5 Safety

This section describes safety outcomes from the published literature that the committee considered as part of the evidence about this procedure. For more detailed information on the evidence, see the interventional procedure overview.

Evidence is based only on studies that reported outcomes by the risk level of the patients.

All-cause mortality and cardiovascular mortality within 30 days

5.1 A randomised controlled trial (RCT) of 358 patients for whom surgical aortic valve replacement (SAVR) was unsuitable (PARTNER 1B) compared transcatheter aortic valve implantation (TAVI; n=179) with medical management (n=179). There were no statistically significant differences in all-cause mortality (5% [9/179] compared with 3% [5/179], p=0.41) and cardiovascular mortality (5% [8/179] compared with 2% [3/179], p=0.22) between TAVI and medical management at 30‑day follow-up.

5.2 In an RCT of 699 patients for whom SAVR was suitable but high risk (PARTNER 1A; TAVI, n=348 compared with SAVR, n=351) there were no statistically significant differences in all-cause mortality (3% [12/348] compared with 7% [22/351], p=0.07) and cardiovascular mortality (3% [11/348] compared with 13% [10/251], p=0.90) between the TAVI group and SAVR group at 30‑day follow-up. In another RCT of 795 patients (the US CoreValve trial; TAVI, n=394 compared with SAVR, n=401) there were also no statistically significant differences in all-cause mortality (3% [13/390] compared with 5% [16/357], p=0.43) and cardiovascular mortality (3% [12/390] compared with 5% [16/357], p=0.32) between the TAVI group and SAVR group at 30‑day follow-up. When data were pooled for both studies, the risk ratio (less than 1 favours TAVI) for all-cause mortality was 0.64 (95% confidence interval [CI] 0.38 to 1.39, p=0.06) and for cardiovascular mortality was 0.90 (95% CI 0.52 to 1.56, p=0.70).

5.3 In an RCT of 2,032 patients for whom SAVR was suitable but intermediate or low risk (TAVI, n=1,011 compared with SAVR, n=1,021) there was a non-significant lower all-cause mortality (3% compared with 4%, p=0.24) and cardiovascular mortality (2% compared with 3%, p=0.72) for TAVI using the femoral route compared with SAVR at 30-day follow-up. For the transthoracic route all-cause mortality (6% compared with 4%, p=0.21) and cardiovascular mortality (5% compared with 3%, p=0.47) were not significantly different. In another RCT of 280 low- and intermediate-risk patients (TAVI, n=145 compared with SAVR, n=135), all-cause mortality (2% [3/142] compared with 3% [5/134], p=0.43) and cardiovascular mortality (2% [3/142] compared with 4% [5/134], p=0.43) were not significantly different. A systematic review of 3,179 patients (with risk scores of 8% or less in 4 RCTs) also reported a non-significant lower all-cause mortality for TAVI compared with SAVR (OR 0.67; 95% CI 0.42 to 1.07).

Cerebral complications

5.4 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, TAVI compared with medical management) the hazard ratio for stroke or transient ischaemic attack (TIA) was significantly higher in the TAVI group (hazard ratio [HR] 2.81; 95% CI 1.26 to 6.26, p=0.004) at 3-year follow-up, whereas at 5-year follow-up there were no significant differences between the TAVI and medical management groups (HR 1.39; 95% CI 0.62 to 3.11, p=0.555).

5.5 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve trial [n=795]) in patients for whom SAVR was suitable but high risk, the incidence of stroke and TIA was reported. Both pooled and individual risk ratios from PARTNER 1A and the CoreValve trial showed no statistically significant differences in the incidence of all stroke in patients for whom SAVR was suitable but high risk at 30 days (risk ratio [RR] 1.26; 95% CI 0.56 to 2.86, p=0.57), 1 year (RR 1.21; 95% CI 0.49 to 2.98, p=0.68), 2 years (RR 1.11; 95% CI 0.51 to 2.41, p=0.78), 3 years (CoreValve, ITT RR 1.14; 95% CI 0.53 to 2.46, p=0.75) and 5 years (PARTNER 1A, ITT RR 1.13; 95% CI 0.68 to 1.87, p=0.65). Both pooled and individual risk ratios for TIA from PARTNER 1A and the CoreValve trial also showed no statistically significant differences at 30 days (RR 3.04; 95% CI 0.62 to 15.01, p=0.17), 1 year (RR 1.46; 95% 0.63 to 3.41, p=0.38), 2 years (RR 1.92; 95% CI 0.90 to 4.11, p=0.09), 3 years (CoreValve, ITT RR 1.53; 95% CI 0.55 to 4.25, p=0.42) and 5 years (PARTNER 1A, ITT RR 1.77; 95% CI 0.75 to 4.15, p=0.19).

5.6 In the RCT of 2,032 patients for whom SAVR was suitable but intermediate risk (PARTNER 2A, TAVI compared with SAVR) there were no significant differences between groups in all strokes at 30 days (55 [6%] compared with 61 [6%], p=0.57), 1 year (78 [8%] compared with 79 [8%], p=0.88) and at 2 years (91 [10%] compared with 85 [9%], p=0.67). An RCT of 276 patients for whom SAVR was suitable but low to intermediate risk (the NOTION study, TAVI compared with SAVR) reported incidence of stroke and TIA at 30 days (4 [3%] compared with 4 [3%], p=0.94), 1 year (7 [5%] compared with 8 [6%], p=0.68) and at 2 years (13 [10%] compared with 10 [8%], p=0.67). The systematic review of 3,179 patients (based on data from 2,576 patients in 3 studies) reported a non-significant reduction in stroke rates (RR 0.80; 95% CI 0.63 to 1.01) for transfemoral TAVI compared with SAVR at 2-year follow-up. Comparing transapical TAVI with SAVR (based on data from 552 patients in 2 studies), the risk ratio was 1.67 (95% CI 0.97 to 2.87) at 2-year follow-up.

Aortic regurgitation

5.7 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, TAVI compared with medical management) there were similar rates of moderate or severe aortic regurgitation in both groups at 30‑day and 1‑year follow-up (15% compared with 17%).

5.8 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve [n=795] trial) the incidence of aortic regurgitation (paravalvular and transvalvular) in patients for whom SAVR was suitable but high risk was reported. Pooled data (RR 4.02; 95% CI 1.99 to 8.11, p=0.0001) at 1 year and individual study data favoured SAVR over TAVI at all follow-up points up to 3 years (PARTNER 1A: at 30 days RR 16.29; 95% CI 3.98 to 66.6, p=0.0001; at 6 months RR 30.26; 95% CI 4.16 to 220.01, p=0.0008; at 2 years p=0.008; CoreValve trial: at 3 years p=0.04).

5.9 In the RCT of 276 patients for whom SAVR was suitable but low to intermediate risk (NOTION, TAVI compared with SAVR) significant differences in moderate to severe aortic regurgitation at 3 months (15% compared with 22%, p<0.001) and at 1 year (16% compared with 1%, p=0.001) were reported. In the systematic review of 3,179 patients (based on data from 3 trials) moderate or severe aortic regurgitation occurred more often after TAVI than after SAVR at 2‑year follow-up (RR 12.22; 95% CI 5.17 to 28.88).

Aortic valve re-intervention

5.10 In the systematic review of 3,179 patients for whom SAVR was suitable but low to intermediate risk (based on data from 3,058 patients in 4 studies) the risk of aortic valve re-intervention was significantly higher after TAVI than after SAVR (RR 3.25; 95% CI 1.29 to 8.14).

Prosthesis-patient mismatch

5.11 In the RCT of 699 patients (PARTNER 1A) for whom SAVR was suitable but high risk, the overall incidence and severity of prosthesis-patient mismatch was significantly better in the TAVI group than in the SAVR group (assessed at first postoperative echocardiogram: 46% [severe 20%] compared with 60% [severe 28%]; p<0.001 and 42% compared with 57%, p<0.001, at 30 days).

Myocardial infarction

5.12 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, comparing TAVI with medical management) there were no significant differences in the occurrence of myocardial infarction (MI) at 2‑year (p=0.69) and 3‑year (p=0.59) follow-up.

5.13 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve [n=795] trial, comparing TAVI with SAVR) the incidence of MI for patients for whom SAVR was suitable but high risk was reported. Pooled data (at 30 days RR 0.72; 95% CI 0.17 to 2.94, p=0.64; at 1 year RR 1.18; 95% CI 0.42 to 3.29, p=0.76; at 2 years RR 0.51; 95% CI 0.06 to 4.05, p=0.52) and individual study data (3-year CoreValve ITT, RR 1.45; 95% CI 0.45 to 2.94, p=0.52 or 5-year PARTNER 1A ITT, RR 0.46; 95% CI 0.16 to 1.31, p=0.14) showed no statistically significant differences between the treatment groups.

5.14 In 2 RCTs (PARTNER 2A and NOTION, comparing TAVI with SAVR) with 2,032 and 276 patients for whom SAVR was suitable but intermediate risk there were no significant differences in incidence of MI between TAVI and SAVR. The systematic review of 3,179 patients for whom SAVR was suitable but intermediate to low risk (based on data from 3,128 patients in 4 studies) found no difference between the treatment groups for MI (RR 0.87; 95% CI 0.59 to 1.29) at 2‑year follow-up.

Endocarditis

5.15 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, comparing TAVI with medical management) there were no significant differences between the groups in endocarditis at 2‑year (3% compared with 1%, p=0.32) and 3‑year (2% compared with 1%, p=0.32) follow-up.

5.16 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve [n=795] trial, comparing TAVI with SAVR) in patients for whom SAVR was suitable but high risk, there were no significant differences between groups in endocarditis (PARTNER 1A: 0% compared with less than 1%, p=0.32, at 1 month; 2 [less than 1%] compared with 3 [1%], p=0.63, at 1 year; 4 [2%] compared with 3 [1%], p=0.61, at 2 years; and 5 [2%] compared with 6 [3%], p=0.65, at 5 years; CoreValve study: 3 [1%] compared with 5 [2%], p=0.346, at 3 years).

5.17 In 2 RCTs (PARTNER 2A study and NOTION, comparing TAVI with SAVR) with 2,032 and 276 patients, for whom SAVR was suitable but intermediate risk, there were no significant differences in endocarditis between TAVI and SAVR (RCT of 276 patients: 1 [1%] compared with 0, p=0.33, at 30 days; 4 [3%] compared with 2 [2%], p=0.47, at 1 year; RCT of 2,032 patients: transfemoral route 6 [1%] compared with 6 [1%], p=0.92, at 1 year; 10 [2%] compared with 6 [1%], p=0.33, at 2 years; transthoracic route 1 compared with 0, p=0.32).

Atrial fibrillation

5.18 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, TAVI compared with medical management) there were no significant differences in the incidence of new atrial fibrillation between the treatment groups (at 30 days less than 1% compared with 1%, p=1.00; at 1 year less than 1% compared with 2%, p=0.62).

5.19 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve [n=795] trial, comparing TAVI with SAVR) in patients for whom SAVR was suitable but high risk, there were significant differences in the incidence of atrial fibrillation (PARTNER 1A: 12% compared with 17%, p=0.07, at 1 year; CoreValve study: 12% [45] compared with 31% [108], p<0.001, at 30 days; 16% [60] compared with 33% [115], p<0.001, at 1 year).

5.20 In 2 RCTs (PARTNER 2A and NOTION, comparing TAVI with SAVR) with 2,032 and 276 patients, for whom SAVR was suitable but intermediate risk, there were significant differences in the incidence of new atrial fibrillation between the treatment groups. In the RCT of 2,032 patients, for transfemoral TAVI compared with SAVR the incidence was 38 (5%) compared with 204 (27%), p<0.001, at 30 days; 45 (6%) compared with 210 (28%), p<0.001, at 1 year and 55 (7%) compared with 211 (28%) p<0.001, at 2 years. In the same RCT, for transthoracic TAVI compared with SAVR the incidence was 53 (23%) compared with 61 (26%), p=0.50, at 30 days; 55 (24%) compared with 62 (26%), p=0.60, at 1 year; and 55 (24%) compared with 62 [26%], p=0.60, at 2 years. In the RCT of 276 patients the incidence was: 24 [17%] compared with 77 [58%], p<0.001, at 30 days; 51 [38%] compared with 79 [60%] p<0.001, at 1 year; 32 [23%] compared with 80 [60%], p<0.001, at 2 years). The systematic review of 3,179 patients (based on data from 3,058 patients in 3 studies) found that the risk ratio for new onset atrial fibrillation at 2‑year follow-up was 0.43 (0.35 to 0.52) for TAVI compared with SAVR.

Need for permanent pacemaker

5.21 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, TAVI compared with medical management) the proportion of patients with permanent pacemaker implantation (PPI) was lower in the TAVI group at 2 years (6% compared with 9%, p=0.47) and 3 years (8% compared with 9%, p=0.75) but these differences were not significant.

5.22 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve [n=795] trial, comparing TAVI with SAVR) in patients for whom SAVR was suitable but high risk, the need for PPI was reported. Pooled data tended to favour the SAVR group. However the differences were not statistically significant at 30 days (RR 1.94; 95% CI 0.70 to 5.34, p=0.20), at 1 year (RR 1.75; 95% CI 0.94 to 3.25, p=0.08) and at 2 years (RR 1.77; 95% CI 0.95 to 3.30, p=0.07). At 3‑year follow-up (CoreValve trial: TAVI, n=394, SAVR, n=401) there were significantly fewer PPI in the SAVR group than in the TAVI group (14.5% compared with 28%, p<0.001). At 5 years (PARTNER 1A) there were no statistically significant differences between the treatment groups (9.7% compared with 9.1%, p=0.64).

5.23 In 2 RCTs (PARTNER 2A and NOTION, comparing TAVI with SAVR) with 2,032 and 276 patients, for whom SAVR was suitable but intermediate risk, the need for PPI was reported. In the RCT of 2,032 patients there were no significant differences between the groups (9% compared with 7%, p=0.17, at 30 days; 10% compared with 9%, p=0.43, at 1 year; 12% compared with 10%, p=0.29, at 2 years). But in the RCT of 276 patients the need for PPI was higher in the TAVI group than in the SAVR group (46 [34%] compared with 2 [2%], p<0.001, at 30 days; 51 [38%] compared with 3 [2%], p<0.001, at 1 year; and 55 [41%] compared with 5 [4%], p<0.001, at 2 years). The systematic review of 3,179 patients (based on data from 3,128 patients in 4 studies) reported an increased risk of PPI (RR 2.46; 95% CI 1.17 to 5.15) for TAVI compared with SAVR.

Acute kidney injury and renal failure

5.24 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, TAVI compared with medical management) there were no significant differences in the occurrence of acute kidney injury (AKI) between the treatment groups at 2-year (3% compared with 8%, p=0.15) and 3-year follow-up (3% compared with 11%, p=0.08).

5.25 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve [n=795] trial, comparing TAVI with SAVR) in patients for whom SAVR was suitable but high risk, the occurrence of AKI was reported. Pooled data (at 30 days RR 0.51; 95% CI 0.27 to 0.98, p=0.04) and data from the CoreValve study at 3 years (RR 0.45; 95% CI 0.29 to 0.72, p=0.0007) significantly favoured the TAVI group, whereas there were no statistically significant differences in the pooled estimates at 1 year and 2 years (RR 0.76; 95% CI 0.23 to 2.59, p=0.67, at 1 year; RR 0.64; 95% CI 0.31 to 1.34, p=0.24, at 2 years) and from PARTNER 1A at 5 years (RR 1.01; 95% CI 0.58 to 1.74).

5.26 In 2 RCTs (PARTNER 2A and NOTION, comparing TAVI with SAVR) with 2,032 and 276 patients, for whom SAVR was suitable but intermediate risk, the occurrence of AKI was reported. The RCT of 2,032 patients reported a lower incidence of AKI in the TAVI group than in the SAVR group at 30 days (13 [1.3%] compared with 31 [3%], p=0.0006). Incidence rates were similar for transthoracic TAVI and SAVR (4% compared with 3%). The incidence rates were significantly lower for transfemoral TAVI than for SAVR (2.2% compared with 5%, p=0.002, at 1 year; 3% compared with 7%, p<0.001, at 2 years) and higher for transthoracic TAVI (7% compared with 4.4%, p=0.18, at 1 year, 8% compared with 6%, p=0.23, at 2 years). The RCT of 276 patients reported a higher occurrence of AKI in the SAVR group than in the TAVI group (9 [7%]) compared with 1 [0.7%], p=0.01) at 30 days. The systematic review of 3,179 patients (based on data from 2,576 patients in 3 studies) reported that the risk ratio of AKI at 2‑year follow-up was 0.38 (95% CI 0.27 to 0.54) for transfemoral TAVI and 1.54 (95% CI 0.77 to 3.07) for transapical TAVI compared with SAVR.

Vascular complications

5.27 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, TAVI compared with medical management) the hazard ratio for major vascular complications at 3‑year follow-up was statistically significantly higher with TAVI (17%) than with medical management (3%, HR 8.27; 95% CI 2.92 to 23.44, p<0.0001).

5.28 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve [n=795] trial, comparing TAVI with SAVR) in patients for whom SAVR was suitable but high risk, there were no statistically significant differences between the treatment groups in the incidence of major vascular complications in either pooled data at 30‑day (RR 3.04; 95% CI 0.63 to 3.41, p=0.17), 1‑year (RR 1.46; 95% CI 0.63 to 3.41, p=0.38) or 2‑year follow-up (RR 1.92; 95% CI 0.90 to 4.11, p=0.09), or in the individual studies at 3‑year (CoreValve study, p=0.42) or 5‑year follow-up (PARTNER 1A, p=0.19).

5.29 In 2 RCTs (PARTNER 2A and NOTION, comparing TAVI with SAVR) with 2,032 and 276 patients, for whom SAVR was suitable but intermediate risk, major vascular complications were reported. In the RCT of 2,032 patients there was a higher incidence of major complications in the TAVI group than in the SAVR group (7.9% compared with 5%, p=0.008 at 30 days; 8% compared with 5.3%, p=0.007 at 1 year and 9% compared with 6%, p=0.006 at 2 years). The incidence rate was lower for transthoracic TAVI than for SAVR. The RCT of 276 patients reported more major vascular complications in the TAVI group than in the SAVR group at 30 days (6% compared with 2%, p=0.10).

Major bleeding

5.30 In the RCT of 358 patients for whom SAVR was unsuitable (PARTNER 1B, TAVI compared with medical management), the risk of major bleeding was statistically significantly higher for TAVI than for medical management (29% compared with 20%, p=0.04) at 2 years, but not statistically significant different (32% compared with 33%, p=0.92) at 3-year follow-up.

5.31 In 2 RCTs (PARTNER 1A [n=699] and the CoreValve [n=795] trial, comparing TAVI with SAVR) in patients for whom SAVR was suitable but high risk, there were no statistically significant differences between the treatment groups in the risk of major bleeding in either pooled data at 30 day (RR 0.67; 95% CI 0.36 to 1.25, p=0.21), 1‑year (RR 0.73; 95% CI 0.48 to 1.12, p=0.02) and 2‑year follow-up (RR 0.78; 95% CI 0.54 to1.13, p=0.19) or in the individual (CoreValve) study at 3‑year follow-up (RR 0.92; 95% CI 0.75 to 1.12, p=0.38). However, major bleeding was significantly lower in the TAVI group than in the SAVR group in the individual (PARTNER 1A) study at 5‑year follow-up (RR 0.73; 95% CI 0.57 to 0.95, p=0.02).

5.32 In 2 RCTs (PARTNER 2A and NOTION, comparing TAVI with SAVR) with 2,032 and 276 patients, for whom SAVR was suitable but intermediate risk, the risk of major bleeding was reported. In the RCT of 2,032 patients there was significantly lower life-threatening or disabling bleeding in patients who had TAVI than in those who had SAVR at 30‑day (10% [105/1,011] compared with 43% [442/1,021], p<0.001), 1‑year (15% [151] compared with 46% [460], p<0.001) and 2‑year follow-up (17% [169] compared with 47% [471], p<0.001). The rates were also significantly lower in patients who had transthoracic TAVI than in those who had SAVR (23% compared with 50%, p<0.001, at 30‑day, 29% compared with 52%, p<0.001, at 1-year and 30% compared with 54%, p<0.001, at 2‑year follow-up). The RCT of 276 patients reported significantly lower bleeding in the TAVI group than in the SAVR group at 30 days (11% [16] compared with 21% [28], p=0.03). The systematic review of 3,179 patients (data from 2,576 patients in 3 studies) reported that transfemoral TAVI was associated with a significant reduction in major bleeding compared with SAVR (RR 0.39; 95% CI 0.29 to 0.54). Transapical TAVI (based on data from 552 patients in 2 studies) also had a reduced risk of bleeding, RR 0.53; 95% CI 0.42 to 0.67.

Rare safety events

5.33 A number of observational studies reported rare safety events with TAVI for severe aortic stenosis including: acute MI, acute myocardial injury from damage to apical epicardial collateral circulation, acute occlusion of right coronary artery, acute severe occlusion of the left main coronary artery, aortic arch rupture, aortic dissection, aortic perforation, aortic rupture (abdominal), aorto-right ventricular defect (lethal), apical left ventricular thrombus, apical tear, balloon rupture, catheter-induced ventricular septum defect, circumflex artery occlusion, cutaneo-pericardial fistula, delayed ventricular apical bleed, distal coronary embolisation, early valve degeneration, elliptic distortion of the aortic prosthesis, false left ventricular apical aneurysm, guide wire thrombus formation, iatrogenic chordal rupture, iliac artery rupture, intercostal artery pseudoaneurysm, interventricular septum rupture, late prosthesis migration and rotation, left ventricular pseudoaneurysm, major bleeding from the apex, mitral valve destruction by wire entrapment, multivessel coronary artery spasm, papillary muscle rupture, perforation of the medial circumflex branch of the common femoral artery, pseudoaneurysm at the left ventricular apical access site, pseudoaneurysm of the apex, ruptured pseudoaneurysm of a renal artery, Takotsubo syndrome and valve embolisation.

5.34 In addition to safety outcomes reported in the literature, specialist advisers are asked about anecdotal adverse events (events which they have heard about) and about theoretical adverse events (events which they think might possibly occur, even if they have never done so). For this procedure, specialist advisers listed the following anecdotal adverse events: valve migration, embolisation, thrombosis, valve or annular trauma during TAVI implants leading to late 'fistulous' connections to adjacent cardiac structures. They considered that the following were theoretical adverse events: haemolytic anaemia, infective endocarditis, structural valve failure, reduced leaflet movement and longer-term problems with device durability needing re-intervention by either SAVR or 'valve-in-valve' TAVI.

  • National Institute for Health and Care Excellence (NICE)