The effectiveness of hand hygiene procedures in reducing the risks of infections in home and community settings including handwashing and alcohol-based hand sanitizers

Infectious GI disease and hygiene 

Other infectious GI disease 

From recent investigations, it is now recognized that a substantial proportion of the total infectious GI disease burden in the community is because of person-to-person spread within households, particularly for viral infections, where it is most often the cause. Person-to-person transmission in the home can occur by direct hand-to-mouth transfer, via food prepared in the home by an infected person, or by transmission because of aerosolized particles resulting from vomiting or fluid diarrhea. Apart from transmission by inhalation of airborne particles, these infections are preventable by good hygiene practice.

The 2003 WHO report9 stated that, of the total GI infection outbreaks (including foodborne disease) reported in Europe during 1999 and 2000, 60% and 69%, respectively, were due to person-to-person transmission. In the United Kingdom, it is estimated that up to 50% of GI infection results from person-to-person tranmsission.11 A study of United Kingdom outbreaks14 suggested that 19% of Salmonella outbreaks and more than half of E coli O157 outbreaks are transmitted by nonfoodborne routes.

National surveillance systems vary in their methods of data collection but mostly focus on foodborne disease. Inevitably, this means that data on GI illnesses relate mainly to large foodborne outbreaks in restaurants, hospitals, and others, whereas sporadic nonfoodborne cases in the general community go largely unreported. In the United Kingdom, even when “household” outbreaks are reported, they mostly involve home catering for parties and other functions and are therefore mainly foodborne outbreaks.15 Because milder cases of GI illness often go unreported, this means that the overall GI infection burden, particularly that which is not foodborne, is unknown; the most informative data on the overall burden of infectious GI illness (both foodborne and non-foodborne) in the community come from various community-based studies, which have been carried out in Europe and the United States and are reviewed below.

Two large community studies have been carried out in Europe: one in the United Kingdom and the other in The Netherlands. The UK study, carried out from 1993 to 1996 involving 460,000 participants in the community presenting to general practice, estimated that only 1 of 136 cases of GI illness is detected by surveillance. The study indicated that as many as 1 in 5 people in the general UK population develop GI illness each year, with an estimated 9.4 million cases occurring annually of which about 50% are nonfoodborne.11, 16 It was estimated that, for every 1 reported case of campylobacter, Salmonella, rotavirus, and norovirus, another 7.6, 3.2, 35, and 1562 cases, respectively, occur in the community; based on the number of laboratory reports, it is possible to estimate the true number of infections occurring in the community (Table 1).

Table 1. Estimated number of cases of infectious gastrointestinal disease in England and Wales associated with campylobacter, Salmonella, rotavirus, and norovirus

	Organism	Number of laboratory reports from fecal isolates in 2005	Ratio of actual reported cases	Estimated number of cases in the community
	Campylobacter	42,679	7.6	324,360
	Salmonella	11,191	3.2	47,763
	Rotavirus	13,306	35	567,790
	Norovirus	2607	1562	4,072,734

From the community study carried out in The Netherlands between 1996 and 1999,17 it was estimated that approximately 1 in 3.5 people experience a bout of infectious GI disease each year. Campylobacter was detected most frequently (10% of cases), followed by Ghiardia lamblia (5%), rotavirus (5%), norovirus (5%), and Salmonella (4%). Relative to the population of The Netherlands (16 million), 650,000 norovirus gastroenteritis cases occur annually.18

The US study of Mead et al,12 which also included data from community-based studies, indicated that the total number of cases of infectious GI illness annually is approximately 210 million (of which approximately 64% are nonfoodborne). They estimated that the number of episodes of acute gastroenteritis per person per year is approximately 0.79. From the available data, the authors were also able to estimate the proportion of total episodes that were nonfoodborne. As shown in Table 2, by far the most frequently reported causes of GI illness were norovirus, rotavirus, and campylobacter. For campylobacter, E coli, and norovirus, a significant proportion of cases was estimated as nonfoodborne, whereas, for hepatitis A (HAV), Shigella, and rotavirus, almost all cases were estimated as nonfoodborne. For Salmonella on the other hand, only 5% of cases were considered as nonfoodborne. Davis et al reviewed outbreaks of E coli O157 related to family visits to animal exhibits.19

Table 2. Estimated annual infectious gastrointestinal illnesses in the United States

		Total infectious GI illnesses	Infectious illnesses (%) that are nonfoodborne
	Norovirus	23,000,000	13,800,000 (60)
	Rotavirus	3,900,000	3,861,000 (99)
	Campylobacter	2,453,926	490,785 (20)
	Salmonella	1,412,498	70,624 (5)
	Shigella	448,240	358,952 (80)
	Hepatitis A	83,391	79,221 (95)
	E coli O157	73,480	11,022 (15)

Indications are that norovirus is now the most significant cause of infectious GI illness in the developed world, both outbreak related and endemic.20, 21 Currently, we are seeing increased outbreaks of norovirus, a major concern in Japan22 and also in Europe.23 Expert opinion is that norovirus strains now circulating are more “virulent” and more easily spread from person to person via hands and surfaces or during food handling.20

Infection with HAV is common worldwide,24 and adenovirus is also a frequent cause of gastroenteritis. C difficile-associated disease now occurs with increasing frequency in the community, in which it usually affects persons receiving antibiotic therapy but also healthy individuals.25 Recently, a new strain (027) of C difficile has emerged in North America, causing infections in the community among individuals with no predisposing factors.25 A recent study26 indicated that exposure to a family member with Helicobacter pylori gastroenteritis was associated with a 4.8-fold increased risk of infection in another family member and that infection most usually involved person-to-person transmission, associated with conditions of crowding and poor hygiene.

Using data from the 2006 E coli O157:H7 outbreak in 2006 in the United States associated with contaminated spinach, Seto et al developed a model that showed that secondary person-to-person transmission was similar to that in previous E coli outbreaks (≈12%). The model suggests that even a modestly effective hygiene promotion strategy to interrupt secondary transmission (prevention of only 2%-3% of secondary illnesses) could result in a reduction of ≈5% to 11% of symptomatic cases.27

Respiratory tract infections and hygiene 

Respiratory tract infections are largely caused by viruses. In the United States, viruses account for up to 69% of respiratory infections.28 The common cold is reported to be the most frequent, acute infectious illness to humans.29 Data from the United States suggest that the mean number of respiratory illnesses experienced per year in adults is approximately 1.5 to 3.0, and, in children under 5 years of age, it is approximately 3.5 to 5.5.28

Approximately 80% of upper RT infections are caused by rhinoviruses. Other species causing acute rhinitis are coronaviruses, parainfleunza viruses (PIV), respiratory syncytial viruses (RSV), and adenoviruses.30 Although colds are generally mild and self-limiting, they represent a significant economic burden because of loss in productivity and medical costs. Furthermore, secondary infections produce complications, such as otitis media, sinusitis, or lower respiratory infections including pneumonia, with its risk of mortality, particularly in elderly adults. Several studies have demonstrated that colds are also a trigger for asthma.31 RSV is the major cause of viral RT infection in young children worldwide. Child day care attendance in North America caries with it a very high risk of RSV infection within the first 2 years of life and accounts for 0.5% to 1.0% of hospitalized infants in the United States.32

Influenza is a more serious RT illness, which can cause complications that lead to increased physician visits, hospitalization, and death, the risks being highest among persons aged >65 years, children aged <2 years, and persons who have medical conditions (eg, diabetes, chronic lung disease).33, 34, 35 Influenza must also be considered in terms of days absent from work and school and pressure on health care services.35 An important aspect of influenza is the threat associated with the emergence of novel subtypes capable of causing an influenza pandemic.36 According to Bridges et al,37 influenza epidemics in the United States result in an annual average of 36,000 deaths and 114,000 hospitalizations; among those with influenza who belong to an “at-risk” group, a significant proportion develop pneumonia, and up to 1 in 10 can die of related complications. In Europe, the 2004-2005 influenza season annual report38 showed that, of 25 countries, 15 recorded what is regarded as high activity (150 up to 3000 influenza-like or acute respiratory illnesses per 100,000 population).

Although data indicating the role of hands and other surfaces in the transmission of colds have been available for some time, it is only in the last few years that there has been any real awareness that hands and surfaces may also be a transmission route for flu viruses.32 Evidence that measures such as hand hygiene can reduce spread of RT infections comes from the SARS outbreaks in Hong Kong, which coincided with the latter part of influenza season, when it was observed that, as extensive personal and community public health measures took place, influenza case numbers fell significantly, more so than usual for the time of year.39

Skin and wound infections and hygiene 

Skin and wound infections are common in the home and community, but most are self-limited. Because these infections, apart from S aureus infections go unreported, little or no data are available on the burden of skin and wound infections in the community. S aureus is the most common cause of infections of skin and soft tissue, which, in a small proportion of cases, lead to the development of bacteremia or pneumonia.40 Serious infections usually occur in health care facilities—in patients who are immunocompromised—in which S aureus is mostly usually associated with wounds and intravenous devices and in which the antibiotic-resistant strain, MRSA, is a major concern. Infected patients discharged from hospitals and health care workers (HCWs) caring for MRSA-infected patients can bring MRSA into the home and pass it on to healthy family members, who become colonized, thereby spreading the organism into the community and facilitating the circulation of these strains.41, 42, 43 MRSA colonization in an individual can persist for up to 40 months.44, 45

In recent years, MRSA has been increasingly found to cause infections in healthy members of the community without apparent risk factors.25 These CA-MRSA strains are different from health care-associated (HCA) MRSA strains and are a concern because they equally infect children and young adults. These strains primarily cause skin and soft tissue infections but can also cause invasive infections such as sepsis, pneumonia, and osteomyelitis, which is some cases can be fatal.25 Some CA strains are known to produce Panton-Valentine leukocidin (PVL), which has been implicated as a virulence factor,46 although opinion is, however, divided as to whether this is the case; whereas some studies support this notion,47 others do not.48 In the United States, CA-MRSA is now a significant concern. CA-MRSA strains have also now been detected in France, Switzerland, Germany, Greece, the Nordic countries, Australasia, The Netherlands, and Latvia.25 In the United Kingdom, cases of CA-MRSA and PVL-producing strains have been reported, but the number of reported cases is still small.25, 49

Health care and “at-risk” groups in the home 

Key factors that contribute to changing ID trends are the social and demographic changes that are occurring within the global population that affect our resistance to infection. “At-risk” groups cared for at home include not only newborn infants whose immune system is not fully developed but also the rapidly increasing elderly population whose immune system is declining. “At-risk” groups include patients discharged recently from hospital, immunocompromised family members, and family members with invasive devices such as catheters. It also includes people whose immunocompetence is impaired as a result of chronic and degenerative illness or because they are undertaking certain drug therapies. All of these groups, together with those who carry HIV/AIDS, are increasingly cared for at home by a caregiver, who may be a household member. A survey of the United States and 3 European countries—Germany, The Netherlands, and the United Kingdom—suggests that up to 1 in 5 of the population belongs to an “at-risk” group (Table 3). The data suggest that between 12% and 18% of the population of these countries are >65 years of age. In an intervention study of 148 patients with AIDS, it was found that patients assigned to the intensive handwashing intervention group developed fewer episodes of diarrheal illness (1.24 ± 0.9 vs. 2.92 ± 0.6 new episodes of diarrhea, respectively, during a 1-year observation period.50

Table 3. Prevalence of “at-risk” persons in the domestic setting

		United States	United Kingdom	Germany	The Netherlands
	Total population	290 million	60 million	82 million	16 million
	Over 65 years of age	35.6 million	9 million	13 million	2 million
	Living with cancer: significant proportion in the community, and undergoing chemotherapy	2 million	1 million	-	160,000
	Under 1 year of age	35.6 million	600,000	800,000	100,000
	Discharged from hospital within previous 2 weeks	1.25 million	200,000	-	60,000
	Hospital outpatients at home	-	-	1,270,000	-
	AIDS cases∗	40,000	15,000	-	91
	People in home care	0.5 million	-	-	-
	Total “at-risk” persons	>1 in 7	>1 in 6	>1 in 5.6	>1 in 6.3

∗ This does not include those who are HIV positive, who may also have lowered resistance to infection.

GI pathogens are now implicated as causative or contributory factors in the development of cancers and other chronic conditions; examples include hepatitis B virus (hepatocelluar carcinoma),51 H pylori (peptic ulcer disease),52 and Campylobacter jejuni (Guillain Barré syndrome).53 Foodborne illness has been estimated to result in chronic sequelae in 2% to 3% of cases54; a European Commission report55 cites evidence of chronic disease, such as reactive arthritis, following 5% of Salmonella cases, with 5% of E coli O157 cases progressing to serious and even fatal complications. Even mild viral infections can be predisposing factors to more severe and possibly fatal secondary bacterial infections.56

Microbiologic studies of the spread of pathogens via hands and other surfaces 

In recent years, a range of studies has been published, many related specifically to the home, which indicate the extent to which ID agents occur and are spread in home and community settings during normal daily activities and their potential to cause infection. These studies include assessments of frequency occurrence of sources of pathogens in the home, their rate of “shed” from an infected source into the environment, their rate of die away on hands and other surfaces, their rate of transfer via the hands to the mouth, nose, conjunctiva, and others and/or to ready-to-eat foods, and their the infectious dose. The infectious dose (ie, the number of particles to which the recipient is exposed), their immune status, and the route by which they are infected are key factors that determine the infection risk. The “infectious dose” varies for different pathogens and is usually lower for people who are “at-risk” than for healthy household members.

Transmission of infectious GI disease 

Risks from exposure to GI pathogens via the hands 

As shown in Fig 3, exposure to GI pathogens can occur by direct hand-to-mouth contact or indirectly via contaminated food. In the home, food can be contaminated either directly by an infected food handler or indirectly by cross contamination via hands and surfaces from another source, which may be contaminated food, another infected household member (or carrier), or a household pet or farm animal. Hand-to-mouth contact is a frequent occurrence, particularly among children; a study of mouthing behavior in 72 young children showed that children <24 months of age exhibit the highest frequency, with 81 events/hour; for children >24 months of age, this was reduced but was still of the order of 42 events/hour.63

The potential for transmission of pathogens from hands to ready-to-eat foods is supported by a number of studies:

•In a model domestic kitchen, 29% of food preparation sessions using campylobacter-contaminated chicken resulted in positive campylobacter isolations from prepared salads, cleaning materials, and food contact surfaces.64

•Bidawid et al65, 66 showed that touching lettuce with finger pads contaminated with HAV and feline calicivirus (FCV), used as a surrogate for norovirus, for 10 seconds resulted in transfer of 9.2% and 18%, respectively, of the virus. Based on the load for HAV in feces (106 to 109 viral particles/g), an estimated 1300 particles were transferred to the lettuce.

•Rusin et al showed that, when volunteers' fingertips were inoculated with a pooled suspension of Micrococcus luteus (M luteus), Serratia rubidea (S rubidea), and bacteriophage PRD-1 and held to the lip area, transfer rates were 40.99%, 33.97%, and 33.90%, respectively.67

As stated above, the infection risk from oral consumption depends on the number of bacterial cells or viral particles that are consumed. Table 4 shows that, for many of the commonly occurring GI pathogens, the infectious dose is relatively small.

Table 4. Infectious doses for gastrointestinal pathogens

	Organism	Infectious dose
	Salmonella species	Up to 106 but could be as low as 10-100 cells.68 Contamination may be amplified by transfer to foods, which are then stored incorrectly.
	Campylobacter species	500 organisms can result in human illness.69
	E coli 0157	Oral dose for E coli 0157 may be as little as 10 cells.70 In one outbreak, a median dose of <100 organisms per hamburger was reported.71
	Norovirus	10-100 units or even less.72
	Rotavirus	May be as few as 10 particles.73 Ward et al showed that 13 of 14 adults became infected after consuming rotavirus (103 particles) picked up from a contaminated surface via the hands.73

Sources and spread of GI pathogens to the hands 

Figure 3 illustrates that the risk of exposure to GI pathogens via the hands depends on the extent to which these pathogens are brought into the home (either by infected people or pets or via contaminated food) and the extent to which they are spread via hands and other surfaces and by airborne transmission. Relevant data from various sources, as summarized below, suggest that exposure to GI pathogens via the hands is a frequent occurrence during normal daily activities and that the numbers of organisms transferred by hand-to-mouth contact can be well within the numbers required to cause infection.

Household members who are infected, or who are carriers, are a primary source of infection in the home. Pathogens that can be carried persistently by otherwise healthy people include Salmonella species and C difficile. Approximately 3% of adults (mainly those >65 years of age), and up to two thirds of babies, are known to carry C difficile in their gut, although it is not known what proportion are toxin producing.25 People or animals that carry GI pathogens shed large numbers of organisms in their feces or when they vomit. A single vomiting incident following norovirus infection may produce 30 million viral particles,72 and, at the peak of a rotavirus infection, >1011 virions may be excreted per gram feces.74 Surfaces in the home may become contaminated by enteric organisms that are aerosolized during vomiting or by transfer of vomitus and fecal matter via hands. Viruses aerosolized from flushing the toilet can remain airborne long enough to contaminate surfaces throughout the bathroom.75

Infectious agents introduced into the home via food include Salmonella, campylobacter, listeria, and E coli O157. A variety of foods can act as a source of these organisms, including meat, fish and poultry products, dairy products, fruits, and vegetables. Organisms in particles, and moisture or juices, from food will contaminate any surface they come into contact with. An EFSA survey76 of Salmonella in chicken indicates significant differences among EU member states, with isolation rates between 0% and 68.2%; the level reported for the United Kingdom was 7.1% to 9.4%. The EFSA also reported that up to 66% of samples from fresh poultry were positive for campylobacter.77 In the United States, more than half of raw chicken is estimated to be contaminated with camplylobacter.76 Chapman et al78 showed that 0.4% to 0.8% of meat products purchased from UK butchers were positive for E coli O157. In a recent study in Canada, C difficile was isolated from 20% of 60 samples of retail ground meat purchased over a 10-month period, and 11 isolates were toxigenic.79

The home is frequently a shelter to a range of different pets; more than 50% of homes in the English-speaking world have cats and dogs, with 60 million cats and dogs in the United States. In the United States, up to 39% of dogs may carry campylobacter, and 10% to 27% may carry Salmonella80; cats are also carriers of these organisms. Carriage of C difficile in household pets is quite common; up to 23% of pets are affected, although these mostly involve noncytotoxigenic strains.25

Kramer et al,81 Sattar et al,82 and Rzezutka and Cook83 reviewed data showing that GI pathogens can survive on surfaces for several hours and, in some cases, days, particularly on moist surfaces, although infectivity depends on the numbers that survive (Table 5).

Table 5. Persistence of gastrointestinal pathogens on dry inanimate surfaces

		Duration of persistence (range)
	Type of bacteria
	Campylobacter jejuni	Up to 6 days
	C difficile (spores)	5 months
	Escherichia coli	1.5 hours to 16 months
	Helicobacter pylori	Less than 90 minutes
	Listeria species	1 day to months
	Salmonella species	1 day
	Shigella species	2 days to 5 months
	Type of virus
	Norovirus and feline calicivirus	8 hours to 7 days
	Rotavirus	6-60 days

Studies to quantify transfer between hands, foods, and kitchen surfaces67, 84 showed that transfer rates were highly variable, ranging from as high as 100% to as low as 1%. Transfer to hands was highest from nonporous surfaces but lower from surfaces such as carrots, sponges, and dishcloths (<1%). Rusin et al67 sampled volunteers hands after touching surfaces contaminated with M luteus, S rubidea, and phage PRD-1. Activities included wringing out a dishcloth/sponge, turning off a faucet, cutting up a carrot, making hamburger patties, holding a phone receiver, and removing laundry from the washing machine. Transfer efficiencies for the phone receiver and faucet were 38% to 65% and 27% to 40%, respectively. Paulson85 showed that, when gloved hands were contacted for 5 to 10 seconds with surfaces such as cutting boards and doorknobs contaminated with FCV (log 5.9 particles), the log number of particles recovered from hands was 4.7 to 5.4.

These laboratory studies are supported by a range of field studies showing spread via the hands and other surfaces during normal daily activities:

•Following preparation of Salmonella and campylobacter-contaminated chickens in domestic kitchens, these species were isolated from 17.3% of hands and hand and food contact surfaces. Isolation rates were highest for hands, chopping boards, and cleaning cloths (25%, 35%, and 60%, respectively, of surfaces sampled).86

•In homes containing an infant recently vaccinated for polio (during which time shedding occurs in feces), virus was isolated from 13% of bathroom, living room, and kitchen surfaces.87 Most frequently contaminated were hand contact sites such as bathroom taps, door handles, toilet flushes, soap dispensers, nappy changing equipment, and potties.

•Following handshaking with a volunteer whose hands were contaminated from touching a virus-contaminated door handle, successive transmission from one person to another could be followed up to the sixth person.88

•Where fingers were contacted with norovirus-contaminated fecal material, the virus was consistently transferred via the fingers to melamine surfaces and from there to hand contact surfaces, such as taps, door handles, and telephone receivers. Contaminated fingers sequentially transferred the virus to up to 7 clean surfaces.89

•A study90 with FCV showed survival for up to 3 days on telephone buttons and receivers, for 1 or 2 days on computer mouse, and for 8 to 12 hours on keyboard keys and brass disks representing faucets and door handles. The time for 90% virus reduction was <4 hours on computer keys, mouse, and brass disks; 4 to 8 hours on telephone receivers; and 12 to 24 hours on telephone buttons.

•In homes of infants with recurrent C difficile infection, 12% of environmental surfaces were positive for C difficile, and 1 of 4 other household members carried C difficile in stool. In a control home with no household carriers, none of 84 environment samples were positive for C difficile.91

•In 4 out of 6 homes in which there was a Salmonella case, the causative species was isolated from fecal soiling under the flushing rim and scale material in the toilet bowl for up to 3 weeks after notification of infection.58 Flushing toilets seeded with Salmonella enteritidis resulted in contamination of hand contact surfaces such as toilet seats and toilet seat lids.

These represent recent examples of studies that have been reported. These and other studies are also reviewed elsewhere.82, 92, 93, 94

In developing hygiene policies for preventing GI infections, one of the difficulties is assessing risks associated with hand transmission relative to other risks such as inadequate cooking or storage of food or inhalation of infected vomit particles. Gillespie et al15 reported an evaluation of reported outbreaks linked to UK households for 1992 to 1999 that suggested, of the 85% of outbreaks designated as foodborne, cross contamination was implicated in 20% of outbreaks compared with 30% and 31% of outbreaks for which inadequate storage and cooking, respectively, were thought to be the cause. There were no data to suggest what percentage of cross contamination events involved the hands, and Gillespie et al15 expressed concern that most of the reported outbreaks were linked to home catering, thus not necessarily representative of normal daily routine. Aerosol transmission can result from settling on hand and food contact surfaces, but, for norovirus, infection can sometimes result from direct inhalation of infected particles of vomit by people immediately adjacent to the person who vomits. The potential for airborne transmission of norovirus was demonstrated in studies in a restaurant and a primary school, in which close proximity to infected persons in the immediate aftermath of a vomiting attack was identified as a risk factor.95, 96

Transmission of RT infections 

The last 2 years have seen an unprecedented global focus on developing strategies for preventing transmission of influenza. The WHO97 is taking a lead on pharmaceutical interventions such as vaccines and antivirals but has also made recommendations for other interventions,98 which include highlighting the importance of hygiene, and in particular hand hygiene, in minimizing spread in the home and community.

Risks from exposure to respiratory pathogens via the hands 

As shown in Fig 3, exposure to RT viruses can occur either by inhalation of infected mucous or inoculation of the nasal mucosa or eyes with virus-contaminated hands, which then cause infection via the mucous membranes and upper RT. Rhinovirus and RSV are deposited into the front of the nose or into the eye (where they pass down the lacrymal duct), either on the end of the finger or possibly sometimes in aerosolized droplets.99 Rubbing the eyes and nose with the fingertips is a common occurrence; Hendley et al found that 1 in 2.7 attendees of hospital rounds rubbed their eyes, and 33% picked their nose, within a 1-hour observation period.100

A review of the data94 (Table 6) suggests that the infectious dose for respiratory viruses is relatively small. Alford et al suggest that aerosolized doses of as little as 1 TCID50 (tissue culture infective dose) of influenza virus could infect volunteers.101

Table 6. Infectious doses for viruses that cause respiratory diseases∗

	Virus	Minimal infectious dose associated with intranasal inoculation
	Respiratory syncytial virus	100-640 TCID50
	Rhinovirus	0.032-0.4 TCID50; also cited as 1-10 TCID50
	Influenza	2-790 TCID50
	Parainfluenza	1.5-80 TCID50

∗ From Boone and Gerba.94

Evidence for transmission of rhinovirus and RSV infections via contaminated hands comes from a number of studies:

•A number of studies have demonstrated that self-inoculation by rubbing the nasal mucosa or conjunctivae with rhinovirus-contaminated fingers can lead to infection.100, 102 Over a period of 10 years, Gwaltny and Hayden performed intranasal challenges on 343 healthy young adults who had no antibody to the challenge, and infected 321 (95%).103 After handling contaminated coffee cups and other objects, more than 50% of subjects developed infection.104 Hall et al showed that volunteers touching contaminated objects and/or the fingers of symptomatic individuals had a higher attack rate of colds if they touched their eyes or nose.105

•In a 4-year family trial, Hendley and Gwaltney104 found that prophylactic treatment of mothers' fingers with iodine reduced the incidence of RT infections. When illness occurred in the family, mothers were instructed to dip their fingers in iodine upon awakening in the morning, then every 3 or 4 hours or after activities that washed the iodine from the skin. The secondary attack rate in mothers was 7% in the iodine group and 20% in placebo families. No infections occurred in mothers after 11 exposures to an infected index case in the iodine group, compared with 5 infections after 16 exposures in the placebo group.

•Hall et al showed that infected infants excrete prodigious amounts of RSV in their nasal secretions for several days105 and that RSV could be recovered from hands that had touched surfaces contaminated with secretions from infected infants.99 Hall and Douglas found that close contact with symptomatic infants who were producing abundant secretions, or their immediate environment, was necessary for infection.106

Sources and spread of RT pathogens to the hands 

Figure 3 illustrates that the risk of exposure to RT pathogens via hands depends on the extent to which these pathogens are spread from an infected person during normal daily activities. Relevant data come from various sources and are summarized below. Taken together, the data suggest that, when a household member is infected, exposure of other household members via hands is likely to occur during normal daily activities and that the numbers of organisms involved are within those required to initiate infection if transferred to the eyes or nose.

People infected with cold viruses shed large quantities of virus-laden mucus. Droplets of nasal secretions generated by coughing, sneezing and talking can travel over a distance >3 m to contaminate surrounding surfaces.37, 98, 107, 108, 109 Up to 107 infectious influenza particles per milliliter has been detected in nasal secretions.110 The mean duration of a cold is 7.5 days. Viral shedding may occur 24 to 48 hours before illness onset but generally at lower titers than during the symptomatic period. Titers generally peak during the first 24 to 72 hours of illness and decline within several days, with titers low or undetectable by day 5. Children can shed virus for up to 3 weeks, whereas immunocompromised people may continue to shed virus for weeks to months.98

Infectious material can also be deposited directly on hands and tissues during sneezing and blowing the nose. Contamination of hands can occur by handshaking or touching contaminated surfaces. Pathogens shed into the environment from these sources can survive for significant periods and are readily spread around the home to and from the hands and via handkerchiefs and tissues, tap and door handles, telephones, or other hand contact surfaces:

•Gwaltney and Hendley demonstrated that most subjects with experimental colds had rhinovirus on their hands and that virus could be recovered from 43% of plastic tiles they touched.104 For people with rhinovirus colds, virus was found on 39% of hands and 6% of objects in their immediate environment.100 Reed demonstrated recovery of virus from naturally contaminated objects in the surroundings of persons with rhinovirus colds.102

•In a recent study, Winther et al111 recruited volunteers suffering from colds to stay overnight in hotel rooms. After checkout, but before room cleaning, 10 objects identified as frequently touched were sampled for rhinovirus. Virus was found on 35% of objects, including door handles, light switches, pens, faucet and toilet handles, and television remote controls. Some people contaminated none or few sites, most contaminated several, and some contaminated almost all (up to 8) sites. In a second study in which the same subjects stayed overnight in a hotel room in which hand contact surfaces (light switch phone button and handset) had been contaminated with rhinovirus-contaminated mucus, 60% of subjects became contaminated with rhinovirus.

•Ansari et al showed that hands readily pick up rhinovirus and PIV by touching contaminated objects.112 As much as 70% of infectious rhinovirus has been shown to transfer to a recipient's fingers after contact for 10 seconds.113 In a study with volunteers who handled contaminated doorknobs or faucets, recovery rates from 3 to 1800 plaque-forming units of rhinovirus were obtained from fingertips.114

•In a study of US day care centers and domestic homes, influenza A virus was detected on 23% of day care center surfaces sampled during the fall of 2003 and 53% of surfaces sampled during the spring. Although no influenza was detected on home surfaces during the summer, influenza was detected on 59% of surfaces sampled during March in 5 homes in which there was an influenza-infected child. No virus was recovered from 3 other homes in which all household members were healthy. Influenza virus was recovered most frequently from telephone receivers (80%) and least frequently from computer keyboards (40%). Other surfaces found to be contaminated included refrigerators; kitchen faucets; light switches; microwaves; TV remote controls; doorknobs; and bath, faucet, and toilet handles. Influenza virus was recovered from 69% of the day care center diaper changing areas, indicating presence of virus in infant feces.94

Transfer of RT infections via contaminated hands depends on the ability of the virus to survive and retain its infectivity outside the human host. The potential for survival varies significantly between nonenveloped rhinovirus and RSV, compared with the enveloped influenza virus and PIV:

•Kramer et al81 and Hendley et al100 review data showing that rhinovirus and RSV can survive for significant periods (2 hours to 7 days for rhinovirus, up to 6 hours for RSV) on dry surfaces and for at least 2 hours on human skin.

•Ansari et al115 and Brady et al116 showed that, although PIV can survive on nonabsorbent surfaces for up to 10 hours, survival on hands was relatively poor (1-2 hours).

•Bean et al117 showed that influenza virus could survive up to 24 to 48 hours on nonporous surfaces and up to 8 to 12 hours on cloth, paper, and tissues. By contrast, virus could be recovered from hands for only 5 minutes and then only if hands were contaminated with high viral titers. Virus could be transferred from nonporous surfaces to hands for 24 hours and from tissues to hands for 15 minutes. Higher humidity shortened virus survival. Virus on nonporous surfaces could be transferred to hands 24 hours after the surface was contaminated, whereas tissues could transfer virus to hands for 15 minutes after the tissue was contaminated. On hands, virus concentration fell by 100- to 1000-fold within 5 minutes after transfer.

Opinion as to the importance of the hands relative to the airborne route for transmission of rhinovirus colds is divided. Some investigators30, 99, 103, 104 maintain that contamination of the hands followed by inoculation of the eyes or nose is of paramount importance; in fact, Gwaltney et al found that it was exceedingly difficult to transmit virus orally or by kissing and found little evidence of droplet or droplet nuclei transmission.100, 113 Others maintain that the evidence favors droplet and droplet nuclei transmission as the most important mode of spread.118 For RSV, there is general agreement that the hands are the primary route for the spread of infection.32, 105, 106

For influenza, although more data are needed, it is increasingly accepted that not only airborne (both true airborne transmission involving droplet nuclei [<5 μm in diameter] and “droplet transmission” involving droplets >10 μm that deposit onto surfaces quite rapidly) but also surface (including hand) transmission come into play.98, 119, 120 The relative contribution of each mode of transmission is unknown but appears to vary depending on the circumstances, symptoms, respiratory tract loads, and the viral strain.121 Data from animal studies and influenza outbreaks suggest that droplets generated when infected persons cough or sneeze are the predominant mechanism of airborne transmission,37 although data supporting droplet nuclei spread are also available.101, 122, 123, 124 It is possible, however, that influenza is less transmissible via hands and surfaces compared with rhinovirus and others because of its lower ability to survive outside a human or animal host. Data suggest that, to some extent, airborne droplets and droplet nuclei cause infection as a result of settling on hand contact surfaces. The frequent occurrence of diarrhea and the detection of viral RNA in fecal samples tested suggest that the H5N1 influenza virus may replicate in the human gut and could be a source of transmission via hands and surfaces.125 At present, however, it is thought that this is unlikely. The growing evidence base related to the survival, transmission, and human exposure to RT viruses via hands and other surfaces is also reviewed elsewhere.32, 35, 37, 82, 94, 99

Intervention studies to establish the causal link between hand hygiene and infectious disease in the home and community 

Both observational and interventional study designs have been used to assess the relationship between hand hygiene and ID transmission. By definition, observational studies are not randomized and must utilize careful methods to preserve internal validity. Control of confounding and the potential for selection, recall, and other biases are also a concern, for example, individuals who wash their hands less frequently are also less likely to report symptoms. Intervention studies on the other hand compare infection rates in groups in which handwashing is, or is not, promoted. Intervention studies employing randomization of treatment groups have been considered the “gold-standard” in terms of reducing selection biases. These studies have the ability to ensure that randomized groups are similar, apart from treatment allocation and differences that occur by chance. For these reasons, we limit discussion to intervention studies, focusing on GI and RT illnesses, because these are the most common infectious illness symptoms in home and community settings.

A range of intervention studies have been carried out to evaluate the causal link between handwashing and ID transmission and have been reviewed in a series of papers to assess the consistency and strength of the link.151, 152, 153 Overall, these studies indicate a strong and consistent link between handwashing and GI disease and a significant link between handwashing and RT illnesses. For the most part, these studies have been carried out in child day care centers, schools, and military and other public settings in which the outcome is often measured against a high baseline level of infection. Relatively few studies have been carried out in household settings in the United States and Europe. Difficulties associated with studying households in developed areas include fewer children under the age of 5 years, higher level of hygiene infrastructure, and difficulties in collecting data. Given that there are likely fewer susceptible individuals clustered within household settings, the prevalence of GI and RT illnesses is relatively much lower, making it more difficult to detect a significant influence of hand hygiene on the occurrence of illness.

Whereas some intervention studies are not relevant to this review and have been omitted, others give useful insight into the potential impact of handwashing in the home and in the general community. Studies that are included have been selected on the basis of whether transmission routes are likely to reflect those in the home, most particularly whether the relative rates of transmission via these routes (as shown in Fig 2) are likely to be similar. For this reason, studies on GI infection in developing countries have been excluded; in these settings, limited access to sanitation means that rates of direct hand-to-mouth transmission from feces is high relative to other routes of transmission (eg, person-to-person transmission via hands, or inadequate food hygiene), compared to settings with adequate water and sanitation in which transmission is more likely to involve person-to-person transmission and transmission via food, rather than direct feces-to-hand-to-mouth. For GI illnesses, we have, therefore, focused on studies carried out in developed country communities, although, even for studies such as those in child daycare centers, in which food preparation is not undertaken by study participants, the data probably reflect mainly the impact on person-to-person transmission. For RT infections, studies conducted in both developed and developing countries are included on the basis that relative rates of airborne transmission versus transmission via hands are likely to be similar regardless of setting.

In a recent review, Aiello et al assessed the relationship between handwashing and GI outcomes154 focusing on studies conducted in North America and Europe. Table 7 summarizes studies providing an effect estimate (risk ratio, rate ratio, and others) as well as 95% confidence intervals (95% CI). In all of the studies, handwashing with soap was the factor studied, although in some, this was combined with hygiene education measures. All studies assessing handwashing and hand hygiene education were conducted in school or day care settings.

Table 7. Intervention studies evaluating the impact of hand hygiene education or handwashing on reductions in infectious illnesses

	Intervention	Setting, country	Outcome measured	Result, % reduction (95% CI)	Statistical significance	Reference
	HW + HE	Child care, United States	Diarrhea	48 (0.23-0.65)	Yes	Black et al, 1981155
	HW + HE	Child care, United States	Diarrhea	12 (−0.10-0.30)	No	Bartlett et al, 1988156
	HW	Elementary school, United States	Gastrointestinal symptoms	57 (0.27-0.75)	Yes	Master et al, 1997157
	HW + HE	Child care, Canada	Diarrhea	−10 (−0.50-0.19)	No	Carabin et al, 1999158
	HW + HE	Child care, Australia	Diarrhea	50 (0.32-0.64)	Yes	Roberts et al, 2000159

HW + HE indicates that the intervention arm received a hygiene education component that included handwashing; HW indicates that increased frequency or a scheduled handwashing was the only difference between the intervention and control group. HW, handwashing; HE, hygiene education; CI, confidence interval.

Among the studies in Table 7, the reduction in GI illness associated with handwashing ranged from −10% to +57%. However, 3 of the 5 studies were not statistically significant, including the study that identified a value of −10%. The studies that gave statistically significant results all describe reductions close to 50%. Overall, these reviews suggest a consistent causal relationship between handwashing and reduction in GI illness, although the findings are less consistent and of a lesser magnitude than in lesser developed settings in which studies considered statistically significant suggested reductions from 26% to 79%.154

In assessing RT infections, the reviews of Aiello and Larson151 and Aiello et al,154 mentioned above, examined both RT and GI illness outcomes, whereas 3 other systematic reviews focused solely on the relationship between hand hygiene and RT infections. A review by Lee et al, assessing the relationship between several nonvaccine interventions and prevention of acute RT infection,160 concluded that the promotion of hand hygiene may be useful for preventing RT disease. Rabie and Curtis in 2006 also published a review of hand hygiene studies involving RT infections.161 They reported that hand hygiene (handwashing, education, and waterless hand sanitizers) can reduce the risk of respiratory infection by 16% (95% CI: 11%-21%). These investigators have now updated their estimate with 2 further, more recent, studies that, when all studies are taken together, give a pooled impact on respiratory infection of 23%.162

Based on these studies, Table 8 summarizes the results of community-based interventions (excluding health care-related and military settings) on RT illnesses. Most studies were conducted in economically developed countries (83%, 5/6). The range of reduction in illness was 5% to 53%, but only 33% (2/6) of the studies were statistically significant. The results suggest that hand hygiene education and promotion of handwashing can reduce rates of RT illnesses, but the impact is less than for GI infections, although it must be borne in mind that the available data are more limited.154

Table 8. Intervention studies evaluating the impact of hand hygiene education or handwashing on reductions in respiratory illnesses

	Intervention	Setting, country	Outcome measured	Result, % reduction (95% CI)	Statistical significance	Reference
	HE	Elementary school, United States	Respiratory illness-related	53 (−1.38-0.91)	No	Kimel,163 1996
			Absenteeism
	HW + HE	Elementary school, United States	Respiratory illness-related	21 (−0.02-0.39)	No	Master et al,157 1997
			Absenteeism
	HE	Child care, United States	Colds	32 (−0.04-0.56)	No	Niffenegger,164 1997
	HE	Child care, Canada	Respiratory	20 (0.07-0.32)	Yes	Carabin et al,158 1999
	HE	Child care, Australia	Respiratory	5 (−0.01-0.11)	No	Roberts et al,165 2000
	HW + HE	Community, Pakistan	Acute respiratory tract infections	51 (0.49-0.53)	Yes	Luby et al,166 2005

HW, handwashing; HE, hygiene education; CI, confidence interval.

There are also several studies of handwashing that do not distinguish between GI and RT outcomes.157, 167, 168 These studies measure outcomes such as illness-related absenteeism, making it difficult to assess the impact on specific disease etiologies. Of these studies, only one reported a significant reduction (25%).157 Two were conducted in day care centers, and 1 was conducted in an elementary school. All 3 studies were conducted in economically developed areas (United States, Sweden, and Israel).157, 167, 168

Several methodologic issues must be considered for these studies. Studies that use randomization are more likely to produce study groups with similar baseline characteristics. Surprisingly, 40% of the 11 studies in Table 7, Table 8 did not randomize. In some studies, randomization may not be an option (eg, in community settings) because the intervention is too complicated to randomize to multiple groups rather than assigning it to a single geographic area. Controlling for potential confounding variables is also an important issue, for example, if a study did not control for age and included adults as well as children, the effect of a hygiene intervention may be diluted because adults are at lower risk for diarrheal disease compared with children. In randomized studies, adjustment for confounding in the statistical analysis may not be required if potential confounders associated with intervention and control groups appear balanced, for example, randomization of households in the same geographic area may produce intervention and control arms with the same age distributions, hygiene habits, and health profiles. As summarized in Table 9, of the 11 studies, only 50% (9/18) reported controlling for at least 1 potential confounding factor. Although masking (also known as blinding) can be difficult to implement in hygiene studies because subjects, observers, and interviewers are usually aware of the intervention status, a few studies (2/18) were able to employ masking to reduce knowledge of the intervention. Masking can reduce biases associated with knowledge of intervention, including changes in behaviors, practices, and data collection methods.

Table 9. Methodologic considerations for the 11 intervention studies evaluating the impact of hand hygiene education or handwashing on reductions in respiratory and gastrointestinal illnesses

	Percentage randomized	Percentage adjusted for confounding	Percentage masked	Percentage controlled for clustering∗
	60	50	10	50

∗ These studies either controlled statistically for clustered units or the unit of analysis reflected the unit of measurement.

For intervention studies, disregarding clustered sample design may cause bias. For example, a handwashing program in a day care center may affect a child's risk of disease through its individual-level effect (the effect of handwashing of a child on his or her own risk of disease) and through its group-level effect (the effect of center-wide handwashing on risk of disease, even if the child is not following the handwashing program). Clustered interventions must take into account the grouped data structure in subsequent analyses or must analyze data at the same level at which it was collected. If the group or cluster is not controlled for, through specialized techniques such as generalized estimating equations, or analyzed at the group unit of measurement (average classroom illness rate), the investigator risks making a type I error (eg, falsely concluding that the effect of the intervention is significant, when there is no significant difference). Although most handwashing studies employ clustered data structures, use of clustered data statistical techniques have only become more prevalent since 2000.

Efficacy of handwashing using soap and water 

In vivo “panel test” studies of the effectiveness of handwashing 

In Europe, the efficacy of handwashing is established by panel tests that determine the reduction in the number of organisms released from artificially contaminated hands. The test applicable to handwash products is the Committee European Normalisation Hygienic Handwash Test EN1499.170 In this test, E coli is inoculated onto the hands and dried. The handwash product is applied to the hands with a rubbing action for either 30 seconds or 1 minute. The residual number of bacteria present on the hands is assessed pre- and postwash by a rinse sampling process and the log reduction determined. To make a claim that a product is a hygienic handwash, it must produce a log reduction in release of E coli from the hands at least equivalent to that produced by a reference soft soap product (mean, 2.76 log in 1 minute; range, 2.02-4.27). In the United States, handwashing is evaluated by a similar panel test using Serratia marcescens as the test organism. The test applicable to consumer handwash products is the American Society for Testing Materials (ASTM) Standard Method for evaluation of Healthcare Handwash Formulations E1174.171 The product, when evaluated by this method, must produce a 2-log reduction after 5 minutes.

A range of studies, based on these methodologies, has been carried out to determine the efficacy of handwashing, and are reviewed by Boyce and Pittet,172 Kampf and Kramer,173 and Sickbert-Bennet et al.169 From their assessment, Kampf and Kramer estimated that handwashing produced a mean reduction of up to 2.4 log within 1 minute. Data from individual studies are summarized in Table 10 and suggest that, for E coli, the greatest reduction is achieved within the first 30 seconds, ranging from 0.6 to 1.1 log after 15 seconds to 1.8 to 2.8 log after 30 seconds. Extending the washing time to 1 minute produces a reduction of 2.6 to 3.23 log, but increasing the process for more than 1 minute does not appear to gain any additional reduction. Relatively few data are available on the effectiveness of handwashing in removal of viruses, but the available data (Table 10) suggest that handwashing may be less effective for viruses compared with bacteria.

Table 10. Effect of handwashing on transient hand flora∗

	Microorganism	Duration of handwash	Mean log reduction
	E coli	10 seconds	0.5
		15 seconds	0.6-1.1
		30 seconds	1.37-3.0
		1 minutes	2.6-3.2
		2 minutes	3.27
	S marcescens	10 seconds	1.87
	S aureus	30 seconds	0.5-3.0
	P aeruginosa	30 seconds	2.0-3.0
	C difficile	10 seconds	2.0-2.4
	Poliovirus	30 seconds	1.9
	Rotavirus	10 seconds	0.14
		30 seconds	1.17-1.19
	Bacteriophage MS2	10 seconds	1.82

∗ From Kampf and Kramer,173 Sickbert-Bennet et al,174 and Schurmann and Eggers.175

Although panel test data suggest that handwashing efficacy is similar across a range of bacterial species, some field-based studies suggest that efficacy can vary quite significantly. In some cases, organisms can be attached to the hands too firmly and may not be removed by handwashing. A study of the spread of Salmonella and campylobacter from contaminated chickens via hands during handling and preparation in a kitchen176 showed that, although campylobacter were efficiently released from the hands by a 30-second rub and rinse process, a 2-minute process was necessary to eliminate Salmonella. The hand rinsing process is also important; Cogan et al86 showed that, following preparation of Salmonella and campylobacter-contaminated chickens in domestic kitchens, 15.3% of hands and hand and food contact surfaces still showed evidence of contamination even after participants had carried out a washing-up routine with detergent and hot water and then used a cloth to wipe surfaces. Sites contaminated most frequently were hands (20%); dishcloths, utensils, and tap handles (25%); and sink surrounds (30%). These results were confirmed in further studies176, 177 in which, after cleaning up with a typical routine involving a bowl of hot soapy water and a cloth, although isolation rates from hands of participants were 5% (1/20) for campylobacter, 45% (9/20) of participants still had Salmonella on their hands, and, on 3 occasions, counts recovered were >103 colony-forming units. In a further study in which participants cleaned up in the same way but then rinsed their hands under running water for 10 seconds, no samples were positive for campylobacter. However, 15% (3/20) still had low numbers of Salmonella isolated from their hands. Larson et al showed that the quantity of soap (1 mL and 3 mL) used can also have an impact on the microbial reduction achieved by handwashing.178

Bidawid et al65, 66 studied the impact of handwashing in preventing transfer of HAV and FCV from artificially contaminated finger pads to pieces of lettuce (Table 11). Touching the lettuce for 10 seconds resulted in transfer of 9.2% and 18%, respectively, of the virus. When finger pads were washed before the lettuce was touched, the amount of virus transferred was reduced to 0.39% and 0.4%, respectively. Amounts of HAV and FCV remaining on treated finger pads were 6.5% and 7%, respectively. Surprisingly, virus transfer to lettuce when the finger pads were rinsed with water alone was between 0% and 0.3%, depending on the volume of water used for rinsing.

Table 11. Efficacy of handwashing and alcohol-based hand sanitizers in preventing transfer of hepatitis A and feline calicivirus from fingertips to lettuce

		Hepatitis A		Feline calicivirus
		Particles transferred to lettuce, %	Particles remaining on hands, %	Particles transferred to lettuce, %	Particles remaining on hands, %
	No treatment	9.2		18
	Handwashing	0.39	6.5	0.4	7
	62% Ethanol, 20 seconds	0.64	64	2.1	13
	75% Ethanol, 20 seconds	0.46	24	1.2	17

Barker et al showed that a thorough 1-minute handwash with soap was sufficient to eliminate norovirus from fecally contaminated hands to levels that gave negative reverse-transcription polymerase chain reaction assays.89 However, Schurmann and Eggers175 concluded that enteric viruses, particularly poliovirus, may be more strongly bound to the skin and that the inclusion of an abrasive substance in handwash preparations is needed to achieve effective removal. Handwashing was also found to be ineffective in eliminating adenovirus from hands of a physician and patients.179

For handwashing, a hand-rubbing time of 15 seconds with soap is generally recommended, although the data in Table 10 indicate that 30 seconds to 1 minute is needed to achieve the optimum of 2- to 3-log reduction. In practice, it is doubtful whether people comply with even a 15-second handwash, although there are few data to confirm this. A study of 224 healthy homemakers in New York180 showed that a single handwash had little impact, with mean log counts of 5.72 before compared with 5.69 after handwashing. Another study with 52 office workers and students showed a mean log prewash count of 4.81 compared with 5.07 postwash.181 Kampf and Kramer173 also reviewed studies from health care settings in which increased bacterial counts were found on the hands after handwashing, and handwashing failed to prevent transfer of bacteria from hands to surfaces. Although there are no data available to confirm this, increases in contamination may result from sweating induced by hot water, which flushes resident bacteria from the sweat glands onto the hand surface or aids detachment of bacteria attached to skin scales.

It is important to bear in mind that, although soap and water removes contamination from the hands, soap itself has a limited antimicrobial effect, which means that contamination will be transferred to the sink. Hospital studies show that Pseudomonas aeruginosa and Burkholderia cepacia182 can form reservoirs of contamination in sink waste pipes and can be a source of infection at times when splashes of contaminated water come in contact with hands. Mermel et al reported that hands of HCWs became recontaminated from faucet handles during a Shigella outbreak.183 Soap bars also have the potential to spread contamination from person to person via the hands.88

Efficacy of ABHS 

ABHS are formulations that contain either ethanol 1-propanol or 2-propanol or a combination of these products. Their antimicrobial activity is attributed to their ability to denature proteins. Although products containing 60% to 95% alcohol are most effective, higher concentrations are less effective because proteins are not easily denatured in the absence of water. A range of in vivo and in vitro studies have been carried out to determine the effectiveness of ABHS and are reviewed by Boyce and Pittet,172 Kampf and Kramer,173 and Sickbert-Bennet et al.169

In vivo panel testing of ABHS 

In Europe, the efficacy of ABHS is established by panel tests that determine the reduction in the number of organisms released from artificially contaminated hands. The test applicable to ABHS is the Committee European Normalisation Hygienic Handrub Test EN1500.184 In this test, E coli is inoculated onto the hands and dried. The sanitizer is applied to the hands with a rubbing action for a specified period. The residual number of bacteria present on hands is assessed pre- and posttreatment by a rinse-sampling process and the log reduction determined. To claim that a product is a hygienic handrub, it must produce a log reduction at least equivalent to that produced by a reference product containing 60% vol/vol 2-propanol (mean, 4.24 log in 1 minute; range, 3.17-6.46).

In vivo panel testing against bacterial strains 

Data from in vivo panel tests, summarized in Table 12, indicate that ABHS show good and rapid activity against bacterial stains such as E coli and S aureus. Efficacy is at least as good, if not better, than that achieved by handwashing with soap (Table 10); log reductions obtained after a 30-second contact period were of the order of 3.4 to 3.7 or more compared with 1.8 to 2.8 for a 30-second handwashing process. Boyce and Pittet172 conclude that, typically, log reductions of the release of test bacteria from artificially contaminated hands average 3.5 log after 30 seconds and 4.0 to 5.0 log after 1 minute.

Table 12. Efficacy of alcohol-based hand sanitizers in reducing the release of bacteria from artificially contaminated hands

				Mean log reduction exposure time (min)
	Agent	Concentration (%)	Test bacterium	0.5	1.0	2.0
	1-propanol	100	E coli	-	5.8	-
		60		-	5.5	-
		50		3.7	4.7	4.9
		40		-	4.3	-
	2-propanol	70	E coli	-	4.9	-
				3.5	-	-
		60	E coli	-	4.0-4.4	-
			S marcescens	-	4.1	-
		50	E coli	3.4	3.9	4.4
	Ethanol	80	E coli	-	4.0	-
		70		3.4-3.6	3.8-4.3	4.5-5.1
			S aureus	3.7	-	-
		60	E coli	-	3.8	-

Paulson et al185 compared the efficacy of ABHS containing 62% ethanol (contact time 5 minutes) with handwashing against S marcescens, which showed that handwashing (20 seconds rubbing followed by 30 seconds rinsing) produced a log reduction of 2.29 compared with 3.83 for the ABHS. Hammond et al186 recorded a 2.84-log reduction for 62% ethanol against S marcescens in 10 seconds using the ASTM method. Sickbert-Bennet et al,174 however, showed that exposure of S marcescens to 60% to 62% ethanol for 10 seconds produced only a 1.15- to 1.55-log reduction compared with 1.87-log reduction for handwashing for 10 seconds, when tested by the ASTM 1147 method.

Leischner et al187 carried out in vivo tests that showed that alcohol gels were significantly less effective against C difficile spores (1.68- to 1.94-log reduction) compared with handwashing with chlorhexidine soap (2.46-log reduction). Residual spores were readily transferred by handshaking following ABHS use.187 The reduction in spore counts is higher than expected in view of their known resistance to alcohol and may result from the friction associated with application of the gel rather than a bactericidal action; Kampf and Kramer173 state that water alone can produce a reduction of 0.5 to 2.8 log within 1 minute for E coli.

Using the standard ASTM 1174 method, Sickbert-Bennet et al evaluated the effect of exposure time and volume of product used on the efficacy of 62% ethanol.169 They showed that the use of 7 g of the ABHS produced a higher log reduction compared with 3 g (2.7- to 3.8-log reduction compared with 1.0- to 1.8-log reduction). Rubbing the hands until dry (3-12 minutes) was more effective compared with a 10-second application (1.0- to 1.6-log reduction compared with 0.6- to 1.1-log reduction).

Two recent field studies indicate that an ABHS is equally or slightly more effective than handwashing in reducing bacterial contamination on hands. Davis et al19 compared the reduction of bacterial counts on hands using soap and water or a 62% ethanol-based hand sanitizer (contact time 30 seconds) after animal handling at a US livestock event. There was no significant difference in the distribution of log reductions obtained using ABHS compared with handwashing; log reductions in total count ranged from −1.4 to 1.4 and −3.0 to 3.5 for total coliforms. Traub-Dargtz et al carried out a study at 2 clinics in Canada to evaluate the efficacy of handwashing compared with use of ABHS (62% ethanol, contact time 30-60 seconds) on veterinary staff performing routine equine physical examinations.188 Mean bacterial load on hands increased by 0.36 and 0.91 log (for the 2 clinics, respectively) as a result of handling the animals, whereas the mean log reduction produced by handwashing with soap was less than 0.6, compared with 1.29 and 1.44 log (for the 2 clinics, respectively) produced by ABHS.

In vivo panel testing against viral strains 

A number of in vivo studies have been carried out to determine the efficacy of ABHS in reducing the release of viruses from hands. Test methods were variants of the method of Ansari et al189 or the ASTM E1174 method,171 in which the virus is applied to the fingertips and the efficacy of the product in reducing the numbers of viral particles recoverable from the hands determined. The residual number of viral particles present on the hands is assessed pre- and posttreatment and the log reduction determined. Data collated by Boyce and Pittet172 (Table 13) indicate that ethanol at 60% to 80% produces a 0.8- to >3-log reduction against a range of viruses, the extent of the reduction depending on the viral strain, the nature and concentration of the alcohol, and contact time.

Table 13. Efficacy alcohol-based hand sanitizers in reducing the release of test bacteria from artifically contaminated hands

				Outcome: log reduction
	Agent	Test strain	Contact time	Product	Soap and water	Reference
	Ethanol 70%	Hepatitis A	10 seconds	0.89†	0.66	Mbithi et al196
		Polio		1.34†	0.97
	Ethanol 80%	Poliovirus	30 seconds	0.4	2.1	Davies et al192
	Ethanol 70%	Rotavirus	10 seconds	2.4	0.89	Ansari et al189
	Ethanol 80%	Polio	10 seconds	1.6		Steinman et al193
	2-propanol 70%			0.8
	1-propanol 70%			0.8
	Ethanol 60%	Rotavirus	10 seconds	>3		Sattar et al194
		Adenovirus		>3
		Rhinovirus		>3
	2-propanol 70%	Bovine rotavirus	30 seconds	3.1	1.2	Bellamy et al195
	Ethanol 70%			2.9
	Ethanol 70%	Feline calicivirus	30 seconds	3.78		Gehrke et al190
	1-propanol 70%			3.58
	2-propanol 70%			2.15
	Ethanol 70%	Feline calicivirus	30 seconds	2.66		Kampf et al191
	1-propanol 70%			1.53

∗From Boyce and Pittet.172

† Sufficient to prevent transfer of virus from fingers to another surface.

Data indicate that activity of ABHS against viral strains is less than against bacterial strains and that ethanol has greater activity against viruses than 2-propanol. However, all of the strains referred to in Table 13 are nonenveloped viruses, which are known to be more resistant to disinfectants than enveloped viruses. As far as hand hygiene in the home and community is concerned, however, this is key because many of the viral strains responsible for hygiene-related ID commonly occurring in community settings (rotavirus, norovirus, rhinovirus, and adenovirus) are nonenveloped. That having been said, the data suggest that, although nonenveloped viruses such as HAV and enteroviruses (eg, poliovirus) require 70% to 80% alcohol to be reliably inactivated, studies by Sattar et al194 suggest that 60% ethanol was sufficient to reduce the titers of rotavirus, adenovirus, and rhinovirus by >3 log within a 10-second contact period. Data indicate that FCV may also be relatively susceptible to alcohols compared with other nonenveloped viruses, although sensitivity depends on the type and concentration of alcohol. Using an in vivo test based on the ASTM 1838-02 method, Gehrke et al190 showed that 70% ethanol was the most effective agent against FCV with a log reduction of 3.78 compared with 70% 1-propanol (log reduction, 3.58) and 70% 2-propanol (log reduction, 2.15) (exposure time 30 seconds). However, a more recent study by Kampf et al191 suggested a log reduction after 30 seconds of only 2.66 and 1.53 for 70% ethanol and 70% 1-propanol, respectively.

In a number of these studies, handwashing with soap was also investigated. These studies175, 189, 196 showed that the action of ABHS against HAV, polio, and rotavirus was significantly better than that achieved by handwashing with soap. However, in the test model used by Ansari et al189 and Mbithi et al,196 inoculated fingertips are exposed to soap solution or ABHS by inverting them over a vial containing the product. In practice, handwashing involving rubbing and rinsing is likely to remove larger numbers of organisms from hands. In a further experiment, Ansari et al115 also demonstrated that 2-propanol (70%) was more effective (98.9% reduction after 10 seconds) than liquid soap (77% reduction) against rotavirus.

Mbithi et al showed that the log reduction of polio and HAV virus (0.89-1.34) by application of 70% ethanol was sufficient to prevent transfer to another surface via the fingertips.196 Using similar methodology, Bidawid et al65, 66 studied the impact of ethanol hand sanitizers in preventing transfer of HAV and FCV from artificially contaminated finger pads to pieces of lettuce. Results (Table 11) show that touching the lettuce for 10 seconds resulted in transfer of 9.2% and 18%, respectively, of the virus. When finger pads were treated with 62% ethanol or 75% ethanol (contact time 20 seconds) before the lettuce was touched, the amount of virus transferred was reduced to 0.64% and 0.46%, respectively, for HAV and 2.1% and 1.2%, respectively, for FCV. Although both 62% and 75% alcohol produced significant reductions in virus transfer, significant amounts of virus were found to remain on treated finger pads. In all cases, treatment with ethanol was less effective than handwashing.

Kampf and Kramer172 and Boyce and Pittet173 suggest that, to achieve satisfactory activity against nonenveloped viruses, higher alcohol concentrations and extended contact times are needed. Absolute ethanol reduced viral release from hands by 3.2 log, 80% ethanol by 2.2 log, and absolute 1-propanol by 2.4 log197 but with a contact time of 10 minutes. Schurmann and Eggers175 concluded that high alcohol-concentration products are effective against enteroviruses only under favorable conditions (large disinfectant/virus volume ratio, low protein load). Other studies also demonstrate superior activity of high ethanol concentrations against nonenveloped viruses such as polio, HAV, and adenovirus.198, 199

In vitro testing against bacteria, viruses, and fungi 

Whereas in vivo tests can be used to indicate the efficacy of products under use conditions, in vitro suspension tests are used to establish whether efficacy extends to a broad range of organisms.

In vitro testing bacterial and fungal strains 

Alcohols have excellent and rapid activity against gram-positive and gram-negative vegetative bacteria and fungi when tested in vitro.172, 173 A study by Fendler et al200 (Table 14) shows the efficacy of an ABHS containing 62% ethyl alcohol against a range of bacterial and fungal species, giving 4- to 6-log reduction in 15 to 30 seconds.

Table 14. In vitro tests to determine the efficacy of 62% ethanol against bacterial and fungal strains∗

	Microorganism type	Species	Log10 reduction
	Gram-positive and gram-negative bacteria	Clostridium difficile; Corynebacterium diphtheriae; Enterococcus faecalis (vancomycin-resistant); Enterococcus faecium (vancomycin-resistant); Listeria monocytogenes; Staphylococcus aureus (methicillin-resistant); Staphylococcus epidermis; Streptococcus pneumonia; Staphylococcus progenies; Pseudomonas aeruginosa; Escherichia coli (including O157;H7); Salmonella enteritidis; Salmonella typhimurium; Serratia marcescens; Shigella dysenteriae; Shigella sonne	>4.20 to >5.00
	Fungi	Aspergillus flavus; Aspergillus niger; Candida albicans; Candida tropicalis; Epidermophyton floccosum; Penicillium citrinum; Trichophyton mentagrophytes	>3.92 to >6.42

∗ From Fendler et al.200

In vitro testing against viral strains 

Data, as reviewed by Boyce and Pittet,172 confirm that enveloped viruses such as herpes, influenza, PIV, and RSV are very susceptible to alcohols. Data from individual studies (Table 15) suggest that activity against enveloped viruses is equivalent to that against bacterial strains. However, in agreement with in vivo data, alcohols tend to be less effective against nonenveloped viruses, although this is not the case for all strains. Fendler et al200 confirmed good activity for ethanol (62%) against PIV and herpes viruses (>4-log reduction in 30 seconds) and some, but relatively less, activity against the nonenveloped rhinovirus, cocksackie virus, adenovirus, and HAV (1- to 3-log reduction in 30 seconds). Hammond et al186 showed >5-log reduction against herpes and influenza virus but also >4.25-log reduction against rhinovirus type 16. There are no data on efficacy against rotavirus in vitro.

Table 15. In vitro tests to determine the efficacy of alcohol-based hand sanitizers against viruses

	Type	Species	Product	Contact time	Log reduction	Reference
	Enveloped viruses	Herpes type A; influenza A2.	Ethanol 62%	30 seconds	>5	Hammond et al186
		Parainfluenza types 2 and 3; hepatitis A; influenza A2	Ethanol 62%	30 seconds	>4.1 to >5.0	Fendler et al200
	Nonenveloped viruses	Adenovirus type 2; rhinovirus types 14 and 37; coxsackie B3.	Ethanol 62%	30 seconds	1.25 to 2.75	Fendler et al200
		Rhinovirus type 16	Ethanol 62%	30 seconds	>4.25	Hammond et al186 Fendler et al200
		Feline calicivirus (surrogate for norovirus)	1-propanol 50%-70%	30 seconds	>4	Gehrke et al190
			Ethanol 50%	30 seconds	2.19
				1 minute	3.65
			Ethanol 70%	30 seconds	2.19
				1 minute	>3.8
			2-propanol 50%	30 seconds	2.31
				1 minute	3.22
			2-propanol 70%	30 seconds	2.31
				1 minute	2.35
		SARS coronavirus	Ethanol 85%-95%	30 seconds	>4.25	Rabenau et al202

In vitro tests suggest that alcohols are relatively effective against FCV, although Gehrke et al190 (Table 15) found that 1-propanol was more effective than 2-propanol and ethanol. It was also found that these alcohols were less effective against FCV at 80% than at 50% and 70%. At this concentration (80%), 1-propanol, 2-propanol, and ethanol produced log reductions of only 1.9, 1.35, and 2.16, respectively. By contrast, Duizer et al201 showed that 70% ethanol produced less than a 2-log reduction for FCV after 8 minutes and a 3-log reduction after 30 minutes.

These data are confirmed by a further study (McNEIL-PPC unpublished) using in vitro suspension test methods as used to generate data in Table 15. The data (Table 16) show that 62% ethanol gave a 3- to 6-log reduction in 30 seconds against a range of nonenveloped viruses including not only RSV, PIV, and influenza A and B but also against some strains of rhinovirus and echovirus.

Table 16. In vitro tests to determine the efficacy of alcohol hand gel containing 62% ethanol: contact time 30 seconds

	Virus	Log reduction
	Adenovirus 5	0.33
	Adenovirus 6	0.33
	Adenovirus 7	0.27
	Adenovirus 8	0.66
	Coronavirus 229E	2.83
	Coronavirus OC43	2.00
	Echovirus 9	5.00
	Echovirus 11	4.83
	Influenza A2	6.0
	Influenza B	6.0
	Parainfluenza 1 (sendai)	6.0
	Parainfluenza 4b	3.0
	Respiratory syncytial virus	3.17
	Rhinovirus 18	3.83
	Rhinovirus 2	5.0
	Rhinovirus 14	6.0

Assessing the efficacy of handwashing and ABHS by Quantitative Microbiologic Risk Assessment 

One of the problems in developing hygiene promotion policies is the lack of quantitative data on the relative health impact of different hygiene interventions. Although intervention studies yield quantitative data on health impact, as discussed in section 4.2, the reliability of these estimates is difficult to confirm. By contrast, in vivo and in vitro tests are more economic to perform and can be used to determine relative efficacy of different procedures but give no assessment of how the contamination reduction on hands correlates with health impact. In an attempt to overcome these problems, Haas et al have applied the technique of Quantitative Microbial Risk Assessment (QMRA) to estimate the relative health benefits resulting from use of different hygiene procedures.204 This approach involves using microbiologic data from the published literature related to each stage of the infection transmission cycle to calculate infection risk.

In a recent study,205 these investigators developed a model for studying the effect of hand contact with ground beef during food preparation, which was used to study the impact of handwashing and use of ABHS in preventing subsequent transference from the hands to the mouth compared with no handwashing. Pathogenic E coli and E coli O157:H7 were selected for this study because it is known from other investigations206 that handling ground beef during home food preparation poses a risk of infection with E coli. To perform the risk assessment, data on the density of pathogens in ground beef, transference from beef to hands, removal by handwashing or ABHS, rate of transfer from hand to mouth, and infectivity of ingested pathogens were obtained from the literature and, after screening for data quality, were used to develop probability distributions. For assessing log reductions produced by hand hygiene procedures, only in vivo panel testing data were considered. The median log reduction used in these calculations was 0.3 (range, 0.2-3.0) for handwashing and 4.3 (range, 2.6-5.8) for ABHS.

Table 17 shows the estimates of the infection risk from handling raw beef, as obtained from the analysis. The authors note that these risks are conditional in the sense that they quantify the risk to an individual who has handled ground beef and who engages in hand-to-mouth activity. The probability that an individual will engage in such behavior is not known, and, therefore, a direct comparison with actual disease rates cannot be made. However, with some plausible assumptions, it was assessed that, assuming that there are 100 million individuals in the United States, each of whom handles ground beef once per month, this results in 1.2 × 109 contacts per year. Assuming that 10% of these individuals contact hand to mouth after handling ground beef, this amounts to 1.2 × 108 incidents per year. For E coli O157:H7, using the median risk, this would result in an estimate ranging from 0.014 infections per year if all individuals washed their hands with soap following contact with ground beef to 0.7 infections per year if no handwashing is done. This would equate to a 98% median risk reduction for handwashing compared with no handwashing. If an ABHS was used, this would result in an estimate of 0.00005 infections per year if all individuals used ABHS following contact with ground beef. This would equate to a 99.9996% median risk reduction for use of ABHS compared with handwashing.

Table 17. Mean and median risks of infection from handling raw beef and subsequent hand-to-mouth contact with or without handwashing intervention

		Escherichia coli			Escherichia coli 0157
	Active	No handwashing	Handwashing	Use of alcohol hand rub	No handwashing	Handwashing	Use of alcohol hand rub
	Mean	8.24 × 105	2.61 × 106	5.46 × 108	1.39 × 102	1.25 × 102	1.15 × 102
	Median	9.57 × 108	1.86 × 109	5.36 × 1012	5.98 × 109	1.18 × 1010	3.71 × 1013
	Standard deviation	9.00 × 104	3.18 × 105	1.33 × 106	7.79 × 102	7.52 × 102	7.26 × 102

This study follows an earlier study by Haas et al207 to calculate risks associated with hand-to-mouth transfer after diaper changing of a baby infected with Shigella. Based on this model, it was calculated that the probability of acquiring infection was between 24 of 100 and 91 of 100 for those who used handwashing with soap after changing diapers. This was based on panel test data indicating a mean log reduction of S marcescens on hands because of handwashing of 2.56. The authors calculated that, by using a hand hygiene formulation that increased the log reduction to 2.91, the infection risk would be reduced to between 15 of 100 and 90 of 100.

Haas et al205 conclude that quantitative microbial and ID risk models offer a useful tool to assess the relative extent to which different hygiene procedures can impact on ID risks. They concede, however, that, although risk modeling represents a promising approach, there are limitations to most models because of the multifactorial nature of infection transmission, the dynamic environment in which transmission takes place, and the paucity of data to specify model parameters.

Intervention studies of the effectiveness of handwashing and ABHS 

In an earlier section, we evaluated intervention study data to assess the strength of the causal link between hand hygiene and ID transmission. In this section, we use these data to evaluate the effectiveness of handwashing as a hygiene measure and in relation to the effectiveness of using ABHS as an adjunct or an alternative to handwashing.

Despite the methodologic limitations, the collective weight of evidence from intervention and microbiologic studies described earlier suggests that handwashing with soap can have a significant impact in reducing the incidence of GI and RT infection. The data, however, show that the health impact from handwashing promotion varies significantly according to the setting and outcome. Statistically significant reductions ranged from 48% to 57% for GI illness and 20% to 51% for RT illness. Although all studies were carried out in settings such as day care centers and schools, we believe that the modes of transmission in these settings and the relative rates of transmission of RT and GI infections are likely to reflect those occurring in the home.

In 2004, Meadows and Le Saux published a review of the effect of rinse-free hand sanitizers in elementary schools over a 20-year period.208 They concluded, however, that the data were of poor quality and that more rigorous intervention trials are needed. In a more recent study, Aiello et al examined the epidemiologic evidence for a relationship between waterless hand sanitizers and infections in the community setting over several decades.154 In Table 18, we present the studies that specifically examined ABHS. Only studies with an effect estimate and 95% CI are presented. The type and content of the ABHS varied across studies, for example, one study reported the use of a 60% isopropyl alcohol rinse and another study utilized an alcohol-based foam sanitizer. However, most studies used alcohol-based gels or other alcohol-based emollients. The alcohol content included ethanol and 2-propanol at concentrations ranging from 60% to 90%.

Table 18. Effectiveness of alcohol-based hand sanitizers on infectious illnesses in the community setting

	Intervention	Control	Setting, country	Outcome measured	Result, % reduction (95% CI)	Statistical significance	Reference
	HEP	No HEP or HP	Child care, United States	Enteric disease	29 (0.07-0.45)	Enteric: Yes	Butz et al209
	HP: HW + use of ABHS as supplement			Runny nose	−5 (−0.16-0.05)	Runny nose: No
	HEP	No HEP or HP	Child care, United States	Diarrhea	16 (−0.06-0.33)	No	Kotch et al210
	HP: HW + use of ABHS where no alternative			Respiratory	−6 (−0.21-0.06)
	HEP	No HEP or HP	Child care, Finland	Enteric disease	Age ≤3: 20 (0.09-0.30)	Enteric ≤3: Yes	Uhari and Mottonen211
	HP: “intensified” handwashing, use of alcohol-based oily disinfectant plus other measures				Age >3: 0 (−0.22-0.18)	Enteric >3: No
				Cold	Age ≤3: 13 (0.07-0.18)	Cold ≤3: Yes
					Age >3: 4 (−0.04-0.11)	Cold >3: No
				Any infectious illness	Age ≤3: 9 (0.05-0.13)	Any ≤3: Yes
					Age >3: 8 (0.02-0.13)	Any >3: Yes
	No HEP	Maintain normal handwashing	Elementary school, United States	Illness-related absenteeism	20 (0.17-0.22)	Yes	Hammond et al186
	HP: maintain normal handwashing, use of ABHS at specified times additionally also, eg, after sneezing
	HEP	No HEP or HP	Elementary school, United States	Illness-related absenteeism	50 (0.38-0.59)	Yes	Guinan et al212
	HP: normal HW + use of ABHS in classrooms
	HEP	No HEP or HP	University, United States	Respiratory illness rates and absenteeism	26 (0.17-0.35)	Yes	White et al213
	HP: HW + regular use of ABHS as supplement
	HEP	HEP	Elementary school, United States	Illness-related absenteeism	44 (0.16-0.62)	Yes	Morton and Schultz214
	HP: HW + regular use of ABHS as supplement
	HEP	No intervention	Home, United States	Gastrointestinal	59 (0.10-0.81)	Gastrointestinal: Yes	Sandora et al215
	HP: ABHS supplied for use in the home			Respiratory	3 (−0.3-0.28)	Respiratory: No

Ages are in years. HEP, hygiene education program, eg, chain of infection explained; HP, hygiene promotion, eg, promotion of handwashing and/or use of ABHS plus other measures; CI, confidence interval.

Of the 8 intervention studies, 7 were conducted in the United States (88%, 7/8) and 1 in Finland (13%). Most were conducted in child day care centers, elementary schools, or universities (88%, 7/8), and one was conducted in the household (13%, 1/8). Outcomes included GI-related illnesses/symptoms and/or upper RT-related illnesses/symptoms examined as separate outcomes or in combination with other infection-related symptoms as part of a school absence-related definition of “infectious-illness.” Of the 8 studies, 88% (7/8) reported significant results for at least one age group or outcome. The effect was stronger in younger compared with older age groups for studies providing age stratified data.

The reduction in GI illness ranged from 0% to 59% for the 4 intervention studies that examined GI illnesses as separate outcomes.209, 210, 211, 215 Of these, all but one showed statistically significant reductions.210 The study by Uhari et al showed a significant reduction in GI illness only among children ≤3 years of age.211 All but one of these studies was conducted in child care settings.215 When evaluated separately, reductions in GI illness appeared more robust compared with the findings for upper RT illness.

For upper RT illness, the reduction in infectious illness/symptoms ranged from −6% to 26%. Only 2 of the 5 studies (44%) examining upper RT illness as a main outcome reported a statistically significant reduction. Uhari et al reported a 13% reduction in RT illnesses among children ≤3years of age, but no significant effect in older children.211 White et al reported a 20% reduction in upper RT illness among students using ABHS in residence halls.213 However, this study suffered from several methodologic shortcomings, including lack of control for clustered units, no randomization, no masking, and no monitoring of product use.

All but one of the intervention studies included a hygiene education component, but, in 7 of these studies, this was only provided in the intervention arm. The level of education varied widely, ranging from basic information on when to use the ABHS (ie, after sneezing and coughing, after use in the restroom, before lunch) to in-depth education programs214 and biweekly instructional material designed to educate families on hand hygiene and infection transmission.215 In all studies, ABHS was promoted as a supplement to handwashing, or as an alternative to handwashing when soap was unavailable, and it is likely that the hygiene education would have had the effect of encouraging more frequent handwashing as well as use of ABHS. Although almost all studies indicated that hygiene education combined with promotion of ABHS can reduce the risks of GI or RT illness, only 2 studies allowed any assessment of the independent effect of the ABHS. Of these 2 studies, the study by Hammond et al186 did not provide any educational component to either arm, and the study by Morton and Schultz214 provided education to both intervention and control arms. Both studies showed a significant reduction in illness-related absences (20% and 44%, respectively), but it is not clear whether the illnesses were predominantly GI or RT because these studies used a loose definition of absence-related illnesses.

Sandora et al215 was the only study carried out in the household setting and, of all the studies, reported the highest reduction for GI illness. The trial involved 292 families with children enrolled in 26 child care centers. Intervention families received a supply of ABHS and biweekly hand hygiene educational materials for 5 months; control families received only materials promoting good nutrition. A total of 252 GI illnesses occurred during the study; 11% were secondary illnesses. The secondary GI illness rate was significantly lower (59%) in intervention families compared with control families (see Table 18). A total of 1802 RT illnesses occurred during the study; 25% were secondary illnesses. Although RT illness rates were not significantly different between groups, families with higher ABHS usage had marginally lower RT illness rate than those with less usage (19% reduction). Sandora et al215 suggested that the difference may be due to heightened diligence associated with using ABHS after a GI-related incident compared with an RT incident, such as sneezing.

Overall, based on relatively limited data available, the results in Table 18 suggest that the impact of ABHS promotion, as part of a hygiene education and promotion program in reducing the incidence of GI infections in young children, is similar to that observed for promotion of handwashing with soap. Promotion of ABHS in this manner also produced some reduction in the incidence of RT infections, which was less than that associated with promotion of handwashing alone. Assessing whether there might be an added health benefit of using ABHS above and beyond the effect of hygiene education is hampered by the fact that most studies used hygiene education and ABHS in the intervention arm but did not provide an educational component to the control arm.

Several important methodologic issues were evident, although more recent studies have improved designs and conduct. In most studies, either parents or school personnel provided information on ID among children in the study populations. In all but one study,34 the parents, participants, or personnel monitoring and reporting infections were not masked as to their own or their child's intervention status. Although masking of participants and interviewers to the intervention status is important, because it might influence reporting, it is often difficult to conduct masked hygiene interventions and may not be ethical. Sandora et al determined that it was neither feasible nor ethical to mask subjects or interviewers because it is difficult to devise a formulation that could act as a “placebo” for ABHS, and using a placebo ABHS product might endanger the control group via inadequate hand hygiene.215

In many of the ABHS studies, especially the recent ones, efforts were made to control for potential confounding factors. However, many of the studies did not collect information on baseline hand hygiene practices (nor methods and frequency of cleaning/disinfecting soiled/contaminated environmental surfaces in homes) as well as ABHS use. The studies also excluded participants who reported current ABHS use in the home. Furthermore, participants were asked to refrain from using ABHS in settings outside the home. These are all important design strategies minimizing bias associated with noncompliance or differential usage. Two of the 8 intervention studies failed to use systematic monitoring for hygiene practices, such as frequency of hand-sanitizing episodes, frequency of handwashing, or duration of handwashing.212, 213 This is especially concerning because the study by Sandora et al suggests that the quantity of ABHS influences the risk of infection in a dose-response manner.215 Moreover, if frequency of handwashing and ABHS use is not recorded, it is impossible to isolate the independent effects of ABHS from that of handwashing on infection rates. In these studies of ABHS use, surveillance measures included calculating use from monthly demand, total amount supplied, observation by research assistants, participant report, and reported use by primary caregivers in households.186, 209, 212, 214, 215

Efficacy of the hand hygiene procedure 

Although panel tests carried out under controlled conditions showed that handwashing can reduce the numbers of bacteria and some viruses on the hands by up to 2- to 3-log within 30 seconds to 1 minute, in practice it is doubtful whether people wash their hands properly, even for the prescribed period of 15 seconds, to achieve this. At present, there is a paucity of data on the efficacy of handwashing in relation to how people actually wash their hands on a day-to-day basis, both in the duration of handwashing and handwashing technique; in most of the intervention studies described earlier no attempt was made to time handwashes or to determine residual levels of contamination on the hands after handwashing. Microbiologic data suggest that, for some pathogens (eg, Salmonella), mechanical removal by handwashing alone is inefficient. These data, together with the results of in vivo panel testing of the effectiveness of handwashing and of ABHS, as described above, question the efficacy of handwashing in community-based groups and suggest that more work is needed to determine the efficacy of hand hygiene procedures under conditions normally encountered in the home and how hand hygiene procedures could be improved.

For ABHS, in vivo and in vitro testing suggest that these formulations are highly effective against bacterial pathogens and can produce a 3.5-log reduction on hands within 30 seconds and 4.0- to 5.0-log reduction after a 1-minute application against a wide range of species including Salmonella. It is possible, however, that the potential for increased benefits against bacterial infections compared with handwashing may be offset by reduced efficacy against important nonenveloped viruses such as rotavirus, some strains of rhinovirus, and possibly also norovirus. It has been argued that the higher impact of ABHS interventions against GI compared with RT infections is due to the fact that RT infections are predominantly viral. However, because the intervention data indicate that handwashing also supports a stronger reduction in GI diseases compared with RT diseases, this seems unlikely. Some studies suggest that, to achieve satisfactory activity that includes all types of viruses, higher concentrations up to 80% ethanol should be advised. Other studies suggest that the efficacy of ABHS may be increased by increasing the volume of agent applied to the hands.215

In formulating policies for hand hygiene, it would be convenient to be able to define what represents a “safe” residual level of contamination on the hands after hand hygiene, ie, sufficient to prevent infection transmission, but, because the infectious dose varies from one species to another and is dependant on the immune status of the recipient, this approach is untenable. The QMRA approach, as outlined earlier in this review, however, demonstrates that strategies that produce an increase in the log reduction on hands from 2 to 3 to 4 are accompanied by a significant incremental reduction in the risk of infection in a given population and could thus be worthwhile. This suggests that the health impact from promoting hand hygiene could be increased by developing and promoting procedures for use in the home and community that increase the log reduction of contamination on the hands. This involves identifying products and procedures (both soap-based and waterless products or wipes) that achieve high levels of removal and/or “kill” (alone or in combination) of the full range of gram-negative and gram-positive bacteria and enveloped and nonenveloped viruses and that deliver hand hygiene under conditions that people are prepared to employ in their busy and mobile daily lives. It also suggests that, particularly for “high-risk” situations as outlined below, there is advantage to be gained by promoting handwashing followed by use of ABHS to increase the log reduction.

Applying hand hygiene at the correct time: the need for hygiene education 

The data presented in this review suggest that the favorable health impact from promoting hand hygiene could be further increased by getting people to practice hand hygiene not just more frequently but also at the right time. A number of studies show the relatively poor understanding of the principles of hygiene that is present in the community. This may be one of the factors responsible for the higher risk reductions observed in intervention studies of GI compared with RT infections. For example, knowledge regarding the need for handwashing after coughing or sneezing may not be as pervasive as knowledge about handwashing after defecation, and it may be that people understand better when to wash their hands during food-associated activities but not, for example, while handling contaminated tissues. Although some intervention studies described in this review involved a component of education, it was not possible to determine the extent to which hygiene education that enhanced people's understanding of infection transmission also enhanced health outcome.

Because visible soiling is an unreliable indicator of the presence of pathogens on the hands, people are unlikely to wash their hands at the correct time unless they have been taught to do so or have some awareness of the chain of infection transmission in the home, ie, they are aware of when their hands may be contaminated. Whereas risks associated with food handling are largely confined to defined periods of time, for RT and skin infections (and person-to-person transmission of GI infection), the risk is ongoing and involves a large proportion of our ongoing daily activities. Thus, whereas it is possible for hand hygiene advice associated with food hygiene to be rule based, this is not the case for other types of infections. In the event of a flu pandemic, the advice issued by the UK Health Protection Agency216 to “wash hands frequently” is unlikely to be effective unless people have some idea of the times when their hands are likely to be contaminated with flu virus.

Although current thinking about hygiene promotion tends toward a view that the most effective way to change behavior is by mass social marketing of single rule-based hygiene messages,4 the data presented in this review suggest that the complexity and shifting nature of the ID threat is such that a rule-based approach to hygiene is inadequate to meet current public health needs. The need is for an approach founded on awareness of the chain of infection transmission and how it differs for different groups of infections. Hygiene education needs to be consistently incorporated as part of hand hygiene promotion programs if people are to properly understand the risks and adapt their behavior accordingly.