Randomised Controlled Trials of Alcohol-Based Surgical Site Skin Preparation for the Prevention of Surgical Site Infections: Systematic Review and Meta-Analysis

(1) Background: Surgical site skin preparation is an important approach to prevent postoperative wound infections. International guidelines recommend that alcohol-based combinations be used, however, the optimal combination remains uncertain. This study compares the effectiveness of alcohol-based chlorhexidine and alcohol-based iodophor for surgical site skin preparation for prevention of surgical site infections (SSIs). (2) Methods: Randomised controlled trials comparing alcohol-based interventions for surgical site skin preparation were included. The proportion of SSIs was compared using risk ratios (RR) with 95% confidence intervals (95% CI). The meta-analysis was performed with a fixed effect model using Mantel-Haenszel methods. As an a priori subgroup analysis SSI risk was examined according to different surgical procedural groups. (3) Results: Thirteen studies were included (n = 6023 participants). The use of chlorhexidine-alcohol was associated with a reduction in risk of SSIs compared with iodophor-alcohol (RR 0.790; 95% CI 0.669, 0.932). On sub-group analysis, chlorhexidine-alcohol was associated with a reduction in SSIs in caesarean surgery (RR 0.614; 95% CI 0.453, 0.831) however, chlorhexidine-alcohol was associated with an increased risk of SSI in bone and joint surgery (RR 2.667; 95% CI 1.051, 6.765). When excluding studies at high risk of bias on sensitivity analysis, this difference in alcohol-based combinations for bone and joint surgery was no longer observed (RR 2.636; 95% CI 0.995, 6.983). (4) Conclusions: The use of chlorhexidine-alcohol skin preparations was associated with a reduced risk of SSI compared to iodophor-alcohol agents. However, the efficacy of alcohol-based preparation agents may differ according to the surgical procedure group. This difference must be interpreted with caution given the low number of studies and potential for bias, however, it warrants further investigation into the potential biological and clinical validity of these findings.


Introduction
Over 300 million surgeries are performed annually worldwide based on a 2012 estimate [1]. Surgical site infections (SSIs) remain a major, costly complication of surgical procedures [2][3][4]. The patient's skin bacteria is the major source of infecting pathogens involved in SSIs and is the target of infection prevention strategies such as surgical site skin preparation [5][6][7][8]. The three main agents used are chlorhexidine gluconate, iodophors or alcohol. Despite longstanding use, the optimal preparation remains an issue of controversy [9]. It is recommended that alcohol-based products, combining chlorhexidine with alcohol or iodophors with alcohol, be used in preference to aqueous-based products, based on improved efficacy demonstrated in randomised controlled trials and meta-analyses [6][7][8][9]. These agents have different mechanisms and duration of activity. Alcohol and chlorhexidine gluconate disrupt the cell wall of microorganisms, whereas iodophors act upon intracellular proteins of microorganisms [6,10]. Alcohol has no residual activity, iodophors exhibit persistence of bacteriostatic activity when on the skin and chlorhexidine gluconate has excellent residual activity [5,6,10].
In the guidelines for the prevention of SSIs published by The Centers for Disease Control and Prevention (CDC) [8] and the World Health Organization (WHO) [7,8], there was consensus that alcohol-based preparations were associated with reduced risk of SSI compared to aqueous-based solutions [7,8]. These guidelines, however, gave conflicting recommendations on the optimal agent to combine with alcohol. In the CDC guidelines, no recommendation for a specific alcohol-based product was made based on high-quality evidence from six randomised controlled trials [8]. In comparison, the WHO specified that chlorhexidine-alcohol preparation should be used, based on data from six randomised controlled trials of moderate quality [7]. The trials included a range of different procedure types and no sub-group analyses were performed to examine whether the efficacy of products differed between different surgery types.
The objective of this systematic review and meta-analysis was to determine the comparative effectiveness of alcohol-based chlorhexidine and alcohol-based iodophor as surgical site skin preparation agents to prevent SSIs based on data from randomized controlled trials. A secondary objective was to determine if the effectiveness of alcohol-based agents differs between different surgical procedure types.

Methods
This systematic review and meta-analysis was developed in keeping with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement [11]. The intervention of interest was surgical site skin preparation with alcohol-based solutions or powders applied to the participant's skin at the site overlying the planned surgical incision. The solutions could be applied as a single step (e.g., combination preparations such as 2% chlorhexidine-gluconate in 70% ethanol) or as two sequential steps (e.g., 10% povidone solution followed by 70% alcohol solution). Classes and sub-classes of surgical site skin preparation included, but were not limited to: chlorhexidine gluconate, povidone, iodine, alcohol and ethanol.
Randomised controlled trials (RCTs) in human participants that compared at least two of the alcohol-based interventions for surgical site skin preparation were eligible for inclusion. Studies that compared aqueous-based solutions to other aqueous-based solutions or to alcohol-based solutions were excluded. Studies comparing other antisepsis techniques (e.g., preoperative showering/bathing, impregnated drapes) were not included. Non-English language publications were excluded.
The primary outcome of interest was surgical site infection (SSI), based on the Centers for Disease Control and Prevention/National Healthcare Safety Network (CDC/NHSN) Surveillance definitions for SSIs [5,8]. In the event the trial did not apply the CDC definition, the study definition was documented and mapped to the CDC definition, where possible. Secondary outcomes of interest included: adverse events including skin irritation or allergic reactions and health economic data, including direct hospital costs, societal and quality of life data.
The following electronic databases were searched: Medline (via PubMed), OVID EMBASE, CINAHL and the Cochrane Library Databases. The strategy for electronic databases search is outlined in Appendix A. There was no time limit on the studies included. In addition, the reference list for included studies were reviewed for literature saturation. Study authors were not contacted, and the Grey Literature was not included. Database searches were completed August 2019.
Two authors (TP and EW) independently screened all titles and abstracts of identified studies using the Covidence ® (Veritas Health Innovation LTD, Melbourne, Australia) web-based platform. The full text was retrieved and reviewed for selected abstracts. Disagreement was resolved through consensus. Systematic reviews and meta-analyses were reviewed to identify additional randomised controlled trials. Data were extracted from the selected studies independently by two authors (TP and EW). Data extraction included the procedure type, number of participants, duration of follow up, the authors' definition of surgical site infection, number of patients experiencing a surgical site infection and the interventions being compared. Arm level data were extracted. Risk of bias was assessed using a domain-based evaluation by two authors (TP and EW) independently [12]. Domains examined included random sequence generation, allocation concealment, blinding, attrition, and reporting biases. Disagreement was resolved through discussion and consensus. These data were extracted into RevMan ® (Review Manager [Computer program]. Version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). The protocol for the systematic review and meta-analysis was registered with PROSPERO (CRD42020148548).
The proportion of SSIs was compared between arms using risk ratios (RR) with 95% confidence intervals (95% CI). Statistical heterogeneity was quantified using the I 2 statistic with a threshold at ≥50%. A negative I 2 statistic was regarded as no heterogeneity. Funnel plots were examined for asymmetry applying the approach outlined by Sterne et al. [13]. Disclosed conflicts of interest were also reviewed.
The meta-analysis was performed with a fixed effect model using Mantel-Haenszel methods to obtain the pooled relative risk (RR) estimate [14,15]. In the event significant heterogeneity (I 2 ≥ 50%) was observed, a random effects model using the restricted maximum likelihood method was performed. To account for trials with zero counts in either or both arms, a fixed continuity correction of 0.5 was added to each zero cell [16]. Sensitivity analyses were performed excluding trials with zero total events and, excluding trials with high risk of bias.
As part of the planned subgroup analysis the difference in risk of infection was examined according to different classifications of SSI (superficial and, deep or organ/space) and different surgical procedural groups. Data were analysed using STATA (16.0 College Station, TX, USA) with the META function.

Figure 3. Forest Plot Comparing Risk of Surgical Site Infection (SSI) with Chlorhexidine-Alcohol to
Iodophor-Alcohol Surgical Site Skin Preparation According to Procedure Groups. Navy square represents the effect estimates from the individual studies, the size of the square is proportional to the weight of the study. The horizontal line represents the 95% confidence interval of the study estimate. The red diamond represents the pooled effect size for the specified procedure group. The green diamond represents the pooled effect size.
Adverse events, including allergic reactions to the preparation, were reported in three studies [18,23,27] (Tables 1 and 2). Adverse reactions were rare, occurring in 0.15% of participants allocated to chlorhexidine-alcohol (2/1354) and 0.29% allocated to iodophor-alcohol (4/1361, p = 0.687). Tuuli et al. [27] reported on healthcare resource utilisation including emergency room visits and did not find any difference between chlorhexidine-alcohol and iodine-alcohol. No other study reported on health economic outcomes.
Adverse events, including allergic reactions to the preparation, were reported in three studies [18,23,27] (Tables 1 and 2). Adverse reactions were rare, occurring in 0.15% of participants allocated to chlorhexidine-alcohol (2/1354) and 0.29% allocated to iodophoralcohol (4/1361, p = 0.687). Tuuli et al. [27] reported on healthcare resource utilisation including emergency room visits and did not find any difference between chlorhexidinealcohol and iodine-alcohol. No other study reported on health economic outcomes.
The risk of bias was assessed as low in two studies [23,27] ( Figure 4A). In nine trials, the risk of bias was unclear, particularly the risk of performance and detection bias [17][18][19][20][21]25,26,28,29]. There was high risk of selection bias determined in the trials by Rodrigues et al. [24] and Ostrander et al. [22] ( Figure 4A). Overall, the risk of reporting bias and attrition bias was low ( Figure 4B). Excluding the trials with high risk of bias did not alter the overall estimates (n = 11 trials, RR 0.765; 95% CI 0.646, 0.907; p = 0.0020: I 2 = 39.72% (95% CI 0 to 70.3)), however, on sub-group analysis of different procedure groups, when the trial by Ostrander et al. was excluded, chlorhexidine-alcohol preparation for bone and joint surgery was no longer statistically significant (RR 2.636; 95% CI 0.995, 6.983; I 2 = 0.00% (95% CI 0 to 66.1)).  The funnel plot ( Figure 5A) was asymmetric. When repeated according to procedu group, the plots were symmetric ( Figure 5B) although the number of the trials was sm thereby limiting analysis for bias. The funnel plot ( Figure 5A) was asymmetric. When repeated according to procedure group, the plots were symmetric ( Figure 5B) although the number of the trials was small, thereby limiting analysis for bias.

Discussion
Overall, our study, including 6023 participants, showed the use of chlorhexidinealcohol skin preparations was associated with a 21% reduction in the relative risk of SSI compared to iodophor-alcohol agents. This equates to an absolute difference of 19 fewer infections per 1000 patients undergoing surgery. A key finding however, suggests that the efficacy of alcohol-based surgical site skin preparation agents may differ according to the surgical procedure group. In caesarean surgery, the use of chlorhexidine-alcohol

Discussion
Overall, our study, including 6023 participants, showed the use of chlorhexidinealcohol skin preparations was associated with a 21% reduction in the relative risk of SSI compared to iodophor-alcohol agents. This equates to an absolute difference of 19 fewer infections per 1000 patients undergoing surgery. A key finding however, suggests that the efficacy of alcohol-based surgical site skin preparation agents may differ according to the surgical procedure group. In caesarean surgery, the use of chlorhexidine-alcohol preparations was associated with a 39% reduction in the relative risk of SSIs. In contrast, chlorhexidine-alcohol skin preparation agents were associated with a 2.7-fold increased risk of SSIs in bone and joint surgery. In the other procedure groups, general surgery and skin and soft tissue surgery, there was no demonstrated difference between the alcoholbased agents.
The observation in the bone and joint surgery cohort differs from the WHO guidelines which recommended that alcohol-based chlorhexidine solutions should be used. The metaanalysis performed by the WHO Guideline writing group include three studies examining bone and joint surgery [19,25,26]. The estimates in bone and joint surgery were influenced by one large study by Peel et al. [23] although examination of the forest plot suggests the majority of trials in this surgical group favoured alcohol-based iodophor skin preparation agents. The larger study by Peel et al. (n = 780) [23] examined skin preparation agents in arthroplasty surgery and was reported after the WHO guidelines were published.
The findings in caesarean surgery were influenced by the trials by Tuuli et al. [27] and Kesani et al. [20]. Of note, the trial by Tuuli et al. was published outside systematic review time limits for the WHO meta-analysis however, was "exceptionally included" [3,7]. This trial also was published after the specified time limits for the CDC meta-analysis [8] and was not included in the meta-analysis, potentially accounting for the differing findings between the two guidelines.
The differences observed, particularly between bone and joint surgery and caesarean surgery, may be due to differences in the populations, particularly age and gender. It may also reflect the differences in infection control approaches, for example, screening and decolonisation for Staphylococcus aureus is a recommended strategy in orthopaedic surgery whereas the role has not been established in caesarean surgery [7,8]. The causative organisms of SSI also differ between these surgical groups: Staphylococcus species are the most common bacteria isolated in bone and joint infections, compared with Gram negative bacteria and anaerobes in caesarean surgery [30,31]. The majority of the cohort in the bone and joint surgery group underwent procedures involving the implantation of prosthetic material. Infections involving prosthetic material differ from other types of SSIs due to the propensity of organisms involved in medical device infections to form biofilm [8]. This observation raises the possibility that the differences between bone and joint and caesarean surgery may relate to different anti-biofilm properties of the skin preparation agents. Both chlorhexidine and iodophors display anti-biofilm properties [32][33][34]. The effectiveness of the agents forming biofilms may be concentration dependent [34,35]. In a trial by Smith and colleagues, lower concentrations of chlorhexidine were less effective at eradicating methicillin resistant Staphylococcus aureus biofilm compared to 2 and 4% concentrations [35]. The trial by Peel et al. used 0.5% chlorhexidine however, conversely there were more cases of Staphylococcus aureus SSI in the iodophor-alcohol arm (3/4 SSIs: 75%) compared with the chlorhexidine-alcohol arm (3/12 SSIs: 25%) [23].
Given that this meta-analysis included a small number of trials (n = 13), the observed difference may be a false positive finding, due to chance. Particularly for the analysis according to procedure group, the estimates must be interpreted with caution given the small number of trials included in each sub-group.
The trials included in this study compared different concentrations and formulations of skin preparation agents. The impact on the estimates of differing concentrations is unclear. There is limited data on the optimal concentration of agents, with no head-to-head trials comparing concentrations [3,7,9,36]. In addition, a range of definitions for SSI were applied in the included studies and participant follow-up differed. The heterogeneity, however, was low. Finally, the majority of trials included had unclear risk of bias, particularly for performance and detection biases.

Conclusions
The results of this meta-analysis suggest that the use of alcohol-based chlorhexidine surgical site skin preparation for caesarean section is associated with a lower risk of surgical site infections compared with alcohol-based iodophors. The opposite finding was observed for bone and joint surgical procedures, raising the possibility that the optimal skin preparation agent may differ with surgical procedures. However, when excluding studies at risk of major bias, this difference for bone and joint procedures was no longer significant. These observations must be interpreted with caution and require further investigation to corroborate these findings and to determine if there is a biological mechanism(s) explaining these findings. Further, larger trials, particularly in other surgical procedure groups, are warranted.
Author Contributions: Conceptualization, T.N.P. and E.W.; methodology, all authors; formal analysis, all authors; data curation, T.N.P. and E.W.; writing-original draft preparation, T.N.P.; writingreview and editing, all. All authors have read and agreed to the published version of the manuscript.