Arsenic, drinking-water and health risk substitution in arsenic mitigation: A discussion paper
Technology options and controlling risks
A short list of technologies has been identified by the National Committee of Experts (NCE) as suitable for consideration in the emergency response to arsenic (that is in villages where the proportion of arsenic contaminated tubewells exceeds 80%). These are:
- Pond-sand filters
- Dug wells (also referred to as ring wells)
- Deep hand tubewells
- Rainwater harvesting
It is envisaged that direct provision of the first three technologies by DPHE and partners in a supply-driven approach, but that Government would only engage in promotion of rainwater harvesting. The NCE excluded arsenic removal technologies from consideration in the emergency response. These are considered here, however, as the rationale for exclusion of these technologies appears to be inconsistent with approach used to accept the other technologies. Household treatment of water to remove microbial hazards is also briefly considered.
General issues regarding risk substitution
In discussing the potential for risk substitution and risk management for emergency response technologies, some general key points emerge. For all options used in the emergency response, hygiene education will be essential to promote safe handling of water to reduce re-contamination and increasing risk from microbial hazards. A WSP for water handling is included in the supporting documents to this report.
The NCE raise the importance of third party audit to ensure that all components of the intervention (including construction quality) have been followed. This is essential and it is recommended that in the first instance this is done on a blanket rather than sample basis. This also requires that standard designs are developed with indicative unit bills of quantities prepared. Standard auditing procedures and forms should be developed and used. Where audits identify failure to comply, there should be a requirement for the constructing agency to make the required changes at their own cost. It is important that auditing is applied to all water supplies constructed, irrespective of whether this is by Government, private sector or NGOs. Although designs and construction can provide control measures for hazards, experience from around the world shows that risk management of microbial hazards in particular is dependent on good operation and maintenance. Even within the short timeframe envisaged in the emergency response, poor operation and maintenance may lead to a significant increase in risk.
Ensuring good operation and maintenance in community supplies requires two key interventions. Operators must be provided with adequate training and provided with the basic tools with which to undertake maintenance tasks. Training should also include basic skills in monitoring the water supply through action-oriented inspection to ensure that incipient problems are resolved. The second activity should be a process of ongoing support through a surveillance programme that ensures that periodic inspection and testing of the water supply is carried out and results used to support communities in ensuring effective operation. This latter point is briefly discussed further below in Section 4.
For options that use groundwater, controlling contamination requires both proper wellhead/sanitary completion and control of contaminant sources around the facility. The latter are termed protection zones and are typically defined for both microbial and chemical contaminants. This includes, for instance, exclusion of on-site sanitation close to the tubewell to prevent contamination of the aquifer. Simple methodologies are available and have been based on work in Bangladesh. Further assessments are proposed to define safe distances between latrines and tubewells in the country through DPHE/UNICEF.
Pond sand filters
Pond sand filters (PSF) were designed based on the principles of slow-sand filtration although actual designs violate several these principles, including depth of filter bed, flow rate and intermittent supply (head) above the top of the filter. As the PSF draw water from surface water sources, the potential for microbial hazards to be present in source waters is very high. In addition to contamination by human faeces, the potential for animal faecal contamination is likely to be significant and hazards of particular concern will include E.coli O157, Cryptosporidium parvum and Camplyobacter spp. In ponds affected by algal blooms or receiving high nutrient loads, the risk from cyanobacteria toxins will also be increased.
Although some reports suggest that PSFs are efficient in removing microbes, this has only addressed thermotolerant coliforms. Other results indicate that in practice performance is commonly poor and that microbial contamination of final waters is common. Concerns have been raised about the ability of pond-sand filters to remove pathogen loads in very heavily contaminated ponds and would require a further disinfection stage to be effective.
Overall, the generally poor performance of the PSFs suggests that these have significant potential for risk substitution and would not be a preferred solution in most cases.
UNICEF are considering trying to modify the PSF design to become slow-sand filters, with potentially use of pre-filtration. In a variety of studies at both bench and field slow-sand filters, removal rates of viral and bacterial pathogens have been shown to be effective. Up to 5-log removals of bacterial index organisms and viruses are recorded in the literature. Interestingly, although removal of Giardia cysts is relatively good, in general the removal of cryptosporodium is far less effective. The influence of turbidity on slow-sand filter performance is well-known and has led to the development of multi-stage filtration units that provide effective turbidity removal prior to the slow-sand filter. Roughing filters should be capable of removing reasonably fine material and algae. However, if the turbidity is principally clay material, removal may not be as efficient, although with microbial colonisation of the media this could be expected to improve. The potential for use of geotextiles to improve the speed of schmutzdecke formation could also be considered.
Slow-sand filtration is effective in removing algal cells, which may be enhanced where there is some form of pre-treatment to reduce algal cell loading to prevent increasing frequency of removal of the schmutzdeke. There have been laboratory-based studies of toxin removal that showed removal of various toxins in the range of 30% to 80%, but reporting of performance in the field is not available. In developing a slow-sand filter, consideration should be given to development of two units in parallel, which would be usual recommended practice. Although there are clearly good financial reasons for use of a single unit, this will lead to a significant increase in risks of microbial breakthrough during the ripening period. Although this risk could be mitigated by applying a final disinfection stage, this may not provide full protection (for instance if breakthrough also includes particles) and would increase both cost and operation requirements. In ponds with algal blooms, the filter ripening period would almost certainly increase the risk of toxin breakthrough, although this would not be likely to be at levels that are acutely toxic.
If slow-sand filters are to be used, clear criteria will be required to determine which ponds are suitable for use, for instance clearly defined set-back distances for animal rearing. The development of a standardised format (similar to those available in the WHO Guidelines for Drinking-Water Quality Volume 3) is an important tool in determining whether PSF is a viable and appropriate option. Such a tool logically needs to include measures for assessing risks of cyanobacteria presence. Simple tools are available for such assessments using visual inspection and, where available, assessment of total phosphorous as a limiting nutrient for biomass development.
Hand-dug wells of various descriptions are a familiar technology in Bangladesh and several standardised designs are available approved by DPHE for use. However, there is evidence emerging that some dug wells are contaminated with arsenic above the Bangladeshi standard of 50 g/l. This makes their use more problematic as investment in dug wells may not result in any reduced risk from arsenic, but an increased risk from microbial hazards - thus a double risk substitution.
The vulnerability of dug well to microbial contamination is significant. The problems of direct ingress may be overcome through the use of designs that include concrete aprons, concrete linings, raised headwalls and covers. Nonetheless, it is often difficult to ensure that the linings are watertight and ingress of water through the lining at the upper levels is common. Where dug wells have been installed in other countries with heavy seasonal rainfall, difficulties have been found in maintaining microbial quality in wet seasons. The few available studies indicate that this more often due to poor maintenance of the headworks than from sub-surface leaching from, for instance, pit latrines.
Risks may be further reduced by ensuring a sanitary means of abstraction from the well, either through use of a handpump or by windlass and bucket systems. In the latter case, however, theory is much better than practice as commonly designs where the bucket need never touch the ground are rapidly modified by users to maximise their user-friendliness. Pumps installed on dug wells have proven in many cases to provide significant improvements in quality.
In addition to the need for good operation and maintenance, risks can be further reduced by installing a system of chlorination or where a handpump is used, to install a filter at the base of the well covering the screen.
Chlorination could well be limited to only the monsoon season when risks would be expected to significantly increase. Projects in other countries have shown that chlorination can be effective, but requires good training and follow-up. In some areas it is likely that dug wells may become inaccessible during flooding and would also become heavily contaminated. In these situations, emphasis during training must be given to the need for disinfection of the well prior to re-starting use in the dry season.
As noted by UNICEF Bangladesh, it may be more appropriate to consider renovation of dug wells rather than construction of new wells. In both cases, evidence would be required that the shallow aquifer was not arsenic contaminated.
Deep hand tubewells
Deep hand tubewells have been identified as an attractive emergency response measure. This is in part because of the limited evidence of arsenic contamination of the deep aquifer and because in parts of Bangladesh there is a significant aquiclude/aquitard between the shallow and deep aquifers that should minimise the potential for leaching provided construction is properly carried out. There is a recognition of the need to ensure designs are effective in preventing leaching within Bangladesh and recommended practice is outlined in available documents.
In terms of risk substitution, deep hand tubewells are attractive, because microbial contamination is relatively easy to prevent through good wellhead/sanitary completion and by restricting pollution within protection zones. Wellhead completion is relatively easy and cheap to assure during construction and require only limited maintenance to prevent rapid contaminant pathways developing.
Despite the generally positive prognosis of the use of deep hand tubewells, reservations remain regarding their use as an emergency measure. In some parts of Bangladesh, notably the coastal area, the deep aquifer has been exploited for many years and has not shown arsenic contamination. Use of deep hand tubewells in these areas is therefore a sensible option.
The same situation was largely assumed to be the case in other parts of Bangladesh, but this is becoming less certain. More recent testing of deep tubewells have indicated a significant proportion with arsenic contamination. One problem with interpreting this data is the significant uncertainty regarding the accuracy of the records on well depth. Therefore, it is possible that some of the deep tubewells are in fact shallow tubewells drawing water from the contaminated aquifer. This needs to be clarified as a matter of some urgency.
There is a current survey being undertaken by the USGS of the deep aquifer in order to develop a better understanding of arsenic movement in the sub-surface and the scale and degree of arsenic contamination in the deep aquifer. Until this study is completed, it would seem unwise to promote deep hand tubewells as an emergency response, although they may become more viable in the longer-term response.
Rainwater harvesting is an attractive emergency response technology because it can be located at the home, thus preventing an additional burden on women and children to collect water and because of the abundant rain in Bangladesh. Good designs of rainwater tanks are available and relatively low cost.
The major risk associated with rainwater harvesting comes from faecal matter that may get washed into the tank (one particular risk is associated with Salmonella from bird faeces). This is easily mitigated through use of a first-flush diversion system and through cleaning of the roof and guttering. The critical time for this is the start of the monsoon, as once this is underway it is unlikely that there will be significant build-up of faecal material. In rural areas it is unlikely that there will be a significant risk related to chemical hazards, but this will increase in urban areas due to air pollution from traffic. It is essential that the designs of rainwater systems have meshing on the overflow pipes in particular to prevent the water in the tank becoming a vector breeding site.
In relation to hazards from ingestion of water, rainwater harvesting is generally a relatively low-risk option, although large-scale studies have not been carried out. However, as rainwater harvesting is unlikely to provide drinking-water to last the entire dry season, unless larger tanks are provided. This may therefore mean that the promotion of rainwater use must be linked to provision of alternative options to provide water security throughout the year. Nonetheless, rainwater harvesting offers significant potential for improvement in water safety with acceptable risks attached.
Arsenic removal technologies
The NCE recommended that arsenic removal technologies should not be considered in the emergency response because none of the technologies had been formally verified through the ETV. The NCE has highlighted a range of benefits and disadvantages in the use of arsenic removal technologies at both household and community levels and these are not reviewed in detail here. However, it is pertinent to note that some disadvantages highlighted would equally apply to the water supply options recommended.
The results of the rapid assessment of arsenic removal technologies showed concerns that the use of most household units were associated with an increase in microbial contamination compared to feed water. As discussed in the report of the assessment, this is primarily due to poor hygiene and handling. However, it is likely that re-contamination of water from communal water sources will also be common and many studies world-wide have shown that this occurs even where households use sources of good microbial quality.
Although the disadvantages noted by the NCE are not insignificant, it is debatable whether these are sufficient to disbar consideration of arsenic removal technologies within an emergency response. Although none of the technologies has been formally verified, for a number of these technologies there is a large body of evidence of their effectiveness in removing arsenic. Although there may some risk substitution for microbial hazards, this is not considered to be any greater than for any technology where water must be transported and stored within the home.
Much of the evidence already available for these technologies (including from the manufacturers and the rapid assessment of arsenic removal technologies) is at least as good as the evidence of risk reduction offered by alternative water sources. In terms of overall health risks, it is far from clear that the risk posed by some of the outstanding questions is greater than those posed by the alternative water sources proposed.
The use of arsenic removal technologies at either a community or household level offers significant efficiencies as an emergency response. In the case of household technologies, the capital investment costs to Government will be negligible and significant risk reductions can be expected to accrue at a household level. If the purpose of the emergency response is in effect to gain additional time to permit the development of longer-term improvements in water supply, such a process is attractive.
The installation of a community level technology would potentially offer not only the short-term response but could feasibly develop into a longer-term solution for communities that were interested in purchasing a unit. In such a scenario, initial installation may be free of charge, but retention of the unit beyond the immediate response would entail the same processes of cost-recovery as alternative supplies. Effectively, the treatment unit would become a further option that could be considered in a demand-responsive approach to long-term water supply.
There remain issues around the operation of community-level removal plants, notably the monitoring of the performance of arsenic removal and the timing of media replacement. This is complicated in some areas where phosphate in groundwater competes with arsenic for adsorption sites. These represent areas where solutions must be found in the short to medium term, but may not be as significant in the context of an emergency response as external monitoring and support could be provided to communities.
As noted at the start of Section 3, when considering the potential for risk substitution for alternative water supply options, the rational for excluding arsenic removal technologies appears inconsistent. Arsenic removal technologies may provide a viable emergency response where the potential risk substitution can be managed.
Household treatment of water for microbial hazards
This was not considered by the NCE, but has been raised as an option by some NGOs and other working on developing emergency interventions. A number of options exist for undertaking household treatment of water, including low-cost chlorination (notably the CDC Safe Water System), solar disinfection (for instance SODIS) and within Bangladesh the development of a system that uses household cooking stoves to pasteurise water is being developed. Operation and maintenance of most systems for improving microbial water at a household level are simple.
The CDC Safe Water System has been shown to be a very effective means of reducing diarrhoea, with a range of between 25% (from a study in slums in Dhaka with no sanitation) to 85% (in a study in Uzbekistan where sanitary conditions were poor). The data for other interventions is less well developed and as these result in increased water temperature may lead to problems with acceptability. However, it is known that the SODIS system may also remove arsenic.
WHO has recently completed a review that concluded that this was an effective interim solution to obtaining water of acceptable microbial quality. There may be risks of disinfectant by-product formation if very organic-rich surface waters are used.
The promotion of low-cost household water treatment could accompany any of the interventions currently considered under the emergency programme. It could potentially also be used as stand-alone intervention to treat surface water, although additional treatment steps will be needed to reduce turbidity for disinfection and there would be a need for ongoing testing to ensure that it remained effective. This approach should be considered for inclusion within the emergency programme and linked to hygiene education programmes.