Immediate Air Exposure vs Dressings Following Primary Closure of Clean and Clean-contaminated Surgical Wounds: A Systematic Review and Meta-analysis of Randomized Controlled Trials

Dressings are commonly applied after surgery as an important method of protection against contamination that could cause infection and delayed healing. SSI is a common postoperative complication worldwide, and an urgent need exists for additional investigation into this problem and advocacy for controlling it.¹ SSI has been shown to prolong hospitalization and contribute to increased medical costs, and patients with infection 30 days postoperatively were more likely to develop long-term infection and have a higher mortality rate than those with infection of shorter duration.^2,3

WHO recommendations published in 2018 suggest that surgical dressings should remain undisturbed for at least 48 hours after surgery.⁴ Nevertheless, it remains unknown whether dressings help reduce the risk of SSI, especially after clean and clean-contaminated surgery.⁵

This systematic review and meta-analysis was conducted to compare the performance of immediate air exposure with dressings following primary closure of clean and clean-contaminated surgical wounds.The authors of this study sought to determine whether immediate surgical site air exposure is an appropriate alternative to dressings for select patients.

Search strategy

Using the search strategies described by Dumville et al,⁵ a comprehensive electronic search of PubMed, Web of Science, and Embase from the time of database establishment through October 2021 was performed. The references of publications in the study of wounds were also manually searched for additional studies.

Study selection

Inclusion criteria comprised RCTs, allocation of patients to either an immediate air exposure group or a dressing group immediately after primary surgical wound closure, and reported incidence of SSI. Exclusion criteria comprised duplicate studies, inaccurate statistical analysis, lack of explanation and strict adherence to the stated SSI diagnostic criteria, or use of immediate air exposure for patients who were treated nonsurgically. Articles in any language were included in the search.

Quality assessment

Study quality was assessed using the Cochrane risk-of-bias tool. The quality of each study was evaluated based on the following items: selection bias (random sequence generation and allocation concealment), performance bias, detection bias, attrition bias, reporting bias, and other bias. The authors of the current study used their own judgment to classify studies as low risk, high risk, or unclear risk. Differences were resolved through discussion.

Data extraction

A standard extraction form was developed, and the following information was extracted from each study: publication year and authors, country in which the main authors resided, surgery types, sample size, interventions and controls, follow-up periods, antibiotic prophylaxis, and outcomes. The primary outcome was SSI. Secondary outcomes included air exposure or dressing preference, cost, scarring, and the incidence of bruising and hematoma. Quantitative analysis was conducted per protocol, and the patients who withdrew from the studies were not included.

Statistical analysis

A meta-analysis to investigate the performance of immediate air exposure in primarily closed surgical wounds was conducted using Review Manager (RevMan) software (version 5.3; The Cochrane Collaboration). For dichotomous results, RR with 95% CI was estimated. The chi-square test and I² statistic were applied to evaluate heterogeneity. An I² statistic greater than 50% or a P value less than .05 on the chi-square test indicated significant heterogeneity. For studies with significant heterogeneity, a random effects model was used. Otherwise, a fixed effects model (Mantel-Haenszel method) was applied. Subgroup analysis was performed to explore potential influencing factors or to trace the potential source of significant heterogeneity. When at least 10 studies were involved in the quantitative analysis, the authors planned to investigate publication bias by Egger’s tests. A two-tailed P value less than .05 was considered statistically significant.

Search process

A total of 1428 studies were identified in the comprehensive electronic search, and 2 additional studies were gleaned from the references of pertinent publications. After removal of duplicates and screening based on article titles and abstracts, 12 studies remained for full-text assessment; 6 RCTs with a total of 1243 cases were included in the final analysis^6-11 (Figure 1). The baseline characteristics of the 6 RCTs are shown in the Table.

Characteristics of included studies

There were 4 RCTs conducted in Europe, and 2 RCTs conducted in Asia. Patients ranged in age from 6 to 84 years. Four RCTs were conducted on patients admitted for abdominal, groin, or elective general surgery, and 2 RCTs were conducted on patients admitted for head and neck surgery. Only the study by Blazeby⁶ involved contaminated or dirty cases (15 cases); all other cases were clean or clean-contaminated. A total of 530 cases received no dressings after surgery; 93 of these cases received sterile petroleum jelly ointment. The remaining cases received dressings such as gauze dressings (including traditional mastoid pressure dressings), film dressings (polyurethane film dressings, standard low-adhesive film with absorbent pad dressings), and tissue glue-as-a-dressing. Grover et al⁸ used occlusive dressings and Blazeby⁶ used simple dressings, but neither study specified the dressing materials used. Antibiotic prophylaxis was administered in most included cases. Duration of follow-up ranged from 5 days to 8 weeks.

Quality of included studies

Details on study quality are shown in Figure 2. Tan et al⁹ mentioned the infeasibility of blinding participants and personnel. The studies by Blazeby,⁶ Law and Ellis,⁷ and Phan et al¹¹ were classified as high risk for attrition bias owing to partial loss of their outcomes data and lack of analysis of cases who withdrew from the studies. Tan et al⁹ reported that per protocol sensitivity analysis did not affect the direction of their intention-to-treat findings, and that study was graded as low risk for attrition bias. All RCTs were graded as low risk for selective reporting bias except for the study by Law and Ellis,⁷ in which descriptive outcomes were reported.

Surgical site infection

Because Blazeby⁶ and Law and Ellis⁷ each investigated more than one dressing type, tissue glue-as-a-dressing and polyurethane film cases were pooled in the dressing material subgroup analysis only, and simple dressing and dry gauze cases were included in the total SSI incidence comparison and corresponding subgroup analysis. Low heterogeneity was found (I² = 0; P =.52), and SSI incidence was 11% and 11.1% in the immediate air exposure group and the dressing group, respectively (RR, 0.95; 95% CI, 0.68-1.33 [P =.76]) (Figure 3).

Subgroup analysis was conducted to investigate whether dressings and immediate air exposure performed differently by SSI depth (superficial [involving only the skin and subcutaneous tissue of the incision] versus deep [involving soft tissues such as fascial and muscle layers of the incision]), SSI risk (clean surgery vs. clean-contaminated surgery), follow-up time (≥ 28 days vs. < 28 days), dressing materials (gauze vs. film dressings vs. tissue glue-as-a-dressing), age (adults only vs. adults and children), and anatomic location of surgery (abdomen and groin vs. head and neck). Low heterogeneity was found in all subgroup analyses (I² < 50%; P >.1), and fixed-effect models were applied.

SSI incidence was 22.4% and 19.9% for immediate air exposure and gauze dressings, respectively (P =.65); 1.5% and 4.7% for immediate air exposure and film dressings, respectively (P =.07); and 18.2% and 17.8% for immediate air exposure and tissue glue-as-a-dressing, respectively (P =.94). No significant difference between subgroups was found across gauze, film dressings, and tissue glue-as-a-dressing (I² = 42.9%; P =.17) (eFigure 4).

For follow-up time greater than or equal to 28 days, SSI incidence was 7.4% and 8.4% for immediate air exposure and dressings, respectively (P =.52). For the subgroup follow-up time less than 28 days, SSI incidence was 16.8% and 15.4% for immediate air exposure and dressings, respectively (P =.86). No significant difference between subgroups was found (I² = 0; P =.54) (eFigure 4).

SSI incidence for immediate air exposure and dressings was 4.4% and 6.9%, respectively, after clean surgery (P =.39), and 12.5% and 10.7%, respectively, after clean-contaminated surgery (P =.64). No significant difference between subgroups was found (I² = 0; P =.33) (eFigure 4).

Superficial SSI incidence was 7.3% and 7.5% for immediate air exposure and dressings, respectively (P =.8), and deep SSI incidence was 0% and 3.1% for immediate air exposure and dressings, respectively (P =.17). No significant difference between subgroups was found (I² = 39.1%; P =.20) (eFigure 5).

SSI incidence for immediate air exposure and dressings was 14.2% and 12.8%, respectively, in adults only (P =.74) and 12.1% and 11.9%, respectively, in adults and children (P =.31). No significant difference between subgroups was found (I² = 14.1%; P =.28) (eFigure 5).

SSI incidence for immediate air exposure and dressings was 22.4% and 19.9%, respectively, after head and neck surgery (P =.65) and 6.6% and 7.9%, respectively, after abdomen and groin surgery (P =.38). No significant difference between subgroups was found (I² = 0; P =.34) (eFigure 5).

Secondary outcomes

Dressing versus air exposure preference was reported by Law and Ellis⁷ and Tan et al,⁹ but Law and Ellis⁷ did not report available data for analysis. No quantitative estimates of scarring could be performed owing to descriptive results by Law and Ellis.⁷ Law and Ellis⁷ and Grover et al⁸ reported intervention cost, and no pooled estimates were calculated owing to data formatting that could not be quantitatively analyzed. As for bruising and hematoma, only Rowe-Jones and Leighton¹⁰ reported available data for analysis, and no quantitative analysis was conducted.

Although WHO recommends dressing application for a minimum of 48 hours following primary surgical wound closure,⁴ the current meta-analysis showed that immediate air exposure is comparable to dressings in terms of SSI occurrence (P =.76). The proportion of immediate air exposure cases in which SSI occurred was 11% in this meta-analysis, which is similar to the 9.9% incidence reported by Woelber et al¹² in 2016. Clean and clean-contaminated surgery subgroup analysis in the current study confirmed the similar performance of immediate air exposure and dressings in preventing SSI. However, the low number of cases involving contaminated or dirty surgery (15 of 1243 cases [1.2%]) precluded analysis of SSI incidence in such cases.

The authors of the current meta-analysis divided the included RCTs into 2 subgroups based on a cutoff follow-up time of 28 days. However, studies in the subgroup with the follow-up time of less than 28 days reported a maximum 3-week follow-up.^7,10,11 For some cases in the studies by Rowe-Jones and Leighton¹⁰ and Law and Ellis,⁷ the maximum follow-up was 1 week.According to CDC criteria, SSI is defined as surgical incision infection occurring within 30 days after treatment.⁵ Although follow-up time did not influence comparison of SSI incidence in this meta-analysis, the authors of the current study recommended a minimum follow-up of 30 days, because Woelber et al¹² reported that slightly more than 60% of SSIs occurred after discharge.

Additionally, most included RCTs in this study did not specify duration of dressing application, so the authors of this study could not perform associated quantitative analysis of that parameter. Termination of dressing application before or at the time of discharge was common in the included RCTs, as was termination at the time stitches were removed.^8,9 The authors of the current meta-analysis believed the influence of duration of dressing application to be minor, because there was no difference in SSI incidence between delayed and early removal of dressings for non-contaminated surgical wounds, which was noted by Zhang et al.¹³

Currently, scores of dressings made of different materials and in various forms are available for surgical wound management.¹⁴None of the gauze dressings, film dressings, or tissue glue-as-a-dressing options outperformed wound exposure in reducing SSI occurrence following non-contaminated procedures in the RCTs included in this meta-analysis. However, different studies comparing dressings have shown that the form and material of dressings results in different SSI incidences. In their study, Brölmann et al¹⁵ reported that the infection rate of patients who received gauze was twice as high as that of patients who received hydrocolloid dressings after split-skin grafting. Ezzelarab et al¹⁶ demonstrated that transparent semipermeable dressings were more effective than conventional occlusive gauze in reducing the rate of SSI after clean and clean-contaminated surgery.¹⁶ Chowdhry¹⁷ found that the use of oxidized regenerated cellulose/collagen/silver-oxidized regenerated cellulose dressings did not result in a significantly lower percentage of patients with signs of inflammation of skin graft donor site wounds compared with use of transparent film dressings followed by gauze dressing. Li et al¹⁸ conducted a meta-analysis and concluded that the application of silver-containing dressings was not associated with a lower SSI incidence compared with silver-free dressings in clean and clean-contaminated surgical wounds. According to Khansa et al,¹⁹ silver confers no benefit for clean wounds and closed surgical incisions.

Concerning other postoperative surgical incision-associated events reported in studies included herein, the incidence of bruising/hematoma was 2% for both dressings and immediate air exposure
after middle ear and mastoid surgery.¹⁰ Scarring was reported by Law and Ellis⁷; the quality of the final scars was not different based on dressing type. According to Law and Ellis⁷ and Tan et al,⁹ no difference in surgical site discomfort or pain was observed after clean and clean-contaminated surgery. Application of dressings did have an advantage in wound healing, however, as has been reported elsewhere. Moist dressings have been shown to facilitate healing and reduce focal pain.²⁰ Film dressings can improve the rate of healing and help achieve good cosmetic appearance.²¹ Occlusive dressings have been shown to aid in surgical site healing after biopsy.²²

When no dressings are used, there are no dressing-associated events such as allergic reaction or effects on daily activities. Immediate air exposure of surgical wounds resulted in reduced time spent on dressing changes.⁸ Additionally, immediate air exposure of wounds did not require additional dressings or gauze to manage exudate and did not result in extended hospital stay.^7-9 According to Law and Ellis⁷ and Grover et al,⁸ costs were lower for patients with wounds exposed to air than for those whose wounds were dressed after non-contaminated surgery. However, Woelber et al¹² noted that dressing costs contribute minimally to overall cost-effectiveness. Due to few related economic studies in this field, cost-effective surgical wound management strategies are not yet clear.²³

Immediate air exposure was found to be as well accepted as dressings after non-contaminated surgery. Law and Ellis⁷ reported no difference in patient preference for immediate air exposure versus dressings. Furthermore, Tan et al⁹ reported a significant patient preference for no dressings the first day after caesarean section (P =.002), and no significant difference in preference for dressing types was observed 28 days after cesarean section (P =.46).

Although CDC definitions of SSI were the major basis for diagnosis, the criteria for SSI were not identical throughout the included RCTs. For example, early studies such as those by Law and Ellis⁷ and Rowe-Jones and Leighton¹⁰ mostly relied on discharge bacterial culture for SSI diagnosis or diagnosed SSI based on clinical manifestations, whereas more recent studies such as those by Tan et al⁹ diagnosed SSI strictly according to CDC criteria.The nonidentical SSI diagnostic criteria could lead to variation in estimated SSI cases across studies, which could potentially influence results. Provided the authors of the included studies specified diagnostic criteria and strictly followed them, however, the current group of authors believed the influence of the nonidentical diagnostic criteria to be minor. For clean and clean-contaminated procedures which infrequently developed severe SSI, culture tests were important but might be subject to false-negative results; in such cases, conservative methods mainly based on clinicians’ judgment could play a more important role.²⁴

This study has several limitations. First, only 6 RCTs were included, and the outcomes may be subject to publication bias. Second, risk of attrition bias was deemed to be high in some of the included RCTs.^6,7,11 SSI after non-contaminated surgery was infrequent and rarely severe, and the authors of this study believed that reported withdrawal cases likely were not caused by SSI occurrence.

The findings presented in this meta-analysis show that dressings (gauze dressing, film dressing, and tissue glue-as-a-dressing) do not outperform immediate air exposure following primary closure of clean and clean-contaminated procedures in terms of reducing the risk of SSI.