83 Efficacy and safety of deep brain stimulation for treatment-refractory anorexia nervosa: a systematic review and meta-analysis Publication bias The classic fail-safe N was 40, Orwin’s fail-safe N was 42 with criterion for a ‘trivial’ standardized difference in means as 0 ∙ 1. This suggested that at least 40 studies without any effect must be reported to decrease the overall effect to a trivial effect. Concerning the Begg and Mazumbar rank correlation test, Kendall’s tau’s with as well as without continuity correction was 0 ∙ 00 (P2-sided = 1 ∙ 00), suggesting no publication bias. Egger’s regression intercept was 0 ∙ 23 (95% confidence interval: −3 ∙ 47 to 3 ∙ 94; P2-sided = 0 ∙ 81), also indicating no publication bias. Duval and Tweedie’s trim-and-fill method used the random-effects model looking for missing studies to the left of the mean, meaning a less favourable effect of deep brain stimulation, showed one study that needed to be trimmed. This resulted in an effect size of 0 ∙ 91 (Q-value; 95% CI = 0 ∙ 79 to 1 ∙ 43). Figure 3 shows the values of the Duval and Tweedie’s trim-and-fill method. Using a fixed effect model, the resulting point estimate did not change. Figure 3. Funnel plot of precision of standard error by Hedges’s g for publication bias. -2,0 -1,5 -1,0 -0,5 0,0 0,5 1,0 1,5 2,0 0,0 0,2 0,4 0,6 0,8 Standard Error Hedges's g Funnel Plot of Standard Error by Hedges's g
RkJQdWJsaXNoZXIy MjY0ODMw