Surgical-site infections (SSIs) complicate 2% to 5% of all surgical procedures, 8% of major abdominal procedures, and 20% to 40% of pancreaticoduodenectomies (PD).1–6 Despite multiple infection control initiatives and quality collaboratives, SSIs remain a major concern for reliable safe surgery.7–9 SSIs are associated with an increased risk of postoperative morbidity, prolonged hospitalization, postponement of chemotherapy, increased healthcare costs, and in some cases poor long-term outcomes.7,10 In the United States alone, the burden of SSIs is estimated at $10 billion per year.11
Several prediction models have been developed for SSI risk prediction for patients undergoing abdominal procedures,5,7 considering factors such as neoadjuvant chemotherapy and preoperative biliary stenting. The rate of SSIs in patients undergoing PD at Johns Hopkins Hospital has historically been 15% to 20%, which is below the national average reported by the American College of Surgeons National Surgical Quality Improvement Program (NSQIP).8,9,12 Despite SSI rates below the national average, our institution has identified a subgroup of high SSI-risk patients, for whom SSIs remain a clinically relevant problem (>30%).8
Negative pressure wound therapy (NPWT) is suggested to reduce risk of SSI by reducing fluid accumulation within the avascular dead space in a closed wound. Recent studies suggest that NPWT can help prevent SSIs after abdominal surgery9,12; however, literature is limited to retrospective studies with mixed results. Given these findings, we conducted a randomized, controlled trial of NPWT in high SSI-risk patients undergoing PD.
Patients were randomly assigned to surgical incision closure using either NPWT (placement of a PREVENA Peel & Place Dressing [Acelity]) or standard closure technique (details on application and management provided in supplementary Appendix, http://links.lww.com/SLA/B517). Records of screened and consented patients were maintained electronically. The principal investigator trained study team members to ensure compliance of study-related procedures and documentation. The outcomes were analyzed according to the intention-to-treat principal.13
Patient Selection and Study Group Assignment
Electronic medical records (EMR) of patients scheduled to undergo PD were screened for eligibility. A research coordinator obtained an informed consent from eligible patients, which included adults (≥18 yrs of age) who had a SSI risk score of ≥1 as defined by the risk score proposed by Poruk et al.8 This included patients who had received neoadjuvant chemotherapy, preoperative biliary stenting, or both. Key exclusion criteria included PD performed minimally invasively or known allergies or sensitivity to silver or acrylic adhesives.
Using the simple randomization method, a random allocation sequence was generated. Allocation concealment was achieved by printing allocation onto a gray-shaded card that was folded and sealed in a secured envelope before initiation of the study. Intraoperatively, when the surgeons had committed to performing a PD, patients underwent randomization to assign them to NPWT or standard closure, in a 1:1 ratio.
PD was performed in the standard manner. Once the surgeon committed to performing a PD by ruling out metastatic disease or inoperable local vascular involvement, the circulating nurse contacted the research staff for randomization. The presealed envelope was opened to randomize the patient. Upon randomization to standard closure, the surgical incision was closed primarily. If the patient was randomized to NPWT, the circulating nurse provided the surgeon with the device maintaining sterility, which was then applied to the surgical incision after the manufacturer's instructions (Figure 1). All patients also received standard infection-prevention measures, including preoperative antibiotic prophylaxis, intraoperative forced air warming devices, and tight perioperative glycemic control.
Postoperatively, patients were followed by their surgeons until their discharge and were subsequently evaluated at the surgical clinics within 30 days after their procedure. Surgeons ruled out SSI daily during the patient's hospitalization and again during follow-up. For patients with NPWT, the surgical site was inspected daily beyond postoperative day 4, when the device was removed. In addition, patients’ EMR were reviewed independently by the principal investigator (MJW) blinded to study-group assignments to determine if SSI was documented at any time during the 30-day postoperative period. Remaining clinicopathological data were extracted from patients’ EMR by the research staff. Data on cost were calculated using hospital charges, which are fixed in the state of the Maryland because of the Maryland-regulated hospital reimbursement rate commission.
The trial was approved by the institutional review board of the Johns Hopkins Hospital (Approval # IRB00109564). A data and safety monitoring board provided regulatory oversight by reviewing data periodically, and performed an interim analysis after 50% of the participants had been evaluated. The principal investigator and participating surgeons were not informed of the results of the interim analysis.
The sponsors of the study had no role in the design or conduct of the study, data collection, management, analysis, or interpretation, or the preparation, review, and approval of the manuscript. The decision to submit the manuscript for publication was made by all authors.
The primary outcome was superficial or deep wound SSIs within the initial 30-day postoperative period. SSI was defined by the National Health Safety Network definition of the Centers for Disease Control and Prevention (CDC), which is provided in the supplementary Appendix (Table S1, http://links.lww.com/SLA/B517). In short, this included the presence of either superficial or deep incisional SSI characterized by cellulitis or erythema, induration around the incision or purulent discharge from the surgical incision site with or without the presence of fever. A sterile wound hematoma, seroma, or skin separation alone in the absence of aforementioned signs did not constitute SSI. A sensitivity analysis was performed to account for the potential undercapturing of SSIs that were thought to be wound hematoma, seroma, or skin separation. The patients’ treating physician and medical staff initially made the diagnosis of an SSI. However, the PI, who was blinded to the study-group assignment, reviewed all patients’ EMR using the discrete CDC definition to confirm the diagnosis of an SSI based on the signs and symptoms detailed in the patients’ EMR.
Secondary outcomes included a composite of length of intensive care unit (ICU) stay, length of hospitalization, need for reoperation, 30-day readmission related to SSI, and allergic reactions. Additional outcomes assessed included delayed gastric emptying (DGE), postoperative pancreatic fistula (POPF), and postoperative respiratory complications.
Previous studies performed at our institution have demonstrated that the rate of SSIs in patients with a SSI risk score of ≥1 is 30%. We hypothesized that the use of NPWT could reduce this to 10%. We determined that a sample size of 124 patients would provide a power of 80% to detect a 20% relative reduction in the primary outcome from a baseline risk of 30%, at a 2-sided alpha level of 0.05 (Kelsey method).14
An interim analysis was planned according to the method proposed by Proschan et al., assuming that the results seen up until the interim analysis continue until the end of the trial (empirical conditional power).15 If the conditional power under the empirical assumption were <0.50, the trial would be stopped for futility.
Analyses were performed by the coauthors (AAJ, DD, MJW). Continuous variables were reported as a mean and standard deviation or median and interquartile range (IQR), as deemed appropriate. Categorical variables were summarized as frequencies and percentages. We used the unpaired Student t test or Wilcoxon test to analyze continuous variables, and the χ2 or Fisher exact test to analyze categorical variables. Relative risks and 95% confidence intervals were calculated for the outcomes. Logistic regression was used for detailed analysis on factors associated with SSIs. A sensitivity analysis was performed by classifying hematoma, seroma, granulomas, or skin separation as a SSI (Table S4, http://links.lww.com/SLA/B517). Cost of hospitalization was defined as charges of the indexed hospitalization and/or any readmission within the initial 30-day postoperative period. Given that data on cost of hospitalization were non-normally distributed, linear regression after taking log transformation (natural base) of cost of hospitalization was used for analysis.16 Covariates with a P <0.05 in the univariate analysis were included in the multivariate analysis. A 2-sided P <0.05 was considered significant. Analysis was conducted using R software, version 3.3.3 (Vienna, Austria).
From January 2017 through February 2018, we assessed 325 patients for eligibility (Fig. 2), and 124 underwent randomization; 62 patients received NPWT, and 62 received standard closure. In the standard closure group, 1 patient's incision was closed intraoperatively, using NPWT at the surgeon's discretion. Therefore, the final cohort used for analysis included 62 patients who received NPWT and 61 who received standard closure. The baseline characteristics of these patients are presented in the supplementary Appendix (Table S2, http://links.lww.com/SLA/B517).
The preoperative and operative characteristics of the patients were similar between the 2 study groups (Table 1). Preoperative antibiotic prophylaxis was performed in 100% of both the NPWT and standard closure group (P >0.99). Neoadjuvant therapy was administered to 70.0% of the patients in NPWT group and 54.1% of the patients in the standard closure group (P = 0.05). Preoperative biliary stenting was performed in 82.3% of the patients in the NPWT group and 83.6% of the patients in the standard closure group (P = 0.84). No significant difference was observed in each participating surgeon's contribution of patients to each study arm (P = 0.628). One patient (1.6%) in the standard closure arm received skin closure using staples as compared to no patients (0.0%) in the NPWT group. Final histopathological examination demonstrated pancreatic ductal adenocarcinoma in 96 patients (78.0%), of whom 49 had received NPWT and 47 had received standard closure (Table 1, S2, http://links.lww.com/SLA/B517).
A total of 6 patients (9.7%) in the NPWT group and 19 patients (31.1%) in the standard closure group were diagnosed with SSI [relative risk (RR), 0.31; 95% CI, 0.13–0.73; P = 0.003) (Table 2). This corresponds to an absolute risk reduction of 21.4% and a relative risk reduction (RRR) of 68.8%. The rate of superficial SSI was 6.5% in the NPWT group as compared with 27.9% in the standard closure group (P = 0.002); the rate of deep infection was similar in each group (3.2% and 3.3%, respectively, P = 0.99). Type of wound closure was the only factor significantly associated with SSI (P = 0.003) (Table 2, S3, http://links.lww.com/SLA/B517). Notably, the SSI rate for all patients undergoing PD at our institution, including both low and high risk, was 16.3% during the study period.
No significant differences found between the 2 groups with respect to length of ICU stay (P = 0.12), length of hospitalization (P = 0.23), need for reoperation (RR, 0.25; 95% CI, 0.03–2.32; P = 0.21), rate of 30-day readmission for SSI (RR, 0.49; 95% CI, 0.13–1.88; P = 0.32), and rate of 30-day readmission (RR, 0.41; 95% CI, 0.15–1.09; P = 0.07). No patients in either group developed allergic reactions to the dressings, and there was no device failure in the NPWT group.
Post hoc analysis revealed no difference in rates of delayed gastric emptying (DGE) (RR, 0.98; 95% CI, 0.34–2.88; P > 0.99), postoperative pancreatic fistula (POPF) (RR, 0.44; 95% CI, 0.11–1.56; P = 0.21), and respiratory complications (RR, 0.37; 95% CI, 0.10–1.33; P = 0.13) between the 2 study groups (Table 2).
The median cost of hospitalization for the study population was $43,823 (IQR, $36,820–$59,352). On univariate analysis, factors associated with increased cost of hospitalization included DGE, POPF, SSIs, and postoperative respiratory complications (all P < 0.05) (Table 3). On multivariate analysis, factors associated with increased cost of hospitalization included DGE [% increase in cost of hospitalization (Δ Cost), 45.8%; 95% CI, 22.2%–74.0%; P < 0.001], POPF (Δ Cost, 74.9%; 95% CI, 38.8%–120.4%; P < 0.001), SSIs (Δ Cost, 23.8%; 95% CI, 8.5%–41.2%, P = 0.002), and postoperative respiratory complications (Δ Cost, 79.2%; 95% CI, 45.1%–121.4%, P < 0.001) (Table 3).
The median cost of hospitalization of patients who did not develop a SSI was $41,085 (IQR, 36,195–$57,326). The additional cost of hospitalization due to development of SSIs in our study was $9,778 (IQR, $3,492–$16,927).
Management of SSIs
The management of patients who developed SSI composed of either one or a combination of antibiotic administration (40%), postoperative placement of NPWT (36%), bedside wound opening with wet-to-dry dressings (44%), bedside wound opening and debridement (16%), or reoperation (12%). Two of the 3 patients requiring reoperations developed complete fascial dehiscence. In our series, none of the patients who developed SSIs in the NPWT group demonstrated such dehiscence. The third patient was found to have a collection at the surgical site and required an incision and drainage. Upon intraoperative examination, there was no fistulous tract and the fascia appeared intact. The pus was expressed and sent for cultures and was positive for streptococcus anginosus. All 6 patients (100%) in the NPWT group who developed SSIs had a Clavien–Dindo grade of ≤2 compared with 16 patients (84%) in the standard closure group. Three patients required reoperation for SSI, all belonging to the standard closure group.
In this single-center randomized, controlled trial focused on high SSI-risk patients, we found that the use of NPWT for closure of surgical incision significantly reduced the rates of SSIs after PD. There was an impressive reduction in the relative risk and closure technique was the only factor significantly associated with SSIs. Furthermore, SSIs were associated with an increase cost of hospitalization.
Even when using best practices, many factors associated with SSIs are not easily modifiable, including use of neoadjuvant therapy and the length of operation.5,17 However, one actionable area is the management of surgical incision. This can be achieved by application of (1) wound protectors, (2) disinfectants to the skin, or (3) specialized dressings. A majority of studies evaluating the impact of using such interventions is composed of retrospective studies, which often suffer from selection bias. A recent RCT by Bressan et al reported a RRR of 52% (P = 0.010) in SSIs, using dual-ring wound protectors in patients with intrabiliary stents undergoing PD.18 Despite an improvement, the rate of SSIs in the intervention arm in this study was 21.1%, which is higher than the overall rate of SSIs at our institution as well as the national average reported by NSQIP.8,9,12 Therefore, at our institution dual-ring wound protectors are not used for PD. Furthermore, Zhang et al in a recent meta-analysis of 12 RCTs on the use of wound protectors for patients undergoing lower gastrointestinal surgery reported significantly decreased odds of developing an SSI with the use of a wound protector (OR, 0.46; P < 0.01).19 It may be that combining such wound protectors with NPWT could further improve SSI rates <9.7% for high-risk patients.
Our finding of a reduction in SSIs with the use of NPWT is consistent with a majority of the prior studies on impact of NPWT on SSIs.9,12 A recent meta-analysis evaluating the effect of NPWT in 10 randomized controlled trials found a significant reduction in the relative risk for SSI (RRR = 51%, P < 0.001).20 A subgroup analysis highlighting abdominal surgery patients also found a significant reduction in SSIs (P < 0.001).20 Similarly, another RCT on patients undergoing abdominal surgery demonstrated a significant reduction in the rates of SSIs (P = 0.03).21 Contrastingly, Shen et al in a phase II randomized trial of NPWT on SSIs in patients undergoing laparotomy for gastrointestinal, pancreatic, and peritoneal surface malignancies reported no significant difference in the rates of SSIs (P > 0.99).22 These data may have been underpowered to detect a significant difference given the variable rates of expected SSIs in the surgical procedures studied and absence of patient-level risk stratification. The 2 studies reporting use of NPWT in PD patients, both reported a significant reduction in SSIs.12,9.
Postoperative complications have been identified as the strongest predictor of increased cost of hospitalization.10,17,23,24 This has been attributed to prolonged length of hospitalization and need for readmissions in patients developing these complications.10 Similar to other studies, in our study SSIs were independently associated with an increased cost of hospitalization. In a retrospective study on patients undergoing PD, Santema et al reported a 32.4% increase in cost of hospitalization in the presence of SSI, which is comparable to the 23.8% increase in our study.25 Another study reported that the additional cost due to prolonged hospitalization and wound care related to SSIs was estimated to be over $1400.8 This is lower than our estimated cost, which may be because this study did not take into account the cost associated with readmissions. Furthermore, a recent study evaluating the American College of Surgeons—National Surgical Quality Improvement Program (ACS-NISQIP) reported the additional estimated costs of readmissions per patient secondary to superficial SSI to be $7661.26 Similar studies on patients undergoing spinal surgery and craniotomies have reported a mean direct cost for SSI to be $16,24227 and £7,000, respectively.28 Because of these findings, it is clear that there could be significant healthcare implications to markedly reducing SSI rates. Based on the findings of our study, we believe that use of NPWT in surgical practice can potentially move the field closer to the goal of eliminating potentially preventable complications and lowering healthcare costs.
This trial has some important limitations. First, it was conducted at a high volume center for pancreatic surgery with a broad experience using NPWT, and therefore there is a question about generalizability of the reductions we observed. However, the technique of application of NPWT is straightforward and has a short learning curve. Furthermore, low volume centers are less likely to operate on patients who have undergone prior exploration or those requiring multivisceral or vascular resection, factors that can potentially contribute directly or indirectly to an increased risk of SSIs because of a prolonged operative time. Second, the lack of blinding among the participants and clinicians could potentially introduce bias. This is a limitation that is difficult to address when performing mechanical RCTs. However, the PI was blinded to the wound closure status of patients when he reviewed their EMR to determine SSIs. This blinding helps reduce any bias that was introduced in the study due to lack of blinding to the intervention. Furthermore, any such bias would be expected to be nondirectional. Also, the standard technique was applied to prep the skin before the incision and similar antibiotic prophylaxis process was used for all patients. Also, the reviewers were blinded to the study-group assignment, and used standardized CDC National Healthcare Safety Network definitions to minimize subjectivity.29 Another limitation of the study is that the surgical wounds of patients in the NPWT group were covered for the initial 4 postoperative days and therefore any erythema or discharge would not have been captured. However, upon removal of the dressing the surgical sites were thoroughly inspected. In addition, the infected wounds were not cultured unless a clinical decision was made to culture them. Thus, data on cultures are not available on all patients who experienced SSIs. Finally, total hospital costs reported here may be different than at other medical centers based on the local price markup and discount applied to payers, a floating price common to sates outside of Maryland's regulated rate system.
Given the impact of SSIs on postoperative morbidity and healthcare costs, we propose incorporating the intervention described in this report. Quality collaboratives can support the use of NPWT after open abdominal surgery as a process measure. Further research is need to explore other surgical procedures where NPWT may optimize patient outcomes. Based on the scientific rationale for NPWT, we believe that it may yield the same SSI risk reduction for patients undergoing a wide array of operations as it did in the population we studied.
1. Anderson DJ, Podgorny K, Berrios-Torres SI, et al. Strategies to prevent surgical site infections in acute care hospitals: 2014 update. Infect Control Hosp Epidemiol
2. Tuuli MG, Liu J, Stout MJ, et al. A randomized trial comparing skin antiseptic agents at Cesarean delivery. N Engl J Med
3. Javed AA, Aziz K, Bagante F, et al. Pancreatic fistula and delayed gastric emptying after pancreatectomy: where do we stand? Indian J Surg
4. Ambiru S, Kato A, Kimura F, et al. Poor postoperative blood glucose control increases surgical site infections after surgery for hepato-biliary-pancreatic cancer: a prospective study in a high-volume institute in Japan. J Hosp Infect
5. Barreto SG, Singh MK, Sharma S, et al. Determinants of surgical site infections following pancreatoduodenectomy. World J Surg
6. Fong ZV, McMillan MT, Marchegiani G, et al. Discordance between perioperative antibiotic prophylaxis and wound infection cultures in patients undergoing pancreaticoduodenectomy
. JAMA Surg
7. Ejaz A, Schmidt C, Johnston FM, et al. Risk factors and prediction model for inpatient surgical site infection after major abdominal surgery. J Surg Res
8. Poruk KE, Lin JA, Cooper MA, et al. A novel, validated risk score to predict surgical site infection after pancreaticoduodenectomy
. HPB (Oxford)
9. Burkhart RA, Javed AA, Ronnekleiv-Kelly S, et al. The use of negative pressure wound therapy
to prevent post-operative surgical site infections following pancreaticoduodenectomy
. HPB (Oxford)
10. Badia JM, Casey AL, Petrosillo N, et al. Impact of surgical site infection on healthcare costs and patient outcomes: a systematic review in six European countries. J Hosp Infect
11. Sayyed R, Rehman I, Niazi IK, et al. Resection-reconstruction of aberrant right hepatic artery during whipple procedure (pancreaticoduodenectomy
). J Coll Physicians Surg Pak
2016; 26 (6 Suppl.):S24–S25.
12. Gupta R, Darby GC, Imagawa DK. Efficacy of negative pressure wound treatment in preventing surgical site infections after whipple procedures. Am Surg
13. Ware JH, Hamel MB. Pragmatic trials—guides to better patient care? N Engl J Med
14. Kelsey JLWA, Evans AS, Thompson WD. Methods in Observational Epidemiology. New York, New York: Oxford University Press, Inc; 1996.
15. Proschan MA, Lan KKG, Wittes JT. Statistical Monitoring of Clinical Trials: A Unified Approach, vol. 26. 1at ed.New York: Springer-Verlag; 2006.
16. Diehr P, Yanez D, Ash A, et al. Methods for analyzing health care utilization and costs. Annu Rev Public Health
17. Ceppa EP, Pitt HA, House MG, et al. Reducing surgical site infections in hepatopancreatobiliary surgery. HPB (Oxford)
18. Bressan AK, Aubin JM, Martel G, et al. Efficacy of a dual-ring wound protector for prevention of surgical site infections after pancreaticoduodenectomy
in patients with intrabiliary stents: a randomized clinical trial. Ann Surg
19. Zhang L, Elsolh B, Patel SV. Wound protectors in reducing surgical site infections in lower gastrointestinal surgery: an updated meta-analysis. Surg Endosc
20. Strugala V, Martin R. Meta-analysis of comparative trials evaluating a prophylactic single-use negative pressure wound therapy
system for the prevention of surgical site complications. Surg Infect (Larchmt)
21. Li PY, Yang D, Liu D, et al. Reducing surgical site infection with negative-pressure wound therapy after open abdominal surgery: a prospective randomized controlled study. Scand J Surg
22. Shen P, Blackham AU, Lewis S, et al. Phase II randomized trial of negative-pressure wound therapy to decrease surgical site infection in patients undergoing laparotomy for gastrointestinal, pancreatic, and peritoneal surface malignancies. J Am Coll Surg
23. Idrees JJ, Johnston FM, Canner JK, et al. Cost of major complications after liver resection in the united states: are high-volume centers cost-effective? Ann Surg
24. Dimick JB, Chen SL, Taheri PA, et al. Hospital costs associated with surgical complications: a report from the private-sector National Surgical Quality Improvement Program. J Am Coll Surg
25. Santema TB, Visser A, Busch OR, et al. Hospital costs of complications after a pancreatoduodenectomy. HPB (Oxford)
26. Fadayomi AB, Kasumova GG, Tabatabaie O, et al. Unique predictors and economic burden of superficial and deep/organ space surgical site infections following pancreatectomy. HPB (Oxford)
27. Blumberg TJ, Woelber E, Bellabarba C, et al. Predictors of increased cost and length of stay in the treatment of postoperative spine surgical site infection. Spine J
28. O’Keeffe AB, Lawrence T, Bojanic S. Oxford craniotomy infections database: a cost analysis of craniotomy infection. Br J Neurosurg
29. Horan TCGR. CG M. Surveillance of nosocomial infections. Hospital Epidemiology and Infection Control
. Philadelphia, PA: Lippincott Williams & Wilkins; 2004. 1659–1702.