Comparing the effectiveness of hand hygiene techniques in reducing the microbial load and covering hand surfaces in healthcare workers: updated systematic review

Background: This review, commissioned by the World Health Organization (WHO), examined the effectiveness of the WHO 6-step hand hygiene (HH) technique in reducing microbial load on hands and covering hand surfaces, and compared its effectiveness to other techniques. Methods: Medline, CINAHL, ProQuest, Web of Science, Mednar, and Google Scholar were searched for primary studies, published in English (1978 - February 2021), evaluating the microbiological effectiveness or hand surface coverage of HH techniques in healthcare workers. Reviewers independently performed quality assessment using Cochrane tools. The protocol for the narrative review was registered (PROSPERO 2021: CRD42021236138). Results: Nine studies were included. Evidence demonstrated that the WHO technique reduced microbial load on hands. One study found the WHO technique more effective than the 3-step technique (P=0.02), while another found no difference between these two techniques (P=0.08). An adapted 3-step technique was more effective than the WHO technique in laboratory settings (P=0.021), but not in clinical practice (P=0.629). One study demonstrated that an adapted 6-step technique was more effective than the WHO technique (P=0.001). Evidence was heterogeneous in application time, product, and volume. All studies were high risk of bias. Conclusions: Eight studies found that the WHO 6-step technique reduced microbial load on healthcare workers’ hands; but the studies were heterogeneous and further research is required to identify the most effective, yet feasible technique.


Background
It is widely acknowledged that effective hand hygiene (HH) among healthcare workers (HCWs) is one of the most important infection prevention strategies available 1,2 ; and therefore, is a key element of infection prevention and control guidelines. 3 However, uncertainty remains concerning a range of issues related to HH. 2,4 One major issue relates to which technique to use when performing HH. [4][5][6][7][8][9][10][11] The World Health Organization (WHO) recommends the adoption of a 6-step technique, 2 which was originally developed in 1978 by Professor Graham Ayliffe to standardize testing of HH products. 12 . Elements 2-7 on Figure 1 shows the areas on hands that should be covered with alcohol-based handrub (ABHR) when performing handrubbing, or with soap when performing hand washing using the 6-step technique.
The 6-step technique has now been adopted globally as the gold standard for hand washing with soap and water and for hand rubbing ABHR in clinical practice; 13 compliance, however, is usually low. [14][15][16] One possible way to increase compliance with the technique is to provide HCWs with evidence of how effective the recommended 6-step technique is in decontaminating their hands and in covering all hand surfaces. 17 A previous review identified the body of evidence in relation to the most effective HH technique offered conflicting findings. 18 Given the current increased interest in improving HH practice in response to the COVID-19 pandemic 19,20 and the continuously growing body of evidence, we updated our previous review.  21 The aims of our updated review, commissioned by WHO, were to examine the effectiveness of the WHO 6-step technique in reducing microbial load on hands and covering all hand surfaces, and to compare its effectiveness to that of other techniques. The other techniques involved adaptions to the order of performance of the WHO's 6-step technique or to techniques involving three steps. The 3-step techniques were based on the Centers for Disease Control and Prevention's (CDC) 3-step HH technique which involves putting product on hands and rubbing hands together until dry while covering all hand surfaces. 22

Methods
This systematic review was an update of the original review conducted in 2018. 18 The findings of the studies included in the original and in the updated review are presented in this paper. The updated review used the same search strategy as the original 2018 review 18 but upgraded the quality assessment methods to use, where appropriate, Cochrane Collaboration's Effective Practice and Organization of Care (EPOC) tools. 23 The updated review protocol was prospectively registered with the international prospective register of systematic reviews (PROSPERO 2021: CRD42021236138) (Available from: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021236138) and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 24

Inclusion criteria
Studies were included in the review if they referred to the WHO 6-step technique, or the technique described was consistent with the WHO technique. Studies focusing on HCWs, (including, but not limiting to physicians, nurses, nursing assistants, allied health professionals, or healthcare students), performing either a hand rubbing or handwashing within any healthcare context, in any country were included.

Exclusion criteria
Studies based in operating theatres using surgical hand preparation protocols were excluded, since the HH techniques and duration differ within this setting. 2 Studies were also excluded if they were not specifically about HH technique but were, instead, investigating the efficacy of HH products or evaluating HH compliance. Studies not conducted with HCWs were excluded as well, as were those that did not measure microbial load.

Outcomes
The primary outcomes required in the reviewed studies were reduction in the microbial load or surface coverage of HCWs' hands following application of the tested HH techniques.
Studies were included if their aims did not include testing the effectiveness of the technique, but quantification of the effect of using the technique was one of the outcomes measured and reported separately. Secondary outcome was a measure of time of hand decontamination.

Types of study
To enable the identification of the current available evidence, the review considered empirical research designs, including randomized controlled trials (RCTs), non-randomized controlled trials (NRTs), before and after studies, case control studies, cohort studies and observational descriptive studies. Reviews, conference proceedings and non-primary research records, such as editorials, opinion-based papers and commentaries were excluded.

Search strategy
The search included sources published between 1978 (the first year we were aware of the technique being used 12 ) and February 2021. A three-stage search strategy was employed.
Keywords and index terms were searched in CINAHL, Medline, ProQuest and Web of Science databases with the search restricted to sources published in English language. The full search strategy applied for Medline (Supplementary file 1) was individualised for the other databases according to their functionality. Secondly, as keyword terms cannot be combined in Mednar and Google Scholar, only the broadest keywords were searched in these databases. Finally, the reference lists of included papers were searched manually to identify any additional relevant articles.

Study selection
The titles and abstracts of all articles retrieved from the search were independently screened for relevance by two reviewers against the eligibility criteria relating to study design, population, intervention, and outcomes, as described above. The full-texts of articles that met the inclusion criteria after the title and abstract search, and those in which there was insufficient evidence in the title and abstract to make a decision, were reviewed independently by the two reviewers. Discrepancies were resolved through discussion.

Quality assessment and data extraction
Two reviewers independently extracted data from the eligible studies, using a standardised data collection tool (Supplementary file 2), which was used previously and described elsewhere. 18 Full-text copies of all articles included in the review were independently reviewed by two reviewers to assess their quality. Studies meeting the EPOC design criteria, 25 i.e. RCTs, NRTs, controlled before-and-after studies, interrupted time series studies and repeated measures studies, 25 were assessed for quality using the recommended EPOC tool 23 ; while design-specific, Joanna Briggs Institute critical appraisal checklists 26 were used for the remaining study designs. Disagreements were resolved through discussion or by the involvement of another reviewer.

Analysis
The results are presented in a narrative summary because it was not appropriate to conduct a meta-analysis nor to use the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach, 27 due to the substantial heterogeneity of the studies.

Search results
As shown in Figure 2, the search resulted in a total of 28,615 records. Of these, 4634 were duplicates, resulting in 23,981 records being eligible for stage two of the selection process.
After screening of titles and abstracts for eligibility, 51 records were selected for full-text review, of which nine 6, 28-35 were included in the review. The characteristics of these studies are shown in Table 1. Full-text articles excluded, with reasons (N =42) Did not relate to technique, did not refer to the WHO technique, non-primary research, did not involve HCWs, not conducted in healthcare setting, did not measure microbial load or coverage Studies included in systematic review (N = 9) • Randomized controlled trial (n=5) • Non-randomized cross-over trial (n=1) • Non-controlled before and after (n=3)  ABHRalcohol-based handrub; CDC -Centers for Disease Control and Prevention; CFUcolony-forming unit; CIconfidence intervals; HCWshealthcare workers; HHhand hygiene; Mdnmedian; NCBAnon-controlled before and after study; NRTnon-randomized controlled trial; RCTrandomized controlled trial; RFreduction factor; SDstandard deviation; WHO -World Health Organization Using the EPOC study design criteria, 25 of the studies included, four were RCTs, 28,30,31,34 one was a crossover RCT, 33 three were non-controlled before and after studies (NCBAs) 6, 32, 35 and one was a crossover NRT. 29 All studies used ABHR to investigate some aspects of the WHO 6-step HH technique. The primary outcome of all studies was the bacterial load on the hands of HCWs measured in colony-forming units (CFUs), apart from Sakmen et al., 31 in which the primary outcome measure was percentage hand surface coverage. Time and hand coverage were assessed in addition to bacterial load outcome in two 28,30 and three 30,32,35 studies, respectively. The participants were doctors and nurses, 28, 30, 35 medical students, [31][32][33] or HCWs, 6,29,34 while the settings included hospitals, 6,28,30,34,35 hospital laboratories 29,33 or university hospital training facilities. 31,32 The studies that compared the 6-step technique to other approaches 28-31, 33, 34 and the studies that investigated the effectiveness of the 6-step technique were analysed and discussed separately. 6,32,35 The studies were further categorised per type of settings, while the studies comparing 6-step technique to other techniques were further grouped according to the comparator.

Studies comparing two hand hygiene techniques
Clinically based studies comparing the 6-step and the 3-step technique Two clinically based RCTs 28, 30 compared the WHO 6-step technique with the CDC 3-step technique amongst 120 doctors and nurses. Reilly et al. 30 and Chow et al. 28 both observed a reduction in bacterial load after application of the WHO 6-step technique. However, findings were inconsistent. Chow et al. 28 found that the WHO 6-step technique was no more effective than the CDC 3-step technique (P=0.07). In contrast, Reilly et al. 30 reported that the WHO 6step technique was more effective than the CDC 3-step technique (P=0.02).
Chow et al. 28 and Reilly et al. 30 both monitored, as a secondary outcome measure, the median time for conducting HH and both observed the CDC 3-step technique required significantly less time to complete than the WHO 6-step technique (P=0.04 28 and P=0.002 30 ). In addition, Reilly et al. 30 also evaluated hand coverage and found that the WHO 6-step technique did not increase the total hand coverage area (P=0.15) and that a reduction in bacterial count was not related to hand coverage (P=0.97).

Laboratory-based studies adapting the 3-step or 6-step technique
Two laboratory-based studies, including a crossover RCT 33 and a crossover NRT 29 compared the WHO 6-step technique with an adapted 3-step 33 or adapted 6-step technique 29 . In Tschudin-Sutter et al. 33 , this adapted 3-step technique was different from that used by Chow et al. 28 and Reilly et al. 30 in that it consisted of covering all surfaces of the hands and, in addition, rotationally rubbing fingertips in the palm of the alternate hand and rotationally rubbing both thumbs. Tschudin-Sutter et al. 33 observed that the adapted 3-step technique was significantly more effective at reducing bacterial load than the WHO 6-step technique (P=0.021) when tested amongst 32 medical students.
Pires et al. 29 recruited 16 HCWs to compare the WHO 6-step technique with its variation, in which step six (rotational rubbing of fingertips against the opposite palm and vice versa) was performed as first in the sequence of steps. Pires et al. 29 reported that this modified "Fingertips First" 6-step technique resulted in a significantly greater reduction in bacterial load than the standard WHO 6-step technique (P=0.002).

Clinically based study adapting the 3-step or 6-step technique
Tschudin-Sutter et al. 34 also performed a clinically based study comparing the effectiveness of the WHO 6-step technique with an adapted 3-step technique in reducing microbial load on hands amongst 113 HCWs. As in their previous laboratory-based study 33 31 reported that while there were no significant differences in the total hand coverage between the two groups directly after training (P=0.584) and after 5-12 weeks (P=0.123); at the end of the week coverage was significantly greater in the "self-responsible application" group (P<0.001).

Clinically based studies demonstrating effect of the 6-step technique
Two clinically based NCBA studies investigated the effectiveness of the WHO 6-step technique in reducing microbial load on hands. 6,35 Widmer et al. 35 studied microbial load reduction on hands of 180 doctors and nurses after the use of ABHR using the 6-step technique prior to, and following HH training, while Laustsen et al. 6 observed 117 clinical procedures and investigated microbial load reduction on HCWs' hands after the use of ABHR prior to and following the completion of the clinical procedure.
Both studies 6,35 reported that when participants performed the WHO 6-step technique either correctly or incorrectly, bacterial load was reduced. But, the correct application of the technique, as opposed to the incorrect application, resulted in a greater reduction (Widmer et al. 35 : P<0.001; Laustsen et al. 6 : P value not reported). This finding was also supported by Reilly et al., 30 who found a significant difference between those who performed the WHO 6step technique with 100% accuracy and compared to those who did not (P=0.01).
In addition, Widmer et al. 35 also reported examining the hand surface coverage using an ultraviolet light box to detect areas missed on the hands after HH, but did not report specific results on this.

University hospitals training facilities-based study demonstrating effect of the 6-step technique
One NCBA study, 32 based in the university hospitals training facilities investigated microbial load on hands of 563 medical students before and after they applied ABHR using the WHO 6-step technique following HH training. Tschudin-Sutter et al. 32 found that the bacterial load on the hands of medical students was reduced after the use of the WHO 6-step technique (P<0.001). Tschudin-Sutter et al. 32 also report investigating hand surface coverage following HH using an ultraviolet light box, but findings of this investigation were not provided.

Studies meeting the EPOC study design criteria
Six of the included studies met the EPOC study design criteria. 25 As shown in Table 2, the overall risk of bias of these studies was assessed as high. High risk of bias was most commonly associated with the lack of, or insufficient protection against contamination resulting from the allocation to study groups occurring at the participant's or ward level within a single hospital 28,30,34 or from crossover design. 29,33 High risk of bias was also related to the lack of randomisation, 29 risk of selection bias resulting from the crossover approach 29,33 and potential bias resulting from imbalanced missing outcome data between the groups. 31 In addition, all studies had at least one item assessed as unclear risk. This was related to the lack of sufficient information regarding blinding, 29,31,33,34 baseline outcome measures 31 or baseline participants' characteristics, 28,30,31 the process of random sequence generation, 31, 33 allocation sequence concealment 31 or protection against contamination. 31 More details on the reasons for risk of bias assessment decisions of the EPOC design studies can be found in Supplementary file 3. Furthermore, a number of other potential biases were identified in relation to the analysis and data collection methods. While four studies 28,30,33,34 were powered, and recruitment targets were achieved, two studies 29, 31 did not mention this, so it is unclear whether their sample size was adequate.
Four studies used the modified glove juice sampling method, 28,30,33 or its variation in which sterile bag was used instead of the glove, 34 while fingertip imprint method was used in one study. 29 Chow et al. 28  technique might have also removed some bacteria from participants' hands before ABHR was applied, leading to an overestimation of the bacterial reduction. This number will vary considerably; and if the comparison of reduction outcomes is valid, evidence is required to show that there is a true random distribution of contamination density across the two groups.
It is unknown whether this can be guaranteed in a relatively small sample of clinicians delivering different aspects of care.
In two laboratory-based studies 29,33 participants' hands were artificially contaminated which standardised baseline bacterial load. However, using a single type of microorganism does not reflect the natural conditions and the actual bacterial flora present on HCWs' hands, 36,37 reducing the external validity of their findings.
Furthermore, apart from Reilly et al. 30 and Sakmen et al., 31 studies were unable to determine whether specific areas of the hand had been missed by the HH techniques, because they did not evaluate or did not report hand coverage and sites missed. Previous studies 8,35 showed that the thumb and fingertips are the most frequently missed areas on the hands. In the study by Reilly et al., 30 correlation between bacterial reduction and hand surface coverage was also a limitation, because these data were collected at two different time points. Therefore, Reilly et al. 30 could not be certain that the technique was performed by participants in exactly the same way each time, although standardization by showing each participant an instruction card with a diagram of the allocated technique should have helped minimise the risk.

Methodological quality of other study designs
Three of the studies included in the review were NCBA design; thus, did not meet the EPOC study design criteria. 25 They are by the nature of their design consider to be high risk of bias.
Nevertheless, quality assessment was performed using the relevant Joanna Briggs critical appraisal tool. 26 As shown in Table 3, two of these studies 32, 35 had at least one item assessed as high risk. In two studies, high risk of bias was related to a lack of clearly defined eligibility criteria. 32,35 In addition, for Tschudin-Sutter et al., 32 high risk of bias also resulted from the lack of details on potential confounding factors and on how participants' compliance with the 6-step technique was monitored. Two studies 6, 32 also had at least one item assessed as unclear risk resulting from a lack of clear description of the study subjects 6,32 and insufficient detail to assess whether confounding factors were identified and appropriately accounted for in the analysis. 6 Further details on the risk of bias assessment decisions of the non-EPOC design studies can be found in Supplementary file 4.
It should also be noted that none of the non-EPOC design studies 6, 32, 35 had control groups, making it difficult to differentiate between the observed effect being due to the HH technique or to other confounding variables, such as application time or ABHR volume, thereby affecting the validity of the outcomes. Although all three NCBAs used appropriate statistical analysis, none of these studies mentioned if they were powered, so it is unclear whether their sample sizes were adequate for the analyses they performed.
In addition, despite all three studies microbial load reduction outcomes being measured using objective, valid and reliable methods, involving collecting fingertip imprint samples and using standard microbiological procedures to assess bacterial growth, a limitation of the finger-imprint technique is it only allows bacterial measurement from the fingertips. As the study by Reilly et al. 30 showed, the back of the hands, the back of the thumbs, and the back of the index fingers were the most frequently missed sites regardless of the technique used.
However, it could be argued that the finger imprint technique is perhaps a more appropriate method of bacterial measurement in terms of transmission of infections because it does not involve any massage of the hands that could remove resident organisms.

Discussion
HH is the single most important intervention to reduce the risk of cross transmission of infection. 2 Despite this, to our knowledge, our previous systematic review 18  Chow et al. 28 found no difference in the effectiveness of the WHO 6-step technique compared to the CDC 3-step technique, whereas Reilly et al. 30 found the WHO 6-step technique to be more effective. Tschudin-Sutter et al. 33 reported that an adapted 3-step technique that focused on the fingertips and thumbs was more effective than the WHO 6-step technique when tested in laboratory settings; however, when the authors compared the two techniques in clinical settings, no difference was found. 34 Finally, Pires et al. 29 provided evidence on the superiority of the modified 6-step technique ("Fingertips First" approach") over the standard WHO 6step technique in reducing bacterial load on hands. However, this was a laboratory-based study limited by the lack of randomization. The remaining evidence comes from studies with poor-quality research designs due to their lack of control or comparator groups. As a result, only limited conclusions can be drawn from these studies.
Of particular note is the study by Chow et al., 28 which found that covering of all aspects of the hands with no instructions was as effective as the WHO 6-step technique and quicker.
Reilly et al. 30  Another limitation is that in all of the included studies, participants' were observed performing hand hygiene. Their performance may differ in clinical practice due to competing workload pressures. These studies can, however, form the basis of more robust future studies.
Therefore, we recommend that RCTs directly comparing the effectiveness of the different techniques be performed in clinical practice.

Implications for Research
Further robust research, using well-designed multi-arm parallel RCTs that specifically focus on the different HH techniques, are required to determine which technique is the most effective and in what context. All of the following measures are required: determining the bacterial load on the hands of HCWs before and after application of techniques, ideally, using the glove juice method during clinical practice in acute-care hospitals; controlling the time of application, the products used and the product volume; including inter-rater reliability testing of data collectors; blinding microbiologists to study protocols; and having adequate sample sizes to power the studies. Studies should also report missing data, baseline outcomes, baseline participant characteristics and accuracy of performing the HH technique and, if necessary, appropriately adjust for these factors in the analysis. Randomisation of the population should help control for differences in participant experiences, previous training, and expectations of the HH technique; but the reporting of these data will demonstrate if this has been achieved. Randomization should occur at the healthcare facility level to avoid contamination of the intervention by contact between groups performing the different techniques.
Thus, the development of a protocol for standardising HH research studies, taking account of HH product and volume, and application time, and using the glove juice method for hand sampling in clinically based HH technique studies could improve future research and should be part of HH research agenda. 48

Implications for Practice
This review provides evidence on the effectiveness of the WHO 6-step HH technique, supporting the use of this approach in clinical practice. There is also some evidence demonstrating the microbiological efficacy of the WHO 6-step, adapted 6-step, CDC 3-step and adapted 3-step approaches, but it is insufficient as a body of evidence to be definitive.
Compliance with correct application of HH technique is usually suboptimal and developing new techniques demonstrating optimal efficacy but being simpler and faster will likely help increase compliance. However, HH is an essential part of infection prevention, and control measures and current practices should be maintained and reinforced while additional evidence is gathered.
Regarding the performance of HH systematic reviews, our search retrieved a large number of articles that were excluded because they were not empirical studies. We recommend that others performing similar searches include study design as one of the search domains.
Finally, when reporting the findings of HH research, this review identified the need to include a thorough description of the HH techniques, sampling strategy, and population/sample in each study. A checklist for reporting HH studies is warranted to help improve the evidence base, similar to reporting templates such as CONSORT, 49 STROBE 50 or ORION statement. 51 This review highlighted current evidence regarding the effectiveness of the WHO HH technique in reducing microbial load on the hands of HCWs. The findings provide direction for current practice and future research. HH research must continue to evolve to inform global action to prevent and control healthcare-associated infections and contain antimicrobial resistance.