82 Chapter 4 Obsessive-compulsive symptoms All studies assessed obsessive-compulsive symptoms. Random effects analyses showed a beneficial main effect of 0 ∙ 72 (Hedges’s g; 95% CI = 0 ∙ 39 to 1 ∙ 06; Z-value = 4 ∙ 19; P = 0 ∙ 00; Supplementary figure 5). Symptoms of anxiety Three non-randomized non-controlled clinical trials [4, 15, 25] assessed symptoms of anxiety. Random effects analyses showed a beneficial main effect of 0 ∙ 85 (Hedges’s g; 95% CI = 0 ∙ 38 to 1 ∙ 31; Z-value = 3 ∙ 55; P = 0 ∙ 00; Supplementary figure 6). Secondary analysis: quality of life Three non-randomized non-controlled clinical trials (15, 25, 33) assessed eating disorder symptoms. Random effects analyses showed a beneficial main effect of 0 ∙ 86 (Hedges’s g; 95% CI = 0 ∙ 44 to 1 ∙ 28; Z-value = 4 ∙ 01; P = 0 ∙ 00; Supplementary figure 7). Adverse events The most frequently occurring adverse events in the four studies were related to surgery or procedure; pain at incision site within <4 days (n = 22), pain at incision site after >4 days (n = 5), cutaneous complications (n = 4), and to stimulation procedure; hypomanic symptoms (n = 3), seizure (n = 3), and auto-intoxication (n = 3). Supplementary table 2 lists all reported adverse events, including those possibly but not probably related to the intervention (15). Other reported adverse events were either defined as unrelated to the intervention and attributed to underlying illness, or had no cause specified. Short-term side-effects like flush and sweating were observed during the programming of the DBS device, but were relieved with adjustment of the parameters. Risk of bias and quality of evidence Risk of bias summary is shown in Supplementary figure 1. The risk of bias of the objective (BMI) and subjective (psychological outcomes) measurements of the included studies are respectively shown in Supplementary figure 1a, b. Overall, the risk of bias was moderate for objective outcome measurements and serious for subjective outcome measurements. This was mainly caused by the lack of a control group in the included studies, potentially leading to biases. Furthermore, the studies did not report blinding of both the participants and the researchers, or assessment of the subjective outcomes by an independent party, leading to serious risk of potential bias in measurement of subjective psychological outcomes. No other clear evidence for biases was found. Supplementary table 3 shows details of the assessment of quality of the resulting evidence using GRADE. For the more objective outcome BMI, the quality of evidence that deep brain stimulation improves BMI in refractory-treatment AN-patients is considered moderate, whereas for subjective outcomes psychiatric symptoms and quality of life the quality of evidence is deemed low.
RkJQdWJsaXNoZXIy MjY0ODMw