Surgical sutureless and interventional transcatheter aortic valve prostheses are nowadays extensively adopted in high-risk elderly patients. An explorative analysis was carried out to compare the clinical outcome and costs associated to these approaches. Since 2010, a total of 626 patients were distributed between transcatheter aortic valve implantation (TAVI; n = 364) and sutureless (n = 262) groups. Patients of both groups were not comparable for clinical and surgical characteristics, but many patients were in a “gray zone”; therefore, a retrospective propensity score analysis was possible and performed. For the matched pair samples, postoperative, follow-up clinical data, and costs data were obtained. In-hospital death occurred in 5 patients in sutureless group and 3 patients in TAVI group (p = 0.36). Blood transfusions were higher in sutureless group (2.1 ± 2.3 vs 0.4 ± 1.0 U). TAVI group had a shorter intensive care unit and hospital stay (2.2 ± 2.7 vs 3.2 ± 3.5 days, p = 0.037; 12 ± 6 vs 14 ± 6 days, p = 0.017). No differences in postoperative neurologic (p = 0.361), renal (p = 0.106), or respiratory (p = 0.391) complications were observed between groups. At follow-up (24.5 ± 13.8 months), 1 patient in sutureless group and 7 patients in TAVI group died (p = 0.032). Paravalvular leakage occurred more frequently in patients in TAVI group (35 [34%] vs 7 [6.9%]; p <0.001) with an impact on follow-up survival rate. The costs associated to the 2 procedures are similar when the cost of the device was excluded (p = 0.217). When included, the sutureless approach resulted a cost saving (€22,451 vs €33,877, p <0.001). In conclusion, the patients in the “gray zone” record a satisfying clinical outcome after sutureless surgery and TAVI. Patients in the sutureless group endure more hospital complications, but TAVI entails a higher follow-up mortality. On the costs aspects, TAVI technologies are more expensive, and it reflects on higher overall hospital costs.
Transcatheter aortic valve implantation (TAVI) has emerged as an alternative treatment to conventional surgery for patients of advanced age who are deemed inoperable. However, studies comparing TAVI with conventional surgical aortic valve replacement (AVR) in elderly patients showed that isolated advanced age per se should not be considered an indication for TAVI. In addition, the use of new sutureless aortic bioprostheses that allow shorter cardiopulmonary bypass (CPB) and cross-clamp times has proved to be associated with good outcome in octogenarians. All these findings together suggest that recently developed surgical and interventional techniques may also be adopted in high-risk elderly patients. This issue has a significant economic relevance for health care systems, given the high costs of the new devices and the limited life expectancy of this patient population. The aim of this observational study was to evaluate the clinical outcome and the associated costs of sutureless AVR versus TAVI in patients with aortic stenosis using propensity score matching analysis to generate 2 homogeneous groups with respect to preoperative surgical risk.
Methods
We collected data of all patients with a diagnosis of severe aortic valve stenosis and an indication for surgery in our center since 2010. Two specific programs were initiated in our institution at the same time: the first program was developed in collaboration between cardiologists and cardiac surgeons for the use of TAVI (Sapien, Sapien XT, and Sapien 3; Edwards Lifesciences Inc., Irvine, California; CoreValve; Medtronic, Minneapolis, Minnesota; Acurate TA; Symetis SA, Ecublens, Switzerland), whereas the second program involved the use of the Perceval sutureless bioprosthesis (Sorin Group, Saluggia, Italy). Every week, during an interdepartmental conference, we evaluated all patients with severe aortic valve stenosis referred to our center from peripheral hospitals, private practices, or our emergency department, considering co-morbidities and surgical risk to determine the best therapy. In all patients aged >65 years with an indication for isolated AVR considered candidates for surgery (irrespective of EuroSCORE), a low frailty score (evaluated by clinical inspection and other factors not included in the EuroSCORE or Society of Thoracic Surgeons scoring system) and compatible echocardiographic findings, a Perceval sutureless valve was implanted. An informed consent for the use of personal data and follow-up contact was also signed by all patients. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution’s Human Research Committee.
All patients with frailty factors judged at very high surgical risk with or without a logistic EuroSCORE >20% underwent a TAVI. The transfemoral approach was considered as the first-line strategy, leaving the transapical approach in case of inadequate vascular access. After 5 years of extensive experience with both procedures, a total of 626 patients were operated on, distributed between the TAVI and sutureless groups (n = 364 and n = 262, respectively). Patients of both groups were not comparable for clinical and surgical characteristics, but many patients were “similar”; therefore, a retrospective propensity score analysis was possible and performed. For the matched pair samples, postoperative and follow-up clinical data were obtained.
Prosthetic valve function was evaluated with transthoracic echocardiography. The presence of paravalvular regurgitation was defined according to current guidelines as none or trace, mild, moderate, or severe. All echocardiographic examinations were performed by echocardiographers of our cardiology department. The Perceval implantation technique has been described previously. General anesthesia with endotracheal intubation was used in both groups. Our heart team prefers this technique because it allows performance of intraoperative transesophageal echocardiography. In patients who underwent TAVI by the transfemoral and transapical route, the endotracheal tube was removed immediately after the procedure if appropriate, and these patients were discharged to an intermediate care unit. In the surgical group, patients were extubated in the intensive care unit (ICU). Transfemoral procedures were performed through a minimally invasive direct vascular access: the access site was chosen according to computed tomographic findings, size of the common femoral artery, amount of calcification of the vessel and iliac arteries, and tortuosity. Before all TAVI procedures, the cardiac apex and intended optimal coaxial alignment were localized by transthoracic echocardiography. The surgical technique for positioning and deploying the Sapien XT valve prosthesis, our main TAVI prosthesis used in 92% of our TAVI population, has been well described and standardized for both the transapical and transfemoral approaches.
In-hospital preprocedural and postprocedural resource use data were retrospectively collected from electronic patient records. Hospital financial department provided the associated costs identified according to the German diagnostic related group matrix. The costs analysis was performed according to 10 cost centers, including normal ward, ICU, dialysis unit, operating room, anesthesia, cardiac diagnostic therapy, endoscopic diagnostics, radiology, laboratory, and other diagnostics, and the 3 main cost element groups of labor, material, and infrastructure.
Costs were subsequently aggregated and reported according to three main categories: (1) operating room, including anesthesia; (2) hospital stay, including ICU and cardiac surgical ward; (3) diagnostic, radiology, and laboratory, including cardiac diagnostic therapy, endoscopic diagnostics, radiology, laboratory, and other diagnostics. A mean cost of the device was calculated from the observations included in this data set, starting from 2009 to 2013. TAVI mean cost is threefold compared with Perceval mean cost, and this is in line with the figures reported by Biancari et al where the device cost difference between TAVI and Perceval was even higher.
Categorical variables were summarized as frequencies (%), and continuous variables were summarized as mean ± SD. Normal distribution of data was assessed with the Shapiro–Wilk test. Continuous variables were compared by paired Student t -test and categorical variables by chi-square test. A propensity score matching (1:1) was performed to control selection bias as a result of nonrandom assignment to the groups. The propensity score was defined as the probability of receiving TAVI. This was estimated by means of a multivariate regression analysis. Twenty-two patient characteristics and major preoperative risk factors were entered into the model. The p value of the Hosmer–Lemeshow test was 0.014, and c -statistic for the fitted logistic regression model was 0.867 (p <0.001), indicating that model fitting was excellent. Pairs were generated with the 5:1 digit-matching approach. Finally, postoperative outcomes of the resulting 102 matched pairs of both groups (TAVI and sutureless AVR) with the same propensity score were compared, including in-hospital and valve-related complications and survival rate. Cumulative survival curves were computed according to the Kaplan–Meier method. Survival curves and freedom from valve-related complications were compared with a Cox proportional hazards model stratified for the matched pairs. A cost comparison between patients treated with Perceval and TAVI on mean and SD were provided. A conditional logistic test was run to detect a statistical significant difference in costs between the 2 cohorts.
Results
Preoperative characteristics of the whole study population and of the matched groups are reported in Tables 1 and 2 , respectively.
| Variable | Sutureless (n =262) | TAVI (n =364) | P value |
|---|---|---|---|
| Age (years) | 78±5 | 82±6 | <0.001 |
| Logistic EuroSCORE | 13±11 | 26±17 | <0.001 |
| Height (cm) | 166±10 | 165±13 | 0.36 |
| Weight (kg) | 77±15 | 74±16 | 0.009 |
| LVEF (%) | 59±9 | 54±14 | <0.001 |
| NYHA class | 2.8±0.5 | 3.1±0.4 | <0.001 |
| Pulmonary hypertension class ∗ | 0.5±0.7 | 0.8±0.9 | 0.001 |
| Hypertension | 228 (87%) | 339 (93%) | 0.01 |
| Male gender | 122 (47%) | 158 (43%) | 0.24 |
| Pacemaker | 9 (3.4%) | 12 (3.3%) | 1 |
| Redo | 24 (9%) | 81 (22%) | <0.001 |
| Poor mobility | 48 (18%) | 75 (21%) | 0.3 |
| Peripheral arterial disease | 61 (23%) | 98 (27%) | 0.18 |
| Diabetes | 76 (29%) | 143 (39%) | 0.002 |
| Diabetes managed with insulin therapy | 17 (6.5%) | 64 (18%) | <0.001 |
| CCS class 4 | 3 (1.1%) | 90 (25%) | <0.001 |
| ACS <90 days | 3 (1.1%) | 4 (1.1%) | 0.59 |
| COPD | 43 (16%) | 105 (29%) | <0.001 |
| Renal insufficiency † | 48 (18%) | 171 (47%) | <0.001 |
| Associated procedure | 88 (34%) | 1 (0.2%) | <0.001 |
| CABG | 67 (76%) | 1 (0.2%) | |
| Surgical approach | |||
| Full sternotomy | 94 (36%) | ||
| Partial sternotomy | 159 (61%) | ||
| Right thoracotomy | 9 (3.4%) | ||
| Transapical | 158 (43%) | ||
| Transfemoral | 205 (56%) | ||
| Transaortic | 1 (0.2%) | ||
| Prosthesis size (mm) | 24±1.6 | 25±2.2 | |
| Cross-clamp time (min) | 44±18 | ||
| CPB time (min) | 73±26 |
∗ 0: ≤30 mm Hg; 1: 31 to 55 mm Hg; 2: >55 mm Hg.
| Variable | Sutureless (n=102) | TAVI (n=102) | P value |
|---|---|---|---|
| Age (years) | 80±4 | 79±7 | 0.27 |
| Logistic EuroSCORE | 17±14 | 18±11 | 0.86 |
| Height (cm) | 165±9 | 166±10 | 0.75 |
| Weight (kg) | 76±15 | 76±16 | 0.79 |
| LVEF (%) | 58±9 | 54±14 | 0.09 |
| NYHA class | 3.0±0.4 | 3.0±0.5 | 0.88 |
| Pulmonary hypertension class ∗ | 0.6±0.8 | 0.6±0.8 | 0.86 |
| Hypertension | 88 (86%) | 94 (92%) | 0.13 |
| Male gender | 42 (41%) | 44 (43%) | 0.44 |
| Pacemaker | 6 (6.5%) | 6 (6.5%) | 1 |
| Redo | 13 (13%) | 14 (14%) | 0.50 |
| Poor mobility | 14 (14%) | 15 (15%) | 0.50 |
| Peripheral arterial disease | 27 (26%) | 16 (16%) | 0.08 |
| Diabetes | 40 (39%) | 37 (36%) | 0.56 |
| Diabetes managed with insulin therapy | 15 (15%) | 13 (13%) | 0.46 |
| CCS class 4 | 2 (2%) | 4 (4%) | 0.34 |
| ACS <90 days | 1 (1%) | 1 (1%) | 1 |
| COPD | 22 (22%) | 24 (23%) | 0.43 |
| Renal insufficiency † | 34 (33%) | 33 (32%) | 0.66 |
∗ 0: ≤30 mm Hg; 1: 31 to 55 mm Hg; 2: >55 mm Hg.
Intraoperative complications occurred in both groups (4 patients [3.9%] in the sutureless group vs 3 patients [2.9%] in the TAVI group; p = 0.5). In particular, in the sutureless group, 1 patient required removal of the prosthetic valve, performed using the “X-movement” technique with subsequent reimplantation; 1 patient with previous coronary revascularization experienced vein graft damage requiring suture repair; 1 patient developed severe paravalvular leakage with annular rupture requiring removal of the sutureless bioprosthesis followed by implantation of a stented valve; and 1 patient, operated on through a right lateral minithoracotomy, required conversion to longitudinal sternotomy due to right ventricular damage and ventricular fibrillation not amenable to cardioversion. In the TAVI group, 1 patient experienced aortic injury during manipulation of the guidewire requiring a thoracotomy to control bleeding; 1 patient experienced valve migration into the left ventricle requiring CPB and conventional AVR through partial sternotomy; and 1 patient with refractory ventricular fibrillation required prolonged cardiopulmonary resuscitation and circulatory support using CPB.
The rate of pacemaker implantation was similar between groups (10 patients [10%] in the sutureless group vs 9 patients [9.4%] in the TAVI group; p = 0.588). Conversely, the need for blood transfusion was significantly higher in the sutureless group than that in the TAVI group (2.1 ± 2.3 vs 0.4 ± 1.0 U). Patients who underwent TAVI had a significantly shorter ICU and hospital stay than patients who underwent sutureless AVR (ICU: 2.2 ± 2.7 vs 3.2 ± 3.5 days, p = 0.037; hospital: 12 ± 6 vs 14 ± 6 days, p = 0.017).
No differences in postoperative neurologic (sutureless vs TAVI: 3 [2.9%] vs 5 [4.9%] patients; p = 0.361), renal (5 [4.9%] vs 1 [1.0%] patients; p = 0.106), or respiratory (6 [5.9%] vs 8 [7.8%] patients; p = 0.391) complications were observed between groups. Postoperative delirium developed more frequently in patients from the sutureless group (10 [9.8%] vs 2 [2.0%] patients; p = 0.017), whereas major vascular complications occurred more frequently in patients from the TAVI group (5 [4.9%] vs 0 [0%] patients; p = 0.03).
In-hospital death occurred in 5 and 3 patients of the sutureless and TAVI groups, respectively (p = 0.36). At follow-up (sutureless vs TAVI: 26 ± 13 vs 23 ± 14 months; p = 0.205), 1 patient of the sutureless group and 7 patients of the TAVI group died (p = 0.032). Overall survival rate of the whole study population was 96% (sutureless vs TAVI: 99 vs 93%; Figure 1 ). Paravalvular leakage (at least 1/4+) occurred more frequently in patients from the TAVI group (35 [34%] vs 7 [7%] patients; p <0.001) with an impact on follow-up survival rate ( Figure 2 ).
