Background: The comparison of evidence of treatment effects in randomized and nonrandomized studies has showed that there existed good correlation between them. Meanwhile, on account of the inherent nature of the surgery trials, it was almost impossible to perform blinding except blinding of the statisticians. And random sequence generation or allocation concealment was often unclear in the included studies. So, as the new approach, the systematic comparisons between randomized and non-randomized studies are needed to perform to add more reliable evidence.

Objectives: The aim is to introduce this systematic comparison using an example on the stents for Pancreaticoduodenectomy (PD).

Methods: A systematic literatures searching were performed (PubMed, EMBASE, ISI web of knowledge, the Cochrane library, and four Chinese databases), as well as other additional resources, which were finished in November 2012. The relevant clinical trials about pancreatic duct stents were included. Data was reviewed and extracted by two reviewers independently and was analyzed via RevMan 5 (version 5.2.0). The primary and secondary outcomes were concerned.

Results: Based on searches of the literature published before November 2012, 5 randomized clinical trials (RCTs) and 11 non-randomized clinical studies (nRCTs) were included. The meta-analysis showed that compared with non-stent, the pancreatic duct stent following PD was associated with a statistically significant reduction on overall postoperative pancreatic fistula rate (POPF), as well as the postoperative morbidity, overall mortality, delayed gastric emptying and intra-abdominal collections (the results were showed in Table 1).

Conclusions: The systematic comparisons between randomized and non-randomized studies could provide the extra evidence from the present data and also make the evidence more reliable.