Meta-analysis

Systematic review and meta-analysis of wound dressings in the prevention of surgical-site infections in surgical wounds healing by primary intention
C. J. Walter1 , J. C. Dumville2 , C. A. Sharp3 and T. Page4
Department of Colorectal Surgery, Division of Gastrointestinal Surgery, Queens Medical Centre, Nottingham University Hospitals NHS Trust, Nottingham, 2 Department of Health Sciences, University of York, York, UK, 3 The Wound Centre, Sydney, New South Wales, and 4 School of Nursing, University of Adelaide and Royal Adelaide Hospital, Adelaide, South Australia, Australia Correspondence to: Miss C. J. Walter, Department of Colorectal Surgery, Division of Gastrointestinal Surgery, E Floor West Block, Queens Medical Centre, Nottingham University Hospitals NHS Trust, Derby Road, Nottingham NG7 2UH, UK (e-mail: catherine.walter@nuh.net)
1

Background: Postoperative surgical-site infections are a major source of morbidity and cost. This study

aimed to identify and present all randomized controlled trial evidence evaluating the effects of dressings on surgical-site infection rates in surgical wounds healing by primary intention; the secondary outcomes included comparisons of pain, scar and acceptability between dressings. Methods: Randomized controlled trials comparing alternative wound dressings, or wound dressings with leaving wounds exposed for postoperative management of surgical wounds were included in the review regardless of their language. Databases searched included the Cochrane Wounds Group Specialised Register and Central Register of Controlled Trials, Ovid MEDLINE, Ovid Embase and EBSCO CINAHL from inception to May 2011. Two authors performed study selection, risk of bias assessment and data extraction, including an assessment of surgical contamination according to the surgical procedure. Where levels of clinical and statistical heterogeneity permitted, data were pooled for meta-analysis. Results: Sixteen controlled trials with 2594 participants examining a range of wound contamination levels were included. They were all unclear or at high risk of bias. There was no evidence that any dressing signicantly reduced surgical-site infection rates compared with any other dressing or leaving the wound exposed. Furthermore, no signicant differences in pain, scarring or acceptability were seen between the dressings. Conclusion: No difference in surgical-site infection rates was demonstrated between surgical wounds covered with different dressings and those left uncovered. No difference was seen in pain, scar or acceptability between dressings.
Presented to the Tripartite Colorectal Meeting, Cairns, Queensland, Australia, July 2011, the International Surgical Congress of the Association of Surgeons of Great Britain and Ireland, Bournemouth, UK, May 2011, and the Annual Meeting of the Association of Coloproctology of Great Britain and Ireland, Birmingham, UK, June 2011; published in abstract form as Br J Surg 2011; 98: 19S, Colorectal Dis 2011; 13: 47 and Colorectal Dis 2011; 13: 83 Paper accepted 12 April 2012 Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.8812

Introduction

An estimated 234 million surgical operations are performed worldwide every year, with the majority resulting in a wound that heals by primary intention1 . Signicant morbidity can result if these wounds become infected. Not only does surgical-site infection (SSI) impact on a
2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

patients recovery, it can also lead to increased hospital stay2 . With total rates of SSI in the developed world estimated at around 5 per cent, SSI is a common and expensive healthcare problem3,4 . Although various patient factors, such as diabetes and steroid use, increase the likelihood of SSI, the type of surgical procedure and level of wound contamination also have a major inuence4,5
British Journal of Surgery

C. J. Walter, J. C. Dumville, C. A. Sharp and T. Page

Table 1

Classication of wound contamination5

Description Non-infective operative wounds in which no inammation is encountered, and neither the respiratory, alimentary or genitourinary tract nor the oropharyngeal cavity is entered Operative wounds in which respiratory, alimentary, genital or urinary tract is entered under controlled conditions and without unusual contamination. Specically, operations involving the biliary tract, appendix, vagina and oropharynx are included in this category, provided no evidence of infection or a major break in sterile technique is encountered Fresh, accidental wounds, operations with major breaks in sterile technique or gross spillage from the gastrointestinal tract, and incisions in which acute, non-purulent inammation is encountered Old traumatic wounds with retained devitalized tissue and those that involve existing clinical infection or perforated viscera. This denition suggests that organisms causing postoperative infection were present in the operative eld before the operation

Level of contamination Clean Clean-contaminated

Contaminated Dirty

Table 2

Summary of dressing types according to their British National Formulary listing7

Description

Dressing type Basic wound contact dressings Low-adherence dressings and wound contact materials Absorbent dressings Advanced wound dressings Alginate dressings

Usually cotton pads which are placed directly in contact with the wound. Can be either non-medicated (e.g. parafn gauze dressing) or medicated (e.g. containing povidoneiodine or chlorhexidine) Applied directly to the wound and may be used as secondary absorbent layers in the management of heavily exuding wounds Highly absorbent; come in the form of calcium alginate or calcium sodium alginate and can be combined with collagen. The alginate forms a gel when in contact with the wound surface that can be lifted off with dressing removal or rinsed away with sterile saline Consist of an insoluble polymer and up to 96% water. Can absorb wound exudate or rehydrate a wound, depending on the wound moisture levels. Supplied as at sheets, as an amorphous hydrogel or as beads Permeable to water vapour and oxygen but not to water or microorganisms Composed of a soft silicone polymer held in a non-adherent layer. Moderately absorbent Occlusive dressing, usually composed of a hydrocolloid matrix on a vapour-permeable lm or foam backing Fibrous hydrocolloids have been developed that resemble alginates and are not occlusive Contain hydrophilic polyurethane foam; designed to absorb wound exudate and maintain moist wound surface Consist of an absorbent core of hydrophilic bres held between two low-adherent contact layers Contain charcoal and used to absorb wound odour. Can be used in conjunction with a secondary dressing to improve absorbency Release free iodine when exposed to wound exudate; iodine is thought to act as a wound antiseptic Used to treat infected wounds as silver ions are thought to have antimicrobial properties. Silver versions of most dressing types are available (e.g. silver foam, silver hydrocolloid) Composed of gauze or low-adherent dressing impregnated with ointment thought to have antimicrobial properties Proposed to alter the activity of proteolytic enzymes in chronic wounds

Hydrogel dressings Films permeable lm and membrane dressings Soft polymer dressings Hydrocolloid dressings Fibrous hydrocolloid dressing Foam dressings Capillary-action dressings Odour-absorbent dressings Antimicrobial dressings Iodine-impregnated dressings Silver-impregnated dressings Other antimicrobial dressings Specialist dressings Protease-modulating matrix dressings

(Table 1). Surgical procedures classied as clean have lower infection rates than those classied as dirty, where infection rates are as high as 30 per cent6 . The common practice of dressing a surgical wound at the end of an operation assumes that by providing a barrier from environmental contamination the risk of SSI is reduced. In addition to preventing infection, dressing use may be justied by additional potential benets such as managing wound exudate, offering physical protection, facilitating wound observation and meeting patients desires for wound coverage. A large number of
2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

wound dressing products are currently available (Table 2; categorized according to their British National Formulary (BNF) listing7 ). To date, there is limited evidence to support their role in SSI prevention (Fig. 1). In 2008, a National Institute for Health and Clinical Excellence review of data from ve randomized controlled trials (RCTs) concluded that current studies did not show convincing differences in SSI rates between different dressings used to treat surgical wounds8 . Although the review methods were robust, the search date was September 2007; thus, more recent studies have not been assessed.
www.bjs.co.uk British Journal of Surgery

Wound dressings and prevention of surgical-site infection

Fig. 1

Rhomboid ap for pilonidal disease: what is the best dressing?

This systematic review aimed to evaluate the effect of dressings on the prevention of SSI in surgical wounds healing by primary intention. Secondary outcomes of dressing costs, effects on scarring, and pain and acceptability were also examined.
Methods

Study selection criteria and literature searches

A prespecied protocol dening the inclusion and exclusion criteria of studies was agreed by the authors. It included all published or unpublished RCTs, in any language, evaluating the effect of one dressing compared with another dressing or with no dressing at all (an exposed wound) on surgical wounds healing by primary intention, where the dressing was applied immediately after surgery. This included studies of children (aged over 2 years) and adults. Wounds of all contamination levels were eligible for inclusion. Procedures involving graft sites or where the patient had an infected wound before surgery were excluded. The wound dressings examined were categorized according to the BNF classication as either: basic wound contact dressings (BWCDs), which included absorbent dressings, low-adherence dressings, impregnated/nonimpregnated gauze and adhesive tape; advanced wound dressings such as hydrogels, hydrocolloids and lms; and other dressings, which included all dressings that did not t these descriptions such as antimicrobial and specialist dressings (Table 2)7 . The primary outcome was occurrence of postoperative SSI, as dened by Centers for Disease Control and Prevention criteria or the authors denition of SSI. There was no differentiation between supercial and deep
2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

infection. Secondary outcomes of interest were: cost (any measure of cost of treatment, or other aspects of resource use such as equipment); pain, reported using a validated scale or as reported by the author; scarring, as reported by the author; acceptability (patient and clinician), as reported by the author; and ease of removal (patient and clinician), as reported by the author9 . The search string for Cochrane Central Register of Controlled Trials (CENTRAL) was adapted for use in other databases, all being searched from inception to May 2011: Cochrane Wounds Group Specialised Register, CENTRAL, Ovid MEDLINE, Ovid Embase and EBSCO CINAHL (Appendix S1, supporting information). The Ovid MEDLINE search was combined with the Cochrane Highly Sensitive Search Strategy for identifying randomized trials in MEDLINE: sensitivity- and precisionmaximizing version (2008 revision)10 . The Embase and CINAHL searches were combined with the trial lters developed by the Scottish Intercollegiate Guidelines Network11 . Bibliographies of all retrieved and relevant publications identied were searched for further studies, and attempts made to contact researchers where additional unpublished data were needed. No additional hand searches to those conducted by the Cochrane Wounds Group (which inform the CENTRAL database) were performed, nor were manufacturers contacted regarding studies for inclusion.

Data extraction
Two authors independently screened the titles and abstracts of all identied studies against the inclusion criteria. Articles with the potential to full the criteria were obtained in full and checked for eligibility. Disagreements with regard to nal inclusion were resolved by discussion and a third, independent, opinion. The following data were extracted: country of origin of trial; type of surgery; wound contamination (Table 1); eligibility criteria and baseline participant data; trial methodology reported; details of the dressing/treatment regimen received by each group; outcome data for primary and secondary outcomes; duration of follow-up; and number of withdrawals.

Risk of bias assessment

If any of the trials reported methodology demonstrating inadequate or unclear quality (as determined by the Cochrane Collaboration tool for assessing risk of bias) for the three key areas of trial design (generating
www.bjs.co.uk British Journal of Surgery

C. J. Walter, J. C. Dumville, C. A. Sharp and T. Page

the randomization sequence, allocation concealment and blinded outcome assessment), the trial was classied as having a high risk of bias12 .

For sensitivity analysis, data were presented by the contamination level of the surgery. A post hoc sensitivity analysis for SSI was also done, pooling data across surgical contamination types.
Results

Statistical analysis
Data were stratied according to dressing type and wound contamination level. Estimates for dichotomous outcomes were reported as risk ratios (RR). Continuous data were reported as mean differences (MD), and overall effect sizes were calculated with 95 per cent condence intervals (c.i.). 2 test with P < 0100 and an I 2 value above 50 per cent were taken to indicate statistical heterogeneity. Where levels of clinical and statistical heterogeneity permitted, data were pooled using meta-analysis (conducted using Review Manager 51, available from http://ims.cochrane.org/revman/download)13 . A randomeffects model was applied if I 2 values exceeded 50 per cent in the absence of clinical heterogeneity. For the primary outcome (SSI) it was assumed that where randomized participants were not included in an analysis, they did not have an SSI (that is, they were considered in the denominator but not the numerator). Given the relatively small number of SSI events anticipated, this seemed the most appropriate assumption. Where a trial did not specify participant group numbers before dropout, only complete case data were presented. Data for all secondary outcomes were presented as complete case analysis.
Potentially relevant RCTs identified and screened for retrieval n = 568 (de-duplicated) RCTs excluded n = 503 (abstract showed that it did not meet inclusion criteria) Articles retrieved for more detailed evaluation n = 65 Articles excluded n = 48 (did not meet inclusion criteria) Potentially appropriate RCTs to be included in meta-analysis n = 17 RCTs excluded from meta-analysis n = 1 Awaiting translation n = 1 RCTs included in meta-analysis n = 16 RCTs withdrawn, by outcome n = 0 RCTs with usable information, by outcome n = 16
Fig. 2

A total of 16 RCTs involving 2594 participants met the inclusion criteria (Fig. 2); their study characteristics are shown in Table 314 29 . Four of the 16 trials were three-armed trials, of which one was treated as a twoarmed trial by combining the results from the two arms that randomized two different brands of lm dressing14 . Only ve of the included trials had been published in the past 10 years14 18 . The remaining 11 trials were between 12 and 31 years old19 29 . Nine of the trials examined clean wounds, ve examined clean and/or clean-contaminated wounds, and two examined clean and/or clean-contaminated and/or possibly contaminated wounds.

Risk of bias assessment

Three trials were deemed to be at high risk of bias18 20 . The quality of the remaining trials was unclear as their reports lacked adequate detail to be condent that they were at low risk of bias14 17,21 29 .

PRISMA ow diagram for the trial. RCT, randomized clinical trial

2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery

Wound dressings and prevention of surgical-site infection

Table 3

Summary of all included trials

Duration of follow-up (days) Wound contamination level Interventions and no. randomized* Primary outcome (SSI)

Reference

Country

Secondary outcomes

BWCD versus exposed wounds (2 trials, 319 participants) Belgium 20 Clean and cleanPhan contaminated et al.22 (1993) Law and Ellis21 (1987) UK NS Clean

Group A: BWCD, 102 (86) Group B: Exposed wound and Vaseline ointment, 105 (93) Group A: BWCD, 59 Group B: Exposed wound, 53

A: 21 of 86 B: 29 of 93

NS

A: 3 of 59 B: 1 of 53

Cost: A, GBP 660; B, GBP 080 Scar: no difference between groups Cost: A, GBP 4200; B, GBP 080 Scar: no difference between groups NS

Film dressing versus exposed wounds (1 trial, 107 participants) Law and UK NS Clean Ellis21 (1987) BWCD versus lm dressing (8 trials, 1283 participants) UK NS Clean Cosker et al.14 (2005) De Win et al.23 (1998) Gardezi et al.24 (1983) Hewlett25 (1996) Belgium 710 Clean

Group A: Film dressing, 54 Group B: Exposed wound, 53

A: 5 of 54 B: 1 of 53

Group A: BWCD, 100 Group B: 2 merged lm dressing groups, 200 Group A: BWCD, 6 Group B: Film dressing, 8

A: 5 of 100 B: 9 of 200

A: 0 of 6 B: 0 of 8

Cost: A, BEL 115; B, BEL 143

Pakistan

NS

Clean and cleancontaminated

Group A: BWCD for 2 days, 50 Group B: Film dressing for 7 days, 50 Group A: BWCD, 39 Group B: Film dressing, 37

A: 6 of 50 B: 3 of 50

Pain: A, 1 of 50; B, 3 of 50

UK

NS

Clean and cleancontaminated

NS

Law and Ellis21 (1987)

UK

NS

Clean

Group A: BWCD, 59 Group B: Film dressing, 54

A: 3 of 59 B: 5 of 54

Cost dressing alone: A, GBP 160; B, GBP 146 Cost dressing and procedure packs: A, GBP 436; B, GBP 284 Cost: A, GBP 660; B, GBP 4200 Scar: no difference between groups Pain patient assessment: A, 51(28); B, 16(15) Acceptability patient assessment: A, 42(25); B, 13(12) Acceptability nurse assessment: A, 54; B, 12 Cost: A, DEM 1040; B, DEM 360

Moshakis et al.19 (1984)

UK

68

Clean

Group A: BWCD, 59 Group B: Film dressing, 61

NS

Rohde et al.26 (1981) Wynne et al.18 (2004)

Germany

10

Clean and cleancontaminated

Group A: BWCD, 46 Group B: Film dressing, 44

A: 24 of 46 B: 14 of 44

Australia

28

Clean

Group A: BWCD, 243 Group B: Film dressing, 227

A: 6 of 243 B: 9 of 227

Cost per person: A, AUD 052; B, AUD 159 Discomfort on dressing removal (as reported by patient): A, 142 of 243; B, 166 of 227 Scar width (mm): A, 23 (15); B, 18 (13)

BWCD versus hydrocolloid dressing (6 trials, 1084 participants) Holm Denmark 90 Clean and cleanet al.28 contaminated, (1998) and possibly contaminated

Group A: 2 days of BWCD, 37 Group B: Hydrocolloid dressing, 36

A: 5 of 37 B: 1 of 36

2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery

C. J. Walter, J. C. Dumville, C. A. Sharp and T. Page

Table 3

(Continued)
Duration of follow-up (days) 183 Wound contamination level Clean Interventions and no. randomized* Group A: BWCD, 28 Group B: Hydrocolloid dressing, 28 Primary outcome (SSI) A: 0 of 28 B: 0 of 28

Reference Michie and Hugill27 (1994)

Country USA

Secondary outcomes Scar at 4 weeks very or considerably supple: A, 23 of 26; B, 24 of 26. Scar at 4 weeks raised: A, 14 of 26; B, 21 of 26 Pain over last 48 h at rst and second postoperative visits on VAS 010: very low (< 1) in both groups Pain patients VAS 0100 (mm): A, 40; B, 32

Persson et al.29 (1995)

Sweden

Until hospital discharge

Clean and cleancontaminated

Shinohara et al.16 (2008)

Japan

90

Clean and cleancontaminated, and possibly contaminated Clean

Group A: BWCD for 24 h followed by exposed, 30 Group B: Hydrocolloid dressing, 31 (7 postrandomized dropouts) Group A: BWCD, 71 Group B: Hydrocolloid dressing, 63

A: 2 of 30 B: 2 of 31

A: 4 of 71 B: 3 of 63

Cost per dressing: A, JPY 780; B, JPY 715 Scar width (mm): A, 23; B, 22 Cost for days 15: A, USD 073; B, USD 360; C, USD 336 Pain no pain at removal on day 5: A, 76%; B, 61%; C, 14% Ease difcult to remove (assessed by clinician): A, 5 of 84; B, 13 of 61; C, 45 of 60 Cost per person: A, AUD 052; B, AUD 393 Discomfort on dressing removal (as reported by patient): A, 142 of 243; B, 202 of 267

Wikblad and Anderson20 (1995)

Sweden

28

Group A: BWCD, 92 Group B: Hydrocolloid dressing, 77 Group C: Matrix hydrocolloid dressing, 81

NS

Wynne et al.18 (2004)

Australia

28

Clean

Group A: BWCD, 243 Group B: Hydrocolloid dressing, 267

A: 6 of 243 B: 6 of 267

Comparison of two different BWCDs (1 trial, 50 participants) Australia 5 Clean Lawrentschuk et al.15 (2002)

Group A: Absorbent BWCD, 25 Group B: Parafn tulle BWCD, 25

A: 3 of 25 B: 0 of 25

BWCD versus brous hydrocolloid (hydrobre) dressing (1 trial, 160 participants) Clean Group A: BWCD, 80 Vogt et al.17 Denmark 42 Group B: Hydrobre dressing, (2007) 80 (24 participants, 14 from group A and 10 from group B postrandomization exclusions)

A: 7 of 66 B: 9 of 70

Cost per patient: A, 10118; B, 203487 Pain no. of patients with composite score of good for discomfort on mobilization, dressing change and skin problems: A, 52 of 66; B, 59 of 70

BWCD versus polyurethane matrix hydrocolloid dressings (1 trial, 173 participants) see Wikblad and Anderson20 ) Comparisons of advanced dressings (2 trials, 652 participants; see Wikblad and Anderson20 and Wynne et al.18 )

*Values in parentheses denote patient attrition. Three-armed trial; includes the three-armed trials of Law and Ellis21 and Wynne et al.18 ; values are mean(s.d.); values are mean (range). SSI, surgical-site infection; BWCD, basic wound contact dressing; NS, not stated; GBP, Great British Pound; BEL, Belgium Franc (now replaced by the Euro); DEM, Deutsche Mark (now replaced by the Euro); AUD, Australian Dollar; VAS, visual analogue scale; JPY, Japanese Yen; USD, United States Dollar).

2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery

Wound dressings and prevention of surgical-site infection

Effects of interventions: prevention of surgical-site infection (primary outcome)

Table 3 shows the SSI rates and secondary outcome results of each trial grouped according to the different dressings compared; in total, eight comparisons between dressings (or wound exposure) were made. There were no signicant differences in SSI rates between dressings or exposed wounds in any of the included trials. Data were available for meta-analysis for two comparisons.

Reference Cosker et al.14 Law and Ellis21 Wynne et al.18 Overall

Risk ratio 090 (031, 261) 182 (046, 726) 161 (058, 444) 134 (070, 255)

P
0846 0396 0361 0373

Risk ratio

I 2 = 0%

001 01

10

100

Favours film Favours basic wound dressing

Fig. 3

Basic wound contact dressings versus lm dressings Data from four of the eight trials comparing BWCDs with lm dressings involved clean surgical wounds and were pooled14,18,21,23 . Although two of these studies were threearmed trials, this was the only meta-analysis conducted with their data, so the complete groups relevant to this pooling were used18,21 . Data from the remaining fth clean study could not be pooled because the article did not report SSI data19 . Pooled data showed there was no signicant difference in the number of SSIs between basic wound contact-dressed groups and lm-dressed groups (RR 134, 95 per cent c.i. 070 to 255) (Fig. 3)14,18,21 . The remaining trials in the BWCD versus lm dressing comparison either failed to report their surgical type and SSI data, or reported unclear surgical wound contamination types24 26 . As it is not clear what impact the level of surgical contamination might have on the potential action of dressings in SSI prevention, a sensitivity analysis was performed with pooled data from all six trials that reported SSI data in this comparison, regardless of contamination level (Fig. 4)14,18,21,24,26 . This showed there was no signicant difference in the number of SSIs between BWCD and lm-dressed groups (RR 084, 058 to 124). Basic wound contact dressings versus hydrocolloid dressings Of the six trials investigating the effect of a BWCD compared with a hydrocolloid dressing16,18,20,27 29 , only four reported SSI data across a mixture of wound contamination groups, so data were not pooled initially. In sensitivity analysis, pooling data from all trials reporting SSI rates, no signicant difference in the number of SSIs between BWCD and hydrocolloid-dressed groups was seen (RR 070, 034 to 144) (Fig. 5)16,18,28,29 . Effects of interventions: secondary outcomes Costs Dressing cost was reported in a number of trials. Although exposed wounds were cheaper than dressed wounds, no single type of wound dressing (BWCD, hydrocolloid or lm) was consistently cheaper than the others, with costs
2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

Meta-analysis of surgical-site infection (SSI) rates for basic wound contact dressings compared with lm dressings in clean surgery using a MantelHaenszel model. Odds ratios are shown with 95 per cent condence intervals. De Win et al.23 was not included as there were no SSIs in either arm

Reference
Cosker et al.14 Gardezi et al.24 Law and Ellis21 Rohde et al.26 Wynne et al.18 Overall

Risk ratio
090 (031, 261) 050 (013, 189) 182 (046, 726) 061 (036, 102) 161 (058, 444) 084 (058, 124)

P
0846 0307 0396 0059 0361 0389

Risk ratio

I 2 = 18%

001 01
Favours film

10

100

Favours basic wound dressing

Fig. 4

Meta-analysis of surgical-site infection (SSI) rates for basic wound contact dressings compared with lm dressings in all trials (mixed contamination levels) using a MantelHaenszel model. Odds ratios are shown with 95 per cent condence intervals. De Win et al.23 was not included as there were no SSIs in either arm

Reference
Holm et al.28 Persson et al.29 Shinohara et al.16 Wynne et al.18 Overall

Risk ratio
021 (003, 167) 097 (015, 644) 085 (020, 363) 091 (030, 278) 070 (034, 144)

P
0139 0973 0821 0869 0333

Risk ratio

I 2 = 0%

001 01
Favours hydrocolloid

10

100

Favours basic contact dressing

Fig. 5

Meta-analysis of surgical-site infection (SSI) rates for basic wound contact dressings compared with hydrocolloid dressings in all trials (mixed contamination levels) using a MantelHaenszel model. Odds ratios are shown with 95 per cent condence intervals. Michie and Hugill27 was not included as there were no SSIs in either arm

appearing to vary based on the individual dressing products used (Table 3)16 18,20,21,23,25,26 .

Pain Few studies provided meaningful data on pain. When comparing a BWCD with a lm dressing, Moshakis and colleagues19 measured pain using a patient-assessed linear
www.bjs.co.uk British Journal of Surgery

C. J. Walter, J. C. Dumville, C. A. Sharp and T. Page

scale (110), where 1 corresponded to no discomfort and 10 to extremely uncomfortable or painful; mean(s.d.) pain levels in the basic wound contact group were 51(28) compared with 16(15) in the lm-dressed group, a statistically signicant nding in favour of lm dressings (MD 350, 95 per cent c.i. 429 to 271). However, all analyses in this trial must be interpreted with caution, as it was deemed to be at high risk of bias. The study of Wynne and co-workers18 reported BWCDs to be less painful on removal than a hydrocolloid dressing, but the difference was not statistically signicant (080, 063 to 101). Wikblad and Anderson20 compared pain on dressing removal for BWCDs and a hydrocolloid matrix dressing, and reported a statistically signicant difference in the proportion of participants with no pain on removal in favour of the BWCD (017, 009 to 034); however, the large number of missing data should be noted.

Scarring, acceptability and ease of removal Results from a variety of ad hoc scarring, acceptability and ease of removal measures were reported from the trials. No clear advantages for any of the dressing categories (or wound exposure) over the others were seen with respect to these outcomes (Table 3).
Discussion

This systematic review found no evidence that dressing surgical wounds reduces rates of SSI in comparison with leaving the wounds uncovered. Furthermore, no evidence was demonstrated to support the use of one type of dressing over another in SSI prevention. Many trials that investigated non-clean surgery evaluated a range of contamination levels, so it was not possible to draw conclusions for each scenario. A number of trials also had low numbers of participants and low event rates. Their results require cautious interpretation as this low statistical power leaves the equivalent ndings at risk of a type II error, where a real difference in a dressings effectiveness may exist but has not been demonstrated. In terms of secondary outcomes, there is only weak evidence to suggest that one dressing type has advantages over another. Moshakis et al.19 suggested that lm dressings might be less painful for patients than BWCDs. This study, however, was deemed to be at high risk of bias owing to inadequate allocation concealment and the absence of evidence that statistical procedures had been employed to deal with the inclusion of some participants as their own controls (bilateral excisions). Wikblad and Anderson20 reported that BWCDs were signicantly less painful on removal than hydrocolloid dressings, but as
2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

a large number of data were missing from this analysis it, too, was deemed to be at high risk of bias. A number of trials suggested that advanced dressings were more expensive than BWCDs; however, these limited cost evaluations did not capture all relevant data on resource use, or consider the costs versus the benets of treatments18,20,21,23 . Consequently, the cost-effectiveness data included in these studies are not suitable to guide a surgeons dressing choice. As the justication to dress surgical wounds is not supported by reductions in SSI rates, decisions to use surgical wound dressings might be better based on the other properties and qualities that dressings can offer. Yet, this review highlights that few trials have sought to examine these properties or produce high-quality cost and benet evaluations. Thus, based on the current evidence, wound dressing choice should be made simply with regard to unit dressing costs and the ability of different dressings to manage specic symptoms such as absorption of exudate. While this uncertainty regarding best dressings for surgical wounds remains, any investment in future research must maximize its value to clinicians. Given the large number of both dressing options and surgical procedures, future trials should focus on evaluating dressings or approaches that surgical teams use most often. It would be important to ask surgeons, nurses and patients what they feel are the most pressing questions, such as which type of dressing, or for what duration, a dressing should remain in place, as well as which outcomes are most important. Understanding these questions would allow research resources to be focused on addressing the most pressing issues.
Acknowledgements

This paper was generated from work that contributed towards the Cochrane Review published as Cochrane Database Syst Rev 2011; (7)CD003091. The authors would like to acknowledge the contribution of all peer referees who commented on that Cochrane Review. They also thank members of the editorial base of the Cochrane Wounds Group for their support. Disclosure: The authors declare no conict of interest.
References
1 Weiser TG, Regenbogen SE, Thompson KD, Haynes AB, Lipsitz SR, Berry WR et al. An estimation of the global volume of surgery: a modelling strategy based on available data. Lancet 2008; 372: 139144. 2 Health Protection Agency. Surveillance of Surgical Site Infection in English Hospitals 19972002. A National

www.bjs.co.uk

British Journal of Surgery

Wound dressings and prevention of surgical-site infection

10

11

12

13

14

15

Surveillance and Quality Improvement Programme. Health Protection Agency: London, 2002. Wick EC, Vogel JD, Church JM, Remzi F, Fazio VW. Surgical site infections in a high outlier institution: are colorectal surgeons to blame? Dis Colon Rectum 2009; 52: 374379. Boltz MM, Hollenbeak CS, Julian KG, Ortenzi G, Dillon PW. Hospital costs associated with surgical site infections in general and vascular patients. Surgery 2011; 150: 934942. Berard F, Gandon J. Postoperative wound infections: the inuence of ultraviolet irradiation of the operating room and of various other factors. Ann Surg 1964; 160(Suppl 2): 1192. McLaws ML, Murphy C, Whitby M. Standardising surveillance of nosocomial infections: the HISS program. Hospital Infection Standardised Surveillance. J Qual Clin Pract 2000; 20: 611. BMJ Group and Royal Pharmaceutical Society of Great Britain. British National Formulary 61. BMJ Group and Royal Pharmaceutical Society of Great Britain: London, 2011; 935939. National Institute for Health and Clinical Excellence (NICE). Surgical Site Infection: Prevention and Treatment of Surgical Site Infection. NICE Clinical Guideline 74, 2008. http://www.nice.org.uk/nicemedia/pdf/CG74NICE Guideline.pdf [accessed 2 February 2012]. Horan TC, Andrus M, Dudeck MA. DC/HNSN surveillance denition of healthcare-associated infection and criteria for specic types of infection in the acute care setting. Am J Infect Control 2008; 36: 309332. Lefebvre C, Manheimer E, Glanville J, on behalf of the Cochrane Information Retrieval Methods Group. Searching for studies. In Cochrane Handbook for Systematic Reviews of Interventions. Version 5.0.2 (updated September 2009), Higgins JPT, Green S (eds). The Cochrane Collaboration: Oxford, 2009. Scottish Intercollegiate Guidelines Network (SIGN). Search Filters. http://www.sign.ac.uk/methodology/lters.html#random [accessed 22 September 2010]. Higgins JPT, Green S (eds). Cochrane Handbook for Systematic Reviews of Interventions. Version 5.0.2 (updated September 2009). The Cochrane Collaboration: Oxford, 2009. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003; 327: 557560. Cosker T, Elsayed S, Gupta S, Medonca AD, Tayton KJ. Choice of dressing has a major impact on blistering and healing outcomes in orthopaedic patients. J Wound Care 2005; 14: 2729. Lawrentschuk N, Falkenburg MP, Pirpiris M. Wound blisters post hip surgery: a prospective trial comparing dressings. ANZ J Surg 2002; 72: 716719.

16 Shinohara T, Yamashita Y, Satoh K, Mikami K, Yamauchi Y, Hoshino S et al. Prospective evaluation of occlusive hydrocolloid dressing regarding the healing effect after abdominal operations: randomized controlled trial. Asian J Surg 2008; 31: 15. 17 Vogt KC, Uhlyarik M, Schroeder TV. Moist wound healing compared with standard care of treatment of primary closed vascular: a prospective randomized controlled trial. Wound Repair Regen 2007; 15: 624627. 18 Wynne R, Botti M, Stedman H, Holsworth L, Harinos M, Flavell O et al. Effect of three wound dressings on infection, healing comfort and cost in patients with sternotomy wounds: a randomized trial. Chest 2004; 125: 4349. 19 Moshakis V, Fordyce MJ, Grifths JD, McKinna JA. Tegaderm versus gauze dressing in breast surgery. Br J Clin Pract 1984; 38: 149152. 20 Wikblad K, Anderson B. A comparison of three wound dressings in patients undergoing heart surgery. Nurs Res 1995; 44: 312316. 21 Law NW, Ellis H. Exposure of the wound a safe economy in the NHS. Postgrad Med J 1987; 63: 2728. 22 Phan M, Van der Auwera P, Andry G, Aoun M, Chantrain G, Deraemaecker R et al. Wound dressings in major head and neck cancer surgery: a prospective randomized study of gauze vs sterile vaseline ointment. Eur J Surg Oncol 1993; 19: 1016. 23 De Win M, De Blaere X, Scheers R. The effect of choice of surgical wound dressings on the direct cost of healing. Proceedings of the 8th European Conference on Advances in Wound Management, Madrid, 1998; 5457. 24 Gardezi SAR, Chaudhary AM, Sial AK, Ahmad I, Rashid M. Role of polyurethane membrane in post-operative wound management. J Pak Med Assoc 1983; 33: 219222. 25 Hewlett L. The evaluation of two post-operative dressings in the management of surgical wounds. In Proceedings of the 5th European Conference on Advances in Wound Management, Harrogate, 2124 November 1995. Macmillan Magazines: London, 1996. 26 Rohde H, Thon K, Stoltzing H, Schirren J. [Transparent adhesive tape as postoperative dressing. Randomized, clinical study for comparison with a conventional dressing technic.] Chirurg 1981; 52: 4650. 27 Michie DD, Hugill JV. Inuence of occlusive and impregnated gauze dressings on incisional healing: a prospective, randomized, controlled study. Ann Plast Surg 1994; 32: 5764. 28 Holm C, Petersen JS, Grnboek F, Gottrup F. Effects of occlusive and conventional gauze dressings on incisional healing after abdominal wounds. Eur J Surg 1998; 164: 179183. 29 Persson M, Svenberg T, Poppen B. To dress or not to dress surgical wounds? Patients attitudes to wound care after major abdominal operations. Eur J Surg 1995; 161: 791793.

2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery

C. J. Walter, J. C. Dumville, C. A. Sharp and T. Page

Supporting information

Additional supporting information may be found in the online version of this article: Appendix S1 Search strategy employed for the systematic review of wound dressings in the prevention of surgical-site infections in surgical wounds healing by primary intention (Word document) Please note: John Wiley & Sons Ltd is not responsible for the functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

2012 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery