ADVERTISEMENT
A Retrospective Review of the Outcomes of Vacuum-assisted Closure Therapy in a Vascular Surgery Unit
Vacuum-assisted closure, or negative pressure wound therapy (NPWT), is being used increasingly to treat chronic and complicated wounds since its effectiveness was documented in a pig model.1,2
The concept of using controlled subatmospheric pressure in the treatment of wounds was initially described by Fleischmann et al in 1993.3,4 Since being pioneered by Dr. Michael Morykwas and Dr. Louis Argenta in 1997 to accelerate wound healing by secondary intention in compromised patients, vacuum-assisted closure has been indicated in a variety of clinical settings: acute wounds, chronic wounds (diabetic foot ulcers, pressure sores, stasis ulcers), salvage procedures (infected wounds, dehisced wounds), and adjuncts to surgery (wound bed preparation, securing skin grafts).1–11 It has been shown to be useful in complicated, deep wounds of mixed etiologies.9 Contraindications for its use include malignancy, untreated infection, osteomyelitis, fistula to organs or body cavities, presence of necrotic tissue, bleeding wounds, and underlying exposed organs or vessels.5,6,9,11,12 Wound problems are common in vascular surgery. They pose a therapeutic dilemma due to multiple etiologies affecting the healing process in a patient population who are often debilitated with multiple comorbidities. Chronic wounds not only cause the patient much pain and suffering, but also place a financial burden on society through increased morbidity and mortality.5,6,13,14 Vacuum-assisted closure removes excess serous fluid, blood, and debris, decreases tissue bacterial levels due to the hypoxic environment, increases angiogenesis and tissue perfusion, provides a closed, warm, moist environment, and an even distribution of negative pressure over the wound’s surface to promote granulation and therefore, leads to improved wound healing.1–3,7–9,11,15–20 This results in granulation tissue development, as well as wound contracture and neo-epithelization.9 Additional advantages of vacuum-assisted closure over conventional dressing methods are reduced wound desiccation and need for frequent dressing changes, which reduces nursing time and increases patient comfort, reduces duration of inpatient stay, and potentially reduces overall treatment costs.5,7,13,21–23 The use of vacuum-assisted closure has increased despite a paucity of good clinical trials regarding its effectiveness. There are only a few randomized controlled trials conducted to date.1,13,17,24 It has been reported that there is no difference in overall healing time and reduction of wound surface area compared to conventional dressings.1The Australian Safety and Efficacy Register of New Interventional Procedures—Surgical (ASERNIP-S)’s accelerated systemic review concluded that while some studies were probably too small to detect significant differences and the quality of all available randomized controlled trials were poor, some did show vacuum-assisted closure achieved better healing than standard care methods.5 A longitudinal review of all vacuum-assisted closure use at the Royal Perth Hospital (RPH) vascular surgery unit was conducted from 2002 to 2006 to assess the effectiveness of the authors’ vacuum-assisted closure usage.
Methods
Study design and patients. A retrospective review was conducted on all vascular inpatients in the vascular surgery ward 6G of RPH for which vacuum-assisted closure device (V.A.C.®, KCI, San Antonio, Tex) orders were placed. The list was obtained from KCI. Royal Perth Hospital is Western Australia’s premier teaching hospital, and is the main tertiary referral center for the state with the second largest trauma workload in the country. The hospital has a high rate of patient turnover and admits 73,000 patients annually. All inpatients were tracked from when vacuum-assisted closure therapy was first used in August 2002 to October 2006 were included. Patient notes were recalled from medical records. The primary outcome measure was wounds that either did or did not heal. Secondary outcome measures were length of hospital stay, time for rehabilitation, outpatient follow-ups, and healing time. Wound healing time was calculated from the time vacuum-assisted closure was first applied to the time that the wound was first documented as healed, which was defined as complete re-epithelialization. The median value was used in place of mean where there is a larger range of value. It was hypothesized that wounds with better vascular supply would heal faster. In the vascular patients, this usually means wounds that are more proximal would heal faster with vacuum-assisted closure therapy. Procedures. The patients’ basic demographic data, wound and vacuum-assisted closure profiles, and outcomes were recorded (see the Appendix). 
Bias. Bias was introduced due to the retrospective nature of the study. By taking a consecutive series of patients, they were considered representative of the RPH patient population. Selection bias occurred from those with missing data. Patients were not chosen randomly. Many of these patients were debilitated and poor candidates for surgery, vacuum-assisted closure was chosen to achieve a more rapid healing by secondary intention. Hence, the factors that influenced the choice of vacuum-assisted closure treatment in these patients, which might influence wound healing under any other treatment, might appear as factors that slowed healing in patients treated with vacuum-assisted closure. This will result in a lower prediction of success rate due to the choice of patient population. When vacuum-assisted closure was ceased, other dressings were used to dress the wounds, which could not be controlled for and is subject to bias based on the experience of the staff caring for the patient. Statistical analysis. The Statistical Package for Social Sciences (SPSS) was used for data insertion and analysis. Student’s t-test, chi-squared, and nonparametric tests were used as appropriate. Multiple linear logistic regression and survival analysis were used for multi-variable analysis; P < 0.05 was considered statistically significant.
Results
There were 74 patients in total: 34 (45.9%) women and 40 (54.1%) men with a median age of 65.5 (range 19–95 years). Three of the 74 patients had 2 wounds resulting in a total of 77 wounds: 25 foot (32.5%), 13 (16.9%) each for toe and groin, 9 (11.7%) each for leg and thigh, 7 (9.1%) trunk, and 1 (1.3%) neck. Table 1 illustrates the length of stay for each wound region. The median length of stay was 36 days (range 6–130 days). 
Sixty-seven (90.5%) patients were discharged to home, 3 (4.1%) to nursing home, 2 (2.7%) to hostel, and 2 died. The median rehabilitation time was 42.5 days (range 2–186); the median outpatient follow up was 85 days (range 4–698). Most of the wounds healed (n = 45, 58.44%), 14 (18.18%) are ongoing, 12 (15.88%) patients died, and 3 (3.9%) had below knee amputation (BKA). Two (2.6%) patients were lost to follow up. The outcomes for each area are shown in Table 2. One neck wound, which had healed completely, was excluded from the study. 
These wounds were reclassified as either healed or not healed, and those in the “lost to follow up” and “ongoing” groups were classified as missing. Forty-five (75.4%) healed and 15 (24.6%) did not heal. The median healing time for all the wounds was 86 days (range 17–374 days). Chi-squared test showed a significant association with healing (P = 0.012) in patients with diabetes who were treated with a combination of oral hypoglycemics and insulin. Of the diabetic complications, there is a significant association with healing in patients with stroke (P = 0.029), PVD (P = 0.018), and retinopathy (P = 0.028); there may be some association in patients on dialysis (P = 0.057). However, this association needs to be interpreted with caution since there were only 2 patients. Interestingly, dietitian input and dietary supplementation is strongly significant (P = 0.003 and P = 0.011, respectively). The total of other vascular procedure(s) before admission had a P value of 0.079, and total inpatient vascular procedure(s) during admission had a P value of 0.061. Multivariate analysis showed that the dietitian input and stroke were independently significant in relation to healing, with an odds ratio of 0.013 (CI: 0.065–0.008, P = 0.05) and 0.009 (CI: 0.004–0.468, P = 0.05), respectively. Survival analysis (Mantel-Cox log rank) showed a significant association between different wound locations and healing (P = 0.025). However, the pattern was difficult to see. The locations were collapsed into 4 main categories: trunk, leg, foot, and toe. A clear pattern can be seen with 2 statistically significant groups (P = 0.019): toe and foot, leg and trunk. These 4 groups were collapsed further into 2 groups (Figure 1). The analysis confirmed the hypothesis that wounds that were more distal take longer (240 days longer, [280%]) to heal. 
There is also significant association in patients on combination therapy (P = 0.015), which healed significantly faster. Analysis of the ankle brachial index (ABI) values showed of all the variables that only left dorsalis pedis (LDP) was predictive (P = 0.008, CI: 0.172–0.772). A 1-unit increase in LDP was associated with a healing time reduction of 64%. However, there was no association with the rest of the ABI. Higher weight gain was associated with a faster healing time (P = 0.02, CI: 0.96–0.996). Healing time was reduced by 2.2% for each kilogram gained. There was no statistical significance of lower weight gain (the lowest weight recorded during admission). However, there was little power in the analysis due to the missing values (n = 44, 59.5%). Time in HDA significantly prolonged healing (P = 0.004, CI: 1.104–1.687). Healing time was prolonged by 37% for each day spent in HDA. A model of time to heal is built by incorporating significant predictors from the univariate Kaplan-Meier and Cox regressions to produce a multivariate model where the coefficient for each variable was controlled for all other variables in the model. This gave the most complete picture of healing time, as illustrated in Figure 2. A patient who eventually healed but took an extensive amount of time to do so, was eliminated from this analysis. Four variables emerged as independent predictors of healing time: combination therapy (hazard ratio [HR] = 6.5, P = 0.003), LDP (HR = 0.36, P = 0.036), highest weight (HR = 0.97, P = 0.029), location below the ankle (HR = 0.35, P = 0.034). Highest weight was also excluded from the analysis, as there are a large number of data missing. This again confirms the hypothesis. Patient demographics. The patients’ demographic data are shown in Table 3. 
Table 4 shows the range and median ABI values of the 74 patients. Two patients had left above knee amputation (AKA), 1 had a right AKA, and 4 had right BKA. Twelve patients had no prior ABI values. The most common vascular procedures the patients had before admission were femoral-popliteal bypass (n = 10), foot debridement (n = 7), above knee amputation (n = 5), great toe amputation (n= 4), and femoral-femoral crossover (n = 4). Thirty-seven (50%) patients had no prior vascular procedures and 5 (6.8%) had ≥ 5 procedures. The most common initial procedure was foot debridement (n = 11), followed by great toe amputation (n = 8), femoral-popliteal bypass (n = 5), and second toe amputation (n = 5), fourth and fifth toe amputation (n = 4 each), and open abdominal aortic aneurysm repair (n = 3), and BKA (n = 3). Two (2.7%) patients were conservatively treated; the majority (n = 44, 59.5%) of patients had 1 initial vascular procedure. Thirteen (17.6%) patients required Intensive Care Unit (ICU) admission, with a mean time of 3.77 days (range 1–10). Nine (69.2%) patients had a < 5 days in ICU. Nine (12.2%) patients required HDA admissions during their stay, with a mean of 2.89 days (range 1–6). Seven (77.7%) had a < 5 days stay in HDA. Complications are common and often numerous in vascular patients. Sixteen (21.6%) patients had a stay without complication. Common complications were hyperglycemia (n = 14), urinary tract infection and pain (n = 11 each), non-ST elevation myocardial infarction (n = 10), anemia (n = 9), acute renal failure (n = 8), congestive cardiac failure and hyperkalemia (n = 7 each), and atrial fibrillation (n = 6). Three patients were diagnosed with osteomyelitis, Pyoderma gangrenosum, and compartment syndrome, and 1 with heparin induced thrombocytopenic syndrome. Fifty percent of the patients had < 5 complications during their stay, 21 (28.4%) had ≥ 5 complications. Wound profiles. Twenty-four (31.17%) were on the right, 48 (62.34%) on the left, and 5 (6.49%) were midline. Twenty-nine (37.7%) were surgical incision breakdown. Majority (n = 42, 54.5%) were infected wounds secondary to causes other than surgery (infected secondary wounds). Six (7.8%) were infected wounds that dehisced after initial closure. The time to break down ranged from 3 days to 48 days, with a median time of 10 days. The overall median wound surface area was 18.7cm2 (range: length 0.5 cm–20 cm, width 0.5 cm–15 cm): toe (15 cm2), foot (20 cm2), leg (32 cm2), thigh (57.5 cm2), groin (20 cm2), trunk (15 cm2), and neck (1.5 cm2). However, 38 and 40 data points were unavailable for the length and width calculations, respectively.
There were 6 (7.79%) wounds with exposed graft, 4 (5.19%) with exposed bones, and 2 (2.60%) with exposed tendons. Thirteen (17.6%) patients had no follow-up inpatient procedures, and the most frequent were foot debridement (n = 30), above knee amputation site debridement (n = 22), split skin graft to foot (n = 10), heel debridement and superficial femoral artery (SFA) angioplasty (n = 8 for each), and superficial femoral artery (SFA) endarterectomy (n = 6). Thirteen (17.6%) patients were treated conservatively. The majority of patients (n = 51, 68.9%) had < 5 total inpatient procedures.
There were 49 organisms grown. The most commonly found were Enterococcus faecalis (n = 27), Staphylococcus aureus (n = 23), Enterococcus sp (n = 22), Pseudomonas aeruginosa (n = 18), Proteus mirabilis (n = 12), and coagulase negative Staphylococcus (n = 9). Six patients had no organism cultured. The majority (n = 45, 60.8%) of patients had ≤ 3 organisms cultured. The median length of total antibiotics use was 30.5 days (range 0–106). Eleven (14.9%) and 3 (4.1%) patients had no oral and intravenous (IV) antibiotics prescribed respectively. Majority (n = 47, 63.5% for both) of the patients had ≤ 2 antibiotics usage. The most commonly used oral antibiotics were clavulanate potassium (Augmentin Duo Forte™, n = 38), ciprofloxacin (n = 20), clindamycin (n = 13), flucloxacillin (n = 10), and amoxicillin and metronidazole (n = 8 each). The most frequently prescribed IV antibiotics were ticarcillin disodium (Timentin™, n = 51), flucloxacillin (n = 25), metronidazole (n = 20), vancomycin (n = 17), and cephazolin (n = 12). Treatment duration. The median duration of vacuum-assisted closure application was 11 days (range 1–38 days). The most common reason for discontinuation was good healing (n = 27, 26%) in terms of healthy granulation tissue in the wound defect, which allows for delayed surgical closure or healing by secondary intention. Other reasons for discontinuation were pain (n = 10, 9.6%), dryness (n = 9, 8.7%), sloughy wounds (n = 9, 8.7%), and excoriation (n = 8, 7.7%). In 6 (5.8%) cases, vacuum-assisted closure was ceased due to poor healing. No problems with erosion secondary to the evacuation tube onto adjacent tissue were observed.
Discussion
Wounds in vascular patients are often chronic and complicated, affected by multiple comorbidities. Almost two-thirds of the patients had diabetes (60.8%). Patients on combination therapy with oral hypoglycemics and insulin had associated impaired healing. Insulin was added in an attempt to achieve better glycemic control in those whose blood sugar is not controlled by oral hypoglycemics alone. Thirty-nine (86.7%) of the patients with diabetes had peripheral vascular disease (PVD), which adversely affected healing. Other comorbidities like stroke and retinopathy were also found to be risk factors for poor healing. This was a reflection of the degree of diabetic control, as well as micro- and macrocirculatory damage due to the disease, which had implication on the vascular supply and ultimately healing. Dietitian input and supplementation were strong predictors of poor healing and was likely to be multifactorial. Clearly it was acting as a proxy variable for the multifactorial patient characteristics that resulted in a referral for dietary advice. Many of these vascular patients were already debilitated from their multitude of medical problems, their recovery was slow, and this might in turn impact their general and nutritional status, hence dietitian input was usually sought early in the authors’ institution to improve nutritional status in these patients. These factors should be examined as prognostic indicators for poor healing. With the advances in critical care and operative techniques, patients can be treated more aggressively. Of note, further analysis on weight gain is associated with a more rapid healing. This therefore justifies dietary input early in vascular patients. Additionally, their inpatient stay was mostly uneventful. Most of the patients required multiple operative procedures (9 patients had ≥ 5 different procedures, with 3 who had 10), had multiple organisms cultured from their wounds (13 patients had ≥ 5 different organisms, with 2 who had 7), and multiple complications (21 patients had ≥ 5 complications, 2 of which had up to 13). HDA stay is associated with 35.6% longer healing time. This reflected on the poorer general medical condition, which contributes to a delayed healing time. Andros et al9 also found that vascular disease has been found to be a major risk factor for patients with diabetes to develop foot ulceration and subsequent amputation, an indicator of poor healing. Armstrong et al24 in their study of well-controlled patients with diabetes with an ABI ≥ 0.7 and £ 1.2, showed a higher proportion of healed wounds, shorter time to wound closure, more robust granulation tissue response, and a potential for reduced risk of a second amputation in those treated with vacuum-assisted closure after partial foot amputation. In vascular patients, distal vascular supply is often compromised, especially among patients with a lower ABI value, which is a reflection of poor tissue perfusion and subsequent poor healing. It was found that wounds with higher LDP values that are more proximal have the best healing rates. There appears to be 2 patterns with an approximate cut-off value at 100 days. The gradient is initially steep early in the course for both groups and this slows down approximately after 100 days for both groups. However, the more distal group is parallel to the proximal group and it never approximates the latter. The goal of wound treatment is to create an environment conducive to healing, and ultimately to provide definitive treatment to cover the wound.6 Vacuum-assisted closure has been shown to be safe and effective in treating complex diabetic foot wounds with reported healing rates of 70% by secondary intention.9 Braakenburg et al1 showed vacuum-assisted closure therapy is most beneficial for patients with cardiovascular disease and/or diabetes mellitus. Additionally, this form of therapy has the potential to reduce overall treatment cost, and is more comfortable for patients and nursing staff. The cost effectiveness has been reported in terms of shortened hospital stays, decreased overall medical cost and limb salvage.6,7,9,10,21,23 Extrapolation from Medicare records suggested that treatment of chronic wounds may be more cost-effective compared to conventional methods.7 However, many of these studies are small, methodologically flawed, and insufficiently powered—interpretations should be made with care.10 In the present study, 75.4% of the wounds healed with vacuum-assisted closure. Of those that did not heal, 3 proceeded to have BKA. Two of the 12 deaths occurred in the hospital and were related to patient comorbidities (1 had a cardiorespiratory arrest, the other was for palliation), whereas the rest died due to other comorbidities unrelated to the vacuum-assisted closure therapy. A large proportion (n = 5, 45.5%) died within 12–26 months of discharge. Vacuum-assisted closure has been reported to be effective at treating various chronic and complex wounds, with significantly greater reduction in wound volume, depth, increased rate of re-epithelialization, and treatment duration.5,9,11 The median healing time in different wounds with vacuum-assisted closure has been reported to be 29 days.1,13 It accounted for a 28.4% reduction of foot ulcers surface area when compared with a 9.5% increase in the control group treated with saline-moistened gauze.5 However, this 
difference was not statistically different due to the small sample size.5,21 It has also been shown to be useful in allowing the planning of elective definitive surgery without compromising the wound or final outcome.4 Vacuum-assisted closure continually decontaminates the wound and drains the wound surface of exudates, which contains a large amount of proteases.3,9,10,19,20 These would normally inhibit fibroblastic division, collagen production, and cell growth.9,19 Fluid removal helps with localized edema that causes an increase in interstitial pressure, the consequent of which is the occlusion of microvasculature and lymphatics, decreased nutrient, and oxygen delivery.9 Protein degradation enzyme is released with metabolic waste accumulation and increased bacterial colonization, which causes capillary damage and hypoxia.9 It also provides a moist environment to promote granulation tissue formation and prevents eschar formation, which allows for a smoother pathway to re-epithelialize the wound surface.9 Angiogenesis is stimulated, which improves tissue oxygenation and tissue reconstruction.10,19 This occurs even in patients with diabetic microangiopathy, as it promotes healing of distal lesions.19 The reported side effects of vacuum-assisted closure use are periwound maceration, skin necrosis, allergies to adhesive dressing, infection, discomfort with application of pressures > 100 mmHg with the initial foam collapse or dressing change, excessive tissue growth into dressing, minor bleeding, and fistula or sinus formation.5,7,10,25 These complications have been limited in the clinical setting and can be reduced if the device is used properly.7,11 For instance, maintaining a good seal is important, as an air leak has the potential to cause wound desiccation.4
In this review, there was no increased bleeding in patients undergoing anticoagulant treatment and had no adverse effect on wounds with exposed underlying structures. The major indications for cessation of vacuum-assisted closure use, apart from achievement of good healing, were pain, dryness, increased slough production, and excoriation. The recommended pressure (-125 mmHg) was utilized based on an experimental animal study where blood flow was shown to be increased 4-fold.4,6,15,21 Pain with application of vacuum-assisted closure can be limited by starting at a low pressure of -50 mmHg on continuous suction rather than the intermittent suction, and then slowly increasing this to the target -125 mmHg.4,25 Many of the patients also received analgesics to ease dressing changes.
Conclusion
The present study included patients with multiple vascular risk factors for wound healing, and who were more often than not debilitated. Vacuum-assisted closure is an effective therapy to promote healing with adequate debridement in a well-vascularized bed, to prepare the wound for closure by secondary intention, delayed primary closure, skin graft, or flap coverage.9 It is not only useful as an adjunct to avoid amputation, but also a good temporizing measure that allows for simple solutions to complex reconstructive challenges.9 It is a useful tool when used judiciously on select patients.18 Vacuum-assisted closure has been deemed by ASERNIP-S to be a promising alternative for wound management.5 The strongest predictors of poor healing were the use of combination therapy for glycemic control, PVD, stroke, retinopathy, dietitian referral, and supplementation. HDA admission is a covariable in predicting delayed healing, while a more proximal wound, better LDP value, and weight gain are predictors of more rapid healing. Vacuum-assisted closure was found to be most helpful in wounds above the ankle where the blood supply is adequate. It is useful for both the in- and outpatient settings to achieve primary closure and as an adjunct to a definitive procedure.
