Transcatheter aortic valve implantation (TAVI) has become the standard of care for many patients with symptomatic severe aortic stenosis who are at increased risk of morbidity and mortality during surgical aortic valve replacement. However, there is still no general consensus regarding the use of general anesthesia (GA) versus local anesthesia with sedation (non-GA) during the TAVI procedure. Using propensity score–matching analysis, we analyzed the characteristics and outcomes of patients who underwent TAVI with either GA (n = 245) or non-GA (n = 245) in the fully monitored, international, CoreValve ADVANCE Study. No statistically significant differences existed between the non-GA and GA groups in all-cause mortality (25.4% vs 23.9%, p = 0.78), cardiovascular mortality (16.4% vs 16.6%, p = 0.92), or stroke (5.2% vs 6.9%, p = 0.57) through 2-year follow-up. Major vascular complications were more common in the non-GA group. Total hospital stay was similar between the 2 groups. Conversion from non-GA to GA occurred in 13 patients (5.3%) because of procedural complications in 9 patients and discomfort or restlessness in 4 patients. Most procedural complications were related to valve positioning or vascular issues. Two of the 13 converted patients died during the procedure. Both GA and non-GA are widely used in real-world TAVI practice, and the decision appears to be guided by only a few patient-related factors and dominated by local and national practice. The outcomes of both anesthesia modes are equally good. When conversion from non-GA did occur, the complication requiring GA affected outcomes.
Transcatheter aortic valve implantation (TAVI) has become standard of care for patients with symptomatic severe aortic stenosis at extreme or high risk for surgery. In practice, even low-risk patients are already being treated, whereas at least 3 TAVI clinical trials are assessing the role of the therapy in patients considered at only intermediate risk from surgical aortic valve replacement (AVR). It is, therefore, likely the number of patients treated with TAVI will increase, requiring additional numbers of operators and hospitals. Concurrently, the procedure is becoming less complex. Smaller sheath sizes, a reduced need for rapid pacing and balloon valvuloplasty, availability of repositionable and recapturable valves, and decreased reliance on intraprocedural transesophageal echocardiography will herald a new era of TAVI. A significant proportion of procedures are already being performed using local anesthesia with sedation. Others have reported potential benefits of using local anesthesia, including shorter intensive care unit and overall hospital stays, less hemodynamic instability, and less need for vasopressors. It is likely that the proportion of patients treated in this manner will increase. In the ADVANCE study, patients were treated according to best local practice in experienced centers, and a significant proportion was treated with local rather than general anesthesia (GA). Local anesthesia was used in approximately 50% of patients in this study, reflecting the real-world practice at the time, and we compared patient characteristics and procedural outcomes in patients administered local anesthesia versus GA for TAVI.
Methods
For this report, we analyzed the characteristics and outcomes of patients who underwent TAVI with either GA or local anesthesia with sedation (non-GA) in the Medtronic CoreValve ADVANCE study. Patients treated through the direct aortic approach were excluded from this analysis. The design, methods, and primary results of the ADVANCE study have been previously described. Briefly, the ADVANCE study is a prospective, fully monitored, nonrandomized, international, multicenter study evaluating the acute and long-term results of implantation of the Medtronic CoreValve System (Medtronic, Minneapolis, Minnesota) in “real-world” patients with severe, symptomatic aortic stenosis who were considered to have an inoperable condition or to be at high risk for conventional AVR. All ADVANCE study centers were required to have performed a minimum of 40 TAVI procedures before joining the study and to use an on-site, multidisciplinary heart team consisting of at least 1 TAVI-experienced interventional cardiologist and 1 cardiovascular surgeon.
The ethics committee at each study center approved the ADVANCE investigational protocol. ADVANCE was conducted in adherence to the Declaration of Helsinki, and all patients provided written informed consent before the CoreValve implantation procedure.
Detailed device description and implant procedures for the CoreValve System have been previously described. The procedures were performed according to standard local hospital practices, which included the selection of access location (transfemoral or subclavian), the type of access (surgical cutdown or completely percutaneous), and the type of anesthesia (GA or non-GA). Procedural characteristics analyzed for the comparisons between anesthesia groups included access type and site, procedure duration, fluoroscopy time, quantity of contrast agent used, procedural complications, and length and type of hospital stay.
Safety outcomes were analyzed at 30 days and at 1 and 2 years postprocedure and included all-cause mortality, cardiovascular mortality, myocardial infarction, reintervention, stroke, stroke or transient ischemic attack, bleeding, vascular complications, acute kidney injury (stage III), and pacemaker implantation.
Death, stroke, myocardial infarction, and reintervention were adjudicated by an independent Clinical Events Committee consisting of TAVI-experienced interventional cardiologists and a cardiac surgeon using the initial Valve Academic Research Consortium definitions. An independent neurologist reviewed the neurologic events and provided a summary of each event to the Clinical Events Committee, which used this information along with any other patient source data to adjudicate all neurologic events. A core laboratory (Cardialysis, Rotterdam, The Netherlands) performed a systematic review and assessment of procedural angiograms and electrocardiograms through 1-year follow-up. Data were recorded on a standardized electronic case report form and sent to a central database (Merge, Chicago, Illinois) over the Internet.
Categorical variables are reported as counts and percentages. Continuous variables are reported as mean and SD except for non-normal data, such as logistic EuroSCORE, Society of Thoracic Surgeons predictive risk of mortality score, procedure duration, fluoroscopy time, amount of contrast given, and length of stay, which are summarized using medians and interquartile ranges.
Comparisons between anesthesia types are based on the chi-square or Fisher’s exact tests for categorical variables and t tests or Wilcoxon tests for continuous variables, as appropriate. Event rates were generated using the Kaplan-Meier method, and log-rank tests were used for group comparisons. For patients without an event, the date of censoring was the latest date of all follow-up visits (including study exit) and events (including death). A p value <0.05 was considered statistically significant.
To identify 2 comparable groups of patients who underwent GA or non-GA, we performed a propensity score–matching analysis. A multivariable logistic regression model with anesthesia type as the outcome was fit, from which predicted probabilities (i.e., propensity scores) were computed for each patient. Unbalanced variables before matching and an additional 10 variables were included in the model to achieve balance in baseline characteristics in the anesthesia groups after matching. The baseline covariates included in the model were women, New York Heart Association class III or IV, diabetes mellitus, previous median sternotomy, previous aortic valve intervention, previous coronary artery bypass grafting, history of aortic aneurysm, creatinine clearance <20 ml/min, baseline left ventricular ejection fraction, moderate-or-severe tricuspid regurgitation, log-transformed age, square root–transformed EuroSCORE, history of myocardial infarction, peripheral vascular disease, baseline pacemaker, cerebrovascular disease, and atrial fibrillation. Characteristics were considered to be in balance if the percent standardized difference was <10%. All statistical analyses were performed using SAS software (version 9.3; SAS Institute, Inc., Cary, North Carolina).
Results
From March 2010 to July 2011, 1,015 patients were enrolled in the ADVANCE study. Of these, 996 patients had undergone an attempted implantation with the CoreValve System. The mode of anesthesia was entirely site selected and guided by best and customary local practice. Considering the whole group of patients, non-GA was used in 551 patients (55.3%) and GA was used in 445 patients (44.7%). Twenty-one patients treated through the direct aortic access were omitted as they are not considered candidates for both anesthesia options, leaving 424 patients in the GA group. Baseline patient characteristics are presented in Table 1 . Significant differences existed between the 2 patient groups in diabetes, previous median sternotomy, previous aortic valve intervention, previous coronary artery bypass grafting, and left ventricular ejection fraction. Despite these differences, the median Society of Thoracic Surgeons predictive risk of mortality score and logistic were similar between the 2 groups. There were national differences in the use of non-GA versus GA ( Figure 1 ), demonstrated by large differences in the use of non-GA among the highest recruiting countries (Germany 78.6%, Italy 70.5%, and the United Kingdom 4.6%).
| Characteristic | Before Matching | After Matching | ||||||
|---|---|---|---|---|---|---|---|---|
| Non-GA (N = 551) | GA (N = 424) | Std. Diff. ∗ (%) | P Value | Non-GA (N = 245) | GA (N = 245) | Std. Diff. ∗ (%) | P Value | |
| Age (years) | 80.8 ± 6.5 | 81.3 ± 6.4 | 7.8 | 0.23 | 81.3 ± 6.2 (245) | 81.6 ± 6.5 (245) | 5.4 | 0.55 |
| Female | 292 (53.0%) | 204 (48.1%) | 9.8 | 0.13 | 126 (51.4%) | 130 (53.1%) | 3.3 | 0.72 |
| STS predictive risk of mortality score, % | 5.2 (N=550) [3.6, 7.8] | 5.2 [3.5, 7.5] | 4.8 | 0.46 | 5.3 [3.7, 7.0] | 5.2 [3.6,7.6] | 3.2 | 0.72 |
| Logistic EuroSCORE, % | 16.0 (N=550) [10.7, 25.6] | 16.0 [10.1, 24.6] | 4.6 | 0.47 | 16.1 [10.7, 24.7] | 16.3 [10.2, 24.9] | 2.6 | 0.78 |
| New York Heart Association class III or IV | 447/546 (81.9%) | 317/412 (76.7%) | 12.2 | 0.06 | 194 (79.2%) | 193 (78.8%) | 1.0 | 0.91 |
| Diabetes mellitus | 185/544 (34.0%) | 116/421 (27.6%) | 14.0 | 0.03 | 60 (24.5%) | 59 (24.1%) | 1.0 | 0.92 |
| Coronary artery disease | 305/550 (55.5%) | 253/422 (60.0%) | 9.1 | 0.16 | 129 (52.7%) | 136 (55.7%) | 6.2 | 0.49 |
| Previous myocardial infarction | 95/540 (17.6%) | 60/410 (14.6%) | 8.1 | 0.22 | 34 (13.9%) | 37 (15.1%) | 3.5 | 0.70 |
| Previous percutaneous coronary intervention | 170/545 (31.2%) | 132/420 (31.4%) | 0.5 | 0.94 | 73/243 (30.0%) | 76 (31.0%) | 2.1 | 0.81 |
| Previous median sternotomy | 63 (11.4%) | 102/421 (24.2%) | 33.9 | < 0.001 | 37 (15.1%) | 38 (15.5%) | 1.1 | 0.90 |
| Previous aortic valve intervention | 11/550 (2.0%) | 29/423 (6.9%) | 23.8 | <0.001 | 7 (2.9%) | 8 (3.3%) | 2.4 | 0.79 |
| Previous coronary artery bypass grafting | 92/548 (16.8%) | 111/423 (26.2%) | 23.2 | <0.001 | 46 (18.8%) | 46 (18.8%) | 0.0 | >0.99 |
| Cerebrovascular disease | 75/544 (13.8%) | 53/415 (12.8%) | 3.0 | 0.65 | 31 (12.7%) | 29 (11.8%) | 2.5 | 0.78 |
| Aortic aneurysm | 9/548 (1.6%) | 13/420 (3.1%) | 9.6 | 0.13 | 7 (2.9%) | 8 (3.3%) | 2.4 | 0.79 |
| Peripheral vascular disease | 100/547 (18.3%) | 79/420 (18.8%) | 1.4 | 0.83 | 38 (15.5%) | 44 (18.0%) | 6.6 | 0.47 |
| Chronic obstructive pulmonary disease | 133/549 (24.2%) | 86/422 (20.4%) | 9.3 | 0.16 | 53/244 (21.7%) | 57 (23.3%) | 3.7 | 0.68 |
| Creatinine clearance <20 ml/min | 69/542 (12.7%) | 64/414 (15.5%) | 7.8 | 0.23 | 33 (13.5%) | 32 (13.1%) | 1.2 | 0.89 |
| Atrial fibrillation | 184/545 (33.8%) | 139/421 (33.0%) | 1.6 | 0.81 | 86 (35.1%) | 79 (32.2%) | 6.0 | 0.50 |
| Permanent pacemaker | 66 (12.0%) | 58 (13.7%) | 5.1 | 0.43 | 28 (11.4%) | 31 (12.7%) | 3.8 | 0.68 |
| Pulmonary hypertension | 75/532 (14.1%) | 45/398 (11.3%) | 8.4 | 0.21 | 32/241 (13.3%) | 34/233 (14.6%) | 3.8 | 0.68 |
| Prior porcelain aorta | 23/548 (4.2%) | 16/422 (3.8%) | 2.1 | 0.75 | 11/244 (4.5%) | 7 (2.9%) | 8.8 | 0.33 |
| Prior cirrhosis of the liver | 6/550 (1.1%) | 4/422 (0.9%) | 1.4 | >0.99 | 4 (1.6%) | 3/244 (1.2%) | 3.4 | >0.999 |
| Prior right ventricular insufficiency | 23/549 (4.2%) | 17/414 (4.1%) | 0.4 | 0.95 | 12/244 (4.9%) | 8/243 (3.3%) | 8.2 | 0.37 |
| Effective orifice area ( cm 2 ) | 0.7 ± 0.2 (N=441) | 0.7 ± 0.4 (N=340) | 8.0 | 0.28 | 0.7 ± 0.2 (N=213) | 0.7 ± 0.3 (N=218) | 9.0 | 0.35 |
| Mean aortic valve gradient (mm Hg) | 45.8 ± 15.1 (N=502) | 44.9 ± 16.0 (N=363) | 6.1 | 0.37 | 45.2 ± 14.7 (N=241) | 46.9 ± 16.7 (N=231) | 10.9 | 0.24 |
| Left ventricular ejection fraction (%) | 54.8 ± 14.1 (N=464) | 51.3 ± 13.3 (N=370) | 25.9 | <0.001 | 51.6 ± 13.8 | 52.7 ± 12.6 | 8.6 | 0.34 |
| Left ventricular ejection fraction <35% | 42/464 (9.1%) | 40/370 (10.8%) | 5.9 | 0.40 | 28 (11.4%) | 22 (9.0%) | 8.1 | 0.37 |
| Moderate or severe mitral regurgitation | 148/534 (27.7%) | 115/408 (28.2%) | 1.0 | 0.87 | 72/244 (29.5%) | 74/244 (30.3%) | 1.8 | 0.84 |
| Moderate or severe tricuspid regurgitation | 103/493 (20.9%) | 63/384 (16.4%) | 11.5 | 0.09 | 49 (20.0%) | 44 (18.0%) | 5.2 | 0.57 |
∗ A percent standardized difference >10% indicates imbalance between groups.
Because several statistically significant differences in baseline patient characteristics existed between the GA and non-GA groups that could have potentially affected the results of the analysis, we performed a propensity score–matched analysis. A standardized difference of 10% was used as the basis for defining successful matching, where a lower standardized difference corresponds to higher degree of achieved balance. Propensity scoring resulted in 245 matched pairs of patients ( Table 1 ). All the following outcomes analyses are based on these 2 propensity-matched anesthesia groups.
Procedural characteristics and outcomes are listed in Table 2 . The vast majority of cases were performed transfemorally. Patients treated using GA had significantly longer median procedure and fluoroscopy times. More patients implanted through the percutaneous approach were treated with non-GA compared with GA, whereas more patients who had surgical cutdown were treated with GA ( Table 2 ). Thus, the method of access may have affected choice of anesthesia. No statistically significant differences were seen in procedural complications. Total hospital stay was similar between the groups.
| Characteristic | Non-GA (N = 245) | GA (N = 245) | P Value |
|---|---|---|---|
| Access type | |||
| Percutaneous | 217 (88.6%) | 195 (79.6%) | 0.007 |
| Surgical cutdown | 28 (11.4%) | 55 (20.4%) | 0.007 |
| Access site | |||
| Transfemoral | 239 (97.6%) | 203 (82.9%) | <0.001 |
| Subclavian/axillary | 6 (2.4%) | 42 (17.1%) | <0.001 |
| Periprocedural outcomes | |||
| Procedure duration (min) | 65.0 (N=230) [48.0, 90.0] | 83.0 (N=206) [60.0, 108.0] | <0.001 |
| Fluoroscopy time during entire procedure (min) | 18.0 (N=229) [13.0, 27.0] | 20.0 (N=222) [15.0, 28.0] | 0.03 |
| Amount of contrast given (cc) | 181.0 (N=243) [140.0, 230.0] | 185.0 (N=233) [148.0, 240.0] | 0.73 |
| Procedural complications | |||
| Annulus rupture | 0 (0.0) | 0 (0.0) | — |
| Valve embolization | 0 (0.0) | 1 (0.4) | >0.99 |
| Conversion to open aortic valve repair | 0 (0.0) | 0 (0.0) | — |
| Coronary compromised | 0/221 (0.0%) | 1/211 (0.5%) | 0.49 |
| Procedural death | 2 (0.8%) | 3 (1.2%) | >0.99 |
| Total hospital stay (days) | 9.0 (N=241) [6.0, 13.0] | 9.0 (N=219) [6.0, 12.0] | 0.95 |
| Intensive care (days) | 2.0 (N=245) [1.0, 4.0] | 2.0 (N=241) [1.0, 4.0] | 0.41 |
| Non–intensive care (days) | 6.0 (N=241) [3.0, 9.0] | 6.0 (N=220) [3.0, 9.0] | 0.72 |
Conversion from non-GA to GA occurred in 13 patients during their procedure. A total of 20 procedural complications occurred in 9 of the 13 patients. The remaining 4 converted patients did not experience a procedural complication and, thus, were most likely converted to GA because of discomfort or restlessness. Most procedural complications were related to valve positioning or vascular issues. The valve was repositioned with snare or retrieved in 3 patients; failure of the vessel closure device requiring surgery occurred in 3 patients; access vessel occlusion (treated with percutaneous balloon) occurred in 1 patient, access vessel perforation (required transfusion) occurred in 1 patient, and 1 patient experienced hemorrhage requiring transfusion and cardiorespiratory arrest. Two of the 13 converted patients died during the TAVI procedure.
Safety outcomes at 30 days and at 1 and 2 years are presented in Table 3 . All-cause mortality ( Figure 2 ), cardiovascular mortality, and stroke were similar between the patients who underwent GA and non-GA through 2 years of follow-up. However, patients who underwent local anesthesia with sedation had significantly higher incidence of major vascular complications at all time points.
