Source or coincidence? Factors contributing to source identification of Legionella infections.

In clinical medicine, "evidence-based" medicine is a common phenomenon. However, in Public Health there is a need for intervention research. The aim of this study is to introduce 'evidence-based' medicine into the working procedure of Municipal Health Services (MHSs), in order to achieve a more rational and cost-effective decision making process in infectious disease control. Difficulty in assessment of a potentially harmful situation is best illustrated with an example from MHS practice. In April 2004, a Legionnaires' disease (LD) patient was potentially infected with the Legionella bacteria during a visit to a large theme park. Two earlier reported LD patients could be related to the same location after consultation of a historic database on potential sources of infection with LD. It took the involved MHS over a month to pursue the theme park's management team to agree with sampling activities by a specialised laboratory. In the mean time, two newly reported patients were potentially related to the same park, albeit doubtfully. The theme park's management team was firmly convinced that there was no relation between the visit of the patients to the theme park and their infection. In their view it was mere coincidence that the patients had visited the theme park, which attracts over one million visitors each year. In this case, the MHS faced a major problem: should they be alert on a starting outbreak? Or was there no reason for control, and would action only cause unnecessary panic? In this specific situation, the involved MHS was not able to convince the management team of the necessity given the available data. To date, no one knows who was right. This is not the only reason that calls for a decision making tool. MHSs have to control a variety of infectious diseases in their region. LD is just one of these diseases. Moreover, it is a rare disease, and thus not part of the MHSs daily activities. Still, MHSs need to control situations in which they are coping with different interests from: population at risk, private owners, management teams, government and the health inspectorate. Since August 2002, all MHSs report LD patients to the national source identification programme. Unlike MHS employees, the experts of this programme work with LD on a daily base. The programme provides a database that contains a systematic collection of potential sources of infection with Legionella bacteria. Detailed data from MHSs and laboratories are merged into this database. Using these data it is only little extra effort to determine which factors indicate true sources of infection. This study aims to disclose the experiences of a specialised Legionella team to MHSs in a user-friendly way. The source identification database contains a diversity of clusters: several patients related to one location. The probability for a location being the true source of infection increases with the number of patients related. Since sources may infect patients over a period of 17 years (1), long-time collection of data on potential sources seems rational. However, the more data are collected, the more so-called "clusters of patients" may come up based on coincidence. Calibration of this source identification database to a flexible matched control database is necessary to enable experts to differ between source and coincidence. In fact, factors contributing to identify true sources of infection can be determined. Relevant factors can be reduced into a model for a decision-making tool. This evidence-based tool will assist MHSs during upcoming clusters of LD patients. The tool will be accessible through a secure web server. Use of the decision-making tool by MHSs will prevent new cases from registered potential sources.