We determined the natural history of the colonization of our hospital's potable water by culturing water approximately biweekly from 20 sites throughout the hospital for 4 years. Overall, 545 (24.7%) of the 2200 samples grew Legionella pneumophila. During hyperchlorination, 11·7% of the samples were positive while 41·6% were positive in the absence of chlorination. There was no seasonal trend towards positivity, but there was marked inter-site variation in the semi-quantitative culture results. However, a single strain of legionella (as defined by plasmid profiling) tended to persist at a site. Such a site was a unique ecological niche in that different sites in the same wing were populated by distinct strains.
The two wings of our hospital had a significantly different distribution of strains of legionella – plasmid profile type III predominated in the Victoria Wing while types II and VI predominated in Centennial Wing. Twenty-four of our 28 cases of nosocomial Legionnaires' disease occurred in the Centennial Wing. Three of the four cases in the Victoria Wing were caused by plasmid profile type III while 18 of the 24 isolates from patients who acquired their infection in the Centennial Wing were type II. We conclude that each water outlet serves as its own ecological niche of L. pneumophila.