Bacterial Species and Load Increase During Negative Pressure Wound Therapy: A Prospective Cohort Study

Negative pressure wound therapy (NPWT) was first described by Fleischmann et al in 1993.^1,2 It entails the introduction of a subatmospheric pressure, produced by suction through a tube, to a dressing foam or gauze, which covers the wound bed. This can be carried out with the use of a NPWT system (V.A.C. Therapy; KCI, San Antonio, TX), that delivers continuous or intermittent negative pressure to the wound bed. The introduction of NPWT took place in Europe in 1994,³ and an overview of studies of its first implications in various wounds were published by Argenta⁴ and Morkykwas⁵ in 1997. Negative pressure wound therapy is used to promote granulation tissue formation and accelerate wound healing; it reduces edema, increases tissue perfusion, and may prevent infection.^6-9 Several studies^10-12 have shown NPWT can be safely applied in infected wounds after debridement. It can be used in a variety of wound types, including acute, chronic, traumatic, and surgical, as well as for flap surgery and skin grafts, in diverse anatomical locations.^13,14

However, the bacterial growth or clearance in the wound during NPWT remains a complex matter. Despite reports indicating NPWT can achieve bacterial clearance and prevent infection, the change of the bacterial species and load during NPWT is still unclear, and studies have published contradictory results.^15,16 Several studies^5,17,18 reported a positive influence of NPWT (eg, clearance or reduction of bacteria), while other studies^19-22 reported a stable or even increase of bacterial growth in NPWT. The systematic review by Glass et al¹⁶ indicates the general belief that NPWT suppresses bacterial growth is an oversimplification, and the change in bacterial load is likely to be species-specific. Of note, 2 clinical studies^20,23 examined the change of bacterial species during NPWT and showed contradictory results in the shifts of bacterial species; Mouës et al²⁰ studied a relatively small group of patients (N = 29), in which the type of antibiotics used was not specified, and the other study by Jentzsch et al²³ only involved patients in a trauma center. 

The primary objective of this study was to determine the change in bacterial species and load, as well as the incidence of infection, before and after NPWT, in patients with various types of wounds. Furthermore, the course of wound infections during NPWT were evaluated.

Study design and participants
In this multicenter, observational, prospective cohort study, patients from the department of surgery at the Red Cross Hospital (Beverwijk, The Netherlands), Noordwest Hospital Group (Alkmaar, The Netherlands), and Leiden University Medical Center (LUMC; Leiden, The Netherlands) were included. Patients 18 years of age or older who had an indication for NPWT, as noted by their treating physician, were eligible for inclusion. Patients were excluded if it was not possible to take a wound swab or if they required NPWT in the head/neck region (due to the divergent culture protocol required for wound swabs taken from these regions). Regular NPWT was conducted according to the local protocol in the 3 aforementioned centers. No formal sample size calculation was performed for this observational study; the targeted number of patients was 100. 

The study was conducted according to the principles of the Declaration of Helsinki (World Medical Association Medical Ethics Manual 2^nd ed, 2009) and in accordance with the Dutch Medical Research Involving Human Subjects Act. The study was approved by the Committee of Medical Ethics at LUMC (P15.227), and local permission of each participating center was obtained.

Study procedure: swab culture
To lower the burden of treatment in patients, a noninvasive assessment of wound colonization with swab cultures was chosen. Superficial wound swabs were taken without the dressing foam in situ at 2 time points: (1) before the start of NPWT and (2) at the end of the study (ie, at the end of NPWT or after 6 weeks of follow-up). Additional wound swabs during the course of NPWT were taken only if ordered by the treating physician for clinical reasons. The wound was cleaned or debrided by the treating wound nurse or physician. A sterile dry swab was taken in a linear line of about 10 cm along the periphery of the wound. If the wound was smaller than 10 cm, the swab was rotated 3 times along the periphery. The swab was placed in a tube containing Stuart Transport Medium (Copan Italia SpA, Brescia, Italy) and then transported to the laboratory. Swabs were cultured on sheep blood agar with gentamicin (GBA) and without gentamicin (BA), cystine lactose electrolyte deficient (CLED), and Columbia agar containing colistin and nalidixic acid (CNA), then incubated overnight twice at 35°C. Blood agar and CLED were incubated in O₂, CNA in CO₂, and GBA in anaerobic atmosphere. 

Study definitions
Bacterial species, load, multidrug resistance, and wound infection. The positive wound swab cultures were subdivided into pathogenic and non-pathogenic bacteria. Pathogenic bacteria are bacteria that have a known negative impact on wound healing and include: Staphylococcus aureus, Pseudomonas aeruginosa, Proteus mirabilis, Streptococcus dysgalactiae, S pyogenes, S lugdunensis, S canis, and Clostridium perfringens. In this study, a wound swab culture was defined as “positive” if there was growth of any of these pathogenic bacteria and as “negative” if there was neither bacterial growth nor growth of only non-pathogenic bacteria. In other words, bacterial species and load of non-pathogenic bacteria were not taken into account in this study. Multidrug-resistant (MDR) bacteria were defined as per the national guidelines of the Dutch Infection Prevention Working Group,²⁴ which was based on the guidelines from the US Centers for Disease Control and Prevention.^25,26 Multidrug-resistant P aeruginosa was defined as having a resistance for at least 3 types of antibiotics.²⁴ 

Bacterial load was determined semi-quantitatively by counting the number of colonies on each plate and categorizing the load as low (+; ≤10 colonies), medium (++; >10 but <100 colonies), or high (+++; ≥100 colonies). S aureus, ß-haemolytic streptococci, P aeruginosa, and Clostridium spp were further analyzed irrespective of the number of colonies. Skin inhabitants such as coagulase-negative staphylococci, Propionibacterium acnes, and Corynebacterium spp were not further analyzed unless grown in pure culture and with more than 100 colonies. 

The presence of a wound infection (ie, identified with yes or no) was based on the judgment of the treating wound nurse or physician. If the presence of a wound infection was observed, the treating nurse or physician noted which of the following symptoms the patient displayed: purulent drainage, redness, localized pain or tenderness, swelling, warmth, and/or abscess. 

Additional study parameters. The additional study parameters included patient characteristics, wound characteristics, and treatments (Table 1). Patient characteristics consisted of age, gender, body mass index (BMI), smoking, presence of diabetes mellitus, presence of cardiovascular disease according to the definitions of the Framingham study,²⁷ and use of immunosuppressive medication (eg, corticosteroids) or presence of an immunosuppressive illness that could impair wound healing (eg, active malignancy). The wound characteristics included the indication for (ie, ending of) NPWT, location, size, depth, and duration. Treatment parameters comprise use of subatmospheric pressure, type of wound coverage, use of silver-containing bandage, and type of antibiotics used.

Statistical analysis
The patient, wound, and treatment characteristics are presented as numerical values for categorical data or as the median and first quartile to third quartile (Q1–Q3) values for continuous data with a non-parametric distribution. Either McNemar’s test or Wilcoxon signed-rank test were used for comparing paired categorical data. A value of P < .05 was used as the threshold for statistical significance. Data were analyzed using SPSS version 24.0 (IBM Corp, Armonk, NY). 

Study group 
A total of 163 patients were included in the study between March 2016 and April 2018. Of these patients, 59 were excluded from the analysis because no swab was taken at the end of their study period, leaving 104 patients eligible for analysis (Figure 1). Due to logistic reasons (including unavailability of personnel), 21% of the eligible patients were not included in 1 of the hospitals. In the other 2 hospitals, this number is not known but not expected to be different. Patient and wound characteristics of the study group are presented in Table 1. The male population comprised 47%, with the the median age at 63.0 years (Q1–Q3, 49.3–76.8) and the median BMI at 26.1 (Q1–Q3, 23.6–30.8). The most common wound types were surgical (36%) and traumatic (12%). The wounds were located on the legs (39%), trunk (33%), feet (15%), sacrum (6%), groin (5%), and arms (3%). At the time of inclusion, the wounds had a median length of 8.0 cm (Q1–Q3, 5.0–17.0), width of 4.9 cm (Q1-Q3, 2.3–8.5), and depth of 1.5 cm (Q1-Q3, 1.0–3.0). 

Bacterial species and load
In total, 26 patients who had a negative wound swab culture pre-NPWT developed a positive wound swab culture post-NPWT, and 36 patients with a positive wound swab culture pre-NPWT had a persistent positive wound swab culture post-NPWT (Table 2). Overall, patients had more positive wound swab cultures post-NPWT compared with pre-NPWT (McNemar’s test, P < .001).

The most common pathogens cultured were S aureus and P aeruginosa (Figure 2). There was a significant increase in the presence of S aureus from pre-NPWT (n = 24) to post-NPWT (n = 41) (McNemar’s test, P < .001). The increase in P aeruginosa was not significant (P = 1.000).

There were mixed cultures (2 or 3 different species) in 9 of the 104 patients in the pre-NPWT wound culture swab and in 16 patients in the post-NPWT wound culture swab (Table 3). A total of 8 different species were cultured pre- and post-NPWT (Table 3). Only 2 patients had MDR bacteria; 1 swab taken pre-NPWT was positive for MDR P aeruginosa, which was not cultured in the swab taken post-NPWT, and 1 swab detected MDR P aeruginosa in a post-NPWT swab, which was not cultured in the patient’s pre-NPWT swab.

The bacterial load of pathogenic bacteria in the pre-NPWT swabs was low in 62%, medium in 22%, and high in 16% of the cultures (Table 3). The bacterial load distribution in the post-NPWT swab cultures of pathogenic bacteria was 38%, 44%, and 18%, respectively. The bacterial load post-NPWT was significantly higher than pre-NPWT (Wilcoxon signed-rank test, P < .0001). 

Infection
Based on the judgment of the treating nurse or physician, 26 patients had a wound infection pre-NPWT. Of these, 21 did not have a wound infection post-NPWT, and 5 had a persisting wound infection; 2 patients with persistent infection cultured the same species, and the others cultured different species. A total of 14 patients who did not have a wound infection pre-NPWT developed a wound infection post-NPWT. There was no difference between the presence of infection post-NPWT versus pre-NPWT (McNemar’s test, P = .30) (Figure 3). 

Wound treatment
Most wound bed dressings were either black foam (79%), antiseptic Kerlix gauzes (6%), or a combination of the 2 (12%). The foam and gauzes were sometimes combined with a silicon or fatty sheet (8%). The median applied negative pressure to the wound bed was 100 mm Hg (Q1–Q3, 80 mm Hg–125 mm Hg). Silver bandage was used in 2.8% of the cases. The median duration of NPWT was 13 days (Q1–Q3, 7.0–21.0 days). Negative pressure wound therapy continued beyond the end of the study period (6 weeks) in 4 patients. The most common reasons for NPWT ending within the study period were that the wound was small enough to close spontaneously or via intervention (eg, skin graft) (54%), or necrosis developed (11%) (Table 4). 

Antibiotics
A total of 46 patients used antibiotics pre-NPWT and 17 patients at the end of the study period. Of these patients, 12 and 4 used antibiotics for an indication other than wound infection (eg, pneumonia) at the start and end of NPWT, respectively. The most prescribed antibiotics were flucloxacillin (16%), clindamycin (13%), and amoxicillin-clavulanic acid (13%) at the start of the study and flucloxacillin (21%) at the end of the study. Of the 26 patients with a wound infection pre-NPWT, 15 received antibiotics as additional treatment. Of the 19 patients with a wound infection post-NPWT, 3 received antibiotics as additional treatment.

The study aimed to assess the change of bacterial species and load in a variety of patients who received NPWT. The number of positive wound swab cultures of pathogenic bacteria increased post-NPWT versus pre-NPWT. The most cultured pathogenic bacteria were S aureus and P aeruginosa. The bacterial load was moderately higher at the end of NPWT than prior to NPWT. The incidence of infection did not differ between pre- and post-NPWT.

S aureus was the most cultured species in the study, which corresponds with other clinical studies^12,28-30 reporting cultured bacteria from wounds with NPWT. A study by Yusuf et al³¹ that used sonication to measure bacterial load and species in the foams of NPWT showed Enterobacteriaceae were the most cultured bacteria, followed by S aureus. In the present study, the presence of S aureus was also the only pathogenic bacteria that increased during NPWT. This is consistent with the results of Mouës et al,²⁰ in which an increase of S aureus was seen with NPWT, but it contrasts with the reported decrease in S aureus from Jentzsch et al.²³ In addition, 2 animal studies^32,33  compared NPWT with wet-to-dry (WTD) dressings and found P aeruginosa was reduced in the NPWT group compared with the WTD dressings. Lalliss et al³² also evaluated S aureus but did not find a reduction. 

The bacterial load was moderately higher post-NPWT than pre-NPWT. An increase in bacterial load with NPWT corresponded with the results reported by Weed et al.¹⁹ Their study¹⁹ retrospectively looked at the quantitative culture samples in 25 patients treated with NPWT and found a trend towards an increase of bacterial load in 43% of the NPWT patients without an apparent effect on wound healing. It also was seen in a randomized trial in acute and chronic wounds by Braakenburg et al,²⁹ who found an 84% increase in bacterial load in the NPWT group. The reported increase exceeded the 58% reported in their control group, which consisted of patients treated with modern dressings.²⁹ Mouës et al²⁰ investigated whether the positive effect on wound healing in patients treated with NPWT compared with conventional gauze therapy could be explained by an effect on the bacterial load. The authors²⁰ did not find a reduction of bacterial load in either group. In contrast, several studies^32-35 have found a reduction in bacterial load. Of note, 4 animal studies^22,32,33,35 examined the bacterial load in either solely S aureus-infected or P aeruginosa-infected animal models and found a decrease in bacterial load in the NPWT group compared with a control group using gauze dressings. A human retrospective study by Jentzsch et al²³ showed a significant decrease in bacterial load during NPWT. In addition, 2 prospective studies^36,37 showed a reduction of bacterial load in patients during NPWT, 1 of which used antiseptic fluid as the instillation with the NPWT.³⁷ Thus, evidence regarding the course of bacterial load remains contradictory. In the present study, the increase did not appear to negatively influence the presence of infection. 

The prevalence of MDR bacteria in the current population was very low; 2 swabs were positive for P aeruginosa. This is lower than the reported^38,39 prevalence of MDR bacteria in other countries, varying between 27% to 45% in wounds to 4.2% in nursing home residents.⁴⁰ The low occurrence of MDR bacteria in the current study in comparison with other countries probably has to do with the strict MDR policy that is applied in the Netherlands. This entails restrictive prescriptions of antibiotics and strict hygienic measures, including isolation of patients with MDR bacteria during hospital stay.⁴¹

Although the literature^12,42,43 shows NPWT can be safely applied in infected wounds, expert panel members in the use of NPWT indicate caution should be taken.⁴⁴ In the present study, NPWT was applied in 26 patients with a wound that, according to the treating nurse/physician, was infected. Notably, in 21 of these 26 patients, according to the nurse/physician, the infection ceased at the end of NPWT. This may indicate NPWT does not maintain a wound infection and supports the notion that NPWT is not necessarily contraindicated in infected wounds.

The use of antibiotics in this study was limited to less than half of the population at the start of NPWT to only 16% at the end of NPWT. Of the patients diagnosed with a wound infection, 40% did not receive antibiotics, possibly based on an adequate drainage of the wound using NPWT. The most prescribed antibiotic was the small-spectrum antibiotic flucloxacillin, which differed from other studies^23,45 describing the use of antibiotics in combination with NPWT. This is probably due to the strict prescription policy in which only S aureus is treated as it is the most pathogenic and holds the highest risk of causing systemic infection.

To the best of the authors’ knowledge, this is the largest reported prospective trial investigating bacterial species and load during the course of NPWT with a pre- and post-analysis to date. Moreover, to the best of the authors’ knowledge, this is the first study to examine the rate and course of infection in patients with a variety of indications for the use of NPWT. 

A control group with wounds treated with conventional dressings was not used, thus the outcomes cannot be compared with such a group. Also, superficial swabs were used as the tool to culture bacteria and bacterial load, whereas it is arguable that swabs are not as reliable as more invasive methods (ie, biopsies³⁶). Though superficial swabs are used as a standard measurement of bacterial flora and load, which renders the current results representative of regular daily practice; the heterogeneity in wound type might also be considered a limitation.

The time to complete wound healing in NPWT was not evaluated, so a statement cannot be made about bacterial species in relation to time to wound healing. In addition, as previously mentioned, the heterogeneity in wound type also might be considered a limitation, because statements for a specific group of patients, with regard to their indication for the use of NPWT, cannot be made. A next step might be to study different subgroups in a larger study with sufficient statistical power.

The relatively high number of patients excluded due to the missing data of the primary endpoint (ie, missing the end wound culture swab) may also be considered a limitation. However, there is no reason to believe the missing data were not at random, thus the cohort to the target number of included patients and data was completed.

Some minor limitations include the following: the study did not focus on other wound factors (eg, size, type of dressing) and their influences on bacteriology and infection.  

This study showed the number of S aureus strains and overall bacterial load increased during NPWT among the patient population, while the incidence of infection remained the same. Future studies should investigate whether the increase in bacterial load has an influence on other wound outcome parameters. 

Note: The authors gratefully acknowledge the contributions of the wound care nurses E. van Urk, R. Koopman-Kuyl, W. Murre, A. Ooijevaar-Kuijs, I. Olijve-Zijlstra, E. Meurs, K. Broekhuijsen, D. Hoedjes, and I. Henkes; physician assistant M. de Jong; trauma surgeon E. Krug; and medical microbiologists W. Rozemeijer, MD, and J. Sinnige, MD. 

Authors: Kelly Aranka Ayli Kwa, MD^1,2; Pieta Krijnen, PhD¹; Alexandra T. Bernards, MD, PhD³; Inger B. Schipper, MD, PhD¹; Annebeth Meij-de Vries, MD, PhD²; Roelf S. Breederveld, MD, PhD^1,2

Affiliations: ¹Department of Trauma Surgery, Leiden University Medical Center, Leiden, The Netherlands; ²Burn Center Beverwijk, Red Cross Hospital, Beverwijk, The Netherlands; and ³Department of Medical Microbiology, Leiden University Medical Center

Correspondence: Pieta Krijnen, PhD, Leiden University Medical Center, Department of Trauma Surgery, K6-50, P.O. Box 9600, 2300 RC Leiden, The Netherlands; p.krijnen@lumc.nl 

Disclosure: The authors disclose no financial or other conflicts of interest.