Surgical hand antisepsis to reduce surgical site infection.

BACKGROUND
Medical professionals routinely carry out surgical hand antisepsis before undertaking invasive procedures to destroy transient micro-organisms and inhibit the growth of resident micro-organisms. Antisepsis may reduce the risk of surgical site infections (SSIs) in patients.


OBJECTIVES
To assess the effects of surgical hand antisepsis on preventing surgical site infections (SSIs) in patients treated in any setting. The secondary objective is to determine the effects of surgical hand antisepsis on the numbers of colony-forming units (CFUs) of bacteria on the hands of the surgical team.


SEARCH METHODS
In June 2015 for this update, we searched: The Cochrane Wounds Group Specialized Register; The Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library); Ovid MEDLINE; Ovid MEDLINE (In-Process & Other Non-Indexed Citations) and EBSCO CINAHL. There were no restrictions with respect to language, date of publication or study setting.


SELECTION CRITERIA
Randomised controlled trials comparing surgical hand antisepsis of varying duration, methods and antiseptic solutions.


DATA COLLECTION AND ANALYSIS
Three authors independently assessed studies for inclusion and trial quality and extracted data.


MAIN RESULTS
Fourteen trials were included in the updated review. Four trials reported the primary outcome, rates of SSIs, while 10 trials reported number of CFUs but not SSI rates. In general studies were small, and some did not present data or analyses that could be easily interpreted or related to clinical outcomes. These factors reduced the quality of the evidence. SSIsOne study randomised 3317 participants to basic hand hygiene (soap and water) versus an alcohol rub plus additional hydrogen peroxide. There was no clear evidence of a difference in the risk of SSI (risk ratio (RR) 0.97, 95% CI 0.77 to 1.23, moderate quality evidence downgraded for imprecision).One study (500 participants) compared alcohol-only rub versus an aqueous scrub and found no clear evidence of a difference in the risk of SSI (RR 0.56, 95% CI 0.23 to 1.34, very low quality evidence downgraded for imprecision and risk of bias).One study (4387 participants) compared alcohol rubs with additional active ingredients versus aqueous scrubs and found no clear evidence of a difference in SSI (RR 1.02, 95% CI 0.70 to 1.48, low quality evidence downgraded for imprecision and risk of bias).One study (100 participants) compared an alcohol rub with an additional ingredient versus an aqueous scrub with a brush and found no evidence of a difference in SSI (RR 0.50, 95% CI 0.05 to 5.34, low quality evidence downgraded for imprecision). CFUsThe review presents results for a number of comparisons; key findings include the following.Four studies compared different aqueous scrubs in reducing CFUs on hands.Three studies found chlorhexidine gluconate scrubs resulted in fewer CFUs than povidone iodine scrubs immediately after scrubbing, 2 hours after the initial scrub and 2 hours after subsequent scrubbing. All evidence was low or very low quality, with downgrading typically for imprecision and indirectness of outcome. One trial comparing a chlorhexidine gluconate scrub versus a povidone iodine plus triclosan scrub found no clear evidence of a difference-this was very low quality evidence (downgraded for risk of bias, imprecision and indirectness of outcome).Four studies compared aqueous scrubs versus alcohol rubs containing additional active ingredients and reported CFUs. In three comparisons there was evidence of fewer CFUs after using alcohol rubs with additional active ingredients (moderate or very low quality evidence downgraded for imprecision and indirectness of outcome). Evidence from one study suggested that an aqueous scrub was more effective in reducing CFUs than an alcohol rub containing additional ingredients, but this was very low quality evidence downgraded for imprecision and indirectness of outcome.Evidence for the effectiveness of different scrub durations varied. Four studies compared the effect of different durations of scrubs and rubs on the number of CFUs on hands. There was evidence that a 3 minute scrub reduced the number of CFUs compared with a 2 minute scrub (very low quality evidence downgraded for imprecision and indirectness of outcome). Data on other comparisons were not consistent, and interpretation was difficult. All further evidence was low or very low quality (typically downgraded for imprecision and indirectness).One study compared the effectiveness of using nail brushes and nail picks under running water prior to a chlorhexidine scrub on the number of CFUs on hands. It was unclear whether there was a difference in the effectiveness of these different techniques in terms of the number of CFUs remaining on hands (very low quality evidence downgraded due to imprecision and indirectness).


AUTHORS' CONCLUSIONS
There is no firm evidence that one type of hand antisepsis is better than another in reducing SSIs. Chlorhexidine gluconate scrubs may reduce the number of CFUs on hands compared with povidone iodine scrubs; however, the clinical relevance of this surrogate outcome is unclear. Alcohol rubs with additional antiseptic ingredients may reduce CFUs compared with aqueous scrubs. With regard to duration of hand antisepsis, a 3 minute initial scrub reduced CFUs on the hand compared with a 2 minute scrub, but this was very low quality evidence, and findings about a longer initial scrub and subsequent scrub durations are not consistent. It is unclear whether nail picks and brushes have a differential impact on the number of CFUs remaining on the hand. Generally, almost all evidence available to inform decisions about hand antisepsis approaches that were explored here were informed by low or very low quality evidence.


Library
Trusted evidence. Informed decisions. Better health.

Cochrane Database of Systematic Reviews
One study (500 participants) compared alcohol-only rub versus an aqueous scrub and found no clear evidence of a di erence in the risk of SSI (RR 0.56, 95% CI 0.23 to 1.34, very low quality evidence downgraded for imprecision and risk of bias).
One study (4387 participants) compared alcohol rubs with additional active ingredients versus aqueous scrubs and found no clear evidence of a di erence in SSI (RR 1.02, 95% CI 0.70 to 1.48, low quality evidence downgraded for imprecision and risk of bias).
One study (100 participants) compared an alcohol rub with an additional ingredient versus an aqueous scrub with a brush and found no evidence of a di erence in SSI (RR 0.50, 95% CI 0.05 to 5.34, low quality evidence downgraded for imprecision).

CFUs
The review presents results for a number of comparisons; key findings include the following.
Four studies compared di erent aqueous scrubs in reducing CFUs on hands.Three studies found chlorhexidine gluconate scrubs resulted in fewer CFUs than povidone iodine scrubs immediately a er scrubbing, 2 hours a er the initial scrub and 2 hours a er subsequent scrubbing. All evidence was low or very low quality, with downgrading typically for imprecision and indirectness of outcome. One trial comparing a chlorhexidine gluconate scrub versus a povidone iodine plus triclosan scrub found no clear evidence of a di erence-this was very low quality evidence (downgraded for risk of bias, imprecision and indirectness of outcome).
Four studies compared aqueous scrubs versus alcohol rubs containing additional active ingredients and reported CFUs. In three comparisons there was evidence of fewer CFUs a er using alcohol rubs with additional active ingredients (moderate or very low quality evidence downgraded for imprecision and indirectness of outcome). Evidence from one study suggested that an aqueous scrub was more e ective in reducing CFUs than an alcohol rub containing additional ingredients, but this was very low quality evidence downgraded for imprecision and indirectness of outcome.
Evidence for the e ectiveness of di erent scrub durations varied. Four studies compared the e ect of di erent durations of scrubs and rubs on the number of CFUs on hands. There was evidence that a 3 minute scrub reduced the number of CFUs compared with a 2 minute scrub (very low quality evidence downgraded for imprecision and indirectness of outcome). Data on other comparisons were not consistent, and interpretation was di icult. All further evidence was low or very low quality (typically downgraded for imprecision and indirectness).
One study compared the e ectiveness of using nail brushes and nail picks under running water prior to a chlorhexidine scrub on the number of CFUs on hands. It was unclear whether there was a di erence in the e ectiveness of these di erent techniques in terms of the number of CFUs remaining on hands (very low quality evidence downgraded due to imprecision and indirectness).

Authors' conclusions
There is no firm evidence that one type of hand antisepsis is better than another in reducing SSIs. Chlorhexidine gluconate scrubs may reduce the number of CFUs on hands compared with povidone iodine scrubs; however, the clinical relevance of this surrogate outcome is unclear. Alcohol rubs with additional antiseptic ingredients may reduce CFUs compared with aqueous scrubs. With regard to duration of hand antisepsis, a 3 minute initial scrub reduced CFUs on the hand compared with a 2 minute scrub, but this was very low quality evidence, and findings about a longer initial scrub and subsequent scrub durations are not consistent. It is unclear whether nail picks and brushes have a di erential impact on the number of CFUs remaining on the hand. Generally, almost all evidence available to inform decisions about hand antisepsis approaches that were explored here were informed by low or very low quality evidence.

What are surgical site infections and who is at risk?
The inadvertent transfer of micro-organisms such as bacteria to a patient's wound site during surgery can result in a wound infection that is commonly called a surgical site infection (SSI). SSIs are one of the most common forms of health care-associated infections for surgical patients. Around 1 in 20 surgical patients develop an SSI in hospital, and this proportion rises when people go home. SSIs result in delayed wound healing, increased hospital stays, increased use of antibiotics, unnecessary pain and, in extreme cases, the death of the patient, so their prevention is a key aim for health services.

Why use hand antisepsis prior to surgery?
There are many di erent points in the care pathway where prevention of SSIs can take place. This includes antiseptic cleansing of the hands for those who are operating on the patient. Surgical hand antisepsis is the focus of this review. The two most common forms of hand antisepsis involve aqueous scrubs and alcohol rubs. Aqueous scrubs are water-based solutions containing antiseptic ingredients such as chlorhexidine gluconate or povidone iodine. Scrubbing involves wetting the hands and forearms with water, systematically applying an aqueous scrub solution using either hands or sponges, rinsing under running water and then repeating this process. Alcohol solutions containing additional active ingredients are used to perform an 'alcohol rub'. Surgical teams systematically apply the alcohol rub solutions to their hands and allow it to evaporate. Alcohol is e ective against a wide range of bacteria and other micro-organisms. Following hand Library Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews antisepsis, operating sta then put on gloves, which provide an important barrier between operating sta and the patient; however, because gloves can become perforated during surgery, it is necessary to have hands as germ-free as possible.

What we found
In June 2015 we searched for as many relevant studies that had a robust design (randomised controlled trials) as we could find and compared di erent types of hand antisepsis before surgery. We included 14 studies that compared a range of methods for performing surgical hand antisepsis. The two measures used to assess the e ectiveness of treatments were the number of cases of SSIs in patients (presented in four included studies) and the number of viable bacteria or fungal cells (known as colony-forming units, or CFUs) on the hand of the person operating before surgery and a er surgery (which is a way of counting the bacteria present on the skin surface). It is not clear whether the method of hand antisepsis influences the risk of SSI, as most of the studies were too small and had flaws. There was some evidence that hand antisepsis with chlorhexidine may reduce the number of bacteria on the hands of health professionals compared with povidone iodine. Importantly, we do not know what the number of CFUs on the hands tells us about the likelihood of patients developing SSIs. There was also some evidence that alcohol rubs with additional antiseptic ingredients may reduce CFUs compared with aqueous scrubs.
Up-to-date June 2015 Library Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews

B A C K G R O U N D
The inadvertent transfer of micro-organisms to patients' wound sites during surgery can result in postoperative surgical site infections (SSIs). SSIs are one of the most common forms of healthcare-associated infections for surgical patients (NICE 2008). Around 5% of surgical patients develop an SSI (NICE 2008), though this incidence can double when surveillance includes active postdischarge follow-up (Leaper 2015). SSIs result in delayed wound healing, increased hospital stays, increased use of antibiotics, unnecessary pain and, in extreme cases, the death of the patient (Plowman 2000).
Micro-organisms that cause SSIs come from a variety of sources within the operating room, including the hands of the surgical team. Members of the surgical team wear sterile gloves to prevent transferring bacteria from their hands to patients. However, gloves can become perforated during surgery, so it is necessary to have hands as germ-free as possible. This is achieved by conducting surgical hand antisepsis immediately before donning sterile gloves prior to commencing surgical or invasive procedures. While handwashing removes transient micro-organisms, surgical hand antisepsis goes a step further to inhibit the growth of resident microorganisms, thereby minimising the risk of a patient developing an SSI (WHO 2009).This is achieved using antiseptic agents that kill and inhibit bacteria, fungi, protozoa and bacterial spores. An ideal antiseptic agent would be fast-acting, persistent (e ective for a number of hours), cumulative (repeated exposure inhibits bacterial growth for a number of days), have a broad spectrum of activity and be safe to use.There are several individual components of surgical hand antisepsis, including the pre-wash; the application technique; the use of sponges, brushes or nail picks; the choice of antiseptic solution and the duration of the antisepsis.

Defining terms -scrub and rubs
Several di erent terms are used when describing surgical hand antisepsis. Antisepsis with running water and an aqueous solution is referred to as a surgical or a traditional scrub. Antisepsis with an alcohol solution is referred to as an alcohol rub or a waterless scrub. In this review, we understand surgical hand antisepsis to encompass both methods of surgical antisepsis: scrubbing and rubbing. The very first antisepsis of the day is referred to as the initial antisepsis. Scrubs or rubs performed therea er but on the same day are referred to as subsequent antisepses.

Surgical hand antisepsis -current practice
The Association for Perioperative Practice (AfPP) recommends a pre-wash prior to the first antisepsis of the day, when hands are washed with soap or an antimicrobial solution under running water (AfPP 2011). The function of the pre-wash is to remove dirt (organic material). AfPP 2011 then recommends cleaning nails using a pick under running water. Clinicians can then perform antisepsis using either an antimicrobial solution with running water, referred to as a traditional scrub, or an alcoholic rub without water. AfPP 2011 suggests alcohol rubs are more e ective in reducing bacteria on the skin but should not be used if there is visible dirt present. The AfPP does not cite any specific antimicrobial solution as being the most e ective, but, like many other organisations, recommends that the solution chosen meets the ideal properties for an antimicrobial solution (ACORN 2012;AORN 2010;WHO 2009). These properties are identified by the Centers for Disease Control and Prevention (CDC) as the solution being: • fast-acting; • persistent (e ective for a number of hours); • cumulative (repeated exposure inhibits bacterial growth for a number of days); • having a broad spectrum of activity; and • safe to use (CDC 2002).
AfPP 2011 recommends a duration of 2 to 5 minutes (depending on manufacturers instructions) for a traditional scrub, but does not provide details on the recommended duration of an alcoholic rub. There is some discrepancy regarding aspects of hand antisepsis between di erent organisations. For example, the World Health Organization (WHO) and the Australian College of Operating Room Nurses (ACORN) only recommend a pre-wash if hands are visibly dirty (ACORN 2012;WHO 2009), and ACORN 2012 recommends that the first scrub of the day last 5 minutes while subsequent scrubs last 3 minutes. Nail brushes no longer appear to be recommended as these can damage skin (ACORN 2012;AfPP 2011;AORN 2010;WHO 2009).
Guidelines for surgical antisepsis also cover topics such as rings, artificial nails and nail polish (AORN 2010;ACORN 2012;AfPP 2011;HIS 2001;Mangram 1999). The impact of these factors on SSI is the focus of another Cochrane review (Arrowsmith 2014). There are concerns that hand antisepsis causes skin damage to sta hands and that some products are more abrasive than others (Larson 1986b). This topic is outside the remit of this review.

Surgical hand antisepsis solutions
Solutions for hand antisepsis are either aqueous (water) based or alcohol based.

Aqueous scrubs
Aqueous scrubs are water based solutions containing active ingredients that are used during traditional handscrubs. The most common solutions contain chlorhexidine gluconate or povidone iodine (see below). Scrubbing involves wetting the hands and forearms with water, systematically applying an aqueous scrub solution using either hands or sponges, rinsing under running water and then repeating the process.

Alcohol rubs
Alcohol-based solutions are used to perform an 'alcohol rub'. Health professionals apply the solution to dry hands and then rub them together systematically before allowing the solution to evaporate. Alcohol rubs do not require water for their application. Some alcohol rub solutions contain additional active antiseptic agents.

Antiseptic agents Alcohol
Alcohols have little or no residual e ect, and the concentration rather than the type of alcohol is thought to be most important in determining its e ectiveness (Larson 1995). Alcohol rubs are usually available in preparations of 60% to 90% strength and are e ective against a wide range of gram-positive and gram negative bacteria, mycobacterium tuberculosis, and many fungi and viruses. The three main alcohols used are ethanol, isopropanol Library Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews and n-propanol, and some rubs may contain a mixture of these. Compared with other common antiseptic products, alcohol is associated with the most rapid and greatest reduction in microbial counts (Lowbury 1974a), but it does not remove surface dirt as it does not contain surfactants or have a foaming action (Hobson 1998). Alcohol-based solutions usually (but not always) contain additional active ingredients to combine the rapid bacteriocidal e ect of alcohol with more persistent chemical activity.

Iodine and iodophors
Iodine has mostly been replaced by iodophors, as iodine o en causes irritation and discolouring of skin. Iodophors are composed of elemental iodine, iodide or triiodide, and a polymer carrier of high molecular weight (WHO 2009). Combining iodine with various polymers increases the solubility of iodine, promotes sustained release of iodine and reduces skin irritation. Iodophors are e ective against a wide range of gram-positive and gram negative bacteria, mycobacterium tuberculosis, fungi and viruses (Joress 1962). Iodophors contain iodine with a carrier such as polyvinylpyrrolidone (PVP). PVP, also known as povidone, is a polymer that detoxifies and prolongs the activities of drugs. PVP prolongs the activity of iodine by releasing it slowly. A combination of PVP and iodine, known as povidone iodine (PI), is less irritating than earlier solutions of iodine tincture (Joress 1962). Iodophors rapidly reduce transient and colonising bacteria but have little or no residual e ect (Larson 1990).

Chlorhexidine
Chlorhexidine is a biguanide. It is e ective against a wide range of gram-positive and gram-negative bacteria, lipophilic viruses and yeasts (Hibbard 2002a). It is not sporicidal. Although its immediate antimicrobial activity is slower than that of alcohols, it is more persistent because it binds to the outermost layer of skin, the stratum corneum (Larson 1990). Over time, repeated exposure can lead to a cumulative e ect where both transient and resident organisms are reduced (Larson 1990). Chlorhexidine gluconate is e ective in the presence of blood and other protein-rich biological materials (Hibbard 2002a).

Quaternary ammonium compounds
Quaternary ammonium compounds (QACs) are composed of a nitrogen atom linked to four alkyl groups. Alkyl benzalkonium chlorides are the most widely used as antiseptics, though other compounds include benzethonium chloride, cetrimide and cetylpyridium chloride. QACs are primarily bacteriostatic and fungistatic, although they are microbicidal against some organisms at high concentrations.They are more active against gram-positive bacteria than against gram-negative bacilli. QACs have relatively weak activity against mycobacteria and fungi and greater activity against lipophilic viruses. Their antimicrobial activity is adversely a ected by the presence of organic material, and they are not compatible with anionic detergents. In 1994, the Food and Drug Administration (FDA) Tentative Final Monograph (TFM) tentatively classified benzalkonium chloride and benzethonium chloride as having insu icient data to classify as safe and e ective for use as an antiseptic handwash (WHO 2009).

Hexachlorophene
Hexachlorophene is a halophenol compound. It is a slow-acting antiseptic that forms a film over the skin (Crowder 1967). The film retains bacteriostatic properties and is e ective against grampositive bacteria but is less e ective with gram-negative bacteria and fungi (Crowder 1967). A report of toxicity in neonates led to restricted usage (Kimborough 1973), and today, hexachlorophene has mostly been replaced by triclosan.

Chloroxylenol
Chloroxylenol, also known as para-chloro-meta-xylenol (PCMX), is a halogen-substituted phenolic compound. It is not as quickacting as chlorhexidine or iodophors, and its residual activity is less pronounced than that observed with chlorhexidine gluconate (McDonnell 1999). In 1994, the FDA TFM tentatively classified chloroxylenol as having insu icient data to classify as safe and e ective (WHO 2009).

O B J E C T I V E S
To assess the e ects of surgical hand antisepsis on preventing surgical site infections (SSIs) in patients treated in any setting. The secondary objective is to determine the e ects of surgical hand antisepsis on the number of bacteria colony-forming units (CFUs) present on the hands of the surgical team.

Types of studies
All published and unpublished randomised controlled trials (RCTs) of surgical hand antiseptic techniques were included. Controlled clinical trials were to be considered in the absence of RCTs. Two possible units of randomisation were considered: the scrub team or individual members of the scrub team.

Types of participants
All members of the scrub team or personnel working within the operating theatre or day case setting. The SSI outcome is measured in participants who have undergone surgery.

Types of interventions
This review included comparisons of the following with each other and/or placebo and/or no antisepsis: • Surgical hand antisepsis; • Aqueous scrub solutions.
• Alcohol rubs containing additional active ingredients.
• Surgical hand antisepsis of di erent durations.

Library
Trusted evidence. Informed decisions. Better health.

Primary outcomes
Occurrence of postoperative SSI, as defined by the CDC (Mangram 1999) or the study authors. We did not di erentiate between superficial and deep-incisional infection.

Secondary outcomes
Number of bacterial CFUs found on the hands of the surgical team.

Search methods for identification of studies Electronic searches
We describe the search methods of the original version and first update of this review in Appendix 1.
For this first update we searched: The

Searching other resources
We handsearched the bibliographies of all retrieved and relevant publications identified by these strategies for further studies.

Selection of studies
For the original review, three authors independently assessed the titles and abstracts of potentially relevant studies identified through the search strategy, retrieving the full text of all studies that potentially met the criteria. If it was unclear from the title or abstract whether a study met the criteria or there was a disagreement over the eligibility, we retrieved the full text of the study. The three authors then decided independently whether or not to include the studies. There were no disagreements among authors regarding which studies to include.
For the update, two authors independently assessed all titles and abstracts using the same methods, seeking assistance from translators where necessary. Again, there were no disagreements among authors about which studies to include.

Data extraction and management
We piloted a standardised data extraction form, and two authors independently used the finalised version to extract the following data from studies:

Assessment of risk of bias in included studies
Two review authors independently assessed the included studies using the Cochrane tool for assessing risk of bias (Higgins 2011). This tool addresses six specific domains: sequence generation, allocation concealment, blinding, incomplete data, selective outcome reporting and other issues (Appendix 5). In this review we recorded issues with unit of analysis, for example where a cluster trial had been undertaken but analysed at the individual level in the study report. We assessed blinding of outcome assessment and completeness of outcome data for each of the review outcomes separately. For this review, we anticipated that blinding of participants (surgical sta ) may not be possible. For this reason the assessment of the risk of detection bias focused on whether trials reported blinded outcome assessment (because wound infection can be a subjective outcome, it can be at high risk of measurement bias when outcome assessment is not blinded).
We presented our assessment of risk of bias using two 'Risk of bias' summary figures; one which is a summary of bias for each item across all studies, and a second which shows a cross-tabulation of each trial by all of the risk of bias items. We summarised a study's risk of selection bias, detection bias, attrition bias, reporting bias and other bias.

Data synthesis
Data were entered into Review Manager so ware (RevMan 2014). Continuous outcomes (i.e. CFUs) were reported as mean di erences (MDs) with 95% confidence intervals (CI). Dichotomous outcomes (i.e. SSIs) were presented as risks ratio (RR) with 95% CI. We reported findings narratively and considered pooling of data a er exploring clinical and statistical heterogeneity. We examined clinical heterogeneity by looking at the type of intervention, the participant population and the type of surgery. For assessment of statistical and related heterogeneity we used I 2 values (Higgins 2003). I 2 examines the percentage of total variation across RCTs that is due to heterogeneity rather than chance (Higgins 2003). Very broadly, we considered that I 2 values of 25%, or less, may mean a low level of heterogeneity Higgins 2003, and values of more than 75%, or more, indicate very high heterogeneity.

Handling of data where the appropriateness of the analysis reported in the paper was unclear.
Where the trial had a cross-over design or was cluster-randomised but the analysis did not appear to take this into account, we reported the available raw data (e.g. mean values) as well as the e ect estimate calculated in the paper and discussed the likely e ect of an incorrect analysis on the e ect estimate.

'Summary of findings' tables
In the update, in line with current Cochrane methods, we planned to present the main results of the review in 'Summary of findings' tables where we had pooled data. These tables present key information concerning the quality of the evidence, the magnitude of the e ects of the interventions examined and the sum of available data for the main outcomes (Schünemann 2011a). The 'Summary of findings' tables also include an overall grading of the evidence related to each of the main outcomes using the GRADE (Grades of Recommendation, Assessment, Development and Evaluation) approach. The GRADE approach defines the quality of a body of evidence as the extent to which one can be confident that an estimate of e ect or association is close to the true quantity of specific interest. The quality of a body of evidence involves consideration of within-trial risk of bias (methodological quality), directness of evidence, heterogeneity, precision of e ect estimates and risk of publication bias (Schünemann 2011b). We planned to present the following outcomes in the 'Summary of findings' tables for each comparison.

• SSI events • Number of CFUs
Where we did not pool data, we decided to conduct the GRADE assessment for each comparison and present this narratively within the Results section without the presentation of separate 'Summary of findings' tables.

Description of studies
Also see Characteristics of included studies and Characteristics of excluded studies.

Results of the search
The search for this update took place in June 2015 and yielded 274 abstracts. We obtained 18 of these as full-text records for further assessment, subsequently excluding 14 (see Excluded studies) and including four (Al-Naami 2009;Nthumba 2010;Tanner 2009;Vergara-Fernandez 2010). The addition of these four new studies to the 10 studies in the previous version of the review brought the total number of studies included in this update to 14.

Included studies
We present an overview of included studies and comparisons in Table 1.

Cochrane Database of Systematic Reviews
We identified and included 14 eligible trials in this review. Four trials reported the primary outcome, namely SSI (Al-Naami 2009;Nthumba 2010;Parienti 2002;Vergara-Fernandez 2010). The remaining 10 trials reported the number of CFUs (on the hands of the surgical team), which is a surrogate outcome that is thought to give an impression of the likelihood of infection. All 14 trials took place in operating departments, with 8 studies involving surgery and 6 testing interventions in surgical sta but without surgery taking place. Of the included studies, four compared di erent durations of scrubs or rubs (Kappstein 1993;Pereira 1990;Pereira 1997;Wheelock 1997).
Some of the included studies had complicated designs: Parienti 2002 was an equivalence, cluster, cross-over trial where the unit of randomisation was the surgical service. Each surgical service carried out one intervention for one month and then switched to the alternative intervention the following month. Nthumba 2010 also used a cluster, cross-over design, whereby operating theatres were allocated to an intervention with cross-over every two months. We also considered Hajipour 2006 to be a cluster trial, as it randomised four surgeons to di erent antisepsis methods which they used prior to surgery on multiple participants. Seven other studies (Gupta 2007, Herruzo 2000, Kappstein 1993, Pereira 1990Pereira 1997;Pietsch 2001;Wheelock 1997) were also crossover trials.

Risk of bias in included studies
We summarise our 'Risk of bias' assessment in Figure 1, Figure 2 and in the 'Risk of bias' assessment tables. Overall, most studies were at unclear risk or high risk of bias for one of the following: selection bias, detection bias and attrition bias. Only two studies were at low risk of bias for all these (Nthumba 2010; Tanner 2009).  Cochrane Database of Systematic Reviews

Figure 2. (Continued)
We considered four studies to be at low risk of detection bias, as they blinded outcome assessors (Al-Naami 2009;Nthumba 2010;Hajipour 2006;Tanner 2009). We considered two studies to be at high risk of detection bias (Parienti 2002;Wheelock 1997).
We also judged two studies to be at high risk of attrition bias. Pereira 1997 reported that 9/32 randomised members of sta failed to complete with no reasons given for withdrawal. Al-Naami 2009 excluded 100 of 600 patients from the analysis for reasons such as their condition was revised on subsequent histopathological examination, they had incomplete forms, or they "failed follow-up".
We judged several studies to be at unclear risk of other bias due to uncertainty as to whether a cluster-randomised or cross-over design had been taken into account in the analysis. Parienti 2002 and Hajipour 2006 were at unclear risk of other bias as a cluster design was detected but it did not seem that clustering had been taken into account in the analyses. Herruzo 2000, Kappstein 1993, Pereira 1990, Pereira 1997Wheelock 1997 were at unclear risk because it appeared that the crossover designs had not been taken into consideration in the analyses.

E ects of interventions
We included 14 trials in this review. In total, the trials evaluated nine basic comparisons related to the type (i.e. scrub or rub, active ingredients), duration and tools used for surgical hand antisepsis. Basic hand hygiene (soap and water) compared with 75% isopropyl alcohol plus 0.125% hydrogen peroxide Nthumba 2010 compared soap and water with an alcohol rub which contained hydrogen peroxide as an additional active ingredient. Surgeons in both groups scrubbed with soap and water for 4 to 5 minutes before the first procedure of the day and subsequently if there was visible soiling. For subsequent procedures, surgeons were randomised in clusters based on operating theatre to soap and water or 7 to 10 ml of a locally produced hand rub based on isopropyl alcohol with hydrogen peroxide, which they applied for 3 minutes and kept wet. The trial used a total of 10 clusters, each defined by six operating theatres, across five two-month intervals, with cross-over a er each two-month period. The trial included 3317 patients undergoing clean and clean-contaminated surgery and assessed SSI at 30 days using modified CDC definitions. There appeared to be a low risk of bias for all domains except the blinding of participants and personnel, which would have been di icult to achieve. Trialists accounted for both the clustering and cross-over in the power calculation and in the analyses.

Surgical site infection (SSI)
Nthumba 2010 collected SSI data for 30 days a er discharge. There was no clear di erence in the number of SSIs between groups. In total, 8% (128/1596) of participants developed SSI in the soap and water scrub group compared with 8.3% (127/1537) in the alcohol rub group (RR 0.97, 95% CI 0.77 to 1.23). This was based on a complete case analysis; data were unavailable for 5.5% (184/3317) of participants and were not imputed. Losses to follow-up were comparable between the two arms, at 5.1% (86/1682) for soap and water scrub versus 6.0% (98/1635) for alcohol rub (Analysis 1.1).

Number of colony forming units (CFUs)
Not reported

Comparison 1 Summary: Basic hand hygiene compared to alcohol rubs containing additional active ingredients
It is not clear whether hand antisepsis with soap and water is more or less e ective in preventing subsequent SSI than antisepsis with an alcohol rub containing hydrogen peroxide (moderate quality evidence).

Comparison 2: di erent aqueous scrub solutions: chlorhexidine gluconate versus povidone iodine (4 trials)
Four studies compared chlorhexidine gluconate versus povidone iodine but used di erent regimens.
Pereira 1990 randomly assigned 34 participants (operating room nurses) to one of four groups. The four interventions (which were used for 1 week) were 4% chlorhexidine gluconate (Hibiclens) or 7.5% povidone iodine (Betadine), using a 5 minute initial and 3 minute subsequent scrub; and 4% chlorhexidine gluconate Cochrane Database of Systematic Reviews (Hibiclens) or 7.5% povidone iodine (Betadine) using a 3 minute initial and 30 second subsequent scrub. Control of the order of interventions was through a Latin square design. Investigators took hand bacterial samples immediately a er the initial scrub, 2 hours a er the initial scrub and 2 hours a er the subsequent scrub. Although the study had a cross-over design, it did not appear that this was taken into consideration in the analysis, thus 95% CIs may be overestimated.
Furukawa 2005 compared 4% chlorhexidine gluconate (Hibiscrub) with 7.5% povidone iodine (Isodine) using a 3 minute scrub. Twenty-two operating room nurses were randomised to one of the two intervention groups. Each nurse took part only once. The nurses did not take part in any actual surgery.
Herruzo 2000 randomised 154 members of surgical teams and compared three intervention groups: relevant to this comparison was a 3 minute scrub of either aqueous chlorhexidine gluconate 4% or aqueous povidone iodine 7.5%. The study had a cross-over design, which appears to have been accounted for. The study also reports repeated measures which does not appear to have been accounted for in the analysis.
Pereira 1997 compared 4% chlorhexidine gluconate (Hibiclens) with 5% povidone iodine plus 1% triclosan (Microshield PVP) using a 3 minute initial and 2.5 minute subsequent scrub. Twenty-three operating room nurses were randomised to carry out each of five interventions for one week each. The order of interventions was controlled through a Latin square design. Participants did not take part in any actual surgery. Investigators took hand bacterial samples immediately a er the first antisepsis, 2 hours a er the first antisepsis and 2 hours a er the subsequent antisepsis. Although the study had a cross-over design, it did not appear that trialists took this into consideration in the analysis, thus the 95% CI may be overestimated.
Because the analysis did not account for the e ects of paired data resulting from the cross-over design, the study may have overestimated the uncertainty of the e ect estimate; the correct 95% CIs for the estimate may be narrower than those reported here. However, the study also used a repeated measures design which the analysis did not account for; this could lead to an underestimation of the uncertainty. The interaction of these two factors makes the true confidence intervals unclear, so we have downgraded the evidence twice due to imprecision.

Very low quality evidence due to imprecision and indirectness of outcome. Downgraded twice for imprecision due to two analytical issues which mean precision estimates may change upon correct analysis of data, and downgraded twice for indirectness as CFU is a surrogate outcome and because the intervention was used in the absence of surgery being conducted. A GRADE assessment of very low quality evidence means any estimate of e ect is very uncertain.
Pereira 1990 compared 4% chlorhexidine gluconate (Hibiclens) with 7.5% povidone iodine (Betadine) using a 3 minute initial and 30 second subsequent scrub. There was evidence that chlorhexidine may be more e ective than povidone iodine in reducing the number of CFUs immediately post scrubbing (MD − 0.17, 95% CI − 0.28 to − 0.06), 2 hours a er the initial scrub (MD − 0.41, 95% CI − 0.72 to − 0.10) and 2 hours a er the subsequent scrub (MD − 0.65, 95% CI − 0.93 to − 0.37 Analysis 2.1).

Very low quality evidence due to imprecision and indirectness of outcome (as above).
In Furukawa 2005, there were fewer CFUs in the chlorhexidine gluconate group a er scrubbing (MD − 2.40, 95% CI − 3.26 to − 1.54; Analysis 2.1).

Low quality evidence: downgraded twice for indirectness as CFU is a surrogate outcome and because the intervention was used in the absence of surgery being conducted.
In Herruzo 2000, there was evidence that a 3 minute aqueous scrub using chlorhexidine gluconate was more e ective in reducing CFUs on hands than a 3 minute aqueous scrub using povidone iodine, both immediately a er antisepsis (MD − 48.00, 95% CI − 50.57 to − 45.4) and at the end of a surgical procedure (MD − 132.0, 95% CI − 141.20 to − 122.80; Analysis 2.1). Because the analysis did not account for the repeated measures, the study may have underestimated uncertainty around the e ect estimate: the true confidence intervals for the estimate may be wider than those reported here.

Low quality evidence; downgraded once as precision estimates may change upon correct analysis of data and downgraded once due to indirectness as CFU is a surrogate outcome.
We considered pooling the three studies (four comparisons) (Furukawa 2005;Herruzo 2000;Pereira 1990); however, in light of the high degree of statistical heterogeneity for each possible comparison (ranging from 100% to 56%) and the di erences in interventions noted, we did not undertake this.
Because the analysis did not account for the e ects of paired data resulting from the cross-over design or for the e ect of using repeated measures, the true confidence intervals for the estimate are uncertain.

Library
Trusted evidence. Informed decisions. Better health.

Comparison 2 Summary : Comparison of di erent aqueous scrub solutions -chlorhexidine gluconate compared with povidone iodine
Data from four trials (five comparisons) of chlorhexidine-containing aqueous scrub solutions with povidone iodine containing solutions (all having initial longer duration of use followed by shorter subsequent use) suggest that chlorhexidine containing agents may reduce the numbers of CFU on the hands to a greater extent than povidone iodine containing solutions. However, overall this is low or very low quality evidence, and the number of CFUs is a surrogate outcome for SSI. Some of the studies included appeared to have been incorrectly analysed and one was at high risk of attrition bias.
No trials reported SSI events so there is no evidence to link the number of CFUs to clinical outcomes.

Comparison 3: comparison of di erent alcohol rubs containing additional active ingredients (2 trials)
Two trials compared alcohol rubs containing additional active ingredients (Gupta 2007;Pereira 1997).
Gupta 2007 compared three 2 ml aliquots of 1% chlorhexidine gluconate in 61% ethyl alcohol (Avagard) against a 3 minute application of zinc pyrithione in 70% ethyl alcohol (Triseptin). The 61% alcohol solution is a waterless product, and the 70% alcohol solution is a water aided product which requires rinsing with water. Eighteen operating room sta used each product for five consecutive days. Testing was carried out immediately before and a er antisepsis on day one, and at the end of days two and five.
Pereira 1997 compared 0.5% chlorhexidine gluconate in isopropanol compared with 0.5% chlorhexidine gluconate in ethanol. The alcohol rubs were used immediately a er an aqueous scrub (with 4% chlorhexidine gluconate) and also as the subsequent antiseptic agent. The active ingredient in both alcohol rubs was the same (i.e. 0.5% chlorhexidine gluconate), and both preparations had 70% strength alcohol, the only di erence being the alcohol (isopropanol versus ethanol). Scrubs lasted for 2 minutes, and the initial and subsequent applications of alcohol rubs lasted for 30 seconds.

Number of colony forming units (CFUs)
Pereira 1997 did not find a di erence in the number of CFUs between the isopropanol-and ethanol-based rubs immediately a er the first antisepsis (MD 0.00, 95% CI − 0.57 to 0.57), 2 hours a er the first antisepsis (MD 0.07, 95% CI -0.45 to 0.59) or 2 hours a er the subsequent antisepsis (MD 0.11, 95% CI − 0.49 to 0.71); see Analysis 4.1. Di erences between groups were consistently small and imprecise. As previously noted, the true confidence intervals for the e ect estimates is uncertain because of the issues with the analysis.

Very low quality evidence due to risk of bias of attrition bias; imprecision and indirectness of outcome.
Gupta 2007 did not present su icient raw data in the trial report to allow us to conduct independent statistical analyses, so we have contacted the author for further information. In the interim, we present Gupta 2007's own analysis. Although the trial used a crossover design, we could not determine if this was accounted for in the analysis. When the CFUs were compared over the duration of the study, Gupta 2007 found no statistically significant di erence between the solutions (P = 0.21). It must be noted that this analysis has not been independently verified.

Comparison 3 Summary: Comparison of di erent alcohol rubs containing additional active ingredients.
The comparative e ects of di erent alcohol rubs (each containing additional active ingredients) on number of CFUs are unclear, as the existing evidence is very sparse and of very low quality.

Comparison 4: aqueous scrubs versus alcohol-only rubs (1 trial)
Al-Naami 2009 compared a traditional 3 to 5 minute scrub with either chlorhexidine or povidone iodine versus 10 ml of a 62% ethyl alcohol handrub, which was allowed to dry. Surgeons in both groups scrubbed with the aqueous solution for the first procedure of the day. The study was described as an equivalence trial. Six hundred patients undergoing clean and clean-contaminated surgery were randomised to the two arms; investigators reported data for 500 patients. Reasons for participants being excluded from the analysis included a revised assessment of their condition on subsequent histopathological examination, incomplete forms or failed follow-up. More patients in the traditional scrub arm (24%, 72/300) were excluded from the analysis than in the alcohol rub arm (9%, 28/300).

Surgical site infection (SSI)
Al-Naami 2009 collected SSI data over 30 days from surgery. There was no clear evidence of a di erence in number of SSIs, with 2.9% (8/272) of participants in the alcohol rub arm having an SSI compared with 5.2% (12/228) in the aqueous scrub group (RR 0.56, 95% CI 0.23 to 1.34) (Analysis 5.1). This trial was considered to be at high risk of attrition bias.

Number of colony forming units
Not reported

Comparison 4 Summary: Aqueous scrubs compared with alcohol only rubs
It is unclear whether there is a di erence in SSIs between aqueous handscrubs and alcohol-only rubs (very low quality evidence from one study).

Library
Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews

Comparison 5: aqueous scrubs versus alcohol rubs containing additional active ingredients (6 trials)
Six studies compared traditional scrubs with alcohol rubs containing additional active ingredients (Gupta 2007;Hajipour 2006;Herruzo 2000;Parienti 2002;Pietsch 2001;Vergara-Fernandez 2010). The six trials used di erent antiseptic solutions, therefore it was not appropriate to perform a meta-analysis. Each trial is considered separately.
Parienti 2002 compared a 5 minute scrub using either 4% povidone iodine (Betadine) or 4% chlorhexidine gluconate (Hibiscrub) versus a 5 minute handrub using 75% propanol-1, propanol-2 with mecetronium ethylsulphate (Sterillium). Participants in the aqueous scrub group could choose between chlorhexidine gluconate or povidone iodine solutions. Participants in the handrubbing group carried out a single handwash for 1 minute with non-antiseptic soap at the start of each day. The entire scrub team in each of six hospitals took part. The trial included 4387 consecutive patients undergoing clean and clean-contaminated surgery and assessed SSI at 30 d using the CDC definition. The study was an equivalence, cluster cross-over trial but did not appear to have accounted for the clustering in the analysis.
Vergara-Fernandez 2010 compared an aqueous scrub with 4% chlorhexidine gluconate (which involved the use of a brush and a sponge) versus a handrub using 61% ethyl alcohol plus 1% chlorhexidine gluconate. Mean duration of the aqueous scrub was 3.9 minutes (SD 1.07), and mean duration of the alcohol rub was 2.0 minutes (SD 0.47). This trial took place in a single institution and involved 400 sta operating on 100 patients undergoing clean and clean-contaminated surgery, who were randomised to the two hand antisepsis groups. Investigators assessed SSI at one month using the CDC definition. Twenty per cent of the included sta had hand samples sent for microbiological examination, and samples were assessed as positive or negative for hand cultures.
Herruzo 2000 compared three intervention groups: chlorhexidine gluconate scrub versus povidone iodine scrub versus an alcohol rub with N-duopropenide. Each scrub or rub lasted 3 minutes. We successfully contacted Herruzo 2000 for additional information regarding sample size. 154 members of the surgical team were randomised for 55 operations. Investigators measured CFUs before antisepsis, immediately a er antisepsis and at the end of the surgical procedure.
Pietsch 2001 compared scrubbing using 4% chlorhexidine gluconate (Hibiscrub) with hand rubbing using an alcoholic solution of 45% propanol-2, 30% propanol-1 plus 0.2% ethylhexadecyldimethyl ammonium ethylsulphate (Sterillium). Seventy-five surgeons from one hospital participated in this randomised cross-over trial, using one product for four weeks then changing to the alternative product following a rest week. CFUs were measured before antisepsis, immediately a er antisepsis and a er the surgical procedure.
Hajipour 2006 compared a 3 minute 4% chlorhexidine gluconate scrub versus a 3 minute chlorhexidine in alcohol rub (Hydrex). We contacted the trial authors, who provided additional study details. Following an aqueous chlorhexidine scrub at the start of each day, four surgeons were randomised to one or other intervention and were evaluated repeatedly in that condition. Testing was carried out using the finger press method at the end of each surgical procedure.
Gupta 2007 compared 7.5% povidone iodine aqueous scrub against two alcohol rubs: three 2 ml aliquots of 1% chlorhexidine gluconate in 61% ethyl alcohol (Avagard) and a 3 minute application of zinc pyrithione in 70% ethyl alcohol (Triseptin). The paper does not provide further details regarding the application of the products. Eighteen operating room sta used each of the three products for five consecutive days. Testing was carried out immediately before and a er antisepsis on day one, and at the end of days two and five.

Aqueous chlorhexidine gluconate compared with 61% ethyl alcohol plus 1% chlorhexidine gluconate
Vergara-Fernandez 2010 found no clear evidence of a di erence in SSI rates between groups. In total 2% of participants (1/50) had a SSI in the aqueous scrub group compared with 4% (2/50) in the alcohol handrub group (RR 0.50, 95% CI 0.05 to 5.34). The study was small and the resulting 95% CI intervals wide, ranging from a 95% reduction in risk of SSI to a 400% increased risk of SSI (Analysis 6.1).

Aqueous chlorhexidine gluconate versus N-duopropenide
Herruzo 2000 reported CFU data (log 10 ) a er antisepsis and a er surgery (Analysis 7.1). We were unable to produce an estimate of treatment e ect for the review. Using bivariate analysis, Herruzo 2000 reports that N-duopropenide is more e ective than chlorhexidine in reducing the number of CFUs on participants' hands immediately a er antisepsis (P value < 0.01) and at the end of a surgical procedure (P value < 0.01); the paper did not provide any further information on estimates.

Aqueous povidone iodine versus N-duopropenide
Herruzo 2000 reported CFU data (log 10 ) a er antisepsis and a er surgery (Analysis 8.1). We were unable to produce an estimate of treatment e ect for the review. Using bivariate analysis, Herruzo 2000 reports that N-duopropenide was statistically significantly more e ective than povidone iodine in reducing the number of CFUs on participants hands immediately a er antisepsis (P value <0.01) and at the end of a surgical procedure (P value <0.01).

Library
Trusted evidence. Informed decisions. Better health.

Very low quality evidence downgraded twice as precision estimates are not available and once due to indirectness of outcome.
Aqueous chlorhexidine gluconate versus 45% propanol-2, 30% propanol-1 plus 0.2% ethylhexadecyldimethyl ammonium ethylsulphate (Sterillium) Pietsch 2001 reported that rubbing using 45% propanol-2, 30% propanol-1 plus 0.2% ethylhexadecyldimethyl ammonium ethylsulphate (Sterillium) was more e ective in reducing CFUs on participants' hands than scrubbing using 4% chlorhexidine gluconate, both immediately a er antisepsis (MD − 1.27, 95% CI − 1.23 to − 1.31) and at the end of the surgical procedure (MD − 1.07, 95% CI − 1.03 to − 1.11); Analysis 9.1. This study had a cluster design, in which each surgeon constituted a cluster, giving two clusters in each trial arm. However, the analysis did not take this into consideration, so the reliability of the e ect estimate reported by the authors is uncertain and could be wider than reported.

Aqueous povidone iodine versus 61% ethyl alcohol and 70% ethyl alcohol
Gupta 2007 did not present su icient raw data for us to be able to conduct independent statistical analysis, so we contacted the author to request additional data. In the interim we present Gupta 2007's own analysis. When CFUs were compared collectively from all the sample times, Gupta 2007 reports 'no statistically significant di erence' between the solutions (P = 0.21). It must be noted that this analysis has not been independently verified and it is unclear if investigators adjusted this analysis to account for the cross-over design.

Comparison 5 summary: Aqueous scrubs compared with alcohol rubs containing additional active ingredients
It is unclear if there is a di erence in numbers of SSIs between aqueous scrubs and alcohol rubs. The CFU outcome data were varied with two studies finding in favour of the alcohol rubs (moderate and very low quality evidence), one favouring the scrub arm (very low quality evidence) and one study reporting a 'nonstatistically significant di erence' with no other data (very low quality evidence).
Wheelock 1997 randomised 25 operating room nurses and surgical technologists to either a 2 minute or a 3 minute scrub. A er carrying out the trial scrub, and following a one-week washout period in which they continued to undertake scrubbing as part of their usual work, the participants switched to the other intervention. Though the intention of the trial authors was for participants to use aqueous 4% chlorhexidine gluconate (Hibiclens), participants with a history of skin irritation (15/25 participants) used either 2% chlorhexidine gluconate or parachlorometaxylenol (PCMX). CFUs were measured 1 hour a er the surgical scrub.
Kappstein 1993 compared a five minute rub with a three minute rub using alcoholic disinfectant. The disinfectant is not identified. Both rubs followed 1 minute handwashes using soap and water. Twentyfour surgeons carried out each of three intervention groups once in a random order. Samples were taken before and immediately a er antisepsis.
Pereira 1990 compared a 5 minute initial and 3 minute subsequent scrub with a 3 minute initial and 30 second subsequent scrub using chlorhexidine gluconate. Thirty-four participants were randomly assigned to one of four groups, and each group was assigned to one of four interventions, each lasting one week.
Pereira 1990 also compared a 5 minute initial and 3 minute subsequent scrub with a 3 minute initial and 30 second subsequent scrub using povidone iodine.
Pereira 1997 compared a 5 minute initial and a 3.5 minute subsequent scrub with a 3 minute initial and a 2.5 minute subsequent scrub using 4% chlorhexidine gluconate. Twenty-three operating room nurses were randomised to carry out each of five interventions for one week each.

Surgical site infection (SSI)
Not reported

Three minute scrub versus two minute scrub
Wheelock 1997 presented paired data, which the review authors re-analysed but did not present. There were fewer CFUs on hands immediately a er a 3 minute scrub compared with a 2 minute scrub (MD − 0.29, 95% CI − 0.06 to − 0.52).

Very low quality evidence downgraded once due to risk of detection bias and twice due to indirectness, once due to indirectness of outcome and once because the intervention was used in the absence of surgery being conducted.
Five minute rub versus three minute rub (Analysis 11.1) Kappstein 1993 favoured the 3 minute scrub over the 5 minute scrub when assessed immediately a er antisepsis (MD 0.26, 95% CI 0.14 to 0.38; Analysis 11.1). Because the use of a cross-over design Cochrane Database of Systematic Reviews was not accounted for in the analysis the true confidence intervals may be narrower than those reported here.

Low quality evidence: downgraded once due to imprecision and once due to indirectness.
Five minute initial and three minute subsequent scrub versus three minute initial and 30 second subsequent scrub using chlorhexidine (Analysis 12.1) Pereira 1990 reported no clear di erence between groups in the number of CFUs immediately a er the initial scrub (MD − 0.19, 95% CI − 0.51 to 0.13) or 2 hours a er the initial scrub (MD − 0.23, 95% CI − 0.52 to 0.06). There was evidence of a di erence in CFUs favouring the 5 minute arm 2 hours a er the subsequent scrub (MD − 0.58, 95% CI − 0.92 to − 0.24); see Analysis 12.1.

Very low quality evidence down graded twice due to imprecision and once due to indirectness of outcome.
Five minute initial and three minute subsequent scrub compared with a three minute initial and 30 second subsequent scrub using povidone iodine (Analysis 13.1) In Pereira 1990, there was no clear di erence in the number of CFUs at any time point, whether immediately a er the initial scrub (MD 0.02, 95% CI − 0.28 to 0.32), 2 hours a er the initial scrub (MD 0.12, 95% CI − 0.21 to 0.45) or 2 hours a er the subsequent scrub (MD − 0.13, 95% CI − 0.37 to 0.11); see Analysis 13.1.

Very low quality evidence downgraded due to imprecision and indirectness of outcome.
Five minute initial and three and 30 second subsequent scrub compared with a three minute initial and two and a half minute subsequent scrub using chlorhexidine (Analysis 14.1) In Pereira 1997, there was no clear di erence in the number of CFUs at any time point reported, whether immediately a er the initial antisepsis (MD 0.08, 95% CI − 0.44 to 0.60), 2 hours a er the initial antisepsis (MD − 0.19, 95% CI − 0.78 to 0.40) or 2 hours a er subsequent antisepsis (MD − 0.17, 95% CI − 0.71 to 0.37); see Analysis 14.1.

Comparison 6 summary: duration of surgical antisepsis
Outcome data were only available for CFUs. One study reported evidence of fewer CFUs on hands a er using a 3 minute rather than 2 minute chlorhexidine scrub. Another study reported fewer CFUs a er a 3 minute alcohol rub compared with a 5 minute alcohol rub; evidence was low quality in both cases. One study reported that 3 minute subsequent scrubs with aqueous chlorhexidine (following initial scrubs) were more e ective in reducing the number of CFUs on hands than 30 second subsequent scrubs; this di erence was not observed with povidone iodine treatments used in the same way: estimates from this study was classed as being of very low quality. Other comparisons reported no clear di erences in number of CFUs.
Comparison 7: surgical hand antisepsis using a nail pick versus surgical hand antisepsis not using a nail pick (1 trial) One three-arm trial compared the e ect of surgical hand antisepsis using a nail pick with surgical hand antisepsis alone (Tanner 2009).
The study randomised 164 participants to one of three groups. All groups scrubbed with two measured doses of 2 ml aqueous chlorhexidine gluconate 4% (Hibiscrub) for 1 minute per dose; the total scrub time, which was observed and timed, was 2 minutes. One group performed only this surgical hand antisepsis. A second group used a disposable nail pick to clean their nails under running water before the hand antisepsis procedure. The third group used a disposable nail brush to clean their nails under running water before the hand antisepsis procedure. Participants then undertook circulating duties in the operating room for one hour but did not participate in any surgeries.

Surgical site infection
Not reported

Number of colony forming units (CFUs)
There was no clear evidence of a di erence between nail pick and no nail pick in the number of CFU detected a er one hour on the dominant hands of participants (MD 0.13, 95% CI − 0.14 to 0.40; Analysis 15.1).

Comparison 8 surgical hand antisepsis using a brush versus surgical hand antisepsis not using a brush (1 trial)
In the same three-arm trial (Tanner 2009) compared the e ect of surgical hand antisepsis and using a brush with surgical hand antisepsis alone. Participants were allocated to groups as described above. There was no clear evidence of a di erence between using and not using a brush during hand antisepsis on the number of CFUs (MD 0.24, 95% CI -0.04 to 0.51; Analysis 16.1).

Very low quality evidence; downgraded once due to imprecision and once due to indirectness of outcome and further again for indirectness as no surgery was performed.
Comparison 9: surgical hand antisepsis using a nail pick versus surgical hand antisepsis using a brush (1 trial) In the same three-arm trial (Tanner 2009) compared the e ect of surgical hand antisepsis using a nail pick with surgical hand antisepsis using a brush. There was no di erence in the number of CFUs detected a er 1 hour on the dominant hands of participants who used a brush before hand antisepsis compared with those who used a nail pick before hand antisepsis (MD 0.10, 95% CI -0.17 to 0.37; Analysis 17.1).

Summary of Comparisons 7 to 9: Surgical hand antisepsis using a nail pick and brush
There was no clear evidence of a di erence in CFUs when a nail pick was compared with a brush or with no pick or brush.

min initial and 3 min subsequent scrub versus a 3 min initial and 30 s subsequent scrub:
Chlorhexidine scrub: evidence of fewer CFUs in the longer scrub duration group after the subsequent scrubs (Pereira 1990) Povidone iodine scrub: no evidence of a difference in CFUs between groups at any stage (Pereira 1990) 5 min initial and 3.5 min subsequent scrub versus a 3 min initial and 2.5 min subsequent scrub using chlorhexidine: no evidence of a difference in CFUs at any stage (Pereira 1997) Cochrane Database of Systematic Reviews sus surgical hand antisepsis using a brush

Summary of main results
Fourteen trials met the inclusion criteria and were included in this review.
Only four trials reported risk of surgical site infection (SSI), the primary outcome for this review (Al-Naami 2009; Nthumba 2010; Parienti 2002; Vergara-Fernandez 2010). These results were inconclusive, and it is not clear whether any method of hand antisepsis is any more e ective than another in reducing subsequent SSI. Most of the evidence was of low or very low quality, mainly due to risk of bias and imprecision. Most of the studies were too small to detect a di erence in SSI rates even if one existed.
The remaining 10 trials measured the number of CFUs on hands.
Where trials were cluster and/or crossover trials they were o en not analysed correctly, which may impact on the precision of estimates.
In four comparisons of aqueous scrub solutions containing chlorhexidine versus those containing povidone iodine, (three studies) there was evidence that chlorhexidine-gluconate containing scrubs reduce numbers of CFU on the hands more than povidone iodine (Furukawa 2005;Herruzo 2000;Pereira 1990). The evidence was of low or very low quality. High levels of heterogeneity precluded meta-analysis.
Results of studies comparing alcohol rubs with aqueous scrubs were mixed. In three comparisons (two studies) alcohol rubs containing additional active ingredients were found to be more e ective than aqueous scrubs in reducing the number of CFUs on hands (moderate and very low quality evidence) (Herruzo 2000;Pietsch 2001). Gupta 2007 reported no di erence between povidone iodine scrubs and ethyl alcohol based rubs of 61% and 70% (very low quality evidence). Hajipour 2006 found chlorhexidine aqueous scrubs more e ective than chlorhexidine in alcohol rubs (very low quality evidence).
In terms of the duration of hand antisepsis, there was evidence that a 3 minute scrub reduced the number of CFUs compared to a 2 minute scrub (Wheelock 1997) (very low quality evidence). Data on other comparisons were not consistent, and interpretation is di icult.
One three-arm trial assessed the use of a disposable nail pick or disposable nail brush under running water prior to scrubbing with chlorhexidine and found no clear di erence in the number of CFUs (very low quality evidence) (Tanner 2009).
Nthumba 2010, which found no di erence in SSIs when comparing surgical hand antisepsis with basic hand hygiene (moderate quality evidence), has interesting implications for practice, which warrant further research.

Quality of the evidence
Surgical hand antisepsis is carried out to reduce the number of bacteria on the hands of the surgical team so that in the event of a breach in glove barrier protection, the risk of transferring infections to patients is reduced. Therefore, the most appropriate outcome measure for a study of surgical hand antisepsis is postoperative SSI. Only four studies used this measure (Al-Naami 2009; Nthumba 2010; Parienti 2002; Vergara-Fernandez 2010), while the remaining 10 trials measured CFUs. It is widely assumed that interventions reducing CFUs may impact on incidence of SSIs but there is little evidence to support this assertion. There was considerable variation in the timing of outcome assessment for measuring CFUs as well as some di erences in the methods used to obtain samples.
All of the studies included in this review were underpowered in terms of small sample sizes, a limited number of outcome events or both. RCTs need to be adequately powered so that they are able to detect treatment e ects of a specified size if they exist, and sample size calculations should be used to help estimate the number of people recruited to a trial. Evidence from limited numbers of participants and/or where limited numbers of events occur can also lead to confidence intervals that are fragile (that is not stable) which is also likely to have been a risk for all comparisons here.
A key issue with the included studies was the incorrect or unclear analysis and presentation of cross-over and cluster data. When these methods are used, correct planning, conduct and reporting are vital to maximise the value of the research. Cross-over studies involve an individual receiving more than one of the treatments of interest, with the randomisation consisting of the sequence in which these are received. The analysis requires statistical tests that take the paired nature of these data into account. If investigators do not use these approaches, they may overestimate standard errors, leading to confidence intervals which are less precise than they might be -that is, there is a risk of a type II error (concluding no significant di erence when there is one). Cluster trials, where allocation of treatment is made at a group level and the unit of assessment is at the individual level, require careful planning and analysis to avoid producing artificially small standard errors which can lead to type I errors -that is, finding a statistical significant di erence when one does not exist.

Potential biases in the review process
We attempted to overcome potential publication bias through rigorous searching. We had two studies translated, one from German and one from Spanish ( Cochrane Database of Systematic Reviews contrary. For the update we attempted to contact one author (Chen 2012); this trial is awaiting assessment.
Four trials acknowledged the role of commercial companies in supplying antiseptic products (Parienti 2002;Pereira 1990;Pereira 1997;Vergara-Fernandez 2010), and one trial was led by a research employee of a commercial company (Pietsch 2001). Tanner 2009 reported funding from a commercial company. As the lead author of this review was also an author on an included study (Tanner 2009), two other authors independently extracted data from it and dra ed related review text to minimise any bias.  1993;Pereira 1990;Pereira 1997;Pietsch 2001), and they excluded Parienti 2002 because it measured SSIs rather than CFUs. Hsieh 2006 draws two conclusions: that surgical handrubs using alcoholbased products are more e ective than 6 minute scrubs using 4% chlorhexidine, and that there is no evidence that a 2 minute scrub is more e ective than a 3 minute scrub using 4% chlorhexidine.

Agreements and disagreements with other studies or reviews
Guidelines for hand antisepsis have changed slightly since the first version of this Cochrane review was published. The main changes are the increasing status of alcohol rubs, which are now considered to have equal standing to traditional scrubs and are acceptable alternatives (AfPP 2011; ACORN 2012); and the use of brushes which has been downgraded from not necessary to not recommended (AfPP 2011; ACORN 2012).
The conclusions one can draw from the evidence in this review are limited for the following reasons. Ten of the 14 trials included in this review measured CFUs rather than SSIs (Furukawa 2005;Gupta 2007;Hajipour 2006;Herruzo 2000;Kappstein 1993;Pereira 1990;Pereira 1997;Pietsch 2001;Tanner 2009;Wheelock 1997). Two of these trials used the finger press method rather than the recognised standard glove juice test (Hajipour 2006;Herruzo 2000). Sample sizes are small in most of the trials. In three trials, participants in the scrub group had a choice of antiseptic solutions (Al-Naami 2009;Parienti 2002;Wheelock 1997), and details of the randomisation are poor in many of the trials.

Implications for practice
Uncertainty remains about the optimal method of hand antisepsis for minimising SSI outcomes: only four studies reported this outcome, and they were heterogeneous, at high risk of bias and mainly greatly underpowered. For the secondary (and surrogate) outcome of the number of CFUs on the hands, there is some generally low or very low quality evidence that quality chlorhexidine aqueous scrubs may reduce CFUs compared with povidone iodine aqueous scrubs and that an alcohol rub with added ingredient reduced the number of CFUs on the hand compared with aqueous scrubs (di erent types) immediately a er antisepsis and at the end of the surgical procedure. The evidence regarding the impact of using a nail pick or brush on CFUs is unclear.
In terms of duration of antisepsis, scrubbing for 3 minutes rather than 2 minutes may lead to fewer CFUs on the hand (low or very low quality evidence). Other data are more di icult to interpret.

Implications for research
Trials should attempt to measure the impact of hand antisepsis on SSIs. Investigators should also adhere to a recognised definition of SSI, which includes a 30 day follow-up. Trials should use adequate sample sizes based on a priori sample size calculations and take account of any data cross-over and/or clustering. The following trials might address important clinical questions -these and related uncertainties should be prioritised in conjunction with health professionals and policy makers. Future studies should be adequately designed and powered.
• Brushes, nail picks and sponges compared with no brushes, nail picks or sponges (outcome SSI).

Characteristics of included studies [ordered by study ID]
Methods Randomised controlled trial (described as an equivalence study) Generation of random number sequence: no details given Allocation concealment: no details given Blinding: blinding of outcome assessors; no other details given A priori sample calculations: yes Antisepsis protocol: yes (partial) Withdrawals: incomplete details given Intention-to-treat analysis: no Clear inclusion or exclusion criteria: no Participants 1 hospital General Surgery division; surgeons (no further information on personnel) 600 patients initially randomised from general surgery; data reported on 500 General Surgery: abdominal (e.g. cholecystectomy) and other; mixture of clean and clean-contaminated operations Interventions Group 1 -traditional surgeons' handscrub for 3-5 min using 7.5% povidone iodine (Betadine) or 4% chlorhexidine gluconate (Hibiscrub) (228 patients) Group 2 -As with group 1 for first case; subsequent antisepsis with alcohol handrub with 62% ethanol (Purrel) 10 ml, allowed to dry (272 patients)

Outcomes
Surgical site infection defined as any one or more of the following: symptoms and signs (pain, swelling, redness, hotness, tenderness, indurations, purulent discharge, opened wound) occurring within 30 d from surgery (examinations before discharge, at 1 week, at 1 month, and C/S results); no further detail supplied.

Notes
All patients had standardised skin preparation. 76% of patients had prophylactic antibiotics (indicated for specified surgeries); no difference between arms.

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias)

Unclear risk No information on sequence generation
Quote: "Participants were randomised to either a routine hand scrub or an alcohol hand-rub upon selecting a sealed envelope for each case." No further information. Low risk Assessment of wound appearance and swab by personnel unaware of allocation.
Quote: "Surgeons who examined surgical sites were unaware of the groups' allocation"; "A swab was sent for C/S from any suspected SSI. Health care personnel taking swabs and interpreting results of C/S were also unaware of how hand disinfection for each group had been allocated." Incomplete outcome data (attrition bias) All outcomes High risk 100 (1 in 6) participants originally randomised were excluded from analysis. The reasons for this are not fully explored.
Quote: "Initially an equal number of cases (300 patients in each group) were randomised to each method. However, more cases were further excluded from each group as they turned out to be non-eligible for inclusion after the original randomization (e.g. acute or chronic cholecystitis on histopathological examination), incomplete forms, failed follow-ups, etc." Selective reporting (reporting bias) Low risk Authors reported all specified objectives.
Quote: "The objective of this study is to determine the equal efficacy of alcohol-based hand-rub as compared to traditional surgical scrub in the prevention of SSI as the primary outcome measure; the compliance of surgical sta and skin tolerance as the secondary outcome measure; also keeping in mind cost effectiveness and the potential change in surgical practice at least in our institution"

Methods
Randomised controlled trial Generation of random number sequence: no details given

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "Twenty-two operating room nurses were randomly divided into two groups as follows: the PVI group (n = 11) and the CHG group (n = 11). All the nurses were examined for bacterial contamination of their hands before and after surgical handwashing".
Comment: Evidence of randomisation, however not enough evidence to suggest truly randomised sequence generation Allocation concealment (selection bias) Unclear risk Quote: "Twenty-two operating room nurses were randomly divided into two groups as follows: the PVI group (n = 11) and the CHG group (n = 11). All the nurses were examined for bacterial contamination of their hands before and after surgical handwashing".

Comment: No evidence that appropriate allocation concealment took place
Blinding (performance bias and detection bias) All outcomes -Blinding participants Unclear risk Quote: "All the nurses were examined for bacterial contamination of their hands before and after surgical handwashing"

Comment: No evidence of blinding of participants or personnel to blinding to intervention
Blinding (performance bias and detection bias) All outcomes -Blinding care givers Unclear risk Quote: "All the nurses were examined for bacterial contamination of their hands before and after surgical handwashing" Comment: No evidence of blinding of participants or personnel to blinding to intervention Blinding (performance bias and detection bias) All outcomes -Blinding outcome assessors Unclear risk Tap water outcome Quote: "We took water samples from 4 faucets just before hand washing and 1 ml of each sample was injected onto Brain-Heart infusion bouillon plats (Eiken K. K) and cultured at 37 Celsius for 48 hours for bacterial detection." Comment: Samples seemingly taken by personnel, however unclear as to the role bias would play in the quantitative study of bacterial colonies in water faucets.

Hands and fingers outcome
Quote: "The samples were collected and pre-treated according to the Glove Juice method. In detail, the sample liquid was taken from the right glove just before hand washing and from the le glove after hand washing". This was then "cultured at 37 Celcius for 48 hours, thereafter the number of bacterial colonies was counted".
Comment: Unclear as to whether those obtaining the samples were blinded to the intervention. It is likely that they were not blinded; however the overall judgement is unclear. Interventions Group 1 -brush application of 7.5% povidone iodine aqueous scrub Group 2 -three 2 ml application of 1% chlorhexidine gluconate in 61% ethyl alcohol Group 3 -3 min application of zinc pyrithione in 70% ethyl alcohol and rinsed with water Outcomes Outcome measure: CFUs on participants' hands Method of testing: glove juice method Timing of testing: before antisepsis and immediately after antisepsis on day 1, after 6 hours on days 2 and 5

Surgical hand antisepsis to reduce surgical site infection (Review)
Notes -

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: Participants "were assigned to at random to one of three groups. Each group used one of the three solutions for five consecutive days. The following week, each group used a different scrub solution, such that all participants used each product over the study duration" Comment: Unclear as to whether a random sequence generator was used.

Allocation concealment (selection bias)
Unclear risk Quote: Participants "were assigned to at random to one of three groups. Each group used one of the three solutions for five consecutive days. The following week, each group used a different scrub solution, such that all participants used each product over the study duration"

Library
Cochrane Database of Systematic Reviews Comment: Unclear as to whether allocation was concealed.

Blinding (performance bias and detection bias) All outcomes -Blinding participants
High risk Quote: "Participants could not be blinded to the three solutions due to differences in their nature and method of application".
Comment: Study describes that blinding of participants was not possible.There is no information given as to whether the investigators were blinded, but it is likely there were similarly unblinded to the intervention for the reasons given for the participants above. The judgement for participant blinding is therefore high risk.
Blinding (performance bias and detection bias) All outcomes -Blinding care givers Unclear risk Quote: "Participants could not be blinded to the three solutions due to differences in their nature and method of application".
Comment: Study describes that blinding of participants was not possible.There is no information given as to whether the investigators were blinded, but it is likely there were similarly unblinded to the intervention for the reasons given for the participants above. The judgement for blinding of caregivers here is unclear.

Blinding (performance bias and detection bias) All outcomes -Blinding outcome assessors
Unclear risk Bacterial sampling Quote: "The subject introduced their hands into this bag and the investigator massaged their hands externally with emphasis on web spaces and subungual areas" Comment: no attempt made at blinding method of obtaining bacterial sample, which would advise a high risk of bias decision

Microbial assay
Quote: "Samples were sent to the microbiology laboratory immediately after collection in a blinded manner" Comment: There was adequate evidence that approach microbial testing was blinded.
Therefore, overall assessment of the risk of bias for outcome assessment is unclear.
Incomplete outcome data (attrition bias) All outcomes Low risk Quote: "Eighteen surgical sta members participated in this study. All members used each of the three scrub solutions over the duration or the study . . . Two volunteers developed a skin rash and a burning sensation on their hands within and a few minutes after their first scrub with the ABWA product. They also reported a metallic taste in their mouth and palpitations. Although none of their symptoms were severe and resolved shortly thereafter without any medical intervention, they were removed from the study. All the remaining sta volunteers completed the study".
Comment: small loss to follow-up and full explanations given as to the reasons for dropout Selective reporting (reporting bias) Low risk No direct quotes, although the efficacy of the product in terms of reduction of log reduction in bacterial counts and product preference by participants are both included in the Results in full and comprehensive manner, as outlined in the Methods. All surgeons washed with chlorhexidine (no further detail) for 5 min for first procedure with thorough cleaning under fingernails.

Outcomes
Outcome measure: CFUs on participants' hands Method of testing: finger press testing with agar plates Timing of testing: at the end of the surgical procedure

Notes
The 4 surgeons, who were not blinded, were randomised once and tested 53 times

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "Orthopaedic surgeons were allocated to one of two different handwashing protocols using a randomisation table" Comment: evidence of random sequence generation, therefore judged as low risk Allocation concealment (selection bias) Unclear risk Quote: "[T]he surgeon was randomised to wash for 5 min with either chlorhexidine or alcohol gel" Comment: no evidence that there was an attempt at allocation concealment Blinding (performance bias and detection bias) All outcomes -Blinding participants Unclear risk Quote: "The hand-washing protocol dictated that all surgeons should wash for 5 min with chlorhexidine for their first case with thorough cleaning under the fingernails. Thereafter, the surgeon was randomised to wash for 5 min with either the chlorhexidine or alcohol gel. Alcohol was allowed to dry on the hands prior to double gloving".
Comment: no evidence that participants or personnel were blinded to intervention Blinding (performance bias and detection bias) All outcomes -Blinding care givers Unclear risk Quote: "The hand-washing protocol dictated that all surgeons should wash for 5 min with chlorhexidine for their first case with thorough cleaning under the fingernails. Thereafter, the surgeon was randomised to wash for 5 min with either the chlorhexidine or alcohol gel. Alcohol was allowed to dry on the hands prior to double gloving".

Other bias
Unclear risk This appears to be a clustered randomised trial; it does not seem that clustering was taken into account in the analysis.

Methods
Randomised cross-over controlled trial Generation of random number sequence: no details given Allocation concealment: no details given Blinding: no details given A priori sample calculations: no Antisepsis protocol: minimal details Withdrawals: no details given Intention-to-treat analysis: no Clear inclusion or exclusion criteria: no Participants 154 members of the surgical teams working in plastic surgery and traumatology Baseline comparability: baseline bacterial counts Interventions Group 1 -3 min scrub using aqueous chlorhexidine gluconate, n = 50 Group 2 -3 min scrub using aqueous povidone iodine, n = 49 Group 3 -3 min rub with N-duopropenide, n = 55 Outcomes Outcome measure: CFUs on participants' hands Method of testing: finger press testing with agar plates Timing of testing: before antisepsis, immediately after antisepsis and at the end of the surgical procedure

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "Healthy volunteers washed with one of the three products for 1 week for 3 consecutive weeks. The order of the washings was randomised" Comment: indication that a randomisation process was undertaken, however not clear how the sequence was generated Allocation concealment (selection bias) Unclear risk Quote: "Healthy volunteers washed with one of the three products for 1 week for 3 consecutive weeks. The order of the washings was randomised" Comment: no evidence of allocation concealment given Blinding (performance bias and detection bias) All outcomes -Blinding participants High risk Quote: "The effect of standard surgical washings with 7.5% iodine povidone or 4% chlorhexidine (both with scrubbing for 3 min following a standard technique) was compared with the effect of washing (without scrubbing with 2.3% N-duopropenide in 60 ispopropranol with dermoprotective substance). The latter solution was poured over the hands, which were then rubbed together, and when it began to dry, it was reapplied over 3 min." Comment: The study design here is a cross-over design, reducing potential effects of bias; however, there is no evidence that participants or personnel were blinded to the intervention given, and the conditions differed sufficiently that blinding would not have been possible.
Blinding (performance bias and detection bias) All outcomes -Blinding care givers High risk Quote: "The effect of standard surgical washings with 7.5% iodine povidone or 4% chlorhexidine (both with scrubbing for 3 min following a standard technique) was compared with the effect of washing (without scrubbing with 2.3% N-duopropenide in 60 isopropanol with dermoprotective substance. The latter solution was poured over the hands, which were then rubbed together, and when it began to dry, it was reapplied over 3 min." Comment: The study design here is a cross-over design, reducing potential effects of bias; however, there is no evidence that participants or personnel were blinded to the intervention given, and the conditions differed sufficiently that blinding would not have been possible. Blinding (performance bias and detection bias) All outcomes -Blinding outcome assessors Unclear risk Quote: "We used fingertips to sample hand bacteria as the standard European Norm (pr EN 1500) except we had not artificially contaminated the hands: five fingertips are rubbed for 1 min on a Petri dish containing 10 ml of TBS plus an antiseptic neutralize ". These were cultured for "48 h at 37 C, and the then the CFU/hand (the five fingertips), were counted and transformed into a decimal logarithm".
Comment: Despite the assessment being quantitative, it is not clear whether those who obtained the bacterial samples were independent of the study or blinded to the intervention.
Incomplete outcome data (attrition bias) All outcomes

Low risk No direct quotes, although no losses to follow-up recorded in Results
Comment: no obvious source of attrition bias Selective reporting (reporting bias)

Low risk No direct quotes
Comment: tabular format of results incorporates the assessment outcomes outlined (CFUs) between the groups, suggesting a low risk Other bias Unclear risk Although the trial had a cross-over design, it did not appear that this was reflected in the analysis. Interventions Group 1 -1 min wash with soap and water followed by 5 min rub with an alcoholic disinfectant Group 2 -1 min wash with soap and water followed by 3 min rub with an alcoholic disinfectant Group 3 -1 min was with chlorhexidine soap followed by two min of rubbing with 0.5% chlorhexidine in isopropanol

Outcomes
Outcome measure: CFUs on participants' hands Method of testing: glove juice method Timing of testing: before antisepsis and immediately after antisepsis

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Just states "random order" with no further information

Cochrane Database of Systematic Reviews
Other bias Unclear risk Although the trial had a cross-over design, it did not appear that this was reflected in the analysis.

Methods
Cluster-randomised controlled cross-over trial Generation of random number sequence: appropriate Allocation concealment: no details given but considered unlikely to be an issue in cluster-randomisa- Interventions Group 1 -plain soap and water: 4-5 min clean running water and plain soap; sterile cotton hand towel dry. 5 clusters (n = 1682 patients) Group 2 -As group 1 before first procedure of day and subsequently in case of visible soiling, then alcohol-based handrub (75% isopropyl alcohol, 1.45% glycerol, 0.125% hydrogen peroxide) for 3 min and kept wet (7-10 ml per preparation) Outcomes SSI (defined using modified US Centers for Disease Control and Prevention definitions for nosocomial infection) detected by tours of hospital wards; reviews in outpatient clinic; telephone contact: diagnosis established jointly by study collaborators

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Surgical hand preparation procedures were assigned randomly to the 6 participating operating theatres by tossing a coin, with a cross-over every 2 months. There was no indication of baseline imbalance for important variables such as type of surgery, contamination level of the surgery or use of antibiotic prophylaxis.

Allocation concealment (selection bias)
Low risk Allocation was decided by toss of a coin. Not clear who undertook this process and if it was concealed from the sites. Given that this was a cluster trial with cross-over, the potential for bias stemming from allocation concealment was limited. As the Cochrane Handbook for Systematic Reviews of Interventions notes, "Cluster-randomised trials often randomise all clusters at once, so lack of concealment of an allocation sequence should not usually be an issue." Blinding (performance bias and detection bias) All outcomes -Blinding participants High risk Personnel were aware of allocation as procedures differed in obvious ways.
Compliance by surgical teams was determined by observation of practices. A trained observer who did not belong to the surgical team checked whether each sink had the correct hand preparation and whether all surgeons (including visiting sta ) followed the recommended hand preparation procedures.

Nthumba 2010
Surgical hand antisepsis to reduce surgical site infection ( "SSI was diagnosed . . . and documented by a trained nurse who visited the surgical wards three to four times each week during the 30 days after surgery. Patients discharged before this were reviewed in the outpatient clinic or contacted by telephone . . . The nurse was blinded to the method of surgical hand preparation. The diagnosis of SSI was established jointly by the study collaborators; differences in SSI ascertainment were resolved by consensus without knowledge of the study allocation of the patient." Incomplete outcome data (attrition bias) All outcomes Low risk 3722 patients underwent a surgical procedure in the operating theatre, and 3317 were included in the 10 study clusters. Postdischarge surveillance data could not be obtained for 184 patients (5.5%).

Low risk
The primary objective of this cluster-randomised, cross-over trial was to compare the efficacy of plain soap and water with alcohol-based handrub, using SSI rates as the main outcome measure. The feasibility and affordability of the local production of an alcohol-based handrub was also investigated, together with an assessment of its acceptability among healthcare workers. All specified outcomes were reported.

Other bias Low risk
There was no evidence of other sources of bias. Clustered nature of the data was taken into account in the trial.

Methods
Cluster-randomised controlled equivalence trial Generation of random number sequence: random number tables Allocation concealment: no details given Blinding: discussed but only conducted during postdischarge follow-up A priori sample calculations: yes Antisepsis protocol:

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Quote: "Each participating surgical service was assigned a 2-digit random number by using a random number table. Surgical services corresponding to the 3 higher numbers were assigned to hand-rubbing with AAS and the remaining 3 services were assigned to traditional hand-scrubbing".
Comment: adequate evidence of random sequence generation Allocation concealment (selection bias) Unclear risk Quote: "Each participating surgical service was assigned a 2-digit random number by using a random number table. Surgical services corresponding to the 3 higher numbers were assigned to handrubbing with AAS and the remaining 3 services were assigned to traditional hand-scrubbing" Comment: no evidence of allocation concealment Blinding (performance bias and detection bias) All outcomes -Blinding participants High risk Quote: "[O]bservers of the clinical outcome could not be blinded to the hand antisepsis protocol." The study also mentions that "compliance observers did not belong to the operating department team but were usually present in the surgical suite. To avoid a Hawthorne effect the surgical teams were not informed of the timing of the evaluations".
Comment: Although the effect of blinding has been considered, in the comparison of different scrubbing protocols it would be difficult to blind the participant or personnel. The risk of bias is still high in this instance, however.
Blinding (performance bias and detection bias) All outcomes -Blinding care givers High risk Quote: "[O]bservers of the clinical outcome could not be blinded to the hand antisepsis protocol." The study also mentions that "compliance observers did not belong to the operating department team but were usually present in the surgical suite. To avoid a Hawthorne effect the surgical teams were not informed of the timing of the evaluations".
Comment: Although the effect of blinding has been considered, in the comparison of different scrubbing protocols it would be difficult to blind the participant or personnel. The risk of bias is still high in this instance, however.
Blinding (performance bias and detection bias) All outcomes -Blinding outcome assessors High risk Surgical site infection outcome Quote: "According to CDC guidelines, all SSIs had to be confirmed by the surgeon or the physician in charge on the patient. Thus, observers of the clinical outcome could not be blinded to the hand antisepsis protocol" Comment: As the surgeon was the participant in this case. it is clear that this could constitute a high risk of bias.
Tolerance and compliance outcome Quote: "[T]he surgical personnel (77 subjects) were asked to estimate the effect of the 2 protocols on their skin. We used 2 10 cm visual analogue scales, at month 0 and after 3 crossovers; 0 cm representing absence of an tolerance problem and 10 cm representing maximal dryness with chapped hands and desquamation or maximal irritation with erythema, burning sensation, and abrasion."

Library
Trusted evidence. Informed decisions. Better health.

Cochrane Database of Systematic Reviews
Incomplete outcome data (attrition bias) All outcomes Low risk Quote: "During the study period . . . 4823 consecutive patients underwent surgery. Among these, 385 patients underwent contaminated or dirty-contaminated surgery, and 51 were lost to follow up at 30 days (17 in the hand-rubbing group). The remaining 4387 patients (68.5% of whom underwent clean surgery) were considered for analysis".
Comment: As the sample size is large, the numbers lost to follow-up are not significant enough to warrant a risk of attrition bias.
Selective reporting (reporting bias) Low risk Quote: "Thirty-day surgical site infections rates were the primary end point; operating department teams' tolerance of and compliance with hand antisepsis were secondary end points".
Comment: adequate evidence in the results that these endpoints were accounted for comprehensively

Other bias
Unclear risk This appears to be a clustered cross-over study; it does not seem that clustering was taken into account in the analysis.

Methods
Randomised controlled cross-over trial (Latin square design -participants took part in each of 4 interventions) Generation of random number sequence: no details given Allocation concealment: no details given Blinding: no details given A priori sample calculations: no Antisepsis protocol: yes Withdrawals: 2 participants withdrew Intention-to-treat analysis: no Clear inclusion or exclusion criteria: no Participants 34 anaesthetic, recovery and ward nurses Baseline comparability: gender, age, ethnicity, hand dominance, baseline bacterial counts Interventions Group 1 -5 min initial scrub and 3 min subsequent scrub using chlorhexidine Group 2 -3 min initial and 30 s subsequent scrub using chlorhexidine Group 3 -5 min initial and 3 min subsequent scrub using povidone iodine Group 4 -3 min initial and 30 s subsequent scrub using povidone iodine Outcomes Outcome measure: CFUs on participants' hands Method of testing: glove juice method Timing of testing: before antisepsis, immediately after antisepsis, 2 h after initial antisepsis, 2 h after subsequent antisepsis Notes Participants did not take part in any surgical procedures

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "Subjects were randomly assigned to one of four groups, and each group was assigned to one of the four scrub regimens each week. Control on the treatment order was achieved through a Latin square design, as described by Winder."

Library
Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews Comment: adequate evidence of an appropriate study design, but on balance not enough evidence of truly random sequence generation Allocation concealment (selection bias) Unclear risk Quote: "Subjects were randomly assigned to one of four groups, and each group was assigned to one of the four scrub regimens each week. Control on the treatment order was achieved through a Latin square design, as described by Winder." Comment: no indication that allocation to each group was concealed to the personnel Blinding (performance bias and detection bias) All outcomes -Blinding participants Unclear risk Quote: "Subjects were randomly assigned to one of four groups, and each group was assigned to one of the four scrub regimens each week . . . Subjects were supervised by the investigator while they scrubbed on all test occasions." Comment: no evidence to suggest that there was appropriate blinding of participants or personnel during the study Blinding (performance bias and detection bias) All outcomes -Blinding care givers Unclear risk Quote: "Subjects were randomly assigned to one of four groups, and each group was assigned to one of the four scrub regimens each week . . . Subjects were supervised by the investigator while they scrubbed on all test occasions." Comment: no evidence to suggest that there was appropriate blinding of participants or personnel during the study Blinding (performance bias and detection bias) All outcomes -Blinding outcome assessors

Unclear risk Bacterial contamination
Quote: "While the glove was still on the hand, a sample of the fluid was taken . . . [S]amples were collected on four occasions for each condition: (1) immediately before scrubbing (both hands), (2) immediately after the initial surgical scrub (non-dominant hand only) (3) 2 hours after the initial surgical scrub, immediately before the consecutive scrub (dominant hand) and (4) 2 hours after one consecutive surgical scrub (dominant hand)." Comment: no indication that those collecting the samples, administering the fluid or those performing the microbial assays were in any way blinded to the intervention or protocol Incomplete outcome data (attrition bias) All outcomes Low risk Quote: "Thirty-six subjects were recruited, but two subjects withdrew from the experiment before completing all four treatments (scrubs) because of skin reactions, including erythema, burning sensations and local swelling" Comment: adequate evidence that losses to follow-up were small and fully accounted for Selective reporting (reporting bias) Low risk No direct quotes, but the results of each 'scrub' are displayed fully at baseline and subsequent time intervals in the Results as laid out in the Methodology.
Other bias Unclear risk Although the trial had a cross-over design, it did not appear that this was reflected in the analysis. Group 1 -5 min initial and 3.5 min subsequent scrub using 4% chlorhexidine Group 2 -3 min initial and 2.5 min subsequent scrub using 4% chlorhexidine Group 3 -3 min initial and 2.5 min subsequent scrub using povidone iodine with triclosan Group 4 -3 min initial scrub using 4% chlorhexidine followed by a 30 s application of isopropanol 70% and chlorhexidine 0.5%, and subsequent scrubs using 30 s application of isopropanol 70% and chlorhexidine 0.5% Group 5 -2 min initial scrub using 4% chlorhexidine followed by a 30 s application of ethanol 70% and chlorhexidine 0.5%, and subsequent scrubs using 30 s application of ethanol 70% and chlorhexidine 0.5%

Outcomes
Outcome measure: CFUs on participants' hands Method of testing: glove juice method Timing of testing: before antisepsis, immediately after antisepsis, 2 h after initial antisepsis, 2 h after subsequent antisepsis Notes Participants did not take part in any surgical procedures

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote: "Subjects who agreed to participate in the study followed each of the scrub protocols in turn, the order controlled by the use of a Latin square design. That is, every nurse was required to complete every protocol but not in the same sequence." Comment: clear that the study design does reduce selection bias; however not clear whether there is enough evidence to suggest random sequence was generated Allocation concealment (selection bias) Unclear risk Quote: "Subjects who agreed to participate in the study followed each of the scrub protocols in turn, the order controlled by the use of a Latin square design. That is, every nurse was required to complete every protocol but not in the same sequence." Comment: no evidence to suggest that allocation was concealed to personnel Blinding (performance bias and detection bias) All outcomes -Blinding participants High risk Quote: "Subjects who agreed to participate in the study . . . followed each scrub protocol each day for one week (referred to as the 'test week') with a week of normal activities between each test week. They were issued with a supply of the specific antiseptic to be used in excess to their requirements for scrubbing so that the appropriate antiseptic could be used exclusively dur-Library Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews ing the test week. Subjects were assessed before commencing each scrub protocol and at the end of the test week to determine changes in the number of colony forming units (cfu) after scrubbing and changes in the condition of the hands." Comment: Although study design allows for repeated testing, there is no evidence that there was effective blinding of participants and personnel to the different interventions, and differences between the conditions would have been clearly apparent to those taking part.
Blinding (performance bias and detection bias) All outcomes -Blinding care givers High risk Quote: "Subjects who agreed to participate in the study . . . followed each scrub protocol each day for one week (referred to as the 'test week)' with a week of normal activities between each test week. They were issued with a supply of the specific antiseptic to be used in excess to their requirements for scrubbing so that the appropriate antiseptic could be used exclusively during the test week. Subjects were assessed before commencing each scrub protocol and at the end of the test week to determine changes in the number of colony forming units (cfu) after scrubbing and changes in the condition of the hands." Comment: Although study design allows for repeated testing, there is no evidence that there was effective blinding of participants and personnel to the different interventions, and differences between the conditions would have been clearly apparent to those taking part.
Blinding (performance bias and detection bias) All outcomes -Blinding outcome assessors Unclear risk Sampling method Quote: "A glove juice sampling method to estimate the numbers of CFU present on the hands" Comment: no indication that those performing the sampling or performing the microbial assays were independent to the study

Skin condition
Quote: "Larson's Weekly Skin Assessment Rating Scale was used to rate the condition of the hands with respect to appearance, integrity, moisture and sensation. Participants rated themselves on a weekly basis. An independent rater, blind to the protocol being followed by the subject, also rated the subject's dominant hand each week." Comment: evidence that reasonable measures were undertaken to blind outcome assessment. Overall assessment unclear for outcome assessment blinding.
Incomplete outcome data (attrition bias) All outcomes High risk Quote: "Of the 32 participants recruited, 23 completed all five scrub protocols (scrubs) because it was discovered [that several participants] were allergic or sensitive to one of the more antiseptics. One felt that her skin was already too dry to be able to participate further. One person could not complete the study because she was unexpectedly o work for an extended period and the other two gave no reason for withdrawing." Comment: fairly high rates of losses to follow-up and no information given as to why some participants did not complete the study. Although it seems that the researchers acted appropriately here, it is difficult to give a judgement other than high risk

Cochrane Database of Systematic Reviews
Selective reporting (reporting bias) Low risk No direct quotes, but reasonable evidence to suggest that both bacterial contamination and skin condition have been appropriately and fully tested and reported upon effectively.
Other bias Unclear risk Although the trial had a cross-over design, it did not appear that this was reflected in the analysis.

Methods
Randomised cross-over trial Generation of random number sequence: no details given Allocation concealment: no details given Blinding: no details given A priori sample calculations: no Antisepsis protocol: no Withdrawals: no details given Intention-to-treat analysis: no Clear inclusion or exclusion criteria: no Participants 75 surgeons Baseline comparability: baseline bacterial counts Interventions Group 1 -surgical scrub using 4% chlorhexidine (details of the duration are not given) Group 2 -alcohol rub using Sterillium (45% w/w of propan-2-01, 30% w/w of propan-l-01 and 0.2% w/w of ethylhexadecyldimethyl ammonium ethylsulfate)(details of the duration are not given)

Outcomes
Outcome measure: CFUs on participants' hands Method of testing: glove juice method Timing of testing: before antisepsis, immediately after antisepsis and after surgical procedure completed

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias)

Unclear risk Dermal tolerance study
Quote: "The first study was performed with a panel of 60 volunteers divided in two subgroups of 30 persons each." Comment: no indication of how the subgroups were divided Surgical hand disinfection study Quote: "In a randomised cross-over clinical trial the antimicrobial efficacy of Sterillium for the surgical hand-rub was tested against Hibiscrub in the Kantonsspital Basel over a period of 11 weeks . . . two weeks were needed for recruitment, four weeks to conduct the Sterillium arm of the study, one week's interval and then four weeks for the Hibiscrub part." Comment: no further information given as to how participants were recruited and how they were assigned and allocated to which intervention Hygienic hand disinfection study Quote: "The first panel started with Hibiscrub, the second with Sterilium. Both products were used over seven weeks after one week of preconditioning. After an interval of four weeks the second run started with a single crossover of products." Comment: although cross-over trial used, no indication that the participants or personnel were blinded to the intervention Surgical hand disinfection Quote: "Organisms were recovered by the glove juice method. Plastic bags with the sampling fluid were placed on the subjects hands. The bag on each hand was secured and massaged for 1 min in a uniform manner by a laboratory technician." Comment: no indication whether participants were blinded to intervention and unclear whether technician was blinded to which intervention had been used Hygienic hand disinfection Quote: "The antimicrobial efficacy of each product was compared with that of 60% (v/v) 2-propanol on artificially contaminated hands (E. coli K 12, NCTC 10 538) using a cross-over design with 15 volunteers. The hands were first washed for 1 min with so soap, dried with paper towels immersed in the contamination fluid up to the mid-metacarpals for 5 s with fingers spread and then allowed to dry for 3 min" Quote: "The first panel started with Hibiscrub, the second with Sterilium. Both products were used over seven weeks after one week of preconditioning. After an interval of four weeks the second run started with a single crossover of products".
Comment: although cross-over design used, no indication that the participants or personnel were blinded to the intervention Surgical hand disinfection Quote: "Organisms were recovered by the glove juice method. Plastic bags with the sampling fluid were placed on the subjects hands. The bag on each hand was secured and massaged for 1 min in a uniform manner by a laboratory technician" Comment: no indication whether participants were blinded to intervention and unclear whether technician was blinded to which intervention had been used Hygienic hand disinfection Quote: "The antimicrobial efficacy of each product was compared with that of 60% (v/v) 2-propanol on artificially contaminated hands (E. coli K 12, NCTC 10 538) using a cross-over design with 15 volunteers. The hands were first washed for 1 min with so soap, dried with paper towels immersed in the contamination fluid up to the mid-metacarpals for 5 s with fingers spread and then allowed to dry for 3 min." Comment: unclear as to whether participants or personnel were blinded to the interventions Blinding (performance bias and detection bias) All outcomes -Blinding outcome assessors Unclear risk Dermal tolerance Quote: "The following parameters were measured before and after application: Clinical assessment by observation of the hands and the forearms of the volunteers by a dermatologist. The number of volunteers which dropped out of the study because of skin damage as decided by a dermatologist" Comment: no indication that the dermatologist was independent to the study Surgical hand disinfection Quote: "The bag on each hand was secured and massaged for 1 min in a uniform manner by a laboratory technician" Comment: no indication that the laboratory technician was blinded to the intervention Hygienic hand disinfection Quote: "Control microbial counts were obtained by rubbing the fingertips for 1 minute in a Petri dish containing a liquid broth using a separate dish for each hand. Either 3 ml of the hand gel or two aliquots of 3 ml of the reference alcohol were applied to the hands. The rub-in period was 30 s for the hand gels and 60 s for the reference alcohol as prescribed by EN 1500." Comment: no mention of whether those supervising the process or the overseeing the microbial assays were independent of the study Comment: This total represented a large proportion of the total participants used. Despite attributing the high rates of dropout to reasons related to the product, the risk of attrition bias is fairly high in this case.

Surgical hand disinfection
No direct quotes, but no indication given as to the total number of participants used or whether there was any observed loss to follow-up. Therefore the judgement remains unclear.

Hygienic hand disinfection
Quote:"The antimicrobial efficacy of each product was compared with that of 60% (v/v) 2-propanol on artificially contaminated hands (E. coli K 12, NCTC 10538) using a cross-over design with 15 volunteers." Comment: No losses to follow-up were discussed within the results; however, there is no evidence to suggest no losses to follow-up occurred. Overall judgement therefore remains unclear for outcome assessment.

Low risk Dermal tolerance
No direct quotes; however, the form used to tabulate the results gives no quantitative figures for variables that were predominantly quantitative in nature. For example "D-squames", "Electrical capacity", "Transepidermal water loss" are expressed as either "−− very poor", "− poor", "+ good" or "++ very good". It is unclear why the outcomes have been reported in this way.

Surgical hand disinfection
No direct quotes, although the results indicate the microbial concentration in the sampling fluid before and after treatment, outlined in the methodology and expressed as logarithm.

Hygienic hand disinfection
No direct quotes, although all the hand disinfectants that were discussed in the methodology as appropriate for testing are discussed and tabulated in an appropriate manner. Overall judgement of low risk.

Other bias Low risk No other sources of bias detected; cross-over design taken into account in analysis
Pietsch 2001 (Continued) Methods Parallel group randomised controlled trial. Unit of randomisation and analysis is individual sta member. Cochrane Database of Systematic Reviews results for two people were spoiled. The findings on 162 participants are presented." Selective reporting (reporting bias) Low risk All specified outcomes reported: "The primary purpose of the trial was to compare any two of the trial groups by measuring the difference in post-intervention CFU per hand. A secondary purpose was to identify any relationships between participants and baseline CFU counts." Other bias Low risk No evidence of other sources of bias

Notes
Only 20% of the 400 enrolled sta were assessed for bacteria on hands; these were classified as having or not having a positive culture. No data on number of CFUs on hands were reported. Participants did not take part in any surgical procedures.

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk Quote "We randomly assigned subjects to one of two study groups (i.e., twominute, three-minute surgical hand scrub times)." Comment: no indication as whether a truly randomised sequence was generated

Cochrane Database of Systematic Reviews
Allocation concealment (selection bias) Unclear risk Quote "We randomly assigned subjects to one of two study groups (i.e., twominute, three-minute surgical hand scrub times)." Comment: no indication as to whether allocation was concealed to the participants or personnel. The study was a cross-over trial, therefore all the participants were likely to have undertaken the same interventions; however, the role of bias is unclear in this case.
Blinding (performance bias and detection bias) All outcomes -Blinding participants High risk No direct quotes given, but no mention as to whether the subjects or the personnel were blinded to the intervention. It is very likely that the personnel were not blinded as they would be able to calculate the time spent handwashing.
Blinding (performance bias and detection bias) All outcomes -Blinding care givers High risk No direct quotes given, but no mention as to whether the subjects or the personnel were blinded to the intervention. It is very likely that the personnel were not blinded as they would be able to calculate the time spent handwashing.
Blinding (performance bias and detection bias) All outcomes -Blinding outcome assessors High risk No direct quotes given, but investigators were responsible for measurement of time and for preparation of solution for bacterial culture, as well as the sampling and measurement of log CFU counts. Although this is an objective measure, it is reasonable to suggest that as the investigators were not blinded to the intervention then the risk of bias here could be considered as high.
Incomplete outcome data (attrition bias) All outcomes Low risk Quote: "Each subject was able to complete both surgical hand scrub trials, which resulted in 300 agar plates for incubation and enumeration." Comment: adequate evidence of no loss to follow-up Selective reporting (reporting bias)

Low risk
The main outcome variable was the log counts of bacterial colonies found on participants' hands after washing their hands for a specified time. This was fully represented in table format in the results section.
Other bias Unclear risk Although the trial had a cross-over design. it did not appear that this was reflected in the analysis

High risk of bias
Participants or investigators enrolling participants could possibly foresee assignments and thus introduce selection bias, such as allocation based on using an open random allocation schedule (e.g. a list of random numbers); assignment envelopes were used without appropriate safeguards (e.g. if envelopes were unsealed or non-opaque or were not sequentially numbered); alternation or rotation; date of birth; case record number; any other explicitly unconcealed procedure.

Unclear
Insu icient information to permit judgement of low or high risk of bias. This is usually the case if the method of concealment is not described or is not described in su icient detail to allow a definitive judgement, for example, if the use of assignment envelopes is described, but it remains unclear whether envelopes were sequentially numbered, opaque and sealed.
3. Blinding (participants, personnel and outcome assessors) -was knowledge of the allocated interventions adequately prevented during the study?

Low risk of bias
Any one of the following.
• No blinding, but the review authors judge that the outcome and the outcome measurement are not likely to be influenced by lack of blinding. • Blinding of participants and key study personnel ensured, and unlikely that the blinding could have been broken.
• Either participants or some key study personnel were not blinded, but outcome assessment was blinded and the non-blinding of others is unlikely to introduce bias.

High risk of bias
Any one of the following.
• No blinding or incomplete blinding, and the outcome or outcome measurement is likely to be influenced by lack of blinding.
• Blinding of key study participants and personnel attempted, but likely that the blinding could have been broken.
• Either participants or some key study personnel were not blinded, and the non-blinding of others is likely to introduce bias.

Unclear
Either of the following.
• Insu icient information available to permit a judgement of low or high risk of bias.
• The study did not address this outcome.

Low risk of bias
Any one of the following.
• No missing outcome data.
• Reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias).
• Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups.
• For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention e ect estimate. • For continuous outcome data, plausible e ect size (di erence in means or standardised di erence in means) among missing outcomes not enough to have a clinically relevant impact on observed e ect size. • Missing data have been imputed using appropriate methods.

High risk of bias
Any one of the following.
• Reason for missing outcome data likely to be related to true outcome, with imbalance in numbers or reasons for missing data across intervention groups. • For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk enough to induce clinically relevant bias in intervention e ect estimate. • For continuous outcome data, plausible e ect size (di erence in means or standardised di erence in means) among missing outcomes enough to induce clinically relevant bias in observed e ect size.

Library
Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews • 'As-treated' analysis done with substantial departure of the intervention received from that assigned at randomisation.
• Potentially inappropriate application of simple imputation.

Unclear
Either of the following.
• Insu icient reporting of attrition/exclusions to permit judgement of low or high risk of bias (e.g. number randomised not stated, no reasons for missing data provided). • The study did not address this outcome.

5.
Are reports of the study free of the suggestion of selective outcome reporting?

Low risk of bias
Either of the following.
• The study protocol is available and all of the study's prespecified (primary and secondary) outcomes that are of interest in the review have been reported in the prespecified way. • The study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were prespecified (convincing text of this nature may be uncommon).

High risk of bias
Any one of the following.
• Not all of the study's prespecified primary outcomes have been reported.
• One or more primary outcomes are reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not prespecified. • One or more reported primary outcomes were not prespecified (unless clear justification for their reporting is provided, such as an unexpected adverse e ect). • One or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta-analysis.
• The study report fails to include results for a key outcome that would be expected to have been reported for such a study.

Unclear
Insu icient information available to permit judgement of low or high risk of bias. It is likely that the majority of studies will fall into this category.

Low risk of bias
The study appears to be free of other sources of bias.

High risk of bias
There is at least one important risk of bias. For example, the study: • had a potential source of bias related to the specific study design used; or • has been claimed to have been fraudulent; or • had some other problem.

Unclear
There may be a risk of bias, but there is either: • insu icient information to assess whether an important risk of bias exists; or • insu icient rationale or evidence that an identified problem will introduce bias.

F E E D B A C K Enquiry about status of any ongoing trials, 11 August 2008 Summary
The Authors conclusions' include suggestions for trials that are needed. Ecolab Ltd would like to know if any of the trials have or are being carried out? Library Trusted evidence. Informed decisions. Better health.

Cochrane Database of Systematic Reviews
Submitter has modified conflict of interest statement: I certify that I have a iliations with an organisation or entity with a financial interest in the subject matter of my feedback.

Reply
Tanner conducted a randomised controlled trial in April to June 2008 comparing nail brushes and nail picks with 164 operating room sta . This study found no di erence in the number of colony forming units on the hands of the scrub sta one hour a er they had scrubbed with antiseptic solution and a nail pick, antiseptic solution and a nail brush or antiseptic solution alone. The authors of this review have not been informed of other studies in this field.
New searches have been conducted for this review which is currently being updated.

Contributors
Author of feedback: Alex Haworth Occupation Business Development Manager, Ecolab Ltd.