C A Joseph, PHLS Communicable Disease Surveillance Centre, London, England.

This interesting paper has analysed the data on cases of legionellosis associated with travel to Spain and highlighted some issues which are relevant to the European Working Group for Legionella Infections (EWGLI) surveillance system, eg the reporting of cases with missing information, particularly in the early years when the scheme first started and the difficulty of interpreting data when information is incomplete. EWGLI's software is currently being rewritten to take account of further developments in the scheme and to include improved methods for analysis. The new software will be distributed to all centres later this year.

The authors argue that the names of hotels should be removed from the database when there is no evidence to link them to a case of legionnaires' disease. This issue has been raised in the past and discussed at length at annual EWGLI meetings. The consensus view of the group has been that such deletion of hotel names would severely limit the scheme's ability to fulfil its objective of detecting hotel outbreaks and clusters. The name and address of any accommodation where a case stayed during the second to tenth day before they became ill should remain on the database indefinitely. The database is dynamic and includes travel histories of all reported cases, not just hotels that have been epidemiologically or environmentally implicated in an episode of legionella infection. For single cases of legionnaires' disease (which make up the majority of cases on the database) it is not recommended that environmental investigations be carried out at the hotel of stay unless the database shows that previous cases have stayed there. The chance of the hotel being the source of infection increases when two or more cases are known to have stayed there during the ten days before becoming ill. Links can be made to hotels where cases have stayed during their incubation period only if they are on the database. Furthermore, pointing out to the country concerned that more than one case has stayed at a particular hotel over a period of time, strengthens the request of the collaborator in that country that the hotel be investigated and its control measures monitored. There are several examples on the database of outbreaks occurring at particular hotels, investigations being carried out and control measures applied, only to be followed by further cases or outbreaks at the same hotel one or two years later. Any subsequent cases are a measure of the success or otherwise of control measures taken by individual hotels in tourist resorts.

The publication of this paper is timely and will encourage debate of these issues when the EWGLI group next meets in June 1997.