Impact of Negative Pressure Wound Therapy on Outcomes Following Pancreaticoduodenectomy: An NSQIP Analysis of 14,044 Patients

Abstract

Background: Despite ongoing efforts to improve the pancreaticoduodenectomy technique and perioperative care, surgical site infection (SSI) remains a contributor to morbidity. Efforts to reduce SSI include the use of negative pressure wound therapy (NPWT), but studies and meta-analyses have been met with conflicting results. We aimed to provide an up-to-date large-scale cohort study to assess the impact of NPWT on SSIs. Methods: Utilizing the National Surgical Quality Improvement Program database, we included patients undergoing a pancreaticoduodenectomy between 2017 and 2021 and divided patients into the NPWT and non-NPWT cohorts. A bivariate analysis was performed to compare baseline characteristics and complication rates between the cohorts. Multivariate logistic regression analysis was performed to assess the independent effect of NPWT on 30-day serious complication, 30-day mortality, and the development of deep or superficial SSI. A priori sensitivity analyses were performed in high-risk and malignancy cohorts. Results: Of the 14,044 included patients, 1689 (12.0%) patients had a prophylactic NPWT device, while 12,355 (88.0%) did not. Patients were more likely to have NPWT if they had higher ASA scores, had diabetes, were dialysis-dependent, or had a hard pancreas, but they were less likely if they were a smoker, had steroid use, or had a bleeding disorder. Most complications occurred similarly between the two cohorts, including superficial and deep SSI, but NPWT patients had a longer length of stay (10.4 d vs. 9.5 d, p < 0.001) and higher organ space SSI (22.6% vs. 17.4%, p < 0.001). Following multivariable modeling to control for demographic differences, NPWT was not independently associated with a difference in likelihood of SSI (aOR 0.94, p = 0.691) or serious complications (aOR 0.958, p = 0.669). Furthermore, the sensitivity analyses of both high-risk and malignant subgroup also did not see an independent association of NPWT on the rate of SSI (aOR 0.98, p = 0.898 and 0.96, p = 0.788, respectively). Conclusion: NPWT is used infrequently and is not significantly associated with improved outcomes including in the high-risk or malignant subgroups based on multivariable analysis for surgical site infections nor did it improve the outcomes of 30-day serious complications in these subgroups. Considering this and other studies showing the limited benefit of NPWT in all-comers and in high-risk cohorts, it remains unclear whether NPWT offers benefits following PD.

1. Introduction

Pancreaticoduodenectomy (PD) offers a surgical option for pancreatic head, periampullary, and duodenal pathology. Despite ongoing improvements, PD comes with significant postoperative morbidity. While there have been many studies focusing on reducing postoperative pancreatic fistulas (POPFs) and bleeding [1], surgical site infections (SSIs) remain a common cause of readmission and morbidity following PD [2]. The rates of superficial SSIs and organ space SSIs in patients undergoing PD have been reported to be as high as 10.0% and 6.2%, respectively [3]. SSIs commonly result in increased rates of fascial dehiscence, prolonged postoperative admission, increased need for readmission, and/or the need for reoperation, all of which ultimately delay adjuvant chemotherapy following PD surgery [4,5,6]. Negative pressure wound therapy (NPWT) can be defined as a device/system composed of a foam sponge, a semi-occlusive adhesive barrier, and a suction pump to provide continuous or intermittent sub-atmospheric pressure to the surface of a wound to create an optimal environment for wound healing [7]. The proposed mechanisms by which NPWT attempts to reduce SSI and their ensuing complications are by acting as a protective layer against external microorganisms from entering the wound bed and by removing wound exudate. NPWT has also been suggested to aid with wound healing by promoting cellular proliferation through micro-deformation, increasing local vascularity surrounding the wound bed, and removing pro-inflammatory cytokines [7,8,9,10]. While NPWT use during certain abdominal procedures has shown definitive benefit, its utility in PD remains unclear.

The use of NPWT to reduce the occurrence of surgical site infections (SSIs) following PD remains controversial. While some studies reported significant SSI reduction [11,12,13,14], others report no significant difference in the SSI rates with NPWT during PD [15,16,17]. The data from these studies are obtained through limited-sized, single-centered retrospective reviews [11,12,15], randomized control trials [13,14], a small meta-analysis [17], and a surgical database review which included a very limited number of patients who used NPWT after PD [16]. As such, it remains unclear if routine NPWT provides any benefit in reducing PD SSI rates. Given this uncertainty, a large multicenter analysis would further inform hepatopancreatobiliary surgeons on the use of prophylactic NPWT in PD surgery to decrease SSI risk.

The aim of this study is to evaluate whether NPWT is associated with improved SSI and overall outcomes following PD using the National Surgery Quality Improvement Program (NSQIP) database. Additionally, we evaluate whether NPWT is beneficial for patients undergoing malignant resections or for patients with high risk of SSI. This study will add to previous studies by providing the largest contemporary analysis of NPWT to date and by evaluating NPWT in high-risk cohorts, which have not been well studied previously. Data from this study will help clarify the use of prophylactic NPWT in PD patients.

2. Materials and Methods

This is a retrospective cohort study of prospectively collected NSQIP data comparing patients undergoing PD with and without prophylactic NPWT. We identified patients that underwent PD within NSQIP from 2017 to 2021 using current procedural terminology (CPT) codes: 48150, 48152, 48153, and 48154. The use of NPWT was not mandatory to report in participating NSQIP facilities. When available, NPWT is coded as used versus not used. If no information regarding NPWT use was identified, these patients were excluded. The remaining patients with information available regarding NPWT use were categorized as non-NPWT and NPWT. The primary objective of this study was to evaluate the effect of NPWT on the rates of SSIs. Secondary outcomes were to explore the impact of NPWT on 30-day serious adverse outcomes and on SSI in high-risk patients undergoing PD surgery.

The NSQIP database is a North American surgical database that collects operative, pre-, and 30-day postoperative data on patients undergoing major surgical procedures. Data are from operative cases from over 800 accredited sites by certified clinical reviewers, ensuring accuracy and reliability [18].

The demographics captured within this study include basic demographics, age, sex, body mass index, American Society of Anesthesiology (ASA) scores, comorbidities (cardiac, respiratory, renal), steroid use, and the presence of a bleeding disorder. Perioperative data extracted included broad-spectrum prophylactic antibiotic use, weight loss, and preoperative sepsis. Key oncologic factors investigated included tumor invasion (defined as tumor status T3 or T4), the use of neoadjuvant chemotherapy, and pancreatic gland texture (hard, intermediate, or soft).

Perioperative outcomes collected included readmission, unplanned intubation, and reoperation. Wound and infection complications collected include surgical site infections (SSIs) (which were classified either as superficial, deep, or organ space), wound disruption, pneumonia, urinary tract infections, sepsis, and septic shock. Other complications extracted included pulmonary embolism, deep vein thrombosis, acute renal failure, cerebrovascular accident, cardiac complications (cardiac arrests, myocardial infarctions), length of stay, and mortality. Postoperative pancreatic fistulas, which were described in accordance with the International Study Group for Pancreatic Surgery (ISGPS) definitions [19], were also assessed.

In addition to standard NSQIP outcomes, we also evaluated a composite 30-day serious complication, defined as unplanned intubation, acute renal failure, stroke/cerebrovascular accident, cardiac arrest requiring cardiopulmonary resuscitation, myocardial infarction, sepsis/septic shock, requiring reoperation, hospital stay greater than 30 days, and death. Finally, this study also evaluated outcomes according to the Clavien Dindo Classification (CDC) [20] and the updated comprehensive complication index (CCI) [21]. CDC was estimated according to the complications defined by NSQIP and was applied to the most appropriate expected treatment for each outcome according to Supplementary Material Table S1. These outcomes have shown evidence for predicting long-term complications [22].

Baseline characteristics and 30-day adverse postoperative outcomes between NPWT and non-NPWT cohorts were compared using regression for continuous variables and Chi-square for categorical variables, with p < 0.05 indicating significance. In addition, a multivariate logistic regression model was created to compare the independent effects of NPWT on 30-day serious complication, 30-day mortality, and the development of a superficial or deep SSI. Variables were included in the model if p < 0.1 in the bivariate analysis or if previous studies suggested the influence of the variable on the outcome according to a hypothesis-driven methodology. The multivariable model’s goodness of fit was evaluated using the area under the curve of the receiver operator characteristic (ROC) curve. Additional sensitivity analyses were performed by including only patients who had an operation for malignancy and another for patients who were deemed to be at a high risk of contracting an SSI, defined as having a soft pancreatic texture, male, smoker, underwent neoadjuvant therapy, steroid use, or diabetic.

3. Results

3.1. Baseline and Demographic Characteristics

Of the patients who underwent PD between 2017 and 2021 (28,083), a total of 14,044 patients with information regarding the use of NPWT were included in this study, with 12,355 (88.0%) being non-NPWT and 1689 (12.0%) using NPWT (

Table 1. Baseline characteristics of patients undergoing pancreaticoduodenectomy with and without negative pressure wound therapy (NPWT).

	No NPWT (n = 12,355)	NPWT (n = 1689)	p-Value
Demographics
Female sex	5772 (46.7%)	789 (46.7%)	0.998
Age	65.5 ± 11.4	65.7 ± 11.3	0.601
BMI	27.3 ± 5.7	27.6 ± 6.0	0.012
ASA class			<0.001
1	27 (0.2%)	0 (0.0%)
2	2195 (17.8%)	203 (12.0%)
3	9085 (73.6%)	1344 (79.6%)
4	1041 (8.4%)	142 (8.4%)
5	2 (0.0%)	0 (0.0%)
None assigned	5 (0.0%)	0 (0.0%)
Functional status			0.131
Partially dependent	102 (0.8%)	11 (0.7%)
Totally dependent	5 (0.0%)	3 (0.2%)
Unknown	11 (0.1%)	1 (0.1%)
Comorbidities
COPD	505 (4.1%)	52 (3.1%)	0.046
CHF	110 (0.9%)	12 (0.7%)	0.455
Hypertension	6498 (52.6%)	895 (53.0%)	0.760
Diabetes			0.009
Non-insulin dependent	1678 (13.6%)	234 (13.9%)
Insulin dependent	1661 (13.4%)	272 (16.1%)
Smoker	1993 (16.1%)	217 (12.9%)	0.001
Dialysis	38 (0.3%)	11 (0.7%)	0.025
Steroid use	465 (3.8%)	45 (2.7%)	0.023
Bleeding disorder	393 (3.2%)	29 (1.7%)	0.001
Perioperative factors
Broad-spectrum prophylactic antibiotics	4881 (41.0%)	418 (25.6%)	<0.001
Preoperative weight loss	1123 (13.6%)	126 (11.7%)	0.089
Preoperative sepsis	184 (1.5%)	12 (0.7%)	0.045
Disease factors
Invasive disease (≥T3)	3029 (25.0%)	408 (24.6%)	0.716
Neoadjuvant therapy	3497 (28.4%)	565 (33.8%)	<0.001
Pancreatic gland texture			<0.001
Hard	4014 (32.5%)	571 (33.8%)
Intermediate	1220 (9.9%)	267 (15.8%)
Soft	4397 (35.6%)	560 (33.2%)
Unknown	2724 (22.1%)	291 (17.2%)

Significant p-values are italicized. Bold designates a subsection heading. Data represented as n (%) or mean ± SD. Abbreviations: ASA, American Society of Anesthesiologist; BMI, body mass index; COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; n, sample size; NPWT, negative pressure wound therapy; SD, standard deviation.

). Patients who used NPWT prophylactically had a small but significantly higher BMI (27.6 ± 6.0) compared to those without NPWT (27.3 ± 5.7; p = 0.012; Table 1). Patients in the NPWT cohort had more comorbidities, resulting in less ASA II and more ASA class III patients (p < 0.001). Each cohort had an equal proportion of their patients with ASA class 4, but patients with ASA class 5 were only seen in the non-NPWT cohort (Table 1). There were significantly higher frequencies of certain comorbidities seen in the NPWT cohort such as dialysis dependence (p = 0.025) and insulin-dependent diabetes (p = 0.009). However, the NPWT cohort tended to have lower rates of COPD (p = 0.046), smoking (p = 0.001), steroid use (p = 0.023), and the presence of a bleeding disorder (p = 0.001, Table 1).

Perioperative demographic data showed a significantly higher proportion of patients who underwent neoadjuvant therapy in the NPWT cohort (33.8%) compared to the non-NPWT cohort (28.4%; p < 0.001; Table 1). Additionally, the presence of preoperative sepsis (p = 0.045) was significantly less common in the NPWT cohort (p < 0.001). Finally, the administration of postoperative empiric broad-spectrum antibiotics was significantly lower in the NPWT cohort (25.6%) compared to the non-NPWT cohort (41.0%; p < 0.001; Table 1).

3.2. Thirty-Day Postoperative Outcome Comparison

On the unadjusted outcome analysis, no significant differences were observed between the non-NPWT cohort and the NPWT cohort regarding the rates of superficial SSI (6.1% vs. 6.2%; p = 0.886) and deep SSI (0.6% vs. 0.7%; p = 0.467). The NPWT cohort had a significantly increased length of stay (10.4 ± 6.0 days vs. 9.5 ± 5.8 days; p < 0.001). Likewise, readmission rates were higher in the NPWT cohort (19.4%) compared to the non-NPWT cohort (17.2%, p = 0.039;

Table 2. Outcomes of patients undergoing pancreaticoduodenectomy with and without negative pressure wound therapy.

	No NPWT (n = 12,355)	NPWT (n = 1689)	p-Value
Postoperative course
Operative time (min)	383.3 ± 127.6	378.2 ± 137.2	0.127
Length of stay (d)	9.5 ± 5.8	10.4 ± 6.0	<0.001
Days from operation to discharge	9.0 ± 5.1	9.9 ± 5.3	<0.001
Hospital stay > 30 d	401 (3.3%)	49 (2.9%)	0.451
Postoperative blood transfusion	2291 (18.5%)	317 (18.85)	0.823
Readmission	2140 (17.3%)	327 (19.4%)	0.039
Unplanned intubation	388 (3.1%)	45 (2.7%)	0.288
Reoperation	651 (5.3%)	91 (5.4%)	0.838
Serious complication	2214 (17.9%)	284 (16,8%)	0.265
Death	222 (1.8%)	40 (2.4%)	0.104
Wound and infectious complications
Superficial surgical site infection	757 (6.1%)	105 (6.2%)	0.886
Deep surgical site infection	70 (0.6%)	12 (0.7%)	0.467
Organ space surgical site infection	2143 (17.4%)	382 (22.6%)	<0.001
Wound disruption	154 (1.3%)	25 (1.5%)	0.422
Pneumonia	506 (4.1%)	65 (3.9%)	0.630
Urinary tract infection	287 (2.3%)	24 (1.4%)	0.018
Sepsis	1062 (8.6%)	161 (9.5%)	0.200
Septic shock	380 (3.1%)	47 (2.8%)	0.511
Other complications
Pulmonary embolism	165 (1.3%)	29 (1.7%)	0.208
Deep vein thrombosis, thrombophlebitis	342 (2.8%)	55 (3.3%)	0.256
Acute renal failure	139 (1.1%)	21 (1.2%)	0.667
Cerebrovascular accident	29 (0.2%)	2 (0.1%)	0.339
Cardiac arrest	151 (1.2%)	19 (1.1%)	0.732
Myocardial infarction	179 (1.5%)	15 (0.9%)	0.064
Disease related
Postoperative pancreatic fistula			0.013
None	9712 (79.2%)	1278 (76.4%)
Grade A	839 (6.9%)	113 (6.8%)
Grade B	1554 (12.7%)	260 (15.6%)
Grade C	152 (1.2%)	21 (1.3%)

Significant p-values are italicized. Data represented as n (%) or mean ± SD. Abbreviations: d, day(s); min, minute(s); n, sample size; SD, standard deviation.

). The remainder of the 30-day postoperative outcomes explored had similar rates between groups. This includes perioperative complications such as the need for reoperation and operative time, infectious complications such as sepsis and septic shock, and surgery-specific complications such as the need for blood transfusion. Notably, the cohort using NPWT prophylactically in PD surgery did have significantly higher rates of organ space SSI (p < 0.001) as well as increased rates of pancreatic fistulas (p = 0.013). There was a significant difference in POPF rates between the groups, with higher grade B rates in the NPWT cohort (15.6% vs. 12.7%; p < 0.001).

3.3. Assessing the Independent Effect of NPWT in Multivariable Models on the Risk of SSIs

Following multivariable regression modeling consisting of 22 covariates, the outcome SSI included both deep and superficial incisional SSIs. The use of NPWT was not associated with a decrease in SSI (aOR 0.94, 95% CI 0.70–1.26, p = 0.691;

Table 3. Multivariate logistic regression model identifying independent factors predictive of incisional SSIs (superficial and deep) in patients undergoing pancreaticoduodenectomy.

n = 6941	Multivariable OR	95% CI	p-Value
NPWT	0.94	0.70–1.26	0.691
Age	1.00	0.99–1.01	0.688
BMI	1.03	1.02–1.05	<0.001
Female	1.17	0.96–1.42	0.107
COPD	1.15	0.72–1.86	0.557
CHF	0.56	0.08–4.17	0.572
Hypertension	1.03	0.84–1.27	0.771
Diabetes
Insulin dependent	0.97	0.72–1.30	0.831
Non-insulin dependent	0.75	0.55–1.02	0.063
Smoker	1.06	0.80–1.39	0.689
Dialysis dependent	1.56	0.46–5.28	0.474
Steroid use	1.38	0.84–2.27	0.202
Bleeding disorder	1.20	0.70–2.06	0.518
Preoperative sepsis	1.01	0.44–2.34	0.976
Functional status	1.55	0.60–3.96	0.362
Neoadjuvant received	1.03	0.83–1.29	0.777
Invasive disease (≥T3)	1.03	0.76–1.20	0.662
Soft pancreas	1.10	0.90–1.34	0.360
Preoperative weight loss	0.87	0.64–1.17	0.346
Broad-spectrum prophylactic antibiotics	0.76	0.62–0.93	0.009
Minimally invasive technique	0.61	0.40–0.91	0.016

Significant p-values are italicized. Values presented as odds ratio (95% confidence interval). Brier score 0.061, ROC area under curve 0.593 (p < 0.001). Abbreviations: BMI, body mass index; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; NPWT, negative pressure wound therapy, ROC, receiver operating curve.

). Factors associated with increased likelihood included higher BMI (BMI 1.03, 95% CI 1.02–1.05, p < 0.001) and with decreased likelihood included broad-spectrum antibiotics (aOR 0.75, 95% CI 0.62–0.93) and minimally invasive surgery (vs open, aOR 0.61, 95% CI 0.40–0.91, p = 0.016).

3.4. Evaluation of NPWT in High-Risk Cohorts

Two sensitivity analyses were performed. The first assessed patients having at least one or more high-risk factors for developing SSIs which included the following: soft pancreas, male sex, insulin- and non-insulin-dependent diabetics, smokers, steroid use, and those who underwent neoadjuvant chemotherapy. Again, no statistically significant difference was found with NPWT use in a multivariate model (aOR 0.98, 95% CI 0.75–1.13, p= 0.898, Supplementary Table S2). The only characteristic that was associated with significantly reduced rates of SSI amongst high-risk patients was those who underwent MIS PD (aOR 0.58, 95% CI 0.47–0.98; p−0.038) Patients with a soft pancreas texture were however noted to have a significantly higher likelihood of developing an SSI (aOR 1.57, 95% CI 1.36–1.82, p < 0.001), as well as those with higher BMI (aOR 1.02, 95%CI 1.01–1.04, p = 0.005)

The second subgroup analysis included patients undergoing resection for malignancy. Similarly, NPWT was not associated with a significant reduction in risk of developing SSIs (aOR 0.957, 95% CI 0.694–1.317, p = 0.788; Supplementary Table S3). BMI was shown to increase the risk of SSIs in patients undergoing PD (aOR 1.03 95%CI 1.01–1.05, p = 0.005), while the use of prophylactic antibiotic use showed a reduction in SSIs in this subgroup (aOR 0.808, 95% CI 0.566–0.886 p = 0.003).

3.5. The Association of NPWT with Serious Complications

A multivariable logistic regression model was created to identify factors associated with the likelihood of a composite 30-day serious complication. NPWT use was not associated with decreased likelihood (aOR 0.96, 95%CI 0.79–1.17, p = 0.669;

Table 4. Multivariable logistic regression model for factors predictive of 30-day serious complications in patients undergoing pancreaticoduodenectomy.

n = 6941	Multivariable OR	95% CI	p-Value
NPWT	0.96	0.79–1.17	0.669
Age	1.01	1.00–1.01	0.020
BMI	1.02	1.01–1.03	<0.001
Female	0.74	0.65–0.85	<0.001
COPD	1.50	1.12–2.00	0.007
CHF	1.20	0.50–2.90	0.684
Hypertension	1.25	1.08–1.43	0.002
Diabetes insulin dependent	0.97	0.80–1.17	0.752
Diabetes non-insulin dependent	0.97	0.81–1.17	0.770
Smoker	1.16	0.97–1.38	0.111
Dialysis	1.26	0.53–3.03	0.539
Steroid use	1.12	0.78–1.60	0.539
Bleeding disorder	1.33	0.94–1.89	0.112
Preoperative sepsis	2.41	1.55–3.75	<0.001
Functional status	1.71	0.91–3.23	0.095
Neoadjuvant chemotherapy	0.91	0.78–1.06	0.211
Invasive (≥T3)	1.09	0.95–1.78	<0.001
MIS (vs. open)	0.88	0.70–1.11	0.295
Soft pancreas	1.56	1.36–1.78	<0.001
Weight loss	1.24	1.03–1.48	0.022
Prophylactic broad antibiotics	0.84	0.74–0.96	0.009

Significant p-values are italicized. Values presented as odds ratio (95% confidence interval). Brier score 0.139, ROC area under curve 0.612. Abbreviations: BMI, body mass index; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; NPWT, negative pressure wound therapy; MIS, minimally invasive surgery); ROC, receiver operating curve.

). Multiple factors were associated with increased likelihood including increasing age, BMI, male sex, COPD, hypertension, preoperative sepsis, preoperative weight loss, and soft pancreas texture. Only female sex and preoperative broad-spectrum antibiotics were associated with a decreased likelihood.

In the sensitivity analysis of high-risk patients, NPWT was associated with a decreased likelihood of serious complications (aOR 0.81, 95%CI 0.67–0.98, p = 0.031; Supplementary Table S4). In the sensitivity analysis of malignancy patients, NPWT did not decrease the likelihood in PD patients (aOR 0.95, 95%CI 0.77–1.19, p = 0.031; Supplementary Table S5).

4. Discussion

This large-scale, multicenter retrospective cohort study assesses a contemporary cohort of over 14,000 patients undergoing PD, demonstrating that the use of NPWT did not decrease the likelihood of developing superficial and deep SSIs. It further adds to the literature by evaluating the impact in populations that may be more likely to benefit from NPWT. However, no reduction in SSIs was noted even after the sensitivity analyses of high-risk cohorts and malignant cohorts were performed. Furthermore, NPWT did not decrease the rates of serious adverse outcomes at 30 days when looking at the whole cohort and malignant groups, albeit a decreased likelihood was noted in the high-risk groups.

When exploring the study’s primary outcome, our data support more recent single-centered studies and a prior NSQIP study showing NPWT devices do not reduce the rate of SSI in patients undergoing PD surgery [15,16]. Despite these findings, our results are from a large number of centers, and it remains possible that experience with NPWT, unique ways to utilize them selectively in specific patients, or centers with optimized protocols could potentially enable SSI reduction. The NSQIP retrospective analysis performed by DeLeon has a similar design to our study; however, our study utilizes a more contemporary cohort (2017–2021 vs. 2005–2019), with a substantially larger number of patients within the NPWT device cohort (1689 using NPWT vs. 131) [16]. Furthermore, we performed a unique subgroup analysis including high-risk subgroup analysis and explored the effects of NPWT devices on SSIs and the rate of serious adverse events in patients undergoing PD for the treatment of a malignancy vs. not; this represents a unique subgroup analysis which has not been previously studied. We found that the use of NPWT was not associated with lower rates of SSIs in a high-risk subgroup analysis nor was there a reduction in rates of SSIs in the subgroup analysis of patients undergoing PD for malignancy. Previous studies looking at NPWT use in high-risk groups show mixed results. Javed et al. performed a single-center retrospective cohort study showing significantly reduced rates of SSI in those undergoing open PD [13]. However, high-risk in this study was defined using Poruk’s model [12] and only included those with preoperative biliary stenting and neoadjuvant therapy [11,12]. Our study identified high-risk patients for developing SSIs as patients with a soft pancreas, male sex, insulin- and non-insulin-dependent diabetics, smokers, steroid use, and those who underwent neoadjuvant chemotherapy. With this broad definition of a high risk, our findings align more closely with Green et al., who found no difference in SSI in those with high-risk features [15]. This suggests that there may not be clear roles of NPWT, even in high-risk patients.

Our study found that NPWT was utilized more in patients with increased comorbidities. The impact of their comorbidities on their immune response to fight against SSIs may explain this trend. However, after the multivariable regression modeling, our data did not suggest a protective effect of NPWT in the rate of SSI for the high-risk cohorts. Some literature suggests that factors often seen as increasing the risk of SSIs, such as neoadjuvant chemotherapy [11,12,15], may not necessarily be associated with a consistent increase in SSI [23,24,25,26]. This could provide rationale on why patients classically deemed high-risk may not necessarily benefit and why the literature has been equivocal regarding NPWT use in PD.

Although there may not be much benefit to using NPWT, there may also be concerns about harm with the use of NPWT in PD. The drawbacks of NPWT include resources required for changing, troubleshooting leakages, and costs related to its use [27]. Our analysis also suggests an increased rate of complication as well as length of stay and readmission (Table 2), which may contribute further to hospital costs. Although these drawbacks were also observed in Green’s retrospective cohort study and Deleon’s NSQIP study, increased length of stay and readmission rates in their studies failed to show statistical significance. Given the relatively small number of participants involved in these studies, the lack of statistical significance may be a result of insufficient power in these studies. A meta-analysis of 22 articles on complex abdominal wall incisions found that the cost effectiveness of NPWT devices is a result of a reduction in the length of stay and requirement for reoperation or readmission [27]. PD surgery has unique complications which contribute significantly to morbidity such as the development of POPF [26]. The results of this study would thus suggest that the cost–benefit of NPWT in PD is not as clear and may be less favorable.

The lack of efficacy of NWPT in this patient population cannot be clearly elucidated from these data; however, the specifics of PD may offer potential explanations. The mechanism by which NPWT devices facilitate wound healing is by acting as a protective layer against external microorganisms from entering the wound bed, removing exudate, promoting cellular proliferation through micro-deformation, and increasing local vascularity surrounding the wound bed [8,9,10]. However, intraoperative contamination during PD from the leakage of static bile during the hepaticojejunostomy, enteric contents during the bowel anastomosis, or pancreatic enzymes during the pancreatojejunostomy, in addition to the high rates of POPF postoperatively may explain the inefficacy of NPWT in PD and why they do not contribute to a decrease in of SSIs in PD [28].

The cost analysis of NPWT can be challenging to interpret. The direct cost of an NPWT device can range in the thousands of dollars (USD) depending on the type and brand of the NPWT device [29]. This does not necessarily include the additional time required for managing the device, both intra- and postoperatively. Though cost-effective analyses show a benefit in certain emergent operations [30,31,32,33], they appear less cost effective than standard wound closure in other elective surgeries such as abdominal wall contouring [34]. Potential for complications such as enterocutaneous fistulas [35] further suggest against routine NPWT use. An economic model was developed in 2021 which showed, in general, that NPWT devices reduce healthcare costs through reduced take-back operations, as well as a reduction in wound care therapy duration. It should be noted that this model did not specify a specific type of wound therapy device, nor did it specify which wounds the NPWT devices were studied on [27], thus highlighting the challenge in determining the cost effectiveness of using NPWT devices. Considering the lack of benefit with NPWT during PD, these cost benefits likely do not apply. Similarly, considering the environmental cost of producing and disposing these single-use NPWT devices, the environmental impact of NPWT likely exceeds that of standard wound dressings [36].

The limitations of this analysis include the use of a database, where pertinent data that can affect outcomes, center-specific practices, and intraoperative nuances may not be available. For example, the lack of mandatory reporting for NPWT use, information on the specific type of NPWT device used, the exact timing (prophylactically placed at time of surgery vs. afterwards), and the level of suction used and how often the NPWT was changed are all factors which can influence the effectiveness of the NPWT. However, these data were not collected using the NSQIP database, and the results of this study should be contextualized within these limitations. As discussed, it remains possible that centers with NPWT expertise or unique NPWT protocols may allow for improved outcomes with their use. The lack of center-specific codes limits the ability to perform a subgroup analysis of PD outcomes in centers who do vs. do not use NPWT and may be of interest in future studies. Nonetheless, the database collects information from over 800 centers, is prospectively collected, and undergoes quality assurance, remaining an insightful assessment of the current use of NPWT. We also performed sensitivity analyses which may be more likely to show a protective effect in PD. Furthermore, additional information such as the duration of wound vacuum use and complications associated with wound vacuum use prevent a detailed analysis on their impact on patient care. Finally, retrospective studies are also limited as they are unable to identify causal relationships. We performed multivariable regressions to isolate the independent effects of NPWT on the outcomes of interest. Future studies may add to the literature by focusing attention to the role of NPWT in postoperative SSI management or after reoperations for deep organ space infections.

5. Conclusions

Utilizing a large multicentered retrospective database, we saw NPWT use did not decrease the risk of SSI or serious complications in a multivariable model. When looking at malignancy or high-risk subgroups, NPWT was again not associated with a decreased risk of SSI patients undergoing PD. These results suggest that surgeons should use caution when implementing NPWT in PD patients as there is a lack of evidence of clear benefit in patients undergoing PD, including those at high risk.