I read with interest the article by Peng et al. entitled: "The long-term efficacy of STN vs GPi deep brain stimulation for Parkinson disease: A meta-analysis". However, this meta-analysis includes several methodological mistakes that worth attention.
In systematic reviews and Meta-analyses, the authors should exclude duplicate reports and select unique studies for inclusion in the evidence synthesis process. This is a common practice in this journal "Medicine" and following the guidelines of Cochrane Handbook of Systematic Reviews and Meta-analysis of interventional studies and the preferred reporting items of systematic review and meta-analysis (PRISMA statement) [2, 3].
In this meta-analysis, the authors included five articles. However, three of these five articles described the same patients; Follet et al., Weaver et al., and Katz et al., are three reports from the same study, Veterans Administration Cooperative Studies Program #468 multicentre study. Because of pooling three reports from the same RCT, the impact of this multicentre RCT in the analysis was augmented and has led to incorrect results.
For example, in Figure 5, the authors pooled the three duplicate reports from the Veterans Administration Cooperative Studies Program #468 multicentre study (ClinicalTrials.gov numbers, NCT00056563 and NCT01076452.). According to Figure 5, the differences in PDQ ADL in the 3 studies were as follows: Follet 2010 (MD -4.5 with 95% CI from -9.11 to 0.11), Katz 2015 (MD -1.6 with 95% CI from -6.63 to 3.43), and Weaver 2012 (MD -4.0 with 95% CI from -10.39 to 2.39). It deserves our notice that the MDs of the three reports were not statistically significant. However, when the three duplicate studies are pooled in the meta-analysis model, the pooled MD was statistically significant (MD -3.36 with 95% CI from -6.36 to -0.36, P=0.03).
Secondly, Rodriguez-Oroz et al. published a long-term report from a non-randomized study (DBS study group 2000 ). There is a methodological concern in pooling randomized and non-randomized studies in meta-analysis. However, this is not the main problem; The authors assessed the risk of bias in Rodriguez-Oroz et al. using the Cochrane ROB tool, a tool for evaluating the risk of bias in RCTs. Alternatively, the authors should have assessed the Rodriguez-Oroz et al. study using the ROBINS-I checklist for non-randomized studies . A more serious mistake here is that the authors marked the allocation concealment as (yes +) and the blinding as (yes +) which seems impossible since the Rodriguez-Oroz et al. study was not RCT and is no concealed allocation or blinding in non-RCT studies.
Another issue in the risk of bias assessment is that the study of Weaver et al. was marked as (no -) in terms of the allocation concealment and (unclear ?) in terms of other source of bias while Follet 2010  and Katz 2015  were considered as (yes +) and (no -) in terms of allocation concealment and other bias, respectively. Given that the three reports are describing the same multicentre RCT (Veterans Administration Cooperative Studies Program #468), the risk of bias in these three reports should be similar as they represent the same clinical trial. However, the authors mistakenly judged the risk of bias scores of Weaver et al. (4 out of 6) while both Follet 2010  and Katz 2015  scored 6.
(3) It is known that higher scores of UPDRS indicate worse PD symptoms while a decrease in UPDRS-III indicates the improvement in motor functions. Therefore, the mean change in UPDRS scores in all included studies was calculated as MD=post–baseline, therefore, the changes in UPDRS scores after DBS include a minus sign highlighting a decrease in the UPDRS score. Unfortunately, the authors ignored the direction of the scale and therefore, the presented effect estimates of individual studies and the pooled effect estimate are in the opposite direction.
It is important to highlight that this meta-analysis was based on several methodological errors and if corrected by removing the duplicate studies, most of the provided plots will not be reliable statistically due to an insufficient number of included studies (one study in some outcomes). The evidence provided by this meta-analysis is not robust and is not conclusive; Further studies are still needed to evaluate the long-term efficacy of STN DBS and GPi DBS in patients with advanced PD.
Ahmed Negida; Faculty of Medicine, Zagazig University, Zagazig, El-Sharkia, Egypt
 Peng L, Fu J, Ming Y, et al. (2018) The long-term efficacy of STN vs GPi deep brain stimulation for Parkinson disease. Medicine (Baltimore) 97:e12153 . doi: 10.1097/MD.0000000000012153
 Higgins J, Green S (2011) Cochrane handbook for systematic reviews of interventions Version 5.1. 0. [updated March 2011]
 Moher D, Liberati A, Tetzlaff J, Altman DG (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 6:e1000097 . doi: 10.1371/journal.pmed.1000097
 Follett KA, Weaver FM, Stern M, et al (2010) Pallidal versus Subthalamic Deep-Brain Stimulation for Parkinson's Disease. N Engl J Med 362:2077–2091 . doi: 10.1056/NEJMoa0907083
 Weaver FM, Follett KA, Stern M, et al (2012) Randomized trial of deep brain stimulation for Parkinson disease: thirty-six-month outcomes. Neurology 79:55–65 . doi: 10.1212/WNL.0b013e31825dcdc1
 Katz M, Luciano MS, Carlson K, et al (2015) Differential effects of deep brain stimulation target on motor subtypes in Parkinson's disease. Ann Neurol 77:710–719 . doi: 10.1002/ana.24374
 Rodriguez-Oroz MC, Obeso JA, Lang AE, et al (2005) Bilateral deep brain stimulation in Parkinson's disease: a multicentre study with 4 years follow-up. Brain 128:2240–9 . doi: 10.1093/brain/awh571
 Deep-Brain Stimulation for Parkinson's Disease Study Group, Obeso JA, Olanow CW, et al (2001) Deep-brain stimulation of the subthalamic nucleus or the pars interna of the globus pallidus in Parkinson's disease. N Engl J Med 345:956–63 . doi: 10.1056/NEJMoa000827
 Sterne JA, Hernán MA, Reeves BC, et al (2016) ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ i4919 . doi: 10.1136/bmj.i4919