shows the different organisms isolated in the three study arms. Enterococcus spp. were the most common bacteria found, followed by Enterobacter spp. shows the effect of handwashing with soap or water only on contamination, compared to no handwashing. Overall, handwashing with water alone reduced the prevalence of bacteria substantially. Handwashing with soap was more effective in reducing the prevalence of contamination and specifically of Enterococcus spp. There was a trend that handwashing with soap was also more effective in reducing the prevalence of other species and of multiple isolates, but the statistical support was low ().
Organisms found after self-contamination of hands, and handwashing with either soap, water only, or no handwashing.
Figure 1 Effect of handwashing with water alone or soap and water compared to no handwashing. P-values derived from logistic regression adjusted for within-person correlation, except * where p-value was derived from Fishers exact test ignoring within-person correlation. (more ...)
The effect of repeated measurements in the same individual was low: the design effect (the factor by which a sample size needs to be increased to achieve the same statistical power as an unclustered study) ranged from 1.2 to 1.3 (depending on the comparison group).
Participants were asked to wash their hands as long and as thorough as they would normally do. The length of time required to carry out handwashing was measured once for each method in all volunteers. Participants took on average 12 seconds (standard deviation 2.8) to wash their hands with water alone, and 14 seconds (standard deviation 2.3) to wash their hands with water and soap (p = 0.02).
Thus, handwashing with soap took them only slightly longer than handwashing with water alone. It seems unlikely that this small difference can explain the large difference in the removal of bacteria. Soap on its own appears to have an effect on the removal of bacteria of potential faecal origin, independent of the possibility that soap use may cause people to wash their hands longer.
Unlike the study by Hoque and colleagues our trial was conducted in an experimental (albeit naturalistic) setting, where volunteers deliberately contaminated their hands. Additional testing showed that this approach increased the prevalence of contamination from around 10% to over 40% of individuals. It also improved control over the conduct of the experiment, but may affect generalisability, as the study primarily aimed at providing a proof of principle. However, we believe that the superior effect of soap on the removal of bacteria compared to water alone as the principal finding of our study is unlikely to depend on the setting.
Not all of the bacteria isolated in our study are known to cause disease in humans. Surprisingly, we found few E. coli on hands which may be due to their short survival time in the environment. Overall, the effect of soap appeared to be independent of the type of bacteria (), a view which is supported by the study by Hoque and colleagues who found a similar effect of hand hygiene on unspecified faecal coliform bacteria [6
]. However, the power of our study to detect differences between species was low.
We used plain non-antibacterial soap for the experiment. Future studies could address whether antibacterial soap is more effective in removing pathogens from hands. However, Luby and colleagues conducted a large double-blind randomised trial in Pakistan and found antibacterial soap no more effective in reducing diarrhoea than normal soap [11
]. It is still not clear whether or in what circumstances anti-bacterial soaps offer a health advantage [12
The bacteriological methods used in this study provide no quantification of bacterial load, unlike a study by Hoque and colleagues [7
]. Quantifying the effect of different hand washing procedures on bacterial load may be particularly helpful for studies in poor settings with poor sanitation facilities, where the environmental contamination with faecal organisms is much higher [13
]. We also tested a semi-quantitative finger-print method used previously in Thailand [15
] not unsimilar to the method used by Luby and colleagues [8
] but found that contamination levels were too low to provide consistent results. Therefore we decided to use a qualitative method.
It seems reasonable to assume that handwashing with soap is also more effective in reducing bacterial load compared to water alone. Future studies could address the effect of different hand hygiene procedures on removing gastro-intestinal or respiratory viruses such as influenza A. Hands have been implicated especially in the spread of Norovirus [16
]. Viral studies are more difficult to conduct as viruses may not be as present in the environment as often as are bacteria of faecal origin, but they may be possible for example if patients with laboratory confirmed infection are recruited as volunteers. Alternatively, healthy volunteers may experimentally contaminate their hands with cultured viruses before undergoing different hand hygiene regimes, as was done in a recent study on influenza A H1N1 [17
]. This study found that handwashing with soap was better at removing influenza A H1N1 than several hand sanitizers. Handwashing with water alone was not tested.