The first nosocomial legionella outbreak in a Scandinavian general district hospital was identified in an epidemiological investigation of a cluster of pneumonia cases in Värnamo in 1991. During 3 months, 28 patients and 3 staff fell ill and 3 died. Legionella pneumophila serogroup (sg) 1, was found in high counts throughout the hospital hot water system and was probably spread by aerosolisation via shower nozzles. The outbreak was arrested when the circulating hot water temperature was raised from <45°C to >55°C.

The nosocomial infections proved to be part of a wider legionella outbreak and 10 cases contracted outside the hospital were also detected. Legionellae were cultured from 7 of 15 community buildings and 31% (109/354) of subjects living in the Värnamo area had an elevated titre (≥6) to L. pneumophila sg 1 in 1991, indicating a temporary spread of legionella in the community. Subclinical infection was demonstrated and it was estimated that only 10% of all infections had been clinically identified. Nosocomial legionnaires' disease should alert physicians to possible legionella transmission in the community.

In 21 patients from the nosocomial outbreak, the median L. pneumophila sg 1 antibody titre fell from 1:256 to 1:16 in 3 years. After 10 years, the titre level in this clinical cohort had reached the same level as observed in the background population 5 years earlier. Current international serological criteria (a fourfold or greater rise in antibody titre to ≥1:128) identified only 40% (21/52) of pneumonia cases caused by L. pneumophila sg 1 in a Swedish population in 1991-2001. When the antibody response was related to the antibody titre in local residents, the sensitivity rose to 87% (45/52).

Keeping the circulating hospital hot water temperature above 55°C, and vigilant clinical surveillance of nosocomial pneumonia as a method for control of nosocomial legionnaires' disease was evaluated after 10 years of practice. Infection with L. pneumophila sg 1 was diagnosed in 4 out of 366 (1.1 %) patients treated for nosocomial pneumonia, representing 1 case per 26,000 admissions. All patients were cured without complications. L. pneumophila sg 1 was isolated in 30 of 251 (12%) cultured hospital water samples during the monitoring period. It was concluded that this approach was safe and effective for long term control of nosocomial legionnaires' disease in a primary referral hospital.

The hospital hot water system was found to be colonised with a single genotype of L. pneumophila sg 1 over a 12-year period. The same genotype, identified using amplified fragment length polymorphism (AFLP) analysis, was also demonstrated in 18/20 (90%) community isolates. The phenotypic variation was confined to the monoclonal antibody subtypes Benidorm and Bellingham. The hospital genotype was identified in 2 out of 6 Swedish hospitals, both located within 100 km of Värnamo. Obviously, an entire municipal water network may constitute a distinct ecological niche for a single legionella strain. Certain clones also seem to be widely spread in the environment. This implies that results from molecular subtyping must be interpreted cautiously in epidemiological investigations of legionnaires' disease.