Povidone-iodine-alcohol is superior to chlorhexidine-alcohol in its antimicrobial efficacy on the anaerobic microbial flora in surgical skin preparation of the shoulder region

Background: Cutibacterium acnes is part of the anaerobic skin microbiome and resides in deeper skin layers. The organism is an important agent of surgical site infections (SSI), particularly in shoulder surgery, and is difficult to target with skin antisepsis. We hypothesized that prolonged preparation with an agent that penetrates deeper into the skin is more effective in shoulder surgery. Thus, we compared two different classes of antiseptics, each combined with alcohol, and each applied with two different contact times. Methods: Shoulders of 16 healthy volunteers were treated for 2.5 min (standard) or 30 min (prolonged) with alcohol-based chlorhexidine (CHG-ALC; 2% w/v CHG, 55% w/v [70% v/v] 2-propanol) or alcohol-based povidone iodine (PVP-I-ALC; 3.24% w/v PVP-I, 38.9% w/v 2-propanol, 37.3% w/v ethanol). Skin sites were sampled before, immediately after, and 3 h after treatment. Results: Aerobic skin flora was reduced more effectively by PVP-I-ALC than by CHG-ALC after 2.5 min and immediate sampling, but not after prolonged contact times and 3-h sampling. Anaerobic flora was reduced more effectively by PVP-I-ALC than by CHG-ALC after standard and prolonged contact times and immediate sampling, but not after 3-h sampling. None of the volunteers reported negative effects from skin antisepsis. Conclusions: PVP-I-ALC is more efficacious than CHG-ALC for topical skin antisepsis regarding the anaerobic flora on the skin of the shoulder. Standard and prolonged contact times demonstrated superiority for PVP-I-ALC for samples taken immediately, but missed significance 3 h after application. The clinical relevance of these findings should be studied with SSI as an endpoint. CA, USA). Mann-Whitney- and Wilcoxon-tests were calculated. A p-value <0.05 was considered to indicate a statistically significant difference. Sample size calculations for testing differences in RFs were based on two-sided t-tests for paired means at p = 0.05. Assuming effect sizes (i.e. differences in RFs divided by the standard deviation of the differences) between 0.75 and 1.0, results indicated that a sample between 10 and 16 cases was required to detect these differences with a power of 80%.


Introduction
The skin flora of patients is one of the most important factors in the pathogenesis of surgical site infections (SSI) [1][2][3][4]. Skin antisepsis constitutes an effective measure to reduce the numbers of microorganisms on skin and has been included as a key measure to prevent SSIs in recent international guidelines and recommendations [5][6][7]. The resident aerobic skin flora consists of organisms such as coagulase-negative staphylococci (CNS), Micrococcus luteus, Corynebacterium spp., Malassezia furfur and Acinetobacter spp. [8]. The anaerobic skin flora is located primarily in hair Betaseptic® (3.24% w/v PVP-I, 38.9% w/v IPA, 37.3% w/v ethanol; Mundipharma, Limburg, Germany): vigorous rubbing for 30 s by using sterile forceps and gauze for 30 s, afterward keeping the area wet with the antiseptic by using forceps and gauze for 2 min or a soaked sterile dressing (Zetuvit®, 20 x 40 cm) for 29.5 min.
For the two CHG-ALC products, the IPA percentage was converted to 55% w/v from the manufacturers' stated 70% v/v for the purpose of uniformity of measurements.
Sampling. The microbial skin counts were obtained before antiseptic application (pre-value), after application and air-drying of the sampling site (immediate post-value), and 3 h after treatment (3-h post-value) following ASTM method E1874-14 [23] using the cup scrub technique on 2.5 cm 2 of skin.
Concurrently with the retrieval of the immediate post-values, another skin area of 4 x 4 cm was covered with a sterile dressing (Hydrofilm® transparent dressing 12 x 25 cm, Hartmann, Heidenheim, Germany) to protect a skin area where the 3-h samples were to be collected later.
Microbiological techniques. After 5 min neutralization, serial dilutions were prepared in the respective neutralization solution. 0.1 mL of each solution was plated onto Columbia agar with 5% sheep blood (Becton Dickinson, Heidelberg, Germany) for aerobic incubation (37°C, 48 h) and onto Schaedler agar (BioMérieux, Nürtingen, Germany) for anaerobic incubation (37°C, 7 days). After aerobic incubation, the colony forming units (CFU) were counted and a random sample of colonies was picked for identification. The anaerobic CFUs were counted after 7 days incubation and all anaerobic plates were analysed by Gram stain and VITEK® Cards (BioMérieux) to identify C. acnes.
We expressed the number of CFU as log 10 values and calculated the reduction factors (RFs) as the difference between the log 10 pre-values and the log 10 post-values. To calculate the reduction factors and transform to log 10 , plates without CFU growth were set to the value 1.
Sampling and validation of neutralization. Skin bacteria from five volunteers were collected by the cup scrub technique [23] and pooled. Using the methodology of ASTM E1054-08 [24], pooled bacteria were used to verify the effectiveness of Lipofundin to inactivate CHG, of sodium thiosulfate to diminish the oxidizing agent iodine, and to ensure that the inactivation solutions did not significantly influence the bacterial counts, quantitatively and qualitatively. Final concentrations of 0.4% CHG and 0.6% PVP-I in 1 ml 0.9% NaCl were tested for neutralizer effectiveness. The concentration of the active agent was calculated using the treated skin area of 300 cm² (17.5 cm x 17.5 cm) with 17 mL of antiseptic solution from the applicator and 3 mL of additional antiseptic solution, which arises from the equilibrium of the soaked dressing with the antiseptic liquid film on the skin. Finally, an area of 2.5 cm² served as the basis for the microbiological examinations after scrubbing.
Validation of neutralization was conducted according to the methodology of ASTM E1054-08 [24]. The suitable neutralizers, lecithin for inactivating biguanides and thiosulfate for quenching iodine, were derived from table 1 in ASTM E1054 and Annex B in EN 1040 [25] and EN 13727 [26]. Lipofundin containing 1.2% egg yolk lecithin inactivated 0.4% CHG without any inhibitory effect on growth of pooled skin bacteria after aerobic and anaerobic incubation, and sodium thiosulfate, the quenching agent for iodine in the neutralization mixture, was effective for 0.6% PVP-I without influencing bacterial counts. Similar results were obtained in tests of neutralizer effectiveness, neutralizer toxicity and organism viability under aerobic and anaerobic culture conditions, using test solutions containing the residual antimicrobial agent which were derived from volunteers after skin antisepsis. Therefore, it was ascertained that CHG or PVP-I were effectively inactivated by the respective neutralization solutions without influencing bacterial growth after aerobic and anaerobic incubation.
Statistical analysis. The analysis of the raw data was performed using Graphpad Prism® (GraphPad, La Jolla, CA, USA). Mann-Whitney-and Wilcoxon-tests were calculated. A p-value <0.05 was considered to indicate a statistically significant difference. Sample size calculations for testing differences in RFs were based on two-sided t-tests for paired means at p = 0.05. Assuming effect sizes (i.e. differences in RFs divided by the standard deviation of the differences) between 0.75 and 1.0, results indicated that a sample between 10 and 16 cases was required to detect these differences with a power of 80%. than CHG-ALC only after 2.5 min contact time and immediate sampling ( Table 2), but not in any other of the tested parameters, including prolonged application time and sampling after 3 h. CNS which multiplies aerobically and to a lesser extend anaerobically, were not completely eliminated by CHG-ALC. This corresponds with findings that after 3 minutes of preoperative skin antisepsis with CHG-Alc always CNS were detected in smears in the surgical field before implantation of hip endoprostheses (unpublished) and also with data showing bacterial growth directly after skin antisepsis (propan-2-ol (70 % v/v) in 35% of operations in traumatology, whereof 53% were identified as CNS which finally caused wound contaminations in 10% of operations with clonally related strains [4]. These data underline the need for better protection against CNS in at least orthopaedic and trauma surgery by adequate potent antisepsis in order to better prevent up to now still unacceptable infection rates predominantly caused by CNS. However, the situation was different for the anaerobic skin flora. PVP-I-ALC, when applied for 2,5 or 30 min, was better than CHG-ALC at all four immediate sampling points (Table 4) Our results are indirectly supported by a clinical trial that compared the treatment of abdominal incision sites. It was demonstrated that 0.7% iodine povacrylex with 74% IPA was more effective in reducing SSI than 2% CHG with 70% IPA [30]. Furthermore, another trial comparing 0.5% CHG with 70% alcohol and 1% PVP-I with 70% alcohol in hip and knee arthroplasty showed no difference in superficial wound complications [31]. However, on secondary end-point analysis, skin antisepsis with CHG-ALC was associated with significantly higher odd ratios for overall SSI, including prosthetic joint infection [31]. In another trial, skin antisepsis with 7.5% PVP-I in aqueous solution did not show a significant difference to 2% CHG in 70% IPA in the SSI rate after neurosurgical spine procedures in adults [32].
The shoulder region was chosen as test area for skin antisepsis because of its known tendency to be colonized with C. acnes, which is due to its high density of hair follicles and sebaceous glands [33].
Furthermore, C. acnes is a major pathogen of SSI in shoulder surgery [34,35]. The standard method for preoperative skin antisepsis consists of initially rubbing the skin with either an applicator, or with forceps and swab, for 30 s and afterwards keeping the skin wet with the antiseptic agent for 2 min.
The efficacy of both application methods is comparable with each other [36]. We did not use the applicator with CHG-ALC for the prolonged application time, because we wanted to avoid a potential mechanical irritation that might occur when keeping the area wet with an applicator.
It is considered beneficial to use an extended application time on areas of skin that has a high density of sebaceous glands [37], and this appears important for surgical procedures performed in those areas. The recommended application time for PVP-I-ALC solution that we used, according to the manufacturer, is 1 min on skin with a low density of sebaceous glands and at least 10 min on skin with a high density of sebaceous glands. The extended contact time of 30 min in the present study was chosen because we assumed that with an extended contact time, the deep skin flora would be reached more effectively. If confirmed, the clinical implications for preoperative shoulder preparation would be substantial. Indeed, after a contact time of 29.5 min under a soaked sterile dressing, the area still appeared to be moist.
It is presumed that the physiological flora of the human skin is regenerated completely after 3 days, because the re-colonization already starts 60 min after alcohol-based skin antisepsis [38]. After 24 hours, the skin flora is nearly completely regenerated [39]. Therefore, the study was performed as a crossover study with an interval of 3 days between the tests. In accordance with our hypothesis, the pre-values on day 1 and day 4 did not differ significantly.
The validity and interpretation of our results depends heavily on the selection of effective neutralizing solutions. A final concentration of 3% (w/v) soy bean or egg yolk lecithin may be used as effective neutralizer for CHG, but also Lipofundin, which contains 1.2 % egg yolk lecithin, may be applied as an effective and non-toxic neutralizer, especially for membrane (phospholipid)-attacking antiseptic agents, such as chlorhexidine, octenidine and poly(hexamethylene) biguanide. Neutralization effectiveness and suitability of Lipofundin was previously confirmed in a study using octenidine-and chlorhexidine-containing mouth rinses [40], and also in a study determining the antimicrobial efficacy of poly(hexamethylene) biguanide coating of titanium implant material [41].
Our study has two important limitations. First, an ideal trial should compare CHG and PVP-I combined with the same alcohol species with identical concentrations, if conclusions concerning the activity of the CHG or PVP-I component are to be made. However, we had to use readily available commercial formulations, due to the fact that PVP-I formulations are too difficult to prepare in-house. The CHG comparator contained 55% w/v (70% v/v) straight IPA, and the PVP-I comparator contained 76% w/v of a mix between IPA and ethanol in nearly equal parts. As a consequence, there is a theoretical possibility that this study's results may be due to the different alcohol compositions. Future studies may be able to address the microbicidal activity of CHG and PVP-I when combined with equal comparator alcohols. Second, a majority of our immediate and 3-h post-values had no detectable CFUs. That means, our study unintentionally mostly captured values located at the bottom end of the measurable range of the experimental setup. This also means that any differences between the two antiseptics, contact times and immediate versus 3-h effects were likely underestimated as a result. A scenario in which most of the measured values would be located well within the measurable range would have a greater chance of showing statistically significant differences if they exist. Future studies may be able to address this with larger numbers of volunteers, larger sampled skin sites, larger sample volumes plated on agar, or a combination of these variables.
Commonly recommended contact times for surgical skin preparation, including the 2.5 min chosen in this study, are neither experimentally nor clinically well founded. Starting from the hypothesis that a prolonged application time achieves better penetration and reaches deeper skin compartments and hair follicles, it was decided to examine a contact time of 30 min in addition to 2.5 min. However, no relevant significant differences were observed between these contact times. In addition, the question arises whether a contact time of 30 min is practicable in a busy operating room setting. This means that contact times shorter than 30 min should be investigated in future studies. For example, some antiseptic preparations, when applied for 2.5 min on areas with high density of sebaceous glands, meet or even exceed the efficacy of the experimental reference antiseptic that is applied for 10 min [42]. The 3-h values in our study aimed at assessing the sustained activity of the antiseptic and the intraoperative skin re-colonization under surgical drapes, as would be expected during typical operations.
Common efficacy testing of skin antisepsis only assesses aerobic flora, and in Europe the samples are typically collected by swabbing the skin surface [43] and do not mobilise the deep resident skin flora to the same extent as the ASTM cup scrub method does [23]. One possibility to address this in future studies would be to take dermal punch biopsies after antisepsis of the shoulder area, so that the effects of prolonged contact times can be measured in deeper skin layers.

Conclusions
PVP-I in combination with alcohol (3.24% w/v with ≥76% w/v alcohol) was superior in its immediate efficacy in reducing the anaerobic skin flora on the shoulder compared to 2% w/v CHG with 55% w/v IPA. No significant difference was seen between standard and prolonged contact times. PVP-I-ALC seems to be a promising option for antisepsis on skin with a high density of sebaceous glands at a contact time of 2.5 min, especially in shoulder surgery. Future studies should expand upon these investigations with greater numbers of participants and contact times closer to 2.5 min, and ultimately should focus on clinical trials with SSIs as the endpoint. a Mean reduction factor (RF) and standard deviation.
b Number of volunteers with "0" cfu counts.   a Mean reduction factor (RF) and standard deviation.
b Number of volunteers with "0" CFU counts.
20 Table 5. P values for the reduction of anaerobic flora by chlorhexidine-alcohol (CHG-ALC) and povidone-iodine-alcohol (PVP-I-ALC) after 2.  Figure 1 The sampled skin area is marked in green on a right male shoulder. Most of the incisional regions in arthroscopic and open shoulder surgery are included within this area.