-nize procedural complications as well as to evaluatethe presence and degree of PVL (23) has been dis-cussed as a potential argument in favor of GA. Indeed,the absence of intraprocedural transesophagealechocardiography may be the putative mechanismbehind a higher incidence of PVL observed with LACSin several studies (5,18) and in a recent meta-analysis(6). Contrarily, in the present study we found nodifferences in PVL between the 2 anesthesiologicstrategies, arguing against the need for routinetransesophageal echocardiography during the pro-cedure and against a higher incidence of PVL withLACS.

FIGURE 6 Subgroup Analysis of Risk for 1-Year Mortality According to Local Anesthesia or Conscious Sedation Versus General Anesthesia

Hazard ratios adjusted for variables associated with 1-year mortality in the univariate analysis and for the propensity score to undergo transcatheter aortic valve

replacement in local anesthesia or conscious sedation (LACS) for the entire population (A) and the matched population (B). See Online Table 1 for results of the

multivariate analysis. Values are % (n). ASA ¼ American Society of Anesthesiologists; BMI ¼ body mass index; CCS ¼ Canadian Cardiovascular Society; CI ¼ confidence

interval; COPD ¼ chronic obstructive pulmonary disease; GA ¼ general anesthesia; HR¼ hazard ratio; LVEF ¼ left ventricular ejection fraction; NYHA ¼ New York Heart

Association; STS ¼ Society of Thoracic Surgeons.

Husser et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8

Conscious Sedation Versus General Anesthesia in TAVR M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8

576

Finally, with respect to mortality, all studiesconsistently have shown comparable results for LACSversus GA in 30-day and long-term mortality. How-ever, as most studies were performed in small pop-ulations or did not correct for baseline differences bypropensity matching, potential mortality differencesmight have been overlooked. In the 2 largest obser-vational studies to date, the Sentinel European TAVRPilot Registry and the FRANCE-2 (French Aortic Na-tional CoreValve and Edwards-2) Registry (5,15), nodifference regarding outcomes between LACS and GAwere observed. In a propensity-matched analysis of490 patients from the ADVANCE study, no differencein safety outcomes up to 2 years with LACS over GAwas reported (18). In the present study, we observed aclear benefit of LACS in terms of 30-day mortality inthe entire population of more than 16,000 patients.This benefit remained present even after rigorouscorrection for potential selection bias by means ofconventional multivariate adjustment and propensitymatching. Still, consistent with other studies(5,15,16,18), this early benefit did not translate into anadvantage in 1-year survival.

PATIENT SELECTION FOR LACS AND GA. Analyzingdata of more than 16,000 patients treated withTAVR in Germany during a time period of 4 yearsallowed us to further analyze the benefit of eachanesthesiologic strategy in important subgroupsof patients. Consistent to the entire populationanalysis, subgroup analyses also showed earlymortality benefit with LACS but no midtermadvantage in most subgroups. This observationlikely reflects the high comorbidity burden in TAVRpatients. However, it must be highlighted that ben-efits of LACS were more pronounced in women andpatients of older age, lower ASA class and withpulmonary hypertension. Especially women derivedbenefit to undergo TAVR in LACS, with a significantdecrease in risk for 1-year mortality. These findings,though necessitating confirmation in other studies,may guide further improvements in TAVR outcomesand guide the use of LACS.

STUDY LIMITATIONS. Although propensity matchingis an accepted approach to reduce selection bias inobservational studies (24), we cannot exclude the

PERSPECTIVES

WHAT IS KNOWN? TAVR can be performed in LACS or under

GA. Potential benefits of LACS in the context of faster proced-

ures and shorter hospital stays need to be evaluated with care.

WHAT IS NEW? In a large population of 16,543 patients

included in the GARY the use and impact of LACS versus GA in

TAVR were analyzed. LACS was associated with reduced post-

procedural complications and lower early mortality, suggesting

that its broader application in TAVR is safe.

WHAT IS NEXT? Additional randomized studies are warranted

to further elucidate the value of LACS versus GA in TAVR.

J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8 Husser et al.M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8 Conscious Sedation Versus General Anesthesia in TAVR

577

influence of residual bias on our results. There is awide range of anesthesiologic regimes that may beused in LACS or GA for TAVR with a potential in-fluence on procedural outcome, and the presentstudy does not account for these. In addition, con-version from LACS to GA is not routinely docu-mented in GARY, but when excluding patients withconversion to sternotomy, the results remainedunchanged (data not shown) and support the con-clusions of our study. Finally, as the present studyincludes patients undergoing TAVR from 2011 to2014, the impact of LACS on outcome as well ascost-effectiveness in contemporary cohorts, whichinclude a higher proportion of lower risk patients,is a pending issue that will need to be addressedin the future, ideally in form of randomizedcomparisons.

CONCLUSIONS

The present study underlines the safety and efficacyof LACS in TAVR, with a less complicatedpost-procedural course and lower early mortality, andconfirms logistic benefits justifying the broad appli-cation by experienced teams. Further randomized

studies are warranted to determine the true value ofLACS versus GA in TAVR.

ADDRESS FOR CORRESPONDENCE: PDDr.med.OliverHusser, Klinik für Herz- und Kreislauferkrankungen,Deutsches Herzzentrum München, Technische Uni-versität München, Lazarettstrasse 36, 80636 Munich,Germany. E-mail: oliver.husser@gmail.com.

RE F E RENCE S

1. Leon MB, Smith CR, Mack MJ, et al. Transcatheteror surgical aortic-valve replacement in intermediate-risk patients. N Engl J Med 2016;374:1609–20.

2. Hamm CW, Arsalan M, Mack MJ. The future oftranscatheter aortic valve implantation. Eur HeartJ 2016;37:803–10.

3. Ben-Dor I, Looser PM, Maluenda G, et al. Trans-catheter aortic valve replacement under monitoredanesthesia care versus general anesthesia withintubation. Cardiovasc Revasc Med 2012;13:207–10.

4. O’Sullivan KE, Bracken-Clarke D, Segurado R,et al. Is local anesthesia the optimum strategy inretrograde transcatheter aortic valve implanta-tion? A systematic review and meta-analysis.Thorac Cardiovasc Surg 2014;62:489–97.

5. Oguri A, Yamamoto M, Mouillet G, et al. Clinicaloutcomes and safety of transfemoral aortic valveimplantation under general versus local anes-thesia: subanalysis of the French Aortic NationalCoreValve and Edwards 2 registry. Circ CardiovascInterv 2014;7:602–10.

6. Maas EHA, Pieters BMA, Van de Velde M, Rex S.General or local anesthesia for TAVI? A systematicreview of the literature and meta-analysis. CurrPharm Des 2016;22:1868–78.

7. Beckmann A, Hamm C, Figulla HR, et al. TheGerman Aortic Valve Registry (GARY): a nation-wide registry for patients undergoing invasive

therapy for severe aortic valve stenosis. ThoracCardiovasc Surg 2012;60:319–25.

8. King G, Imai K, Stuart EA. MatchIt: nonpara-metric preprocessing for parametric causal infer-ence 2007. Available at: http://gking.harvard.edu/publications/matchit-Nonparametric-Preprocessing-Parametric-Causal-Inference?page¼17. AccessedApril 9, 2017.

9. Eltchaninoff H, Prat A, Gilard M, et al.Transcatheter aortic valve implantation: earlyresults of the FRANCE (French Aortic NationalCoreValve and Edwards) registry. Eur Heart J2011;32:191–7.

10. Durand E, Borz B, Godin M, et al. Transfemoralaortic valve replacement with the Edwards SAPIENand Edwards SAPIEN XT prosthesis using exclu-sively local anesthesia and fluoroscopic guidance:feasibility and 30-day outcomes. J Am Coll CardiolIntv 2012;5:461–7.

11. Covello RD, Ruggeri L, Landoni G, et al.Transcatheter implantation of an aortic valve:anesthesiological management. Minerva Anes-tesiol 2010;76:100–8.

12. DehédinB,GuinotP-G, IbrahimH,et al. Anesthesiaand perioperative management of patients who un-dergo transfemoral transcatheter aortic valve im-plantation: an observational study of general versuslocal/regional anesthesia in 125 consecutive patients.J Cardiothorac Vasc Anesth 2011;25:1036–43.

13. Motloch LJ, Rottlaender D, Reda S, et al.Local versus general anesthesia for transfemoralaortic valve implantation. Clin Res Cardiol 2012;101:45–53.

14. Yamamoto M, Meguro K, Mouillet G, et al.Effect of local anesthetic management withconscious sedation in patients undergoing trans-catheter aortic valve implantation. Am J Cardiol2013;111:94–9.

15. Dall’Ara G, Eltchaninoff H, Moat N, et al. Localand general anaesthesia do not influence outcomeof transfemoral aortic valve implantation. Int JCardiol 2014;177:448–54.

16. Attizzani GF, Alkhalil A, Padaliya B, et al.Comparison of outcomes of transfemoral trans-catheter aortic valve implantation using a mini-mally invasive versus conventional strategy. Am JCardiol 2015;116:1731–6.

17. D’Errigo P, Ranucci M, Covello RD, et al.Outcome after general anesthesia versus moni-tored anesthesia care in transfemoral trans-catheter aortic valve replacement. J CardiothoracVasc Anesth 2016;30:1238–43.

18. Brecker SJD, Bleiziffer S, Bosmans J, et al.Impact of anesthesia type on outcomes of trans-catheter aortic valve implantation (from themulticenter ADVANCE study). Am J Cardiol 2016;117:1332–8.

Husser et al. J A C C : C A R D I O V A S C U L A R I N T E R V E N T I O N S V O L . 1 1 , N O . 6 , 2 0 1 8

Conscious Sedation Versus General Anesthesia in TAVR M A R C H 2 6 , 2 0 1 8 : 5 6 7 – 7 8

578

19. Miles LF, Joshi KR, Ogilvie EH, et al. Generalanaesthesia vs. conscious sedation for trans-femoral aortic valve implantation: a single UKcentre before-and-after study. Anaesthesia 2016;71:892–900.

20. Mayr NP, Hapfelmeier A, Martin K, et al.Comparison of sedation and general anaesthesiafor transcatheter aortic valve implantation oncerebral oxygen saturation and neurocognitiveoutcome. Br J Anaesth 2016;116:90–9.

21. Bergmann L, Kahlert P, Eggebrecht H, Frey U,Peters J, Kottenberg E. Transfemoral aortic valve

implantation under sedation and monitoredanaesthetic care—a feasibility study. Anaesthesia2011;66:977–82.

22. Covello RD, Landoni G, Zangrillo A. Anestheticmanagement of transcatheter aortic valve im-plantation. Curr Opin Anaesthesiol 2011;24:417–25.

23. Hahn RT, Little SH, Monaghan MJ, et al.Recommendations for comprehensive intra-procedural echocardiographic imagingduring TAVR. J Am Coll Cardiol Img 2015;8:261–87.

24. Elze MC, Gregson J, Baber U, et al. Compari-son of propensity score methods and covariateadjustment: evaluation in 4 cardiovascular studies.J Am Coll Cardiol 2017;69:345–57.

KEY WORDS conscious sedation, generalanesthesia, mortality, outcome,transcatheter aortic valve replacement

APPENDIX For supplemental tables andfigures, please see the online version of this paper.