2.5. Statistical Analysis
Stata version 16.0 was used for statistical analyses. Continuous variables are displayed as means ± SD, and categorical variables are presented as frequencies and percentages. For observational studies with two arms, we calculated the relative risk (RR) and corresponding 95% confidence intervals (CI) for each outcome, whereas for single-arm analysis, we calculated the incidence of events (number of events divided by number of patients) and 95% confidence intervals. P< 0.05 was considered statistically significant.
Meanwhile, chi-square tests and I-squared (I2) are used to quantify and assess statistical heterogeneity among studies. If the I2 value was more than 50% and/or P< 0.05 for the chi-squared test, we considered the between-study heterogeneity to be significant, and we would adopt a random-effect model. Otherwise, we would adopt fixed-effect model. Sensitivity analysis was performed by sequentially omitting one study at a time to assess the effect of a single study on the overall risk, and potential publication bias was also evaluated via Egger’s test.
In addition, subgroup analysis was conducted to screen potential determinants of LAAO outcomes between ICE and TEE groups. According to the characteristics of eligible studies, a total of eight subgroup factors were identified, including study design, elderly age cutoff, ICE group sample size, AF type, male proportion, hypertension proportion, device types and duration of follow up. If the study design included more than one center, it was defined as a multicenter subgroup; otherwise, it was defined as a single-center subgroup. According to age cut-off values of 75, it was divided into ≥ 75 years subgroup and <75 years subgroups. If more than 50% of patients have paroxysmal AF, they are classified as ≥50% PAF group, otherwise they are classified as <50% PAF group. According to the proportion of the male, they were divided into subgroups with ≥ 70% and subgroups with < 70%. Similarly, the proportion of hypertension with “≥ 90%” and “< 90%” subgroup also was defined. According to the sealing position, the existing sealers can be roughly divided into plug type and disc type. Plug type sealers, also known as single sealers, including Watchman, Plaato, and Lefort. Disc sealer is also called dual sealer, including ACP, Lambre, Lacbes, and Leftear. If the LAAO devices included only dual-seal mechanism devices, it was assigned to dual-seal mechanism subgroup, and if the LAAO devices included only single-seal mechanism devices, it was assigned to the single-seal mechanism subgroup. In addition, studies using both dual-seal mechanism devices and single-seal mechanism devices are divided into muti-seal mechanism subgroup. Follow-up time was divided into two subgroups (≥12 months and <12 months).