RESUMO
BACKGROUND: The superiority of early drain removal (EDR) versus late (LDR) after pancreaticoduodenectomy (PD) has been demonstrated only in RCTs. METHODS: A meta-analysis was conducted using a random-effects model and trial sequential analysis. The critical endpoints were morbidity, redrainage, relaparotomy, and postoperative pancreatic fistula (CR-POPF). Hemorrhage (PPH), delayed gastric emptying (DGE), length of stay (LOS), and readmission rates were also evaluated. Risk ratios (RRs) and mean differences (MDs) with a 95% confidence interval (CI) were calculated. Type I and type II errors were excluded, comparing the accrued sample size (ASS) with the required sample size (RIS). When RIS is superior to ASS, type I or II errors can be hypothesized. RESULTS: ASS was 632 for all endpoints except DGE and PPH (557 patients). The major morbidity (RR 0.55; 95% CI 0.32-0.97) was lower in the EDR group. The CR-POPF rate was lower in the EDR than in the LDR group (RR 0.50), but this difference is not statistically significant (95% CI 0.24-1.03). The RIS to confirm or exclude these results can be reached by randomizing 5959 patients. The need for percutaneous drainage, relaparotomy, PPH, DGE, and readmission rates was similar. The related RISs were higher than ASS, and type II errors cannot be excluded. LOS was shorter in the EDR than the LDR group (MD - 2.25; 95% CI - 3.23 to - 1.28). The RIS was 567, and type I errors can be excluded. CONCLUSIONS: EDR, compared with LDR, is associated with lower major morbidity and shorter LOS.