Efficacy of surgical skin preparation with chlorhexidine in alcohol according to the concentration required to prevent surgical site infection: meta-analysis

Abstract Background A combination of chlorhexidine gluconate and alcohol (CHG–alcohol) is recommended for surgical skin preparation to prevent surgical site infection (SSI). Although more than 1 per cent CHG–alcohol is recommended to prevent catheter-related bloodstream infections, there is no consensus regarding the concentration of the CHG compound for the prevention of SSI. Methods A systematic review and meta-analysis was performed. Four electronic databases were searched on 5 November 2020. SSI rates were compared between CHG–alcohol and povidone-iodine (PVP-I) according to the concentration of CHG (0.5 per cent, 2.0 per cent, 2.5 per cent, and 4.0 per cent). Results In total, 106 of 2716 screened articles were retrieved for full-text review. The risk ratios (RRs) of SSI for 0.5 per cent (6 studies) and 2.0 per cent (4 studies) CHG–alcohol were significantly lower than those for PVP-I (RR = 0.71, 95 per cent confidence interval (c.i.) 0.52 to 0.97; RR = 0.52, 95 per cent c.i 0.31 to 0.86 respectively); however, no significant difference was observed in the compounds with a CHG concentration of more than 2.0 per cent. Conclusions This meta-analysis is the first study that clarifies the usefulness of an alcohol-based CHG solution with a 0.5 per cent or higher CHG concentration for surgical skin preparation to prevent SSI.


Introduction
Surgical site infection (SSI) is the third most common category of healthcare-associated infections, with a prevalence of 15.7 to 31.0 per cent among all healthcare-associated infections 1,2 . In one systematic review, most studies revealed an economic benefit associated with prevention of SSI 3 . Several guidelines for the prevention of SSI have been published [4][5][6] . Approximately half of SSIs are estimated to be preventable by application of evidence-based strategies 7 . A bundle approach is suggested to decrease SSIs, and surgical site skin preparation is one of the essential elements used in bundle approaches [8][9][10] . In particular, Leaper and Ousey 11 recommended the use of a 2 per cent chlorhexidine gluconate in alcohol (CHG-alcohol) skin preparation, postoperative negative-pressure wound therapy, and antiseptic surgical dressings because of the high quality of evidence. Two meta-analyses suggested a significant benefit of using CHG-alcohol compared with aqueous povidone-iodine (PVP-I) with moderate-quality evidence 6,12 ; however, a low quality of evidence was shown when the risk of SSI was compared between CHG-alcohol and PVP-I in alcohol-based solutions.
The concentration of the CHG solution ranges from 0.5 per cent to 4.0 per cent, but there are no data that specify the ideal concentration for surgical preparation to prevent SSI. In contrast, the Centers for Disease Control and Prevention guidelines for the prevention of intravascular catheter-related bloodstream infection recommend preparing clean skin with a higher than 0.5 per cent chlorhexidine (not 0.5 per cent preparation) in alcohol solution. This recommendation is based on the fact that when 0.5 per cent chlorhexidine preparation was compared with 10 per cent PVP-I, no differences were seen in either central venous catheter colonization or catheter-related bloodstream infection 13,14 . Pages et al. 15 reported that compared with PVP-I in alcohol, the incidence of catheter-related infection was lower with 2 per cent chlorhexidine-alcohol and similar with higher than 1 per cent CHG-alcohol after controlling for potential confounders.
A 2 per cent CHG-70 per cent isopropyl alcohol solution has become widely used for both preparation during central venous catheter insertion and preparation of the surgical site 16,17 ; however, such compounds are not available in some countries, including Japan, and they often have irritating effects on the skin when a 2 per cent or higher concentration is used [18][19][20] . The efficacy of a 0.5 per cent or 1.0 per cent chlorhexidine skin preparation with alcohol for the prevention of SSI remains unclear. The present study was performed to identify the concentration of CHG in alcohol-based solution for skin preparation to prevent SSI compared with PVP-I. The secondary aim was to compare the SSI rate between skin preparations of CHG-alcohol and PVP-I in alcohol.

Selection of studies
Randomized clinical trials (RCTs) that met the following criteria were included in the meta-analysis: comparison of the SSI rate after skin preparations for surgery using CHG-alcohol and PVP-I; use of antiseptics for preparation of the surgical site in the operating room, not for washing (bathing and showering) separately outside of the operating room; and availability of detailed information in English. At least two authors (T.H., T.M., J.K., I.S., Y.T., and T.M.) independently screened the literature. During the screening, disagreements were resolved through discussions with a third reviewer (S.T.).

Data extraction
Two authors (T.H. and S.T.) independently extracted data from the included studies. Disagreements were resolved by discussions. The following information was extracted: study design, country, study interval, detailed information of each antiseptic (concentration and solution), reported outcome, skin preparations before use of the two antiseptics, observation interval, definition of SSI, patients included, exclusion criteria, number of participants based on intention-to-treat (ITT) (or per-protocol set if ITT was unavailable), number of participants with SSI, and number of participants with adverse events.

Outcomes analysed
The primary purpose of this study was the efficacy of decreasing the SSI risk by CHG-alcohol versus PVP-I (alcohol-based/aqueous solution) according to the CHG concentration. The following terms in each study were defined as SSI: wound infection, postoperative infection, wound complication, and postoperative  surgical wound infection. As the secondary purpose of this study, the following analyses for SSI risk were performed: overall comparison between CHG-alcohol and PVP-I; comparison between CHG-alcohol and PVP-I-alcohol; comparison between CHG-alcohol and PVP-I according to the PVP-I concentration; comparison of the two antiseptic groups stratified by wound classification (clean, clean-contaminated, and contaminated wound) 22 ; and comparison of the two antiseptic groups stratified by SSI type (superficial incisional, deep incisional, and organ/ space SSI) 22 . The comparative risk of adverse events between CHG-alcohol and PVP-I was also evaluated.

Assessment of risk of bias and publication bias
The two authors (T.H. and S.T.) independently assessed the risk of bias using the Cochrane Handbook for Systematic Reviews of Interventions 23 . The bias assessments performed in the present study were random sequence generation (selection bias); allocation concealment (selection bias); blinding of participants and personnel (performance bias); blinding of outcome assessment (detection bias); incomplete outcome data (attrition bias); selective reporting (reporting bias); and other bias. The risk of other bias was judged to be low when the trials received no financial support from pharmaceutical companies.  If sufficient information for assessment was not described, the risk of bias was judged to be unclear. In addition, publication bias was assessed by visual examination of a funnel plot and statistical analyses using Egger's test.

Results analyses and statistical analyses
The extracted data were analysed using Review Manager for Windows (RevMan version 5.4.1; The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark), and forest plots were prepared. The Mantel-Haenszel random-effects model was used to calculate the risk ratios (RRs) and 95 per cent confidence intervals (95 per cent c.i.). Between-study heterogeneity was quantified using the I 2 statistic, which was assessed according to the following criteria: I 2 less than 25 per cent, no heterogeneity; I 2 of 25-50 per cent, moderate heterogeneity; and I 2 greater than 50 per cent, high heterogeneity. A P value of less than 0.050 was considered to indicate a significant difference.

Results
Literature search results Figure 1 shows the screening and selection of studies. From the four electronic databases, 2716 articles were obtained to be screened and 382 duplicate articles were excluded. After screening the titles and abstracts, 2228 articles were excluded, and 106 articles were retrieved for full-text review. Of these 106 articles, 91 did not meet the inclusion criteria. Finally, 15 studies 24-38 were included in the meta-analysis. Table 1 summarizes the characteristics of the studies included in the present meta-analysis. The CHG concentration ranged from 0.5 per cent to 4.0 per cent, and the PVP-I concentration ranged from 1.0 per cent to 10.0 per cent. Data on the CHG and PVP-I concentrations were unavailable in two studies 31,33 and four studies 28,31,33,38 respectively. Of 15 included studies, 14 studies described the follow-up interval, and nine studies had a follow-up of 30 days. Table 2 summarizes the characteristics of the participants. In each article, the number of participants was extracted based on ITT. The number of ITT participants was unavailable in three studies 25,30,37 ; therefore, the number of per-protocol set patients was extracted instead of ITT participants in these three studies. In total, 6974 participants were involved in the studies: 3472 participants were disinfected with CHG-alcohol and 3502 participants were disinfected with PVP-I. Figure 2 indicates the risk of bias for each study. No studies were judged to have a high risk of bias among all included studies. The blinding of participants and personnel (performance bias) was unclear in all included studies except for Srinivas et al 32 .

Overall comparison of SSI rate between CHG-alcohol and PVP-I
SSIs were detected in 516 participants (208 in the CHG-alcohol group and 308 in the PVP-I group). CHG-alcohol was significantly more effective than PVP-I (RR = 0.69, 95 per cent c.i. 0.56 to 0.84, P = 0.0002, I 2 = 18 per cent) (Fig. 3).

Comparison of SSI rate between CHG-alcohol and PVP-I (alcohol-based/aqueous solution) according to CHG concentration
The CHG concentrations in CHG-alcohol were stratified into 0. to 0.97, P = 0.03, I 2 = 21 per cent) (Fig. 4a). Four RCTs compared 2.0 per cent CHG-alcohol and PVP-I [34][35][36][37] . A significantly lower SSI rate was observed in the 2.0 per cent CHG-alcohol group than in the PVP-I group (RR = 0.52, 95 per cent c.i. 0.31 to 0.86, P = 0.01, I 2 = 55 per cent) (Fig. 4b). No significant difference in the SSI rate was found between the greater than 2.0 per cent CHG group and the PVP-I group ( (Fig. 4c,d), possibly because of the lack of power caused by an insufficient number of included studies. Other PVP-I concentrations, commonly using concentrations such as 5 per cent and 7.5 per cent, could not be analysed because only one study was available at each concentration.

Other comparisons of SSI rate between CHG-alcohol and PVP-I
In the comparison among the alcohol-based solutions (Fig. 5), CHG-alcohol was associated with a significantly lower SSI rate than alcohol-  (Fig. S1a,b). Subgroup analyses of the effectiveness of preventing SSI between CHG-alcohol and PVP-I were conducted according to the wound classification (Fig. S2a-c) and SSI type (Fig. S3a-c). Four studies showed data for clean wounds, four for clean-contaminated wounds, and one for contaminated wounds. The effectiveness of CHG-alcohol over PVP-I was demonstrated only for clean-contaminated wounds (RR = 0.56, 95 per cent c.i. 0.37 to 0.85, P = 0.006, I 2 = 50 per cent) (Fig. S2b); no significant difference was observed in the other wound classes (Fig. S2a,c). In the analyses of SSI type, significant benefits in reducing the SSI risk with CHG-alcohol compared with PVP-I were observed for superficial incisional SSI 26,30-38 and deep incisional SSI 30-37 (RR = 0.71, 95 per cent c.i. 0.54 to 0.93, P = 0.01, I 2 = 9 per cent and RR = 0.47, 95 per cent c.i 0.24 to 0.91, P = 0.03, I 2 = 0 per cent respectively) (Fig. S3a,b). No significant difference between the two antiseptic groups was observed for organ/space SSI (RR = 1.23, 95 per cent c.i. 0.54 to 2.82, P = 0.62, I 2 = 0 per cent) 30,[32][33][34][35]38 (Fig. S3c).

Comparison of adverse events between CHG-alcohol and PVP-I
Allergic reactions were the only reported adverse events in the included studies, and no significant difference was observed between the CHG-alcohol group and PVP-I group (RR = 0.75, 95 per cent c.i. 0.17 to 3.29, P = 0.70) (Fig. S4) 32,34,36 .

Assessment of publication bias
Publication bias was assessed using funnel plotting and Egger's test (Fig. 6). No statistically significant publication bias was found (P = 0.5703).

Discussion
This is the first meta-analysis to compare the risk of SSI between CHG-alcohol and PVP-I according to the CHG concentration. The results showed that both 2.0 per cent compound and 0.5 per Berry et al. 24 Brown et al. 25 Cheng et al. 27 Gezer et al. 38 Kesani et al. 36 Ngai et al. 33 Paocharoen et al. 28 Perek et al. 31 Ritter et al. 37 Rodrigues et al. 30 Salama et al. 34 Sistla et al. 29 Springel et al. 35 Srinivas et al. 32 Veiga et al. 26 Random sequence generation (selection bias) Allocation concealment (selection bias) Blinding of participants and personnel (performance bias) Blinding of outcome assessment (detection bias) Incomplete outcome data (attrition bias) Selective reporting (reporting bias) Other bias

PVP-I
Berry et al. 24 Brown et al. 25 Veiga et al. 26 Cheng et al. 27 Paocharoen et al. 28 Sistla et al. 29 Rodrigues et al. 30 Perek et al. 31 Ngai et al. 33 Srinivas et al. 32 Salama et al. 34 Springel et al. 35 Kesani et al. 36 Ritter et al. 37 Gezer et al. 38 Total (95% c.i.) Berry et al. 24 Brown et al. 25 Veiga et al. 26 Cheng et al. 27 Rodrigues et al. 30 Srinivas et al. 32 34 Springel et al. 35 Kesani et al. 36 Ritter et al. 37 Sistla et al. 29 1.2.4 (D) CHG-alcohol = 4.0% cent compound more effectively prevented SSI than PVP-I. This is an important step in clinical practice for countries in which 2 per cent CHG-alcohol is not available. Because of the risk of anaphylaxis, Japanese pharmaceutical regulations prohibit the application of CHG to mucosal surfaces, including in dental care, and limit the CHG concentration in skin antiseptics to a maximum of 1 per cent.
To determine the recommended CHG concentration in CHG-alcohol, direct comparison with different CHG concentrations is required. In one study, the antimicrobial activity of a 2.0 per cent CHG-alcohol solution was superior to that of a 0.5 per cent CHG-alcohol solution when challenged with a Staphylococcus epidermidis biofilm 39 ; however, significantly increased preventative effects against SSI have not been demonstrated by head-to-head RCTs between different CHG concentrations. Three studies compared the efficacy of CHG-alcohol with different CHG concentrations by skin cultures. In a study that evaluated the mean bacterial count reductions for the use of surgical skin preparation, the antimicrobial effectiveness of 1.0 per cent CHG-alcohol was superior to that of 0.5 per cent CHG-alcohol, particularly at the abdominal site 40 . CHG-alcohol has immediate and persistent activity, with the alcohol having a rapid mode of action and the CHG offering residual activity. CHG binds to anionic cutaneous protein, resulting in a prolonged antiseptic effect. Hence, a higher concentration might be required for surgical skin preparation.
Similarly, Casey et al. 41 compared 0.5 per cent CHG-alcohol with 2.0 per cent CHG-alcohol for skin antisepsis in patients undergoing vein graft harvesting for coronary artery bypass graft surgery. There was a significant difference in the culture-positive rate between 0.5 per cent CHG-alcohol and 2.0 per cent CHG-alcohol after incision closure, which occurred at approximately 90 min after application of the skin antiseptics in each group (33.3 versus 12.5 per cent respectively). In addition, significantly fewer microorganisms within the adhesive dressings removed 24 h after application were observed in the 2.0 per cent CHG-alcohol group than in the 0.5 per cent CHG-alcohol group, which might indicate that 2.0 per cent CHG-alcohol more effectively kills microorganisms located in the lower layers of the skin. In contrast, Nishihara et al. 42 reported that there was no significant difference in the log reduction of the bacterial count among CHG preparations of 0.5 per cent, 1.0 per cent, and 2.0 per cent.
Although the precise prevalence of CHG allergy is unclear, the numbers of case reports describing such allergy have recently increased, especially in the perioperative setting 43 26 Cheng et al. 27 Rodrigues et al. 30 Perek et al. 31 Ngai et al. 33 Salama et al. 34 Ritter et al. 37    concentrations of CHG (2-4 per cent) possibly have irritant effects on the skin, leading to an impaired skin barrier and increasing the risk of allergy 18,43,48 . Nishihara et al. 40 reported the mean visual scores of skin irritation and the total cumulative irritation scores after repeated exposure to test products, and lower scores were found in the 1 per cent than 2 per cent CHG-alcohol group. The potential risks versus benefits should be considered before proposing an adequate CHG concentration.
Previous meta-analyses have shown that alcohol-based antiseptic solutions are more effective than aqueous solutions in reducing the risk of SSI 6 . The present meta-analysis demonstrated that CHG was more protective than PVP-I in the evaluation limited to alcohol-based solution. Skin preparation is performed to prevent wound infection, and this meta-analysis confirmed that CHG-alcohol was significantly more protective than PVP-I against both superficial and deep incisional SSI but not against organ-space SSI. Surgical skin preparation with CHG-alcohol was superior to skin preparation with PVP-I for preventing SSI only after clean-contaminated surgery. An additional RCT is required to evaluate the effectiveness of CHG-alcohol in clean surgery.
This study had some limitations. First, washing the patient's skin with antiseptics, which was performed separately outside the operating room, might have impacted the results. Some studies adopted the same antiseptic compound for both preoperative body washing (CHG soap or PVP-I soap) and skin preparation at the surgical site. Second, bias caused by the heterogeneity of SSI prevention protocols, including antimicrobial prophylaxis, and normothermia, should be considered. Of the 15 studies, 12 described the use of prophylactic antimicrobials; however, it was not possible to compile detailed information on criteria for use. Third, comparison with PVP-I in alcohol should be performed to confirm the effectiveness of 0.5 per cent CHG-alcohol. Fourth, considering the time course between the first 24 and the last 38 studies included in this meta-analysis, the improvement of medical care and medical technology during this interval should be considered as an important confounder. Last, only three studies 26,33,34 described the time of exposure. Although application time of 3-5 min is recommended in PVP-I solution, a shorter drying time is permitted in CHG-alcohol 49,50 .
An alcohol-based CHG solution with a CHG concentration of 0.5 per cent or higher can be used for surgical skin preparation to prevent SSI. CHG-alcohol was more effective than PVP-I irrespective of the type of solution (alcohol versus aqueous). Additional studies are required to propose an adequate CHG concentration by head-to-head comparison of the SSI rate and skin complications according to the CHG concentration.

Funding
The authors have no funding to declare.