Evidence-based approach for intraabdominal drainage in pancreatic surgery: A systematic review and meta-analysis

Abstract

BACKGROUND

Historically intraoperative drains were employed after pancreatic surgery but over the last decade, there has been debate over the routine usage of drains.

AIM

To assess the necessity of intra-abdominal drain placement, identify the most effective drain type, and determine the optimal timing for drain removal.

METHODS

A systematic review of electronic databases, including PubMed, MEDLINE, PubMed Central, and Google Scholar, was conducted using Medical Subject Headings and keywords until December 2023. From an initial pool of 1910 articles, 48 were included after exclusion and screening. The primary outcomes analyzed were clinically relevant postoperative pancreatic fistula (CR-POPF), delayed gastric emptying (DGE), overall morbidity, and mortality. Subgroup analyses were performed for pancreaticoduodenectomy and distal pancreatectomy.

RESULTS

Routine use of drains is associated with a statistically significant increase in the risk of CR-POPF and DGE. Conversely, patients who did not have drains placed experienced a significant reduction in morbidity, readmission rates, and reoperations. No significant differences were observed between active and passive drain types. Early drain removal (< 3 days) yielded favorable outcomes compared to delayed removal.

CONCLUSION

Analysis of randomized controlled trials and cohort studies did not demonstrate an advantage of routine drain placement following pancreatic resection, potentially contributing to increased morbidity and mortality. The decision to use drains should be left to the discretion of the operating surgeon. However, early drain removal can substantially reduce morbidity.

Key Words: Intraabdominal drain; Pancreatic resection; Post-operative pancreatic fistula; Delayed gastric emptying; Early drain removal; Drainage duration; Post pancreatectomy drainage

Core Tip: Routine intraoperative drain placement after pancreatic surgery increases the risk of clinically relevant postoperative pancreatic fistula (CR-POPF) and delayed gastric emptying. Patients without drains had lower morbidity, readmission rates, and reoperations. The necessity of routine drain placement is questionable. No clear recommendation can be made between active suction and passive gravity drainage methods. Early drain removal is supported to reduce the occurrence of CR-POPF and associated morbidity and mortality.

INTRODUCTION

Pancreatic resection, encompassing procedures like pancreaticoduodenectomy (PD) and distal pancreatectomy (DP), has evolved to become safer especially with advancements in surgical techniques and perioperative care[1,2]. Despite a mortality rate of less than 5%, pancreatic surgery still presents a considerable challenge due to high morbidity rates ranging from 30% to 60%, primarily attributed to complications such as postoperative pancreatic fistula (POPF)[3]. Given the significant impact on patient’s quality of life and healthcare costs, there is a need to focus on strategies to mitigate POPF occurrence and improve postoperative outcomes[4,5].

The routine placement and management of intra-abdominal drains after pancreatic resection have garnered significant attention and discussion in recent years[6-10]. The practice of inserting prophylactic intra-abdominal drains has roots dating back to the 19^th century, driven by the belief that these drains could evacuate fluids such as blood, bile, and pancreatic juice that might accumulate post-surgery[11]. In modern pancreatic surgery, the management of intra-abdominal drains has become a vital consideration, as indicated by earlier research highlighting its substantial impact on the frequency of postoperative complications[12]. Additionally, drains play a role in mitigating complications related to POPF and aid in the detection of other intra-abdominal hemorrhage. However, the contrarian view of routine use of abdominal drains can lead to retrograde infection, discomfort, foreign body reactions, and prolonged hospital stays[13]. Furthermore drains, which generate substantial negative pressure may contribute to the development of POPFs.

One of the pivotal questions concerning the necessity of placing drains after pancreatic resection was addressed by Conlon et al[7] which marked the initial evidence-based approach, reinforcing the idea that intra-abdominal drains should not be deemed mandatory. However, subsequent randomized controlled trials (RCTs) yielded conflicting results on the efficacy of drains[6,10]. Several systematic reviews and meta-analyses have since been conducted on the role of routine drainage after pancreatic resection. Unfortunately, the available evidence remains limited, underscoring the need for establishing evidence-based guidelines for prophylactic intraabdominal drainage in pancreatic surgery[13,14]. Thus, this systematic review and meta-analysis were aimed to analyse the most recent evidence based data and to answer the following questions: (1) Is the drainage at the operative site more beneficial after pancreatectomy compared to no drain; and (2) If drainage is used, how long should it remain, and what type of drainage should be preferred.

MATERIALS AND METHODS

Search strategy

A comprehensive literature search was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines[15]. PubMed, MEDLINE, PubMed Central, and Google Scholar databases were systematically reviewed using Medical Subject Headings terms and keywords up to December 2023. The search strategy outlined in (Table 1) aimed to identify the relevant studies. After applying inclusion criteria and removing duplicates, abstracts of shortlisted articles were screened, followed by independent full-text review by two authors.

Table 1 Search strategy for electronic databases.

Search strategy	Database
(("Drainage" [Mesh] OR "Drainage, Surgical" [Mesh] OR "Suction" [Mesh] OR "Drainage, Closed" [Mesh]) AND ("Pancreatectomy" [Mesh] OR "Pancreatic resection" [Mesh] OR "Distal pancreatectomy" [Mesh] OR "Pancreatic tail resection" [Mesh] OR "Pancreaticoduodenectomy" [Mesh] OR "Whipple procedure" [Mesh] OR "Duodenopancreatectomy" [Mesh])	PubMed, PubMed Central, and MEDLINE
“Intraabdominal drainage”, “Pancreatic resection”, “Distal pancreatectomy”, “Pancreaticoduodenectomy” separately and in combination	Google Scholar

Mesh: Medical Subject Headings.

Inclusion criteria

Studies comparing the following parameters were included: (1) Intrabdominal drainage vs no drainage; (2) Active (suction) vs passive (gravity) drainage; (3) Early vs late (traditional) drain removal; and (4) Selective drain usage. Studies in which the full text is available in English, and studies that were conducted on humans.

Exclusion criteria

Review articles, letters to the editor, case reports, case series, systematic reviews, meta-analyses, and animal studies were excluded.

Data extraction and quality assessment

Data extraction was carried out by two researchers independently using standardized forms. The quality of the RCTs was evaluated using the Cochrane risk of bias tool while the quality of the non-RCTs was evaluated using the Newcastle-Ottawa scale.

Outcome of interest

The main outcomes considered were the clinically relevant POPF (CR-POPF), delayed gastric emptying (DGE), and overall morbidity rate. The outcome was measured in terms of the reoperation, readmission, length of hospital stay, and overall mortality rate.

Statistical analysis

For continuous variables, we utilized the inverse variance method to determine the standardized mean differences and 95%CI. Dichotomous outcomes were assessed using risk ratios and 95%CI were calculated through the Mantel-Haenszel model. The results were visually displayed in forest plots, and a random effects model was applied to estimate pooled odds ratios (OR) for postoperative complications. Heterogeneity was assessed using the χ² test and I² statistic, while publication bias was examined through Egger’s test and funnel plots. The analysis was conducted using R programming version 4.2.

RESULTS

Study selection

Initially, 1910 studies were identified from major databases (Figure 1). After removing 40 duplicates and excluding 1749 studies not meeting inclusion criteria, full texts of 121 remaining studies were retrieved. Following further screening, 73 studies were eliminated for inconsistency with inclusion criteria, and two inaccessible full-text studies were removed. Ultimately, 48 articles were included, focusing on comparisons between presence or absence of drainage, active (closed suction) vs passive (gravity) drainage, and early vs traditional removal of drainage.

Figure 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses Flowchart.

Intra-abdominal drain and without drain

In the current systematic review and meta-analysis, we meticulously examined a total of 22 studies, comprising five RCTs, and the remaining were retrospective, or cohort studies as shown in Supplementary Table 1[4,6-10,16-31]. These studies collectively encompassed a broad population, involving a total of 30029 participants.

CR-POPF

Twelve studies, comprising 16754 patients in the drain group and 3402 patients in the no-drain group, examined CR-POPF outcomes in PD[4,6,8-10,18-27,29-31]. Substantial heterogeneity was observed (I² = 90%, P < 0.01) (Figure 2A)[4,8,10,18,20-22,24-27,30]. The pooled analysis revealed a significantly higher CR-POPF rate in the drain group (OR = 1.58, 95%CI: 1.04-2.38), supported by a symmetrical funnel plot and non-significant Egger test (P = 0.7613) (Figure 3A). Additionally, seven studies involving 2329 patients in the drain group and 899 in the no-drain group examined CR-POPF outcomes in DP[6,8,9,19,29-31]. Significant heterogeneity was noted (I² = 55%, P = 0.04), with the drain group exhibiting a significantly higher CR-POPF rate (OR = 2.71, 95%CI: 1.73-4.24), supported by a symmetrical funnel plot (Figure 2B, Figure 3B).

Figure 2 Forest plot. A: Forest plot showing the association of clinically relevant postoperative pancreatic fistula (CR-POPF) incidence between the drain and no drain groups among patients who underwent pancreaticoduodenectomy (PD); B: Forest plot showing the association of CR-POPF between the drain and no drain groups among patients who underwent DP; C: Forest plot showing the association of delayed gastric emptying (DGE) between the drain and no drain groups among patients who underwent PD; D: Forest plot showing the association of DGE between the drain and no drain groups among patients who underwent DP; E: Forest plot showing the association of overall morbidity between the drain and no drain groups among patients who underwent PD; F: Forest plot showing the association of overall morbidity between the drain and no drain groups among patients who underwent DP; G: Forest plot shows the association of mortality between the drain and no drain groups among patients who underwent PD; H: Forest plot shows the association of mortality between the drain and no drain groups among patients who underwent DP. POPF: Postoperative pancreatic fistula; DGE: Delayed gastric emptying.

Figure 3 Funnel plot. A: Funnel plot of clinically relevant postoperative pancreatic fistula (CR-POPF) between the drain and no-drain groups among patients who underwent pancreaticoduodenectomy (PD). Egger test result: T = 0.31, df = 10, P value = 0.7657; B: Funnel plot of CR-POPF between the drain and no-drain groups among patients who underwent DP; C: Funnel plot of delayed gastric emptying between the drain and no drain groups among patients who underwent PD; D: Funnel plot of overall morbidity between the drain and no drain groups among patients who underwent PD; E: Funnel plot of mortality between the drain and no-drain groups among those who have undergone PD.

DGE

In the pooled analysis of eight studies focusing on DGE post-PD, the drain group (n = 13276) exhibited a significantly higher incidence compared to the no-drain group (n = 2365), with OR of 1.36 (95%CI: 1.18-1.55)[4,10,20,21,23,26,27,30]. Moderate heterogeneity was observed (I² = 33%, P = 0.16), and the funnel plot was symmetrical (Figure 2C, Figure 3C). Conversely, pooled analysis of two studies on DGE post-DP showed no significant correlation between the drain (n = 861) and no-drain (n = 214) groups, with a pooled OR of 3.79 (95%CI: 0.6–23.8). No heterogeneity was present among the included studies (I² = 0%, P = 0.44) (Figure 2D)[30,31].

Morbidity

Pooled analysis of nine studies reporting the outcome of overall morbidity in PD showed no significant difference between the drain group (n = 16167) and the no-drain group (n = 2847) with OR of 1.18 (95%CI: 0.82-1.7) (Figure 2E)[4,10,20-22,24-27]. There was considerable heterogeneity present among the included studies (I² = 90%, P < 0.01). The funnel plot was symmetrical (Figure 3D).

Pooled analysis of four studies reporting the outcome of overall morbidity in DP showed no significant difference between the drain group (n = 1314) and the no-drain group (n = 489) with OR of 0.95 (95%CI: 0.45-2.02) (Figure 2H)[6,9,19,29]. There was substantial heterogeneity present among the included studies (I² = 83%, P < 0.01) (Figure 2F).

Mortality

Pooled analysis with seven studies reporting mortality post-PD showed a significantly lower incidence of mortality among the drain group (n = 13623) than in the no-drain group (n = 2904), with an OR of 0.59 (95%CI: 0.44–0.79) (Figure 2G)[8,19-21,25-27]. No heterogeneity was present among the included studies (I² = 0%, P = 0.63). The funnel plot was symmetrical (Figure 3E).

Pooled analysis of four studies reporting the outcome of mortality in DP showed no significant difference between the drain group (n = 2118) and the no-drain group (n = 697) with OR of 1.99 (95%CI: 0.51-7.87) (Figure 2H)[8,19,29,31]. No heterogeneity was present among the included studies (I² = 0%, P = 0.62).

Active closed suction and passive gravity drainage

A comprehensive analysis of ten studies, including three RCTs and seven retrospective or cohort studies, involving 24475 patients as shown in Supplementary Table 2[32–41]. While four studies reported a higher incidence of CR-POPF with closed suction drainage[35,37,39,41], others employing gravity drainage showed no significant difference in complication rates. Secondary outcomes indicated slightly more adverse events with active closed suction drainage but no notable impact on 30-day postoperative mortality, overall morbidity, or re-intervention rates. These findings suggest that the choice of drainage method has limited effect on specific outcomes in pancreatic surgery.

Early and late drain removal

A review of sixteen studies, including five RCTs and eleven retrospective cohort studies, focused on the timing of drain removal following pancreatic surgery as shown in Supplementary Table 3[13,42–56]. Definitions of "early removal" varied, with most studies considering removal within 3 days. Late removal (> 5 days) was associated with higher surgical complication rates, though not statistically significant. Delayed removal correlated with increased rates of CR-POPF, bile leak, Clavien Dindo grade ≥ 3 complications, reoperations, readmissions, and 30-day mortality. These findings highlight the importance of timely drain removal for better postoperative outcomes in pancreatic surgery.

DISCUSSION

The historical tradition of employing prophylactic peritoneal drainage following gastrointestinal surgery, epitomized by the adage "when in doubt, drain", lacks compelling contemporary data validating its efficacy[57]. While drains placed near anastomoses serve to remove pancreatic juice or bile and can indicate complications, they come with associated risks such as infections, abdominal pain, and prolonged hospital stays. Despite their routine use for preventing POPFs, the clinical impact can still be significant, and timely identification is crucial for mitigating severe complications[33]. Thus, the ongoing debate within the surgical community questions the necessity of prophylactic drain placement and emphasizes the need for evidence-based practices in this regard.

Intra-abdominal drain and without drain

In our comprehensive review, comparing outcomes between patients with and without drainage after pancreatic resection revealed significant differences. The drainage cohort exhibited higher rates of CR-POPFs and DGE, along with prolonged hospital stays and increased intervention requirements. Despite higher rates of CR-POPFs and DGE in the drainage cohort, this group exhibited a lower mortality rate. The presence of a drain allows for early detection and effective management of complications, preventing severe outcomes like sepsis and organ failure. Consequently, the benefits of early intervention in the drainage group have likely contributed to the observed reduction in mortality. Subgroup analysis of PD and DP revealed consistent findings indicating increased risks of CR-POPFs in both PD and DP groups, and increased DGE specifically in the PD group. These findings underscore the potential disadvantages of routine drainage in pancreatic surgery.

A meta-analysis of 15290 patients from ten studies found a higher incidence of CR-POPF in the drainage group compared to the no-drainage group[58]. However, there was no substantial correlation between drainage and DGE. Subgroup analyses for PD and DP showed comparable outcomes. Additionally, comprehensive analysis of eleven studies on postoperative complications, including PPH, intra-abdominal abscess, wound infection, and reoperation, revealed no notable differences between the groups[4,7,8,16,19,20,24,27,29,31,32]. However, the group with drains exhibited a significantly higher rate of readmission compared to the group without drain[58].

Conflicting findings from recent studies suggest varied impacts of drainage in pancreatic surgery. Brubaker et al[22] found higher rates of POPF, reoperation, and serious morbidity in patients without drainage, while Fisher et al[16] reported increased incidences of POPF, DGE, and readmission with drain usage. Other studies, such as those by Paulus et al[9] and Van Buren et al[10], found no significant differences in outcomes with or without drainage. Behrman et al[6] and Mangieri et al[29] observed increased morbidity and readmission rates associated with drain placement, and Liu et al[58] suggested a potential elevation in POPF incidence with use of drainage. These findings suggest caution in routine drainage employment during pancreatic resection.

Active and passive drainage

Patients undergoing surgical drain placement commonly experience two main drainage methods: (1) Closed-suction drains (CSDs); and (2) Passive gravity drains. CSDs function by generating a negative pressure gradient, ensuring continuous suction within the abdominal cavity irrespective of the patient's position. In contrast, gravity drainage devices establish a route for fluid extraction by leveraging the pressure difference between intra-abdominal and atmospheric pressure. In our review, we found significant differences between CSDs, and passive gravity drains for patients undergoing pancreatic resection. CSDs were associated with higher rates of CR-POPF and DGE, while passive gravity drains led to longer hospital stays and a greater need for intervention but lower mortality. Čečka et al[34] found no significant differences in rates of CR-POPF or overall morbidity between groups using different drainage methods after PD. Additionally, there were no notable variations in reoperation rate, readmission rate, length of hospital stay, or incidence of post-pancreatectomy hemorrhage. Lee et al[32] demonstrated that utilizing a closed-suction drainage system with an external pancreatic duct stent led to a significant reduction in CR-POPF following PD. The system's negative pressure effectively diverted pancreatic juice away from the anastomotic site, potentially enhancing healing and long-term patency of the pancreatic duct.

Veziant et al[3] found no significant differences in outcomes between active and passive drainage methods, including rates of CR-POPF, overall morbidity, and length of hospital stay. Similarly, Aumont et al[35] reported that gravity drainage was independently associated with lower rates of CR-POPF, DGE , and readmission following PD. Overall, the literature suggests that the choice of drainage method does not significantly impact postoperative outcomes, particularly after PD, and may be left to surgeon preference.

Early and late drain removal

The absence of a universally agreed-upon definition for early removal of abdominal drainage following pancreatic surgery has led to a typical interval ranging from 3 days to 5 days. Late removal is generally considered when drains are in place for at least five days. In our systematic review, we found significant differences between early and late removal of abdominal drainage following pancreatic surgery. Late removal (≥ 5 days) was associated with higher rates of complications such as CR-POPF, DGE, severe morbidity, prolonged hospital stays, and increased need for intervention and readmission. However, mortality rates did not differ notably. Therefore, if drainage is necessary, early removal within three days seems to offer more benefits.

Xourafas et al[50] found that early drain removal in patients undergoing PD, especially those with POD1 amylase levels of 5000, was associated with improved perioperative outcomes regardless of the Fistula Risk Score (FRS). Both high-risk and low-risk modified FRS patients showed reduced rates of CR-POPF, shorter hospital stays, and overall morbidity with early drain removal. In the RCT by Bassi et al[42] which included both PD and DP, the results posed some interpretative challenges, despite the notable reduction in POPF and overall morbidity associated with early drain removal. Conversely, Dembinski et al[44] focusing on PD patients did not show statistically significant differences between groups, although there was a trend toward reduced rates of complications and shorter hospital stays with early drain removal, potentially limited by statistical power of the study. Overall, while the definitive benefits of early drain removal remain uncertain, evidence suggests it may facilitate earlier recovery without increasing complications.

The meta-analysis faced limitations due to predominantly nonrandomized studies, potentially biasing results. Several studies lacked comprehensive data on secondary outcomes. High heterogeneity, indicated by I² statistic and low P-value, urges cautious interpretation. Further well-designed randomized trials with robust data collection are crucial to enhance understanding of intra-abdominal drains in pancreatic surgery.

CONCLUSION

After a comprehensive review of 48 articles encompassing patients with PD and DP, our study refrained from making specific recommendations regarding the routine use of drainage. However, an increasing body of evidence suggests that the routine placement of primary drains is not mandatory. Our systematic review does not yield a definitive recommendation concerning the choice between active suction or passive gravity drainage following pancreatic resection. The choice of the drainage method can be at the discretion of the surgeon. Furthermore, our findings support the safe practice of early drain removal for patients undergoing pancreatic surgery which may decrease the occurrence of post operative pancreatic fistula and mitigate the associated morbidity and mortality. Recognizing the potential contribution of intra-abdominal drainage to increased morbidity, further research should be done in this area. This approach will aid in refining the guidelines to optimize drainage practices, balancing the benefits and risks, and promoting evidence-based decision-making in pancreatic surgery.

ACKNOWLEDGEMENTS

We extend our sincere gratitude to Naga KG, John AG, Anto JV and Yadav S for their invaluable contributions and support in this study. Their expertise and dedication have been instrumental in shaping our research, and we are deeply appreciative of their guidance and collaboration throughout the project.