To the Editor: Recycling water from sewage into drinking water was recently discussed in the Journal.1 Although this is technically feasible, we need to be very wary. Such recycling is associated with very high ongoing monetary and energy costs, but, most importantly from a health perspective, is a “very high-risk”2 proposal that reverses 150 years of good public health policy of striving to keep sewage out of our drinking water supplies.

When we need to recycle water from highly contaminated sources, it is much safer to do so for industrial purposes using separated pipelines (as is done in Singapore and Brisbane). The most extensive scientific review on this issue concluded that putting it into drinking water should be a “last resort”, that

Sewage contains very high concentrations of pathogens and drugs. Viruses (the most difficult pathogens to remove) can occur in concentrations higher than 106 per litre — orders of magnitude higher than in even the most polluted rivers. The technical and human performance needed to remove viruses safely will have to be proportionately higher than current practice — difficult to achieve, as we already have skills shortages. We would also need to ensure that the system will work all the time.

Reverse osmosis (RO) is the most effective way to remove viruses and drugs from sewage, and should remove virtually all viruses and drugs. Surprisingly, few in-use data are available to check this. RO membranes seem to leak. One study found that RO only removed 92% of antibiotics.4 Recent safety reviews, including an Australian review5 (based on the previous study3), showed viruses were still detected post-treatment at three of seven sites on some occasions. The calculated virus removal ranged from 87% to > 99.995%, which equates to a “log reduction” of 1 to 5. However, to produce safe drinking water from sewage, we need a consistent 9.5-log reduction for enteroviruses.2 Even Giardia was not always removed. This less than optimal performance was when the system was not known to be malfunctioning; lowered performance might occur as often as 5 days a year.6

Current surrogate testing (eg, organic carbon) can only detect a membrane leak (or bypass) of at least 1%, which is well short of meeting the 9.5-log reduction we need for virus removal and reasonable safety.2 We need real-time tests to show that there is adequate virus removal, rather than none at all or only becoming aware of a problem after processed but contaminated water is already in our reservoirs.

In reply: We agree that augmentation of drinking water sources with recycled sewage goes against the traditional policy of separating the two, and that many factors including cost and energy use need consideration in securing future water supplies. Our editorial1 was not written to promote potable recycling, which ultimately is a political and societal decision, but rather to point out that the carefully considered Australian guidelines for water recycling2 have been developed to ensure that, if this form of recycling is contemplated, it is done in a manner that safeguards public health.

Main messages in the recycling guidelines include the importance of risk assessment for each individual scheme; avoidance of complete reliance on any single technical step (including reverse osmosis) for removal of contaminants, via a “multiple barrier” approach; adequate operational and water quality verification monitoring; and optimising training and skills management within water treatment facilities. Importantly, credits given for “log removal” for each treatment step are based on verifiable on-line performance, not theoretical values.2

These practical messages are also fundamental to the way we manage our conventional drinking water supplies.3 Consequently, discussions about recycling help reinforce the importance of continual assessment of water management, regardless of the source, and help ensure we do not become complacent.