Surgical Drain Placement in Distal Pancreatectomy is Associated with an Increased Incidence of Post-Operative Pancreatic Fistula and Higher Readmission Rates

Abstract

Background

Postoperative pancreatic fistula (POPF) can result in significant morbidity after distal pancreatectomy (DP). It is common practice to place prophylactic surgical drains during DP to monitor and minimize POPF complications; however, their use is controversial.

Objective

The aim of this study is to determine if drainage helps to prevent adverse outcomes and decrease the need for additional interventions after DP.

Methods

All patients who underwent DP without vascular resection were identified in the 2014 Targeted Pancreatectomy ACS-NSQIP Participant Use File. Patients undergoing emergency procedures, ASA 5, or diagnosed with preoperative sepsis were excluded. Univariate and multiple variable analyses were performed to evaluate postoperative outcomes based on use of surgical drain.

Results

1158 patients (age median 62, IQR 16; female 58.6%) underwent elective DP with 85.1% (n=985) having drain placed at time of operation. Laparoscopic technique was used in the majority of patients (54.1%, n=619). POPF occurred in 201 patients (17.5%). Additional percutaneous drain was required in 106 patients (9.2%). POPF was higher in surgical drain group, 19.4% vs. 6.9% (p<0.001). Need for percutaneous drain was similar between drain and no drain groups, 9.3% vs 8.1% (p=0.600). Postoperative sepsis, shock, major complication, reoperation, and 30-day mortality was similar between drain and no drain groups (all p > 0.05). However, readmission was higher in the surgical drain group, 17.8% vs 10.4% (OR 1.9, 95% CI 1.1–3.1, p=0.018). After adjusting for age, ASA, and operative time, readmission remained higher in the surgical drain group (OR 1.9, 95% CI 1.1–3.2, p=0.016).

Conclusion

The use of surgical drainage during DP was associated with increased incidence of readmission and POPF. Drainage showed no effect on outcomes of postoperative sepsis, shock, major complications, reoperation and 30-day mortality. Based on these results, routine prophylactic drainage should be reconsidered for patients undergoing DP.

Background

For decades the routine practice of many surgeons performing pancreatic resection has included placement of a closed suction drainage catheter in the surgical bed. Whether during pancreaticoduodenectomy (PD) or distal pancreatectomy (DP), these drains are placed to assess, manage and evacuate pancreatic fluid that accumulates, and theoretically to control and mitigate the effects of postoperative pancreatic fistula (POPF). Considering that POPF persists as a common complication following pancreatectomy despite advancements in surgical technique and patient management, much attention has been placed on the utility of surgically placed drains in pancreatic resection. Studies have focused on the early discontinuation of closed suction drainage catheters, as well as assessing the effects of omitting them altogether¹. Few randomized controlled trials (RCTs) have been performed to establish the value of pancreatic bed drainage, and those that exist have produced conflicting results^2,6,17,20.

As a result of the limited evidence and emergence of enhanced recovery pathways, pancreatic surgeons have questioned routine prophylactic drainage. Van Buren et al. reported that the absence of intraperitoneal drains in PD increases the frequency and severity of complications.² A single institution experience reported by Yamashita et al. observed increased complications with drainage after DP³. Additional observational studies of patients undergoing DP have shown that the use of prophylactic drainage shows no benefit in terms of decreasing postoperative fluid collections or preventing the need for further percutaneous drainage^4,5. A randomized control trial by Conlon et al. reported that routine drainage following pancreatectomy is associated with an increased rate of postoperative fluid collections, abscesses, and POPF⁶.

Outcomes from routine surgical drain placement for DP at institutions across the United States is limited with our best studies restricted to single institution or limited clinical trials. We hypothesized that drainage of the DP surgical bed at the time of the index operation would decrease adverse events and obviate the need for additional drainage interventions. Therefore, the aim of the current study was to understand the postoperative outcomes of routine surgical drain placement for patients undergoing DP utilizing the Targeted Pancreatectomy American College of Surgeons National Surgery Quality Improvement Program (ACS-NSQIP) registry. Specifically, seeing if drain utilization decreased postoperative adverse events.

Methods

Our Institutional Review Board approved this retrospective comparative cohort study using the 2014 ACS-NSQIP Participant Use Data File (PUF). ACS-NSQIP is a nationally-validated database that provides risk-adjusted, outcomes-based data provided from participating hospitals²⁵. It generates an annual PUF that provides aggregate data regarding patient outcomes, using information provided directly from patient charts, without identifying patients, surgeons, or institutions. Beginning in 2014, procedure targeted cases were captured for vascular, colectomy, gynecology, hepatectomy, and pancreatectomy²⁶. The procedure targeted PUF provides procedure-specific variables and outcome measures. The 2014 procedure targeted pancreatectomy PUF includes 5,187 cases submitted by 106 hospitals.

Patient Selection

All patients who underwent DP (Common Procedural Terminology code 48140 pancreatectomy, distal subtotal, with or without splenectomy, without pancreaticojejunostomy) were identified in the 2014 ACS-NSQIP Targeted PUF (n =1535).²⁶ The following cases were excluded: any procedures where the chart included jaundice, preoperative biliary stent, American Society of Anesthesiology (ASA) grade 5, emergency, or serious preoperative condition (coma, pneumonia, wound infection, ventilator dependent, renal failure, blood transfusion, sepsis). In addition, operations requiring pancreatic anastomosis, vascular resection, enteric anastomosis, or biliary drain placement were excluded. (Figure 1).

Figure 1 –

Selection criteria for inclusion in the analysis.

Study Variables

Two study cohorts, surgical drain and no surgical drain, were derived from the 2014 Pancreatectomy Targeted Procedure PUF. The targeted pancreatectomy PUF includes procedure specific variables to include drain utilization; cases that did not contain drain information were excluded. Clinicopathologic variables were captured including sex, race, ASA, functional status, comorbidities, nutritional status, pathology, tumor stage, tumor size, and disposition. Outcome measures included length of stay, major complication, morbidity, reoperation, readmission, and mortality. Major complications were defined as Clavien–Dindo class III/IV complications (life threatening or requiring intensive care management: unplanned intubation, failure to wean from ventilator >48 hours, acute renal insufficiency requiring dialysis, new-onset neurologic deficit or coma, myocardial infarction, cardiac arrest, pulmonary embolism, graft failure, organ space infection). Minor complications were defined as Clavien-Dindo class I/II complications (surgical site infection, pneumonia, deep vein thrombosis, urinary tract infection, peripheral nerve injury, postoperative bleeding requiring transfusion). Mortality was defined as death within 30-days of surgery.

Procedure specific complications included percutaneous drain placement, delayed gastric emptying, and postoperative pancreatic fistula as defined by the ACS-NSQIP Pancreatectomy Targeted Procedure definitions.²⁶ POPF was defined by the most current International Study Group of Pancreatic Fistula (ISGPF) classification scheme revised in 2016 to biochemical leaks (previously grade A fistula) and true POPF (grade B and C fistulas)

Statistical Analysis

The sample means and standard deviations were computed for the continuous descriptive characteristics, and the count and proportions were calculated for the discrete descriptive characteristics, by drain or no drain groups. Clinicopathologic characteristics were compared between patients with surgical drain and those without surgical drain using chi-squared statistics for categorical and Wilcoxon testing for continuous variables. Univariate analysis using logistic regression, identified clinically significant factors associated with development of any complication, major complications and mortality and were expressed as odds ratios. Multiple variable models were created to evaluate the risk associated with surgical drain placement. The covariates included age, ASA, smoking status, operative time, and albumin (≥ 3 vs. <3). All analyses were performed using SAS version 9.4 (SAS Inc., Cary, NC).

Results

1158 cases of elective DP were identified in the 2014 Targeted Pancreatectomy ACS-NSQIP PUF that met inclusion criteria. The median age was 60.3 years (IQR 16) with 58.6% of the patients being female. In all, 985 of 1158 (85.1%) patients had a drain placed in the surgical bed at time of DP. Postoperative drain amylase level was available in 715 patients (71.6%) with 483 patients (41.7%) having a drain amylase level above 300. The NSQIP targeted pancreatectomy database records the highest drain amylase level within the first 30 days postoperative. Demographic and operative variables for patients who did or did not undergo drainage are presented in Table 1. Surgical drain placement did not vary significantly with age, sex, BMI, obesity, ASA class, smoking status, or history of COPD. Surgical drain was more common in patients with normal albumin (≥ 3) compared with patients with low albumin (<3), 84.5% vs 74.0%, respectively (p < 0.001).

Table 1–

Characteristics of Distal Pancreatectomy (DP) Cohort (n = 1158).

Variable	No Drain (n = 173)	Drain (n = 985)	p value
Age (range)	58 (49–70)	62 (52–70)	0.74
Female Sex (%)	108 (62.4)	571 (58)	0.27
BMI >30 (%)	62 (36.1)	368 (37.4)	0.735
ASA Class 3–4 (%)	118 (68.2)	654 (66.4)	0.635
Smoker (%)	31 (17.7)	154 (15.6)	0.257
COPD (%)	7 (4.1)	31 (3.1)	0.49
†Preoperative Albumin >3 (%)	128 (74)	832 (84.5)	<0.001
Open Approach (%)	78 (45.3)	454 (46.1)	0.855
Operative Duration, minutes (range)	212 (167–258)	206 (160–275)	0.804

^†Statistically significant

Using univariate analysis the clinical outcomes after surgical drainage versus no surgical drainage were compared after DP Table 2. There was no statistically significant difference between 30-day mortality, major/minor complication, or sepsis rates between the drain and no drain groups. Drain placement was not associated with lower proportion with postoperative sepsis, shock, or discharge to nursing facility (all p > 0.05). Placement of percutaneous drain was similar between surgical drain and no surgical drain groups, 9.3% vs 8.1% (p = 0.6). Patients who underwent drainage of the pancreatic bed developed a POPF in 19.4% of cases, whereas those without drains only developed a POPF in 6.9% of cases (p = 0.001). Only Grade B and C fistulas are included in the POPF rates per the most current ISGPF criteria. More than 10% of patients still had their surgical drain present more than 30 days from operation.

Table 2–

Comparison of Clinical Outcomes after Distal Pancreatectomy (DP) with Surgical Drain or No Surgical Drain.

Variable	No Drain (n = 173)	Drain (n = 985)	p value
Mortality (%)	1 (0.57)	5 (0.51)	0.993
Sepsis (%)	8 (4.6)	47 (4.8)	0.888
Shock (%)	3 (1.7)	11 (1.1)	0.493
Discharge to Nursing Facility (%)	9 (5.2)	52 (5.3)	0.800
Minor Complication (%)	30 (17.5)	143 (14.5)	0.331
Major Complication (%)	11 (6.4)	55 (5.6)	0.685
Bleeding Complication (%)	6 (3.5)	35 (3.6)	0.955
Need for Percutaneous Drain (%)	14 (8.1)	92 (9.3)	0.600
Reoperation (%)	5 (2.9)	27 (2.7)	0.912
†POPF Development (%)	12 (6.9)	191 (19.4)	0.001
†Unplanned Readmission (%)	18 (10.4)	175 (17.8)	0.016
Length of Stay (range)	5 (4–6)	5 (4–7)	0.965

^†Statistically significant

Minor complication defined as Clavien-Dindo classes I-II. Major Complication defined as Clavien-Dindo classes III-IV. POPF defined by 2016 update of the International Study Group of Pancreatic Fistula (ISGPS) classification scheme

Similarly, placement of a surgical drain resulted in an unplanned 30-day readmission of 17.8% compared to 10.4% when a drain was not placed (p = 0.016; Fig 2). Univariate analysis of risk factors associated with readmission are reported in Table 3. Using multiple variable analysis, placement of a pancreatic bed drain remained as a statistically significant risk for unplanned readmission after adjusting for age, BMI, ASA class, and operative time, (OR 1.9, 95% CI 1.1–3.2, p= 0.016).

Figure 2 –

Differences in surgical outcomes according to presence or lack of a surgicaUy placed drain at time of distai pancreatectomy. POPF, post-operative pancreatic fistula. * signifies P < 0.05 relative to no drain.

Table 3–

Univariate Analysis of Risk Factors Associated with Unplanned 30-day Readmission.

Variable	Odds Ratio	95% Confidence Interval	p value
†Percutaneous Drain Placement	15.46	9.89–24.17	<0.001
†Any Complication	2.59	2.15–3.12	<0.001
†Major Complication	14.81	10.30–21.31	<0.001
†Minor Complication	2.31	1.33–4.00	0.003
†Sepsis	9.17	5.23–16.08	<0.001
†POPF Development	5.46	3.87–7.72	<0.001
†Reoperation	3.60	1.75–7.41	<0.001
†Surgical Drain Placement (Index Surgery)	1.86	1.11–3.11	0.018
†Smoker	1.57	1.07–2.31	0.021
Obesity (BMI >30)	1.27	0.93–1.74	0.134
Open Approach	1.26	0.93–1.72	0.141
Male Sex	1.26	0.93–1.72	0.143
ASA Class 3–4	1.20	0.86–1.68	0.286
Race (White/Caucasian)	1.11	0.73–1.67	0.633
Age (per year)	1.00	0.99–1.00	0.630

^†Statistically significant

Minor complication defined as Clavien-Dindo classes I-II. Major Complication defined as Clavien-Dindo classes III-IV. POPF defined by 2016 update of the International Study Group of Pancreatic Fistula (ISGPS) classification scheme

Discussion

Clinically significant grade B and C pancreatic fistula remains a common complication following DP. The current study of a large cohort of patients in the United States reports that drain placed at time of distal pancreatectomy is associated with higher POPF. Surgical drain placement was also associated with high incidence of readmission.

Numerous strategies to prevent and control the sequela of postoperative pancreatic fistula have been evaluated.^2,6,17,20 POPF can differ in their presentation and clinical significance, with amylase level of a surgically placed drains commonly used as an important data value in categorization pancreatic healing following DP.⁷ Incidence of POPF after DP ranges widely from 5–29%.^8–11 Whether or not prophylactic surgical drains play an important role in the managing POPF is controversial. How to best manage drains when they are placed also remains controversial.

While there has been considerable movement in surgery towards leaving less intraabdominal drains for elective cases, there is still hesitancy on the part of many surgeons to forego prophylactic drainage with the pancreas. Indeed, Brooke-Smith and colleagues demonstrated in an international multi-center study that prophylactic drainage after major hepatic resection did not decrease the incidence of clinically significant bile leaks, nor the need to perform subsequent percutaneous drainage.¹² Likewise, Emile et al. concluded that routine drainage of pelvic anastomoses in colorectal surgery is of no demonstrable benefit.¹³

For pancreatectomy, several studies have argued that routine drainage should not be required, but the data that exists is controversial and has focused on PD. Mehta and colleagues found that, for 709 patients undergoing PD, routine drainage resulted in no change in readmission or secondary drainage procedures but did lead to a longer length of stay overall.¹⁴ In a separate meta-analysis by Nitsche that included both PD and DP, omission of a surgical drain was associated with increased mortality.¹⁵

The studies that pertain specifically to DP are limited. Paulis et al. found that in 69 patients who underwent routine drainage after DP the drain did not decrease morbidity or need for a further intervention.¹⁶ The study further suggests that drains were rarely useful in diagnosing complications¹⁶. In the most recent RCT regarding drainage in DP performed by Van Buren et al, 344 patients were randomized to either drain or no drain after DP¹⁷. The authors observed no difference in the rate of POPF between groups, but did note an increase in clinically significant fluid collections in patients who did not receive a prophylactic drain.¹⁷ Unfortunately, these studies do not conclusively define the utility, or lack thereof, for routine drain placement following DP.

Several studies have investigated drain management, specifically comparing outcome differences based on timing of drain removal when drain placement is routinely utilized.^18,19 Wang et al. presented a review article with evidence supporting early drain removal following pancreaticoduodenectomy based on drain amylase levels on postoperative day one.¹⁸ More specific to DP, Adachi and colleagues performed a RCT drain removal trial including 71 patients who underwent DP. They found that for those patients who had their drain removed on postoperative day one had a 0% POPF rate compared to 16% in those patients whose drains were left until on or after postoperative day five.¹⁹ These studies seem to provide reasonable evidence that early drain removal is better than delayed removal, but do not address whether or not drain omission altogether can improve postoperative outcomes.

A selective approach to drain placement has also been advocated for patients assessed to be at high-risk for POPF development. McMillan and colleagues performed an RCT for patients undergoing PD in which they created a fistula risk score that might provide insight into which cases should be drained.²⁰ Wang et al. presented data showing that the risk of POPF is higher in patients receiving high-fluid volume during their operation.²⁷ It is important to note, however, that these studies were restricted to patients who underwent PD.

Drain placement itself may contribute to POPF development. Grobmyer et al. demonstrated that intraabdominal closed suction drains can apply pressures of −71 to −175mmHg alongside the tissue bed they are meant to drain.²¹ Even more, they showed that these pressures increase significantly when the drain is “stripped”, as is common surgical practice with the pressure changes being sustained for several minutes.²¹ While no studies have shown drain placement to directly cause POPF, the pressure dynamics associated with drains could potentiate an environment that promotes fistula development.

The current study demonstrated a significant increase in readmission rates with the use of surgical drain placement following DP. Currently, major pancreatic resection is a high outlier for unplanned readmissions with 15–20% of patients requiring readmission within 30 days of operation.²³ Several centers have focused efforts to decrease readmission rates. One such effort published by Ceppa et al. included multi-disciplinary interventions to decreasing readmissions from major pancreatectomy and over a 5 year period their 30-day readmission decreased from 23% to 11%.²⁴ Importantly, surgical drain placement or avoidance of drain placement was not included as an intervention to decrease readmission.

Previous analysis of NSQIP data has demonstrated an association between placing a drain for DP and morbidity. A 2012 NSQIP study by Behrman and colleagues found that drains were placed in 80% of the DPs included in the database.²² This was associated with a 22% POPF rate compared to just 7% documented POPF in those patients who did not receive a drain.²² Unlike the current study the Behrman study used the main NSQIP database as opposed to the Targeted Pancreatectomy database. The Targeted Pancreatectomy database provides a more detailed analysis of POPF.

While the current study presents outcomes from a large number of institutions across the United States, it does have several limitations. This is a retrospective review of a large deidentified national dataset and not a prospective study. Also, indication for surgical or postoperative drain placement is not available in NSQIP. Whether it is the routine practice of the surgeon to place prophylactic drains or there was an intra-operative assessment, for example a poor staple fire, which lead to deliberate drain placement is unknown. The most likely scenario is routine prophylactic drain placement since nearly 90% of cases utilized a drain; however, we are unable to be unequivocally confirm that within NSQIP data. An additional limitation is for patients who did not have a surgical drain placed there is the possibility that biochemical leaks, previously classified as Class A POPF, went unrecognized.

Conclusions

In the current analysis of 1158 patients who underwent elective DP, drain placed at the time of DP is associated with a higher incidence of POPF and unplanned 30-day readmission. The overall rate of drain placement for DP in the study cohort was 85% indicating that placement of surgical drains is common practice for pancreatic surgeons. Avoiding routine placement of a surgical drain at the time of the index operation may improve clinical outcomes in patients undergoing DP.

Synopsis.

This study evaluated the impact of surgical drain placement during distal pancreatectomy, specifically focusing on adverse post-operative events. The 2014 NSQIP targeted pancreatectomy database was analyzed. Review of nearly 1,200 cases revealed that drain use resulted in increased POPF and readmission incidence. Drain use had not effect on any other evaluated adverse events.