Systematic review and meta-analysis: Gastric electrical stimulation for gastroparesis.

BACKGROUND: Controlled trials of gastric electrical stimulation (GES) for gastroparesis reported no significant improvement in symptoms, while open label studies suggested substantial clinical benefits. AIM: To determine if GES is effective in reducing symptoms in patients with gastroparesis. METHODS: We searched PubMed and Embase for articles published in English (1990-2014) using "gastroparesis" as a search term restricted to "clinical trial". We included studies describing repeated patient-based symptom ratings before and during standardized treatments of at least one week duration. RESULTS: Five studies randomly allocated patients to periods with or without GES. Total symptom severity (TSS) scores did not differ between these periods (0.17 [95% confidence interval: -0.06 to 0.4]; P=0.15). However, sixteen open label studies of GES showed a significant TSS decrease (2.68 [2.04-3.32]; Q=39.0; P<0.001). Other treatment modalities similarly improved TSS by 1.97 [1.5-2.44] for medical therapy (MED), by 1.52 [0.9-2.15] for placebo arms (PLA), and by 2.32 [1.56-3.06] for botulinum toxin (BTx). There were significant differences in baseline TSS ratings among these studies (GES: 6.28 [6.28-7.42]; MED: 4.76 [4.09-5.42]; PLA: 4.59 [3.77-5.42]; BTx: 6.02 [5.3-6.74]; Q=35.1; P<0.001). Meta-regression analysis showed these baseline differences to significantly impact TSS ratings during treatment (Q=71.8; P<0.001). CONCLUSION: Independent of the treatment modality, baseline symptom severity impacts treatment results in gastroparesis. Considering the skewed population with refractory symptoms, regression to the mean likely contributes to the substantial discrepancies between the reported results of controlled and open label GES studies, raising questions about the use of GES outside of defined clinical trials.