Effectiveness of hand hygiene practices in preventing influenza virus infection in the community setting: A systematic review

Background Hand hygiene is known to be an effective infection prevention and control measure in health care settings. However, the effectiveness of hand hygiene practices in preventing influenza infection and transmission in the community setting is not clear. Objective To identify, review and synthesize available evidence on the effectiveness of hand hygiene in preventing laboratory-confirmed or possible influenza infection and transmission in the community setting. Methods A systematic review protocol was established prior to conducting the review. Three electronic databases (MEDLINE, Embase and the Cochrane Library) were searched to identify relevant studies. Two reviewers independently screened the titles, abstracts and full-texts of studies retrieved from the database searches for potential eligibility. Data extraction and quality assessment of included studies were performed by a single reviewer and validated by a second reviewer. Included studies were synthesized and analyzed narratively. Results A total of 16 studies were included for review. Studies were of low methodological quality and there was high variability in study design, setting, context and outcome measures. Nine studies evaluated the effectiveness of hand hygiene interventions or practices in preventing laboratory-confirmed or possible influenza infection in the community setting; six studies showed a significant difference, three studies did not. Seven studies assessed the effectiveness of hand hygiene practices in preventing laboratory-confirmed or possible influenza transmission in the community setting; two studies found a significant difference and five studies did not. Conclusion The effectiveness of hand hygiene against influenza virus infection and transmission in the community setting is difficult to determine based on the available evidence. In light of its proven effectiveness in other settings, there is no compelling evidence to stop using good hand hygiene practice to reduce the risk of influenza infection and transmission in the community setting.


Introduction
Hand hygiene is a commonly recommended infection prevention and control measure to reduce the risk of influenza infection and transmission in health care and community settings. Routine hand hygiene protocols that indicate the use of soap and running water to wash hands (1) and/or alcohol-based hand sanitizers to rub hands (1,2) are effective at physically removing influenza virus from human hands.
Hand hygiene practices have been found to be effective in reducing infection and transmission of healthcare-associated pathogens in the health care setting (3); in reducing non-pathogen-specific gastrointestinal and respiratory illnesses in the community setting (4)(5)(6)(7); and for disinfection, removal of contaminants and reduction of the incidence of hospital-acquired infections in the health care setting (3).
Less frequently studied has been the degree of protection against influenza virus infection and transmission afforded by hand hygiene practices in the community setting. An initial scoping search of the literature identified two systematic reviews that came to different conclusions. A review of randomized controlled trials found that hand hygiene as a co-intervention with facemask use in the community setting was efficacious against laboratory-confirmed influenza infection or influenza-like illness, but hand hygiene alone was not (8). Another review of intervention trials and observational studies found evidence of a reduction in influenza infection with hand hygiene interventions in schools, but no effect on secondary transmission of influenza in households in the community that had already experienced an index case (9).
A systematic review was undertaken to identify, review and synthesize the latest evidence on the effectiveness of hand hygiene as an intervention in preventing laboratory-confirmed or possible influenza infection and transmission in the community setting. The term "possible influenza infection" was defined as non-laboratory-confirmed cases, including influenza-like illness or an acute respiratory illness.

Methods
The systematic review parameters, search strategy and analysis plan were established prior to the conduct of the review. Hand hygiene was defined as handwashing, hand antisepsis and actions taken to maintain healthy hands and fingernails (10). The search strategy (Appendix 1) was developed in collaboration with a research librarian. MEDLINE, Embase and the Cochrane Library electronic databases were searched from inception until June 5, 2017 using search terms for influenza and hand hygiene. Searches were restricted to articles published in English or French.
Studies were included for review if they met the following criteria: • They were conducted in a community setting, which is defined as a non-health care, open setting without confinement and without special care for the participants (e.g., school, workplace, household) (8) • They were observational studies that assessed hand hygiene as an exposure of interest (e.g., observed or reported hand hygiene practice) or clinical trials that could include combinations of education, promotion and provision of products to do with hand hygiene, but assessed a hand hygiene intervention that could be reasonably expected to exert an independent influence • They assessed the impact of hand hygiene on: ○ laboratory-confirmed or possible influenza infection or ○ laboratory-confirmed or possible influenza transmission Studies were excluded if they met one or more of the following criteria: • They were conducted in the health care setting only • They assessed a multicomponent intervention for which hand hygiene could not be reasonably expected to exert an independent influence • They were not clinical research studies (e.g., literature reviews, editorials, opinion pieces or news stories, or non-human or in vitro studies) Study selection was completed independently by two reviewers. Reference lists of included studies and relevant secondary research articles retrieved through the search were also searched to identify relevant publications. One reviewer (KM) performed data extraction and quality appraisal and a second reviewer performed validation (LZ). Data were extracted on study design, population, setting, hand hygiene intervention (i.e., from clinical trials) or practice (i.e., from observational studies) and outcomes of interest. Study quality was assessed using the Cochrane Collaboration Risk of Bias Tool for randomized controlled trials (RCTs) (11) and the Effective Public Health Practice Project Quality Assessment Tool for observational designs (12). Disagreements between the two reviewers were resolved by discussion and reaching a consensus.
Narrative data synthesis and analysis were planned to summarize the direction, size and statistical significance of reported effect estimates for various study-defined outcomes and to explore overall patterns in the data extracted from included studies. If possible, meta-analyses were planned to assess the association of hand hygiene with influenza outcomes by income level of country of study, study design, setting, intervention evaluated and outcome assessed.

Results
After database searching, handsearching and removal of duplicates, 998 records remained. After screening, 115 records were identified for full-text review. When all inclusion and exclusion criteria were applied, 16 studies-seven RCTs and nine observational studies-were available for review. Figure 1 summarizes the study selection process.
Observational studies assessed using the Effective Public Health Practice Project Quality Assessment Tool found seven of nine observational studies as weak in quality (20-26) and two as moderate in quality (27,28). The reviewers made a post-hoc decision to not perform a meta-analysis as the limited number of included studies were not adequate for grouping by the study characteristics of interest.

RCTs on hand hygiene interventions
Of the seven included RCTs, six assessed the provision of hand sanitizer or soap with instructions on their use (13)(14)(15)(16)18,19). One RCT delivered an internet-based intervention educating and promoting handwashing without provision of any hand sanitizer or soap to participants (17). None of these RCTs reported the instructions or education given to participants on handwashing or hand antisepsis in sufficient detail to compare the appropriateness of these interventions to best practices.

Observational studies on hand hygiene practices
Of the nine included observational studies, four collected self-reported handwashing frequency (23,(26)(27)(28). Of the remaining five studies, one study dichotomized observed handwashing behaviour as observed or not observed (20) and one as frequent or infrequent (21). These studies did not specify or report the use of handwashing criteria in estimating handwashing frequency or counting handwashing events. Two studies assessed self-reported quality of hand hygiene practice, that is, good or poor (22), and optimal or suboptimal (25), and of these, one defined optimal hand hygiene practice according to published best practices (22). Another study collected self-reported information on adoption of various non-pharmaceutical interventions, including washing hands more often and hand sanitizer use (24).

Hand hygiene and influenza infection
Nine studies evaluated the effectiveness of hand hygiene interventions or practices in preventing laboratory-confirmed or possible influenza infection in the community setting, including two RCTs (15,17), one cohort study (27), three case-control studies (21,23,28) and three cross-sectional studies (20,22,25).
Study findings were mixed; six of nine studies found that some form of hand hygiene intervention or practice reduced laboratory-confirmed (23,28) or possible (17,20,22,25) influenza infection, while three studies found hand hygiene to be not statistically significantly associated with a decrease in influenza infection (15,21,27). For the two RCTs, one found a significant association between handwashing and decreased risk of influenza-like illness (15) and the other found no effect on self-reported clinically diagnosed influenza for a workplace hand sanitizer intervention (13). For the observational studies, which relied on self-reported (22,23,25,27,28) or observed (20,21) hand hygiene practice, most found statistically significantly lower likelihood of possible infection (20,22,23,25,28). The limited number of heterogeneous studies did not allow for more granular qualitative analysis of findings. The results are summarized in Table 1.

Hand hygiene and influenza transmission
Seven studies assessed the effectiveness of hand hygiene practices in preventing laboratory-confirmed or possible influenza transmission in the community setting, including five RCTs (13,14,16,18,19), one cohort study (24), and one case-control study (26). A majority of these studies assessed influenza transmission in the community setting by estimating secondary attack rates (SARs) at the household level (e.g., the proportion of susceptible individuals who became ill) for laboratory-confirmed or possible influenza (13,14,16,18,19).
Five of seven studies did not find a statistically significant association between hand hygiene intervention or practice and influenza transmission (13,14,16,18,24). An RCT found a statistically significant difference in SARs for influenza-like illness across handwashing, handwashing and facemask, and control interventions (0.17, 0.18 and 0.09, respectively), but not in SARs for laboratory-confirmed influenza (19). A case-control study found that handwashing at least three times per day was statistically significantly associated with reduced likelihood of household transmission of pandemic influenza A (H1N1) (26).
In four of five cluster RCTs conducted at the household level, hand hygiene intervention was implemented after the identification of the index case (13,14,18,19). Two of these four studies assessed a subgroup of households where the intervention was implemented within a defined period after the onset of symptoms in the index case (e.g., less than 36 or 48 hours); one of the two studies did not find a statistically significant difference between hand hygiene and control groups (14) while the other study found mixed results, depending on influenza type and determination of influenza (19). Four of five cluster RCTs did not find statistically significant differences in SARs for laboratory-confirmed or possible influenza between hand hygiene and control groups (13,14,16,18) and one found mixed results depending on outcome (19). The results are summarized in Table 2.

Discussion
The present systematic review identified 16 studies that assessed the impact of hand hygiene intervention or practice on influenza infection or transmission in the community setting. Two-thirds of studies suggested hand hygiene practices may help prevent influenza infection. Most studies that looked at influenza transmission, however, had non-statistically significant results. Most studies had design elements associated with the potential for bias. The studies were too heterogeneous in design for meta-analysis. Our findings were similar to the two other systematic reviews conducted on this issue despite methodological differences in study selection. Whereas we found both positive and negative studies, the Wong et al. review (8) found that hand hygiene intervention alone was not efficacious against laboratory-confirmed influenza and the Warren-Gash et al. review (9) found some evidence of influenza risk reduction with hand hygiene intervention, depending on the community setting. Warren-Gash et al. also found no evidence of effectiveness of hand hygiene on secondary transmission of influenza in households that had already experienced an index case (9).

Limitations
There are a number of important limitations to consider when interpreting the findings of this review. In general, the majority of studies investigated outcomes that were not specific to influenza virus infection, but were influenza-like illness and acute respiratory illness, which could be caused by other respiratory viruses. Findings from lower income settings (e.g., rural Bangladesh) may not be generalizable to high-income settings and vice versa. Moreover, in controlled clinical trials conducted in high-income settings, there may already be high baseline levels of hand hygiene practice rendering intervention and control groups more similar irrespective of hand hygiene intervention. The effectiveness of hand hygiene interventions is dependent on mode of influenza transmission and may be attenuated when the mode of transmission is not through contact. The present review restricted its scope to hand hygiene interventions independent of other public health measures; therefore, these interventions may not be reflective of real-world, multicomponent public health measures. Finally, a search of the grey literature was not undertaken, so some studies may have been missed.
There were also limitations inherent to both types of study. Some of the included RCTs lacked statistical power (13,15,16). None of the included RCTs presented information on hand hygiene interventions in sufficient detail to allow the comparison of these interventions to best practices. Possible non-compliance with the intervention and contamination of control participants may underestimate possible effects of hand hygiene. Adoption of effective hand hygiene practice may take longer than the intervention period of a clinical trial. For RCTs investigating influenza transmission in households with an index case, it is possible that the hand hygiene intervention was implemented too late in the course of illness of the index case to be effective in preventing intra-household transmission. In household studies, direct and indirect protection conferred by hand hygiene practice for more susceptible individuals (e.g., children) cannot be readily assessed due to a lack of information on hand hygiene practice collected at the individual level.
For the included observational studies, where hand hygiene practices were either self-reported or observed, measurement of hand hygiene practice may be influenced by response bias (e.g., social desirability bias), recall bias or the observer effect (29). Although most observational studies collected exposure data on self-reported handwashing frequency, these studies did not specify or report the use of criteria for counting handwashing events; therefore, optimal and suboptimal hand hygiene practices cannot be differentiated in the overall reported handwashing frequency. Observational studies may also be susceptible to residual confounding, selection bias and other biases that may further complicate the interpretation of findings. Although the cross-sectional studies included for review found statistically significant results (20,22,25), the cross-sectional design cannot determine whether the reported hand hygiene behaviour preceded influenza illness.

Implications and next steps
These numerous limitations of the existing body of evidence highlight the difficulties of conducting research on this topic in the community setting for both experimental and observational designs (8,9,30). Hand hygiene is a non-invasive, non-pharmaceutical intervention without adequate comparator interventions (31). There are also challenges in conducting RCTs with appropriate sample sizes to establish the relative importance of hand hygiene (6). In the community setting, it is also difficult to implement interventions and assess outcomes.
In light of the robust body of evidence on the benefits of hand hygiene practices with respect to general infectious disease prevention and control (7), the mixed results and limitations of current studies, there is no compelling evidence to stop using good hand hygiene practice to reduce the risk of influenza infection and transmission in the community. Hand hygiene practices are non-invasive and have broad applicability as an infection prevention and control intervention with no demonstrated evidence of harm.
Further research would help to clarify whether, and under what circumstances, hand hygiene interventions in the community are effective in preventing influenza infection and transmission.

Conclusion
Available evidence on the effectiveness of hand hygiene practices in preventing influenza infection and transmission in the community is inconsistent and insufficient in both quality and quantity. However, in light of its efficacy in general infectious disease prevention and control, there is no compelling evidence to stop using good hand hygiene practice to reduce the risk of influenza infection and transmission in the community.