Electronic databases, including the Cochrane Library, PubMed, Embase, and the Web of Science, were searched from January 2000 to Aug 2022. We focused on peer-reviewed publications of clinical trials. The following searched combination of keywords was used: [(Transcatheter Aortic Valve Replacement OR Transcatheter Aortic Valve Replacement OR Transcatheter Aortic Valve Implantation OR TAVR OR TAVI)] AND [(aorta OR aorta OR aortas OR aortae OR Femoral Artery OR Femoral Arteries OR Common Femoral Artery OR Common Femoral Arteries OR Internal Carotid Arteries OR Internal Carotid Arteries OR Internal Carotid Artery OR transapical OR transapical OR apical OR ventricular apex OR apex OR apex cordis OR cardiac apex)] AND[(Aortic Valve Stenosis OR Aortic Valve Stenoses OR Aortic Stenosis OR Aortic Valve Insufficiency OR Aortic Regurgitation OR Aortic Incompetence OR Aortic Valve Incompetence)] in the title/abstract. At the same time, try to collect all relevant literature and search references to supplement possible omissions. The detailed search strategy is presented in the Supplementary Material in the form of a word document.

Two reviewers (LZK and DHS) initially screened independently at the title and abstract level and retrieved all eligible full-text studies for further screening. Disagreements were resolved by discussion with BM and ZYQ. All trials that reported relevant outcomes in men and women after TAVI were considered. If they respectively reported the outcomes of prognosis and survival after TAVI in males and females, such as bleeding, vascular complications, 30-day mortality, etc. Studies were excluded if: (i) case reports, conference abstracts, comments, etc.; (ii) studies that did not report both male and female outcomes after TAVI; (iii) studies that could not find the full text; (iv) TAVI was used in combination with any other cardiac surgery.

For each included study, data were extracted by one reviewer (JPF) and checked for accuracy and completeness by another reviewer (ZXQ). Any differences are resolved through discussion, if necessary, in consultation with a third reviewer (BM and ZYQ). First, we extracted gender, age, body mass index (BMI), and comorbidities (hypertension, pulmonary hypertension, etc.) for inclusion in the study. Secondly, for outcome measures, cardiovascular mortality, bleeding, vascular complication, and stroke etc. were pooled for analysis.

The quality of all studies was assessed using the Newcastle-Ottawa Scale (NOS) by two independent authors (LZK and DHS). Two authors conducted the quality assessment of each included studies from three items: selection bias, comparability bias, and exposure bias. There are evaluation items under each item, and each item is indicated by a star when appropriate. The highest score for the comparability bias item is two stars. Any discrepancies were resolved by consensus.

The primary outcomes were the thirty-day mortality, one-year mortality and perivalvular leakage (PVL). PVL in this context refers to new onset aortic valve leak due to surgery. Thirty-day mortality and one-year mortality can reflect the short-term and medium-term survival of patients, respectively. We selected conversion to open heart surgery during operation, ejection fraction (EF) after operation, reintervention and other common postoperative complications including bleeding, vascular complication, myocardial infraction (MI) etc. as secondary outcomes to assess the prognosis in patients of different sexes after TAVI.

All data were analyzed by Review Manager 5.4 (The Cochrane Collaboration, Copenhagen, Denmark) and Stata SE 16.0 (Stata Corporation, Texas, USA). The risk ratio (RR) with 95% confidence intervals (CI) were estimated for dichotomous data and standard mean difference (SMD) with 95% CI for continuous data, respectively. Heterogeneity was tested by the I${}^2$ test and Q test. If p 0.1 or I${}^2$ $>$ 50%, the heterogeneity test between the research results is statistically significant, random effects model analysis is used. If I${}^2$ 50%, the fixed-effect model was used for analysis [17]. If I${}^2$ $>$ 50%, severe heterogeneity will be considered. Subgroup and meta-regression analysis were conducted to explore the possible source of heterogeneity. Use funnel plot, Begg’s test and Egger’s test to assess the risk of publication bias of studies when there are at least 8 studies. A significance level of $\alpha$ = 0.05 was set for all analyses. Sensitivity analysis was used to assess whether the results were robust and also to assess sources of heterogeneity.

We retrieved 580 articles from Pubmed, 355 from embase, 209 from the cochrane library, and 1673 from web of science. A total of 2817 articles were retrieved from the database, and after deduplication, 2665 article titles and abstracts were evaluated. After initial title and abstract screening, 47 articles remained with downloaded full text. The authors evaluated the full text independently, and 24 eligible articles were included in the meta-analysis. Fig. 1 shows the whole process of literature retrieval and screening.

Fig. 1.

Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flowchart of selection.

We sorted out the essential data characteristics of 24 included articles. The features of the included studies are shown in Tables 1,2 (Ref. [10, 11, 13, 14, 15, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36]). Due to the large number of articles included and the large sample size, some trials lacked the baseline data of some men and women. These trials were published between 2011 and 2022. Sample sizes for individual experiments ranged from 55 to 35,470. A total of 92,499 patients were involved, and the overall mean age of the patients included in the study was 81.7 years. Most of the studies used CoreValve and Edwards SAPIEN devices with prosthesis sizes of 23 mm–29 mm. Individual studies did not mention relevant conditions.

After the evaluation of quality, we found that all included studies were on the upper-middle quality by NOS. Each has a score greater than five stars and meets the criteria for inclusion in the meta-analysis. The final result is shown in Table 3.

The results show that men had a lower risk of death at thirty days (Fig. 2A; RR, 0.87; 95% CI, 0.81 to 0.93; p = 0.0001; I${}^2$ = 47%) and a higher risk of death at one year (Fig. 2B; RR, 1.20; 95% CI, 1.08 to 1.33; p = 0.0008; I${}^2$ = 59%). Meanwhile, men have a higher risk of PVL (Fig. 2C; RR, 1.42; 95% CI, 1.15 to 1.75; p = 0.001; I${}^2$ = 68%). In addition, men had a lower risk of intraoperative conversion to open heart surgery (Fig. 3C; RR, 0.61; 95% CI, 0.51 to 0.74; p 0.00001; I${}^2$ = 0) and had a lower postoperative EF (Fig. 4C; SMD, –0.42; 95% CI, –0.48 to –0.37; p 0.00001; I${}^2$ = 25%) than women.

Fig. 2.

Forest plot of (A) thirty-day mortality. (B) One-year mortality. (C) Perivalvular leakage (PVL).

Fig. 3.

Forest plot of (A) stroke. (B) Kidney injury. (C) Conversion to open heart surgery.

Fig. 4.

Forest plot of (A) reintervention. (B) Atrial fibrillation. (C) Ejection fraction (EF).

We explored the risk of stroke, bleeding, vascular complications, atrial fibrillation kidney injury and MI to evaluate the prognosis of patients after TAVI. The risk of stroke in male group was lower than that in female group (Fig. 3A; RR, 0.86; 95% CI, 0.80 to 0.93; p = 0.0001; I${}^2$ = 12%). In the statistics of 90,691 patients, we found that men have a lower risk of reintervention (Fig. 4A; RR, 0.86; 95% CI, 0.80 to 0.93; p 0.0001; I${}^2$ = 12%). The risk of major bleeding in male group was significantly lower than that in the female group (Fig. 5A; RR, 0.69; 95% CI, 0.61 to 0.79; p 0.00001; I${}^2$ = 82%). As for major vascular complications, we analyzed data from 20,632 male patients and 20,131 female patients. The risk of major vascular complications in male group was also significantly lower than that in the female group (Fig. 5B; RR, 0.56; 95% CI, 0.52 to 0.61; p 0.00001; I${}^2$ = 48%). We also found that men had a lower risk of atrial fibrillation after surgery (Fig. 4B; RR, 0.76; 95% CI, 0.61 to 0.93; p = 0.009; I${}^2$ = 76%). However, no significant gender differences were shown in postoperative MI (Fig. 5C) and kidney injury (Fig. 3B).

Fig. 5.

Forest plot of (A) bleeding. (B) Vascular complication. (C) Myocardial infraction (MI).

We performed subgroup analyses of thirty-day mortality and one-year mortality according to different age groups, different proportions of patients with hypertension (HT) and diabetes mellitus (DM). The results show that among patients younger than 80 years old, women had a higher 30-day mortality risk (RR, 0.69), while there was no significant difference in the 30-day mortality risk between women and men among patients aged 82–84 years (RR, 0.96). Meanwhile, age was the main source of heterogeneity in thirty-day mortality (Fig. 6). Conversely, the subgroup analysis of one-year mortality found that women had a higher risk of death among patients under the age of 80, and men over the age of 80 had a higher risk of death (Fig. 7). A subgroup analysis of DM also showed an interesting result. Women with a lower prevalence of DM had a lower thirty-day risk of death, while those with a higher prevalence of DM had a higher thirty-day risk of death. And DM may be one of the major sources of heterogeneity in thirty-day mortality (Fig. 8). The subgroup analysis of the risk of PVL in different age groups showed that there was no significant difference in the risk of PVL among different age groups, but age was one of the sources of heterogeneity in PVL (Fig. 9). Subgroup analyses on hypertension were not statistically and clinically significant, and the results are presented in the Supplementary Material.

Fig. 6.

Subgroup analysis for different ages in thirty-day mortality.

Fig. 7.

Subgroup analysis for different ages in one-year mortality.

Fig. 8.

Subgroup analysis for DM in thirty-day mortality. DM, diabetes mellitus.

Fig. 9.

Subgroup analysis for age in Perivalvular leakage.

We also performed subgroup analyses of PVL by different age groups, different proportions of patients with peripheral artery disease (PAD). The results show that the risk of PVL was higher in patients older than 84 years, twice as high as in women. Also, age is one of the main sources of PVL heterogeneity.

Age, DM, HT, PAD, pulmonary hypertension (PH) and percutaneous coronary intervention (PCI) were included in the random-effect univariate meta-regression analyses for bleeding. Results suggest PH as a possible source of heterogeneity. We also performed a meta-regression analysis of one-year mortality and PVL, including age, myocardial infarction (MI), DM, HT, chronic obstructive pulmonary disease (COPD), PAD, stroke, diagnosis, PH, PCI and other variables, but major sources of heterogeneity were not found. Details are shown in Supplementary Material.

The funnel plot of bleeding, vascular complications, stroke, PVL, thirty-day mortality and one-year mortality has no obvious publication bias, the details were shown in Supplementary Material. Begg’s test and egger’s test also showed that there was no obvious publication bias in bleeding, vascular complications, stroke, PVL, conversion to open heart surgery, one-year mortality and thirty-day mortality. Detailed results are presented in Supplementary Material. For other outcomes, due to the small number of included studies, no begg’s test or egger’s test was performed. Sensitivity analysis shows that the combined results are robust and reliable. The results of the sensitivity analysis are presented in the Supplementary Material.