Fig. 2: Forrest and funnel plots and sensitivity analysis of LESP (r = 0.1).

Forest plot (Top): Horizontal lines emerging from the square represent 95% confidence intervals (CI) for the study’s result with arrows indicating that a CI extending beyond the boundaries of the graph’s x axis and would go beyond the displayed range preventing it from being completely visualized on the plot. A shorter horizontal line means the study’s estimate is more precise. The size of the square is proportional to the weight of that study, which often reflects its sample size or precision. The pooled weighted mean difference (WMD) combines all the MDs of each study using the random effects model with the Der Simonian and Laird method, depicted by the diamond at the bottom represents the overall pooled estimate of the effect from all included studies. The position of the diamond and its width (representing the 95% CI) indicate the overall effect and its precision. Significant reduction was observed in LESP of 3.82 mmHg post LVSG compared to pre LVSG (WMD 3.82, 95% CI 1.74 to 5.90, Z = 3.60, p < 0.001). Funnel Plot (bottom left): Each dot in the plot represents an individual study. The x-axis shows the effect size of each study expressed as WMD. The y-axis is the standard error and represents the study precision. Larger studies typically have greater precision, whereas smaller studies are less powerful and have lower precision. The overall effect and 95% CIs are denoted by dotted black lines (central and funnel lines, respectively). Significant difference between all studies was noted (Q = 131.65, p < 0.001, \({I}^{2}\) = 88.6%). The asymmetric distribution of studies around the overall effect indicates the presence of publication bias (study heterogeneity), however, there was no publication bias demonstrated by Egger’s test (p = 0.49). Sensitivity Analysis (bottom right): In this sensitivity analysis, “a leave one out method” is used, where an individual study is omitted at a time from all studies in this analysis and the weighted effect size and heterogeneity are recalculated to assess the robustness of our analysis. Horizontal lines in the plot represent the 95% CIs for each study, with squares representing the weight of individual studies. The largest decrease in the \({I}^{2}\) value was observed when Valezi et al. was omitted from our analysis. However, our results across multiple iterations i.e., when we removed one data point/individual study did not significantly change the outcome, and the pooled WMD and heterogeneity (>80%) remained relatively unchanged, indicating that the overall conclusions are robust.