Our study found that an intervention consisting of school-based hygiene promotion and water treatment did not impact pupils' risk of having E. coli hand contamination. However, girls had a significantly increased risk of having high levels of E. coli contamination on their hands. The addition of new sanitation facilities to the HP&WT intervention greatly increased children's risk of having any E. coli and high levels of E. coli on their hands. This effect was significant and of the highest magnitude among girls. These findings suggest a lack of sufficient improvement in handwashing behavior in intervention schools coupled with an undetermined source of increased contamination risk in Sanitation + HP&WT schools.
There are several potential reasons for these unexpected results. Hand contamination levels are likely to vary depending on several factors, including the degree to which hands were contaminated during defecation, whether the individual washed his/her hands with water and soap before sample collection, the quality and duration of handwashing, the level of environmental contamination with feces on surfaces, and the length of time since last handwashing or defecation. Given the rapid decline in E. coli
survival on skin, we assume that the contamination we detected on hands was recently acquired while at school.16,31
Our data show that handwashing materials were more frequently available following the intervention. Therefore, children in the intervention schools had increased opportunity to wash their hands, although some schools did not have handwashing materials on the day of data collection. Alternative behavioral indicators suggest that pupils may not have increased their practice of regular handwashing or of thorough handwashing. There was no significant change in the percentage of girls or boys that used soap during a handwashing demonstration, which, in one study, was shown to be the closest correlate (albeit imperfect and prone to overestimation) with observed behavior among caregivers in Indian households compared with other self-reported indicators.25
The hygiene promotion intervention relied on a simple curriculum and training of teachers to pass on messages to pupils, which may not have been sufficient to change behavior. There is evidence that health message-based hygiene promotion efforts alone are not always sufficient to motivate behavior change among adults in developing countries, but it is not known whether this strategy improves hygiene practices among children6,19
; an evaluation of an intervention in Kenyan schools found no evidence that teacher trainings and school health club activities improved handwashing behavior.13
The sharp increase of contamination in Sanitation + HP&WT schools may be caused by increased usage of school latrines for defecation without concurrent improvement in hand hygiene after using them. Usage of school toilets is associated with their level of cleanliness12,13
; we observed, and pupils confirmed, a perception that latrine cleanliness in the Sanitation + HP&WT schools improved significantly. Additionally, indicators of comfort in using school latrines in this intervention group suggest that pupils probably increased their usage of the new latrines, which may have provided an appealing alternative to their home latrines or lack thereof. Kenyans tend to habitually defecate in the morning upon waking up,32
but children in Sanitation + HP&WT schools may have chosen to delay defecation until they arrived at school. We did not conduct observations that would detect such changes in defecation habits. Girls in these schools reported a greater increase in usage of the latrines than boys, which may in part explain the greater proportion of girls with E. coli
hand contamination in these intervention schools if this is indeed a risk factor. However, there are likely other unmeasured behavioral factors among girls and boys that may explain the increased risk of contamination in both intervention arms. This merits future research. We did not collect data to determine whether girls were engaged in latrine cleaning more often than boys; however, anecdotal evidence from numerous school visits indicates that both girls and boys in Nyanza Province schools tend to be equally responsible for cleaning the latrines designated to their gender. It is important to note that separate analysis of results for each intervention arm was an unplanned exploratory analysis and should therefore be interpreted cautiously.
Anal cleansing materials, such as toilet paper, are almost never provided by Kenyan primary schools and were not provided as part of the intervention. Lack of toilet paper at schools and dirty toilets have been shown to be associated with diarrhea.33
A study by McMahon and colleagues34
was performed following this trial, which examined anal cleansing habits among pupils in our study area. It was discovered that children in this area use a variety of materials for anal cleansing after defecation including leaves, paper from schoolbooks, stones, corncobs, and their hands. Toilet tissue is seldom used because of the high cost, or for some a lack of awareness, and cleansing with water is not commonly practiced among most people in the local Luo culture. Some school children explained that commonly used materials are often inadequate and feared disease transmission as a result. Anal cleansing habits did not appear to differ by gender. According to anecdotal reports, in some cases, smaller children without anal cleansing materials may smear fecal matter on the walls of the latrine. Our study did not include observations of these conditions, but this may be another location where hand contamination can occur. If the lack of sufficient anal cleansing materials at school leaves a child with extremely high levels of fecal contamination on his/her hands, a cursory handwashing—particularly if done without the use of soap–may not remove all pathogens. We did not observe the proximity of handwashing facilities to latrines or the time of day when soap and water for washing were set out. It is possible that they were not available before the start of classes, when pupils may have been using the latrines. Given the lack of anal cleansing materials, if pupils in Sanitation + HP&WT schools used school latrines more often, and if their handwashing practices did not improve substantially, it is conceivable that they would have an increased risk of fecal contamination on their hands. Further research should be conducted to test these hypotheses.
Environmental contamination on surfaces may have also contributed to soiled hands in our study. This appeared to be a factor in research by Ram and colleagues35
in Bangladesh, where 80% of women's hands were found to be contaminated 2 hours after thorough handwashing with soap. A study among street vendors in Guatemala observed a similar trend.20
Likewise, Pickering and colleagues36
discovered that Tanzanian women had substantial hand recontamination following observed typical household activities, suggesting that environmental contamination with feces was pervasive. One trial found that cleaning desks and other surfaces in a United States elementary school reduced episodes of gastrointestinal illness.37
Environmental contamination should be explored in future studies in low-income settings.
Measurement of diarrhea or absence more directly explain the impact of the intervention on key outcomes of interest, and these data are reported elsewhere.24
This study sheds light on the potential mechanism by which this school WASH intervention might have influenced rates of illness and absence. Results from the main impact study revealed that the impact of Sanitation + HP&WT on pupil absence was no different relative to controls than in schools that received HP&WT without sanitation.24
Though it is difficult to draw a direct link between these hand rinse results and our measure of self-reported absence in the prior 2 weeks, this suggests a possibility that some of the benefit conferred by new latrines at school may have been offset by increased illness from elevated risk of hand contamination. The main study findings also suggested a reduction in absenteeism among girls.24
Although comparison of the two indicators is challenging, the increased hand contamination among girls in contrast to this finding suggests that perhaps only a fraction of diarrhea and consequent absenteeism may be attributable to hand contamination and that patterns of school absence may relate in part to the value placed on latrines and handwashing facilities as amenities. Our data show that girls in particular may have been drawn to the amenity value of school latrines, given indictors of acceptability and usage in Sanitation + HP&WT schools ().
Although other studies have measured the impact of school WASH interventions on hygiene awareness or reported behavior, to our knowledge, this is the first study that examines the impact on pupil hand contamination in low-income settings. However, the limitations of our study should be considered. Although we can informally assess the influence of specific WASH conditions and behaviors on hand contamination outcomes, these factors could not be tested in regression models, as our analysis was conducted according to intention to treat (i.e., intervention status), and such conditions and behaviors are presumed to be artifacts of the intervention itself. In addition, we did not have a direct measure of handwashing behavior to confirm whether lack of change in handwashing may be an explanation for the findings.
Hand rinse sampling has been shown to be a valid measure of handwashing effectiveness.5,16,18
However, some studies suggest it is not likely to accurately reflect whether subjects washed their hands, particularly because of the high variability found in repeated hand rinse measures and lack of correlation with other hand hygiene indicators, potentially a result of environmental recontamination.29,35
It is also important to note that our data do not quantify risk of diarrhea, as there is no well-defined relationship between levels of fecal indicator bacteria on hands and risk of enteric illness, and there is considerable variability in individual host susceptibility and pathogen virulence. Ideally, children would not be exposed to any fecal pathogens on their hands, and our presence/absence indicator was chosen to reflect this. Although E. coli
have been used in similar studies, it has been suggested that fecal streptococci, Clostridium perfringens
, or enterococci are better fecal indicator bacteria because of their longer survival on skin.17,36,38,39
We chose to incubate laboratory samples at 44.5°C rather than the 35°C temperature recommended for m-ColiBlue24 medium to reduce the growth of non-specific background colonies to facilitate more accurate counting of the target E. coli
colonies. The higher incubation temperature has been previously validated for this medium in tropical water samples, which we processed simultaneously with hand rinse samples for a different component of the study.40
Our data may be a conservative estimate of E. coli
concentrations because of the higher incubation temperatures and cannot be directly compared with other studies that enumerate E. coli
at 35°. Finally, the small number of schools sampled from the Sanitation + HP&WT intervention arm (N
= 5) limited the precision of the results.