Water treatment at source

Three intervention trial studies in Africa conducted preliminary investigations into well chlorination [15–17]; and were all assessed as low quality, primarily because they only presented descriptive water quality data without any statistical analysis. Well chlorination using liquid bleach was assessed following its implementation during the 2008 cholera outbreak in Guinea-Bissau [15]. The intervention was considered ineffective in maintaining the WHO recommended emergency level (>1mg/l) of free residual chlorine (FRC), with only 15% of inspected wells having met the standard 24 hours following chlorination. The study further reported that families had discontinued their household water treatment following well chlorination, as they believed their water to be safe. This action may have paradoxically increased their risk of cholera, and the paper concluded that the intervention was an ineffective, costly and impractical method of water treatment and should be discouraged.

The two remaining studies evaluated chlorinators; chlorine-containing devices that when suspended in well water gradually release chlorine leaving a continuous FRC [16, 17]. Both studies demonstrated the functionality of locally constructed chlorinators as adequate FRC could be maintained for approximately 3 days, though the studies did not report what the maximum period was, under which the devices could guarantee this. Both studies were preliminary investigations, and health education and training related to the use of chlorinators was not provided. Only one of the studies [16] evaluated the appropriateness of the chlorinators according to the local availability of materials, the ease of operation and maintenance, the acceptability by communities, the cost and logistic issues. The study found an effective and appropriate chlorinator by using a pierced plastic bag filled with sand with a locally pressed chlorine tablet placed at its centre. Well users apparently found the system both easy to use and maintain, however as no follow-up data was presented it is difficult to draw conclusions on its long term effectiveness and acceptability.

Household water treatment (HWT)

Eight studies presented a point of use water quality intervention; which included: filtration (2), solar disinfection (1) and chlorination products with, or without storage vessels (5).

Hygiene promotion

Four cross-sectional studies, two in Africa and two in Latin America, were identified as evaluating community knowledge, and awareness of cholera prevention, and all reported a positive effect.

One assessed the effectiveness of interventions implemented during the 2010 cholera outbreak in Haiti, which included mass media campaigns, community health worker activities, and distribution of water purification tablets, soap and oral rehydration solution [27]. The preferred method of communication of those receiving cholera messages was TV (71%) and radio (69%). Knowledge of common symptoms was high (>80%) and 86% of respondents cited hand washing as a prevention method. Water treatment practices increased from 30% before the outbreak to 74% after the outbreak (p<0.05). The most common methods were water purification tablets (67%) and bleach (58%). The majority (64%) of water sources tested positive for E. coli, but 60% of stored water samples showed detectable FRC levels. The results of the survey suggest that the public health messages were effective, and promoted behaviour changes with regard to water treatment practice.

A study in rural Guinea Bissau during the 1994 cholera outbreak explored local views about cholera and the diffusion of health messages, and assessed whether the messages contributed to behaviour change [28]. Interventions included messages to boil water, or drink water with lemon in it, wash hands with soap, take precautions with food, sweep the family compound, keep children away from dirt, keep flies away, and build and use latrines. The study reported that 94% of 53 interviewees had heard, or seen a cholera prevention message, and that 68% of those, recalled at least one preventative method. Radio was the most common source of information (45%) followed by word of mouth at 41%. While 70% of the respondents would consume boiled water, or water with lemon juice added to it, not a single person could explain the transmission of cholera. The absence of baseline data made it impossible to draw a conclusion regarding a change in behaviour.

Mahadik reported that 99% of 442 respondents in 16 villages exposed to a cholera education programme in Zambia showed awareness of cholera compared to 92% of 131 respondents in two control villages in a different district, however no test of significance was reported [29]. The paper provided no information on how the results were reached, nor was there a description of the study population. Further results show that the control group had better knowledge of the main cholera transmission routes than the intervention group, but these were not discussed.

Quick et al. conducted a knowledge, attitudes and practice survey in urban and rural Amazon communities, to assess the impact of a nationwide prevention campaign, 6 months after the beginning of the Peruvian cholera epidemic in 1991 [30]. The study found that 93% of rural, and 67% of urban respondents believed they could prevent cholera transmission. Sources of cholera information in urban areas were radio (71%), TV (64%) and group health talks (50%) compared to rural areas where health workers (70%), government authorities (61%) and radio (58%) were the most common. Whilst knowledge and perceived importance of cholera prevention methods was high, this did not translate into practice. For example 88% of the urban group and 92% of the rural group knew that drinking treated water was a cholera prevention method but only 25% and 23% respectively, always practised this. Several limitations of the study were reported, including the groups not being representative of the population, use of different sampling methods and time constraints preventing the possibility of verifying reported responses through direct observation.

Storage vessel disinfection

A study in a camp in Kitgum, Uganda (2007), assessed the effectiveness of jerry can cleaning using a strong disinfectant, and the recontamination potential after cleaning [31]. Two different methods were used to clean selected jerry cans from 13 households. The container was either half-filled with chlorine solution and shaken for 1 minute (9 jerry cans), or completely filled and allowed to sit for 1 minute (2 jerry cans) or 5 minutes (2 jerry cans), before all being rinsed and filled with source water (free of coliforms). Samples were taken for microbiological testing before cleaning, after cleaning and at 3 and 5 days thereafter. The coliform count was reduced in 85% of the cleaned jerry cans but no method was deemed more effective than the other. However more strikingly, these methods did not prevent recontamination, at the household, in 46% of the jerry cans.

Household disinfection

One recent study measured the uptake of household disinfection kits as an additional prevention method in the Haiti cholera outbreak [32]. Disinfection kits comprising a 14 litre bucket, scrubbing brush, cloth, chlorine bleach, 10 litre jerry can and soap were distributed to the caretakers of patients admitted to a treatment centre, who also attended a hygiene promotion session. In a follow-up survey 98% of 208 households had already used the kit with the two most popular items being the chlorine and soap. A significant increase in use of the kits was observed after the date when hygiene promotion messages were strengthened. Households used the kits to clean floors (73%) and dishes (62%) followed by laundry and latrines. The study acknowledges several limitations; the lack of a control group, the sampling method may have influenced representativeness, and the lack of measurement of time of ownership of the kits may have biased the results.

Improvements to WASH infrastructure

A study conducted in the Philippines attempted to evaluate the impact of improvements to WASH infrastructure on cholera incidence by comparing a control community with communities with: i) safe water supply, ii) shared toilet facilities, and iii) both water supply and shared toilet facilities [33]. Bacteriological surveillance of confirmed cholera, conducted from 1968–1972 suggested that cholera incidence could be reduced by 68%, 73%, and 76%, by implementing low cost sanitation, water supply or both, respectively. Unfortunately the study suffered from several key methodological shortcomings: i) one to one comparison of interventions implemented at the community level without controlling for intra-cluster correlation, effectively the same as comparing one person with just one other, ii) actual use and functionality of the interventions by the population was not assessed, nor were water quality or water use tests conducted, and iii) no statistical test was provided, nor was the data controlled for other risk factors.

Lack of evidence

The lack of good quality studies, with well defined health outcomes, and consequently a relaxing of exclusion criteria has left our search open to a number of potential biases and limitations which will have had an effect on the results reported. Whilst no time limit was set on the search, and hand searches were done of recent reviews for additional references a decision was made not to include unpublished literature. There is likely to be a wealth of useful information available in operational and evaluation reports by non-governmental organisations (NGOs) working in emergencies and cholera outbreaks, however it is unclear how much is available as not all is in the public domain, while some might contain sensitive information that NGOs are unwilling to share. The grey literature warrants a separate review, as it could provide valuable lessons learnt, not only regarding the approaches used, the types of intervention chosen and implemented, and their successes or failures, but could also provide valuable information regarding the setting in which the outbreaks occurred. Ideally a review should separate lessons learnt from natural disasters, and conflict responses, and distinguish between urban and rural settings, and of course between interventions implemented for those internally, or externally displaced, and those still living in their own housing. Currently the available literature does not allow for this distinction.

This study found a clear lack of evidence to help guide implementers decide what approach and intervention to select during a cholera outbreak. This is, to a very large extent, a result of the difficulties associated with collecting epidemiological evidence and time constraints during epidemics, which often occur during, and following natural disasters or conflicts. The main desire of those involved in cholera control is to limit the speed of transmission and to reduce mortality, whilst establishing baselines, and research in general, is often seen as taking valuable time and resources away from life-saving activities. There is further a belief held by some involved in emergency response that only randomized controlled trials (RCT) can provide evidence, and therefore that any study looking to gather evidence is going to be prohibitively expensive, time consuming and will require a lot of man power. Although RCTs are unlikely to be feasible in emergency settings, and could raise serious ethical concerns, this does not mean that good evidence cannot be collected. Most of the studies presented in this survey failed to adhere to the minimum evaluation procedures (MEP] recommended by the WHO for the evaluation of WASH studies [34], which recommends to measure functionality, adherence to interventions, and the use of intermediate outcomes like: water use, water quality and behavioural indicators, when disease data cannot be collected.

Water source and household water quality interventions

The large majority of studies identified focussed on water quality interventions, even though no study provided evidence that the water quality route was the dominant route of transmission in the reported study areas. This seems to highlight the generally held belief that cholera is exclusively waterborne, thereby ignoring other routes of transmission. Since the cholera outbreaks in London in the 1850s, and the ground breaking work of John Snow, cholera seems to be synonymous with water quality, and the associated interventions. However, John Snow’s conviction that hand washing with soap could dramatically slow down, or even stop the rapid outbreak of cholera epidemics [35] is often forgotten.

Lakes, estuaries and to a lesser extent rivers are considered the main reservoir for V. cholerae and have been implicated in past outbreaks of cholera [36–38]. V. cholerae tends to favour slightly brackish to saline water, temperatures of up to 30°C and association with algae, zooplankton and copepods tends to promote its survival and pathogenicity [39]. These ‘water quality’ conditions tend not to prevail in well water, and wells are not considered reservoirs for cholera. Notwithstanding this, the studies of source water quality interventions in this review focussed exclusively on chlorination of public wells, possibly due to fear of local contamination by latrines or poor hygiene by those collecting water. However, as past studies have shown; water quality improvements at source are often ineffective as good quality source water becomes re-contaminated during collection or use within the household as a result of poor (hand) hygiene [40–42].

The well chlorination studies all reported problems with maintaining adequate levels of chlorine for prolonged periods, and finding an appropriate design that could be locally made and affordable. The acceptability of the devices and chlorinated water by the local population was not rigorously tested. Furthermore, interventions in Guinea-Bissau appeared to show that well chlorination without proper promotion and education led to a false sense of security [15, 43]. This highlights that water quality interventions are likely to be futile if not accompanied by adequate training, health education and hygiene promotion. In the case of well chlorination, implementers must be trained how to treat wells and monitor residual chlorine levels, while users must be informed of the contact time necessary for disinfection. Lastly and most importantly, hygienic water handling practices must be promoted as poor hygiene is likely to undermine an intervention as demonstrated in the study investigating methods of jerry can disinfection [31]. The findings suggest that well chlorination may be difficult to execute in emergency settings because of a lack of trained individuals, the number of wells within a given setting may be large, and unknown resulting in poor coverage, and the amount of chlorine required for each well may differ due to contamination with organic matter. In fact, the authors of the study in Guinea-Bissau clearly judged well chlorination to be an unsuitable intervention [15]. In light of this evidence we would not recommend well chlorination as a measure to control cholera.

Options for water treatment at the household level (HWT) were the most reported interventions in this review with distribution of chlorination products the most popular. However, half of the six studies that distributed a water quality intervention and conducted a health education programme, reported inconsistency of product use throughout the year, with chlorination in particular seen as an emergency measure, and used sporadically depending on its affordability and accessibility. Furthermore, mass distributed chlorine products were poorly used even where prevention knowledge was high [23, 25, 26]. Use and adherence of HWT is considered a challenge even during non-emergency settings [7]. Effective implementation of such interventions in emergencies remains a challenge, however it is clear that simply handing out chlorine, or other forms of HWT, and expecting sustainable behaviour change is unrealistic. For chlorination in particular, evidence from non-cholera emergencies indicates that effective use was highest where households with contaminated water were targeted, the treatment method effectively treated the water and the population was familiar with the method and was willing to use it [25]. Water quality interventions must be preceded by formative research and accompanied by health education so that appropriate products for HWT can be selected and sustainable behaviour change can be achieved. These lessons should be applied to future cholera interventions with more rigorous evaluation.

Water treatment versus other WASH interventions

With cholera largely perceived as a waterborne infection, other transmission routes like the consumption of contaminated food as a result of poor hand hygiene, and person-to-person transmission appear to be overlooked in the literature. However, outbreaks in which waterborne transmission has not been implicated are plentiful. An extensive review of epidemiological data of cholera outbreaks in India, Bangladesh and the Philippines in the 1980s, found that 35–80% of cholera infection in the contacts of index cases occurred after more than 2 days, thereby suggesting that these contacts were infected by person-to-person transmission within the family [35]. The author further argues that even though introduction of cholera within a community is likely to be waterborne, transmission within the community is likely to occur through several routes at the same time. Hence HWT systems, as they focus exclusively on transmission via drinking water, are not by themselves, suitable interventions in every cholera outbreak and should not be employed as a universal remedy.

A review of WASH interventions in 1991 concluded that those for sanitation and water supply for improved hygiene yielded greater reductions in diarrhoeal disease than those for water quality [44]. This agreed with the conventional wisdom, and since then, sanitation and water supply interventions have largely dominated the focus of diarrheal disease prevention. However, since the beginning of this century there has been growing evidence in support of HWT [45]. Meta-analysis of HWT studies shows a combined reduction in diarrhoeal disease of 35% making it as effective on its own as improving access to water and sanitation [6]. These findings, as a result of study limitations, are certainly not without discordance [4, 5], and a more recent meta-analysis controlling for bias estimated that HWT could reduce diarrhoeal disease by a more modest 15% [8]. A recent review investigated the effect of hand washing with soap, water quality improvement and excreta disposal on reducing diarrhoeal disease [3] found that for HWWS, reductions of 42–48% were consistently shown, compared to interventions for water quality (17%) and excreta disposal (36%).

Hygiene promotion and hand washing with soap

In light of the abundance of evidence supporting the promotion of hand washing to reduce diarrhoeal disease, and since the same WASH interventions apply to cholera it is surprising that so few studies specifically evaluate HWWS to prevent cholera. Only one study [27] presented results for self-reported use, and access to soap, which in itself can be subject to bias. The remaining studies [28–30] focussed on the effectiveness of the intervention in eliciting behaviour change and water treatment practices. It should also be appreciated from these findings that improved knowledge does not necessarily translate into improved practices, and as such there is a need to find ways to evaluate hand washing interventions in more depth. Even where hand washing has been ascertained by observation, the use of self-reported diarrhoea incidence as an outcome measure still introduces bias [5, 42]. In future, more objective outcomes such as pathogen presence will need to be used.

The reductions that meta-analysis have shown can be achieved, speak in support of HWWS as a cholera prevention measure, but more research and evaluation are needed to confirm its effectiveness. Furthermore, epidemiological investigations have reported a protective effect of hand washing when food was associated in cholera outbreaks, as well as in outbreaks where the source remained unidentified [46–48]. In an camp in Kenya during an outbreak investigation the source could not be identified, but the authors commented that the epidemic curve was not reflective of a point source outbreak, and found hand washing with soap to be protective [48]. In Zambia, an outbreak investigation identified the consumption of raw vegetables as the source of the epidemic, more precisely the poor hygiene practices of market vendors. Hand washing with soap was a strong protective factor, in contrast with disinfection of household water with a sodium hypochlorite solution [46]. We question the emphasis on water quality, when proper hand hygiene practice can prevent person-to-person transmission as well as food and household water contamination. Hygiene promotion, more precisely hand washing with soap, should be an integral component of any cholera control program. The promotion of HWWS requires formative research among the target population, and the development of tools to achieve and measure behaviour change, creating the illusion that it is expensive and time consuming. Given the rapid onset and high mortality associated with cholera outbreaks, water quality interventions are perceived as being the easiest, quickest and cheapest to deploy, especially when compared to sanitation and water supply programmes which require large investments in infrastructure. However, the distribution of soap is a simple and relatively affordable intervention that can be practised in the household while other interventions in the public domain are being executed [46], while good tools are available to help conduct formative research and plan behaviour change interventions for HWWS [49].

Evidence for diarrhoeal disease suggests that hygiene behaviour is sustained following implementation and is best delivered using small groups and frequent personal contact with a hygiene promoter [50]. The studies in this review all suggest that radio and TV are popular dissemination methods. Therefore the role of mass media should be explored further in comparison with more traditional methods, in those contexts where it is feasible.

Intra-familial transmission

The study of disinfection of jerry cans [31] used for water transport and storage raises the question of the extent to which household hygiene practices must also be improved in order to reduce the excess risk of disease associated with contaminated water at the household level. Recontamination of water storage vessels will most likely be caused by members of the household bringing pathogens into the household and passing them to other family members.

In the past, it was a common practice in cholera outbreaks to spray the houses of cholera patients with disinfectant. This had the potential of causing damage to domestic property and stigmatising the patients and their families, putting in jeopardy the reporting and detection of cases and thus the whole outbreak control effort. Presently, household disinfection kits that normally include a bucket, bleach, soap, cloth and scrubbing brush, are increasingly being used as an alternative to household spraying. The distribution of kits places the responsibility with the family which reduces the stigma attached to the disease and may encourage sustained and improved hygiene at the household level [32]. There is limited evidence for the role of family contacts in the transmission of infection, although studies have found that having a case at home is a risk factor [51, 52] and that family members can spread V. cholerae in stored water and food through contaminated fingers [53]. The practice of sending spraying teams to disinfect houses of cholera patients is now discouraged as there is no evidence that one-off disinfection has any impact on transmission. Furthermore the resources required as well as the time delay before the disinfection is carried out make it less attractive as a practical intervention [54]. The use of kits should be evaluated further in order to inform future best practice.