Tool speeds up source detection in legionella outbreak

Our country experienced another legionella outbreak recently. Two years ago, a 'legionella cloud' was released in the Ghent port area, striking thirty people (employees of port companies but also people living in the vicinity and passers-by) with the feared Legionnaire's disease. Two of them died of the lung disease. 
Legionellosis is a notifiable infectious disease. As usual when several legionella infections are reported in the same period and the same area, the Flemish government's Care and Health Agency (AZG) immediately opens an investigation. Experts tried to trace the source of the contamination as quickly as possible. They interviewed the patients to find out as much information as possible about their whereabouts in the days and weeks before the outbreak began. To do so, they had to go back almost three weeks, because the incubation period - the time between infection with the bacteria and the first symptoms - of Legionnaires' disease can be very long sometimes. 

Source detection

"It was a stressful period," recalls Liesbeth Lejon from the AZG. "New infections were still being reported several weeks after the first reports. This suggested that the source might still be active." The survey ruled out the possibility that the source was a temporary activity visited by the patients – as had been the case in the previous fatal outbreak in Kapellen in 1999, where the source was located in a trade fair for bubble baths. The source was narrowed down to an industrial plant that had spread a legionella cloud over a larger area. 
Based on the wind direction during the weeks before the first contamination, the source was located in the Ghent canal zone. "Based on those data, but also on experience, we defined a search zone," says Lejon. "Fortunately, our assessment was correct." It was also fortunate that the factory where the legionella source was found at the end of May 2019 complied with the notification requirement for cooling towers. "This immediately provided us with accurate contact details." 

Already during the source detection, the AZG had called in air quality experts from VITO, and more specifically on modelling fine dust, exhaust gases and other pollutants. Their input confirmed that Lejon and her colleagues were on the right track, allowing them to fully concentrate on the defined area. "Legionella spreads via aerosols, small moisture droplets that often behave like the substances in our air quality models," says Wouter Lefebvre of VITO. Using inverse modelling, Lefebvre was able to reconstruct the distribution of the legionella cloud backwards over time. But the way the modelling work had to be done was not ideal. "There was no established protocol for source detection by modelling and the necessary data transformation caused delays." 

The experience with the outbreak in the Ghent canal zone proved to be an incentive and at the same time a good test case for the development of a fully-fledged legionella tool. It could then be used as standard in future outbreaks with a potential environmental source. The AZG turned to VITO because of its rich and broad expertise in air quality modelling combined with the necessary IT knowledge to transform complex scientific models into usable tools. 

Reverse wind direction 

Inverse modelling involves completely inverting air quality models. The patients infected with legionella are then regarded as sources. An infection region is reconstructed based on their physical location, time and the reverse wind direction at the time. "We generate a separate map for each hour for each patient," says Denis Caeyers of VITO. "At the end, we put the maps together, and where the overlap is largest, that is where the source of legionella is most likely to be found." This method is actually not new, except that modelling used to be done with pen and paper and the maps were printed on transparencies. These hand-drawn maps were put on top of one another on a projector to make the overlap visible. Today, of course, that is no longer necessary. Furthermore, the VITO modellers can now combine the maps and models they generate with other data, for example through the mandatory cooling tower reporting or satellite images. Automatic detection based on satellite images by the VITO Remote Sensing unit can also be integrated, allowing unreported cooling towers to be visualised and investigated. 

Early and reliable source detection is important, not only to stop a legionella outbreak as quickly as possible but also to be able to carry out highly targeted sampling at suspected facilities (e.g. cooling towers) during the search, which is very time-consuming and expensive. In the event of a possible outbreak in the future, the AZG can now use this new tool very quickly. 

More info 
liesbeth.lejon@vlaanderen.be 
denis.caeyers@vito.be 

Guidelines for restart 

Legionella bacteria thrive best in stagnant water at a temperature of between 20 and 50 °C. The measures to limit the spread of the corona virus required the (partial) closure of sanitary installations in sport facilities, holiday parks and other accommodations and in public buildings. A (temporary) standstill and the restart of these installations is a known risk moment for the growth and spread of Legionella. That is why the AZG (in cooperation with the Scientific and Technical Centre for the construction industry) issued precautionary guidelines for the restart of these public sanitary installations. 

Preventive monitoring of Legionella

The PREMOLEG project was selected as a demonstration and dissemination project as part of the Vlakwa Open Call 2020, allowing the innovative technology behind it to be tested with end users for a year. It is financed via Vlakwa by the Province of West Flanders and the Province of Antwerp. 
These two technologies can both detect Legionella preventively (i.e. long before problems arise) in both drinking water and cooling water systems. Specifically, they consist of an AI monitoring system and a quick offline Legionella concentration determination in collected water samples. 
The preventive monitoring can lead to more economical use of chemicals for disinfection of the pipes and to lower water consumption (because flushing only takes place when it is really necessary). "This makes it a very promising technology that fits in nicely with the broader sustainability picture," says Veerle Depuydt of Vlakwa. "If results confirm it works well, it deserves to be rolled out more widely among legionella-sensitive companies." 
The technologies were developed by the Flemish start-up Liquisens. They will be tested in four organisations between now and the end of the year 2021: a coffee roaster with a small cooling water system, two care institutions with a complex drinking water system and a factory equipped with a cooling tower. 

More info
vd@vlakwa.be