Background

Breast cancer is the most common malignancy among women in western countries, and one out of eight women will develop it during their lifetime. In the Netherlands, nearly 15,000 new cases are identified each year [1]. Breast cancer is the second cause of cancer-related deaths among women and the leading cause of disability-adjusted life years (DALYs) globally [2]. The majority of patients with breast cancer are treated surgically, amongst other treatment modalities. Although surgery of the breast is regarded as a clean procedure [3], a high incidence of surgical site infections (SSIs) is found, making it the most common complication [4]. Previous studies on SSIs in women after breast cancer surgery showed incidences ranging between 3% and 19% [4–7]. This is much higher than the expected 3.4% infection rate associated with clean surgical techniques [8].

SSIs are associated with considerable morbidity and reduced quality of life for patients. SSIs lead to extended hospital stays, re-admissions and re-operations, poor cosmetic results, delay in commencing adjuvant treatment, and they consequently result in additional costs and poorer outcomes [9–11]. Therefore, prevention of SSI has recently gained attention. A recent meta-analysis [12]identified several significant risk factors for SSI after breast cancer surgery, but the type of wound dressings was not evaluated in the included studies. A recent Cochrane Review [13]on the role of wound dressing in the prevention of SSIs revealed that in the current literature, there is no evidence that covering surgical wounds healing by primary intention with wound dressings reduces the rate of SSI, nor that any particular wound dressing is superior to another in this regard. Studies included in this review were mainly outdated and of poor quality. The authors concluded that decision-making should be based on dressing cost and the ability to deal with specific symptoms. High-quality research on the role of wound dressings in the prevention of SSIs is needed. As a result, the CDC guideline has no specific recommendation on the type of dressing or wearing time, except that primarily closed wounds should be sterile dressed for at least 24 to 48 hours [13, 14].

All sorts of wound dressings are available presently, containing different materials and using different application techniques. Characteristics of an ideal wound dressing are the ability to absorb and contain exudate without leakage, a lack of particulate contaminants left in the wound, thermal insulation, impermeability to water and bacteria, suitability with different skin closures (sutures, staples), avoidance of wound trauma on removal, little need for dressing change, provision of pain relief, cosmesis, comfort, and a positive effect on scar tissue formation [15,16].

AQUACEL Ag Surgical (Aquacel) is a type of wound dressing that is thought to meet these characteristics more than others: in-vitro tests showed that the silver in the dressing inhibits aerobic, anaerobic, gram-negative, and gram-positive bacteria, as well as yeast and fungi within 30 minutes [17,18]. The antibacterial activity lasts for 14 days [18]and it is occlusive. Several studies found that less changing of the dressing is needed. Furthermore, patient satisfaction was higher when a wound was treated with Aquacel [19, 20]. Despite Aquacel’s favourable characteristics, a randomized comparative study of Aquacel with other wound dressings after breast cancer surgery has not yet been performed.

Objectives

Our aim in this study was to compare Aquacel with standard gauze wound dressing in the occurrence of SSI among women after breast cancer surgery. We hypothesise that Aquacel will reduce the occurrence of SSI in this particular group of patients.

Trial design

This study was a prospective, open label, randomized, single center active controlled clinical study with a two arm 1:1 parallel group design. It was designed to assess the effectiveness of Aquacel in reducing the risk of SSIs in women after breast cancer surgery. The trial was set to establish the superiority of Aquacel to standard wound care. It was performed in a large secondary teaching hospital (Franciscus Gasthuis, Rotterdam), in a combined inpatient and outpatient setting. The inclusion period was between June 2013 and May 2016, with the last patient completing the 90-day follow-up in August 2016. The study protocol (dossier nr. NL42892.101.12) was reviewed and approved by the Medical Ethics Committee: Toetsingscommissie Wetenschappelijk Onderzoek Rotterdam (TWOR). After approval of the protocol the following changes in the protocol were made: Extension of the randomisation list to have the option to extend patient inclusion; addition of two exclusion criteria because of the anticipated differences of a priori SSI risk in neo-adjuvant chemotherapy and immediate reconstructive surgery; planning of an exploratory subgroup analysis of breast conserving surgery and mastectomy, for being clinically relevant.

The study was conducted according to the principles of the Declaration of Helsinki (version 10, October 2013) and in accordance with the Medical Research Involving Human Subjects Act (WMO).

Participants

All women of age 18 years or above who were diagnosed with breast cancer, needing uni- or bilateral ablative or breast conserving surgery in our hospital, were considered eligible for inclusion in our study. Patients were only included after giving a written informed consent.

Patients were excluded if they had local inflammation or ulceration of the breast, previous breast surgery in the previous 3 months, use of antibiotics in the previous 2 weeks, neo-adjuvant chemotherapy, intended immediate reconstructive surgery, a known allergy for Aquacel or silver, and the inability to read or understand the Dutch language to give informed consent or fill out questionnaires.

Interventions (surgical procedures, wound management and follow-up)

Included patients underwent ablative or conservative breast surgery, with or without an axillary procedure. All procedures were performed or supervised by senior surgeons with a case load of more than 50 per year. All patients received a single dose of intravenous antibiotics as recommended [21](cefazolin 1 gram, by hospital protocol) 1 hour before surgery. In accordance with the CDC guidelines, we considered bilateral procedures as two separate observations [22]. Drain management was performed according to the surgeon’s preference.

After surgery, wounds were cleaned with normal saline and patients received their allocated wound care: standard wound dressing, consisting of an eight layer woven cotton gauze fixed with adhesive tape, or Aquacel. AQUACEL Ag Surgical is a hydrocolloid dressing with hydrofiber technology that delivers ionic silver when it comes in contact with the wound (e.g., exudate). The occlusive dressing protects the skin surrounding the wound, by moisture retention. The material is soft and pliable and can therefore be adjusted to the size of the wound [17, 18].

Both standard wound dressing and Aquacel were kept in place for 7 days by protocol, unless saturated by excessive exudate. Between the 7^th and 10^th days after surgery, follow-up was scheduled at the outpatient clinic. Unblinded, the wound(s) were inspected on signs of an infection following CDC criteria [14, 22] by the independent surgeon or attending physician and any clinical diagnosis of SSI was made. Patients, who were unblinded, filled out a questionnaire on patient satisfaction. Re-admissions/operations, the occurrence of SSI diagnosis after the clinical assessment, and the use of antibiotics were scored 30 days after surgery in two ways: a blinded review of patients record by an independent physician and a telephone consult with the patient by an independent blinded nurse. Patient records were also checked for deep infections on the 90th postoperative day, by an independent blinded physician, to fulfil reporting guidelines [22].

Outcomes

The primary outcome of this study was the incidence (risk) of SSI. Secondary outcome measures were patient satisfaction, the re-admission and re-operation rate, antibiotic use within 30 days, the need for changing the dressing within the first 48 hours, wearing time of first applied dressing, and costs. For managerial purposes, we added a cost analysis.

Aquacel costs €23.25 per dressing. Standard wound care costs €1.28 per dressing. Mean wound-related treatment costs were calculated with the following equation: with attributable cost of SSI after breast surgery in a low (€510/$574[9]) and high(€3634/$4091[10]) cost model, based on the existing literature.

Assessment of SSI.

SSI outcome was scored by a blinded and independent physician, using the CDC criteria following the reporting instructions after breast procedures: 30 days follow-up for superficial incisional SSI, and 90 days for a deep incisional SSI [14, 22] (Fig 1). Final scoring was based on the information captured from 1) the unblinded clinical observation recorded during the follow-up visit at day 7–10 after surgery, 2) the blinded review of patients record and telephone consult on day 30 after surgery, and 3) the blinded review of patients record after completing 90 day follow-up.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. CDC criteria for an SSI [14,22].

^aDiagnosis of ‘cellulitis’ alone does not meet criterion 4 since 2010, but this change underestimates the infection rate and is not recommended to be used by Degnim [5].

https://doi.org/10.1371/journal.pone.0195715.g001

Sample size

A 10% (12.5% to 2.5%) absolute reduction of occurrence of wound infection was considered to be clinically relevant. To reject the null hypothesis (risk of SSI is equal between the wound dressing strategies) with an accepted type 1 error of 5% (two-sided) and type 2 error of 20%, at least 106 patients per treatment arm would be required (randomization ratio 1:1) (www.sealedenveloppe.com, Chi²-test). No interim analysis was planned.

Randomization (sequence generation, allocation, implementation) and blinding

A randomization list for up to 400 patients with an allocation ratio of 1:1 was computer generated (www.sealedenveloppe.com) with stratification by age>60 years of age, smoking, diabetes, use of corticosteroids, and the type of operation (lumpectomy vs mastectomy). The independent nurse created 400 instead of 212 numbers, under supervision of the study supervisor to compensate for and anticipate on the possibility to extend patient inclusion, any unplanned exploratory subgroup analysis and to guarantee a sufficient number of study numbers in each stratum. Patients were enrolled by physicians and group assignment was performed by the independent nurse. Allocation was performed on the day of surgery; for concealed allocation, the operation department was informed by the independent nurse just before applying the dressing. Surgeons were blinded for treatment allocation during surgery until the moment of applying the dressing, not during follow-up. Patients could not be blinded because of the nature of the intervention.

Statistical methods

Statistical analysis was performed according to the intention-to-treat principle. Because protocol compliance was high, per-protocol analysis was avoided. As bilateral cases only occurred in two patients, adjusted statistics through repeated measurements analysis were not performed. Efforts were made to minimise missing data, by recalling patients not attending follow-up. If complete follow-up data were missing, patients were excluded from the analysis. Otherwise, patients were analysed only on available data. Differences in the baseline characteristics and the primary and secondary outcome measures between the allocated study groups were compared with the chi-squared test for nominal/ordinal variables (e.g. proportions of SSI, re-operations, early dressing change), the independent Student’s t-test for continuous variables with normal distributions (age), and the nonparametric Mann-Whitney U test for continuous variables with skewed distributions (e.g. operation duration, patient satisfaction, mean costs). Differences in outcome measures between groups were estimated using a logistic regression analysis (enter analysis) with the respective outcome measure as dependent variable and randomization factors and type of wound dressing as independent variables. Adjusted odds ratios with 95%CIs and p-values are reported. Differences in wearing time of the first dressing between groups were estimated using Cox regression analysis with dressing change as event, randomization factors as covariables and wearing time as time to event. Associations between the potential risk factors and the presence of SSI were quantified in terms of odds ratios (ORs with 95%CIs) and tested using binary multiple logistic regression analysis. Risk factors with a p-value <0.1 in univariate analysis were included in the logistic regression model (enter analysis). An unplanned exploratory subgroup analysis, as well as an unplanned effect modifier analysis was performed to detect differences in the primary outcome measure between breast conserving and ablative surgery, because of its high clinical relevance. Effect modification was modelled as an interaction effect of mode of breast surgery (lumpectomy vs. mastectomy) x allocated wound dressing (Aquacel vs. standard). A p-value < 0.05 (two-sided) was considered to be significant. Statistical analyses were performed using IBM-SPSS version 24 (IBM Corporation, Armonk, New York, USA).

Participant flow

A total of 295 patients underwent breast cancer surgery in our hospital, of which 59 patients were excluded. Fig 2 shows the study profile: 236 patients were randomized, 107 patients to the Aquacel group and 129 patients to the control group. Total loss to follow-up was 6/236 (2.5%), these patients were excluded from the analysis. Based on the available data in the medical records of these patients, no SSI occurred in these patients. Protocol compliance was 227/230 (98.7%). Finally, 230 patients were analysed on a intention-to-treat basis. There were no missing data in the primary outcome and secondary outcome measures (except dressing change), from which we conclude that analysis of available data only is not likely to have caused bias in the results.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Patient flowchart (CONSORT).

https://doi.org/10.1371/journal.pone.0195715.g002

Table 1 shows the baseline characteristics of patients. As expected, there were no differences between the groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics.

https://doi.org/10.1371/journal.pone.0195715.t001

Outcomes and estimation

Table 2 displays the outcome measures between the groups. A total of 7 patients (6.6%) developed an SSI in the Aquacel group, and 16 patients (12.9%) developed an SSI in the control group (RR 0.51 [95% Confidence Interval (CI): 0.22–1.20]; adjusted Odds Ratio (OR) 0.49 [0.19–1.25] p = 0.135). The majority of SSIs were superficial in both groups. In the Aquacel group, only one patient (0.9%) had a deep SSI, compared to four patients (3.2%) in the control group (RR 0.29 [CI: 0.03–2.58], adjusted OR 0.28 [0.03–2.54] p = 0.257).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Outcome measures and comparison between the groups.

https://doi.org/10.1371/journal.pone.0195715.t002

Furthermore, the Aquacel group scored significantly better than controls on patient satisfaction (median score of 8 vs. 7 [p = 0.006] and need for changing the dressing within the first 48 hours (9.6%, vs. 45.9% [RR 0.21 (CI: 0.11–0.40),adjusted OR 0.12 [0.05–0.27] p<0.001). For the outcome measure dressing change, a sensitivity analysis was performed showing an observed difference between the allocated groups of 9/94 (9.6%) and 45/98 (45.9%); a minimal estimated difference of 21/106 (19.8%) Aquacel vs 45/124 (36.3%) control; and a maximal estimated difference of 9/106 (8.5%) vs 71/124 (57.3%).

For the whole group lowering the need for early changing of the dressing was associated with higher patient satisfaction (median score of 8 in the ‘no early change’ vs. 7 in the ‘early change’ group; p = 0.003. A non-significant reduction in re-operation rate was found (0% vs. 3.2%, p = 0.062). Of the 23 patients with an SSI in this study, 15 had a positive bacterial culture result. Among these patients, Staphylococcus aureus and Pseudomonas aeruginosa were found most frequently, in 10/15 cases (67%) and in 3/15 cases (20%), respectively. Other culture results can be found in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Microbiological culture results of SSI cases.

https://doi.org/10.1371/journal.pone.0195715.t003

Ancillary analyses

Harms

There were no important harms or unintended effects in both groups.

Strengths and limitations

To our knowledge, this is the first well-designed RCT that investigates the effect of a silver-containing dressing on SSI rates after breast cancer surgery. Major strengths of this study, apart from its randomized design, are the inclusion of all types of breast cancer surgery, the use of very strict criteria for SSI (CDC) as the primary outcome, complete (90 days) follow-up, very few patients being lost to follow-up, and high protocol compliance.

One study weakness is the fact that during follow-up, patients and surgeons or attending physicians were not blinded when assessing satisfaction and infectious signs. This could potentially lead to optimistic satisfaction scoring by the patient for the new therapy, resulting in an incorrect significant difference. The lack of blinding by the physicians during follow-up could potentially lead to underreporting the clinical diagnosis of infection in the interventional group. However, scoring of the outcome measure SSI itself was done by a blinded physician, not involved in the surgical procedures or follow-up clinical observations. Furthermore, we aimed to minimize the risk of bias in outcome assessment by the physician and nurse by using very strict criteria for both clinical observation scoring and SSI outcome scoring (CDC). A second weakness is that although the SSI rate in the control group was estimated quite accurate, detecting a 10% absolute reduction starting from 12.5% was rather ambitious. The definition of a minimal clinically relevant reduction was extensively debated in our study group at the time of protocol development. Eventually we opted for a 10% absolute reduction. It is a rather conservative estimate in the sense that every surgeon will support that a 10% absolute reduction is clinically meaningful. The disadvantage of this conservative approach is that the study is likely to overlook smaller, but maybe also clinically relevant, risk reductions. Our study showed a relative risk reduction of approximately 50%. A total of 694 patients would have been needed to demonstrate a significant reduction of this size. A third limitation is that the large randomisation list resulted in a slight skewness of treatment allocation. However, comparison of baseline characteristics showed that treatment groups were comparable and the randomisation was not subverted. Lastly, SSI can result in a delay in adjuvant treatment and consequently an impaired oncological outcome[23, 31]. However, we did not specifically analyse the impact of this delay.

Interpretation and generalisability

Generalisation of our findings should be done with caution, as we acknowledge the fact that there is a lack of clear evidence about the value of dressings in surgical practice, and some surgeons use glue or no dressings[32], as opposed to the simple dressing which we have used as control intervention.

Interpretation of SSI reduction rates should be balanced against the nature of the intervention, the setting and the related morbidity, quality of life and costs. There is no guideline providing any strict recommendation on how this interpretation can be achieved. In recent years it has been increasingly recommended by several authors [33–36] to also judge the clinical relevance of study findings. In our study, there is no disadvantage/harm for patients through the intervention and there is proven benefit in terms of patient satisfaction and costs. Therefore, given the fact that our study is underpowered for the detection of a minimal clinically meaningful difference, i.e. 5%, we consider the rather large effect size of SSI reduction in our study to be relevant.

The discrepancy in treatment effect found by the exploratory subgroup analysis between breast conserving surgery and ablative surgery might be explained by the fact that other factors than the type of wound dressing contribute more to the development of SSI in ablative surgery: compromised vascularisation of skin and subcutaneous tissue by extensive dissection, seroma and hematoma formation, and prolonged drainage time [4, 12, 37]. It seems that with the importance of these factors, the type of wound dressing plays a negligible or modest role in reducing the risk of infection after breast ablative surgery. Research in that patient group should focus on reducing these specific risk factors.

As expected, using Aquacel lowered the need for changing the dressing. Early change of the dressing (within the first 48 hours after surgery) was shown to be an independent risk factor for SSI occurrence in this study. This could be an explanation of the lower SSI rate in the Aquacel group, besides the antibacterial effect of the silver. Recommendation of the CDC to sterile dress primarily closed wounds for at least 24–48 hours could be extended to not change the dressing in the first 48 hours, as early change of the dressing was shown to be an independent risk factor for SSI in our study. Furthermore, lowering the need for early changing of the dressing was associated with an improvement in patient satisfaction of more than 5%, which in literature on quality of life and utility is considered to be a relevant difference [38, 39].

Our findings are highly relevant for healthcare providers, with significant differences in favour of the Aquacel group on two of the recently proposed outcome measures to assess wound management by Elliot[32]: patient satisfaction with the dressing and dressing removal. Based on our exploratory subgroup analysis, the treatment effect of silver-containing dressings on SSI rates might be different between breast conserving surgery and mastectomy. Our results stimulate early drain removal and discourage the early change of dressings. Furthermore, reduced costs and improved patient satisfaction are very relevant in healthcare nowadays. Finally, our study confirms the findings of studies in orthopaedic surgery [20, 40], that found that Aquacel improved patient satisfaction and reduced dressing changes, but could not confirm a significant reduction of SSI.

In summary, clinicians should be aware of the difference in risk factors for SSI between breast conserving and ablative surgery and the role of Aquacel dressings in improving patient satisfaction, reducing dressing changes and possibly reducing SSI after breast cancer surgery.