Revman 5

Meta-analysis was performed using RevMan 5.3 and STATA 16.0 (STATA Corp, College Station, TX, USA). For continuous variables, the weighted mean difference (WMD) was used as the effect index, and each effect size was expressed with a 95% confidence interval (95% CI), P<0.05 indicates a statistically significant difference between the two groups. The I² statistic was used to test the heterogeneity among different studies. When P>0.1 and I²≤50%, it means that there was good homogeneity among all studies, and the fixed effect model was used to combine the effect size. When P≤0.1 and I²>50%, it means that there was heterogeneity between the studies, and the random effect model was used to combine the effect size. The source of heterogeneity was identified, and sensitivity analysis was conducted by article by article elimination. If I²≤50% after deleting a single study, it was considered that the study might be the source of influencing the combined effect size, and it was excluded from the meta-analysis. In addition, The Begg’s and Egger’s tests were performed to assess publication bias. P<0.05 was statistically significant unless otherwise specified.

The effect sizes between stem/progenitor cell-EV-treated and control groups were reported as a pooled standardized mean difference (SMD) according to Cohen’s d statistic [57 ], with the 95% confidence interval (CI). SMD values of 0.2, 0.5, 0.8, and 1.0, respectively, correspond to small, medium, large, and very large effect sizes. The SMD was used because the different measures were employed by many included studies to assess the same outcomes. The random-effect analytical model was used for the analyses. Statistical heterogeneity across studies was assessed using the I² statistic. I² > 50% indicated significant heterogeneity [58 (link)]. The sensitivity analysis was conducted to assess the robustness of results. The subgroup analysis based on cell origins of EVs (mesenchymal stromal/stem cell and progenitor cell) was performed to investigate potential sources of between-study heterogeneity. Meta-regression analyses were carried out focused on cell origins of EVs, injection doses, delivery routes, and therapy and outcome measurement time. A cumulative meta-analysis was performed to explore changes in the results over time. To detect the presence and extent of publication bias, we use the funnel plot, Egger tests, and trim and fill. The data were pooled and analysis using RevMan 5.3 and Stata 12.0/SE statistical software.

Weighted mean difference (WMD) and 95% confidence interval (95%CI) were used as the effect size indicators of continuous variables (including NT-proBNP, LVEF, 6MWT, and other cardiopulmonary function indicators) to evaluate whether the intervention effect of Baduanjin on the cardiopulmonary function of CHD patients was statistically significant. The heterogeneity test was then generated using Cochran's Q test and I² test [19 (link)]. The meta-analysis was performed using a random effect model when a significant heterogeneity exists at P < 0.05 and/or I² > 0.5. If heterogeneity was not significant (P ≥ 0.05 and I² ≤ 0.5), the fixed effect model was used for meta-analysis [20 (link)]. The effects of follow-up time, intervention plan, and disease course on heterogeneity and combined outcomes were then assessed by subgroup analysis. Finally, the funnel plot and Egger test were used to evaluate whether there was significant publication bias among included studies [21 (link)]. Statistical analysis was generated using RevMan 5.3 and Stata12.0.

The impact of SOX2OT expression on overall survival, TNM stage and progression, distance metastasis and lymph node metastasis was examined by HRs and 95% CIs. An observed HR >1 indicated poorer prognosis in patients with elevated SOX2OT expression and should be statistically significant when the 95% CI did not overlap with 1. The random-effects model was conducted to analyze the relationship between SOX2OT expression and clinical outcomes when calculated I²>50%[13 (link)–15 (link)]. Probable publication bias was examined by a funnel plot and Begg’s bias test[16 (link)]. P values <0.05 was considered statistically significant. All statistical analyses were performed using Stata SE 12.0 (Stata Corporation) and RevMan 5.3 software.

The association between the expression of VISTA and patients’ prognosis was evaluated by meta-analysis by collecting data from all included studies. We calculated outcome endpoints including OS, DSS and TSS via pooled HRs and 95% CIs. HRs > 1 indicated a poor prognosis. The correlation between the expression of VISTA and the clinicopathological characteristics was evaluated by pooled RRs and 95% CIs. Cochrane’s Q statistic and the I² statistic were used to assess the heterogeneity among the included studies. A random effects model was used to calculate pooled HRs and 95% CIs when there was substantial heterogeneity (Q test: P < 0.1 or an I² > 50%). If not, a fixed effects model was used. Studies of high quality(NOS scores ≥ 7) were selected for sensitivity analysis. RevMan 5.3 and Stata 12.0 statistical software (Stata Corporation, College Station, TX, USA) were used to perform all statistical analyses. A difference was considered significant with a two-tailed p < 0.05.

The effect of breastfeeding on childhood cancer was analyzed using OR and 95%CI as the effect measure. Between-study heterogeneity [22 (link)] was assessed using Cochran’s Q and I² statistics. Initial analyses with I² < 40% were performed using a fixed-effects model; otherwise, a random effects model was adopted. Potential confounders, including the modes of breastfeeding (exclusive, mixed, and formula with different durations of breastfeeding), duration of breastfeeding (including ≥1 month vs. < 1 month, ≥6 months vs. < 6 months, and ≥ 12 months vs. < 12 months), different countries, and cancers of different systems were regarded as the main sources of heterogeneity. Subgroup and sensitivity analyses were employed. We used the one-stage robust error meta-regression model to establish the potential dose-response relationship between duration of breastfeeding and risk of cancer [24 (link), 25 (link)]. This was a one-stage method that treated each study as a cluster combined with robust error estimation as a solution to deal with potential correlations within each study [26 (link)]. The restricted cubic spline function with three auto-generated knots was used to fit the potential non-linearity trends [27 (link)]. The remr command of Stata software was used to run the dose-response meta-analyses [28 ]. All statistical analyses were performed using RevMan 5.3 and STATA 15.0.

The odds ratios (ORs) and 95% confidence interval (CIs) were applied to assess the strength of the correlation between SNPs and cervical cancer susceptibility. A Z-test revealed statistical significance when p < 0.05. I² and Q statistics were employed to detect heterogeneity among different studies. There was no heterogeneity if I² < 50% and p > 0.1 and a fixed effects model was used; otherwise, we thought that heterogeneity existed in the incorporated populations and a random effects model was used instead. Subsequently, we conducted a subgroup analysis according to race. Hardy-Weinberg equilibrium (HWE) was evaluated by χ² test in control groups with p < 0.05 indicating a deviation from HWE. Sensitivity analysis was utilized to estimate the robustness and stability of the meta-analysis results by deleting all the studies one by one. Next, Begg's funnel plot and Egger's test were used to evaluate publication bias. For each SNP, five genetic models were evaluated to assess the correlation with cervical cancer susceptibility: the allele model, dominant model, recessive model, heterozygote model, and homozygous model. The statistical analyses were performed using RevMan 5.3 and Stata 12.0 software. All p values were two sided, and p < 0.05 was considered to be statistically significant.