Background The incidence of Legionnaires' disease is increasing globally and artificial water environment is becoming a common source of outbreaks. Molecular typing techniques can help prevent and control Legionella.

Objective To understand the molecular epidemiological characteristics of Legionella pneumophila in artificial water environment of Shanghai hospitals, and provide a scientific basis for the prevention and control of Legionnaires' disease.

Methods Water samples were collected from artificial water environment in 14 hospitals from May to October each year from 2019 to 2020 in Shanghai. A total of 984 water samples were collected from 8 Grade-A tertiary hospitals and 6 non-Grade-A tertiary hospitals, including 312 samples of cooling water, 72 samples of chilled water, and 600 samples of tap water. The water samples were isolated and serotyped for Legionella pneumophila and preserved, and the positive rate of Legionella pneumophila in the samples was used as an indicator of contamination. The preserved strains were resuscitated and 81 surviving strains were obtained for pulsed field gel electrophoresis (PFGE) typing analysis.

Results A total of 124 Legionella pneumophila positive water samples were detected, with a positive rate of 12.60%. The positive rate was higher in the Grade-A tertiary hospitals (16.54%, 87/526) than in the non-Grade-A tertiary hospitals (8.08%, 37/458) (χ²=15.91, P<0.001). The positive rate of cooling water (23.40%) was the highest among different types of water samples, and the difference was statistically significant (χ²=61.19, P<0.001). The difference in positive rate of tap water was statistically significant among different hospital departments (χ²=11.37, P<0.05). The positive rate in 2019 (15.06%) was higher than that in 2020 (9.84%) (χ²=6.23, P<0.05). From May to October, August had the highest annual average positive rate (16.46%) and October had the lowest (8.54%), but the difference in positive rates among months was not statistically significant (χ²=5.39, P=0.37). The difference in positive rate among districts was statistically significant (χ²=24.88, P<0.001). A total of 131 strains of Legionella pneumophila were isolated, with serotype 1 (80.15%, 105/131) predominating. Among the 81 surviving strains of Legionella pneumophila subjected to PFGE typing, the band-based similarity coefficients ranged from 41.30% to 100%. Among the 29 PFGE band types (S1-S29) recorded, each band type included 1-10 strains, and S28 was the dominant band type. Four clusters (I-IV) of PFGE band types were identified, accounting for 66.67% (54/81) of all strains and containing 13 band types.

Conclusion Legionella pneumophila contamination is present in the artificial water environment of hospitals in Shanghai from 2019 to 2020, and the contamination in tap water deserves attention. The detected serotype of Legionella pneumophila is predominantly type 1, and PFGE typing reveals the presence of genetic polymorphism. Therefore, the monitoring and control of Legionella pneumophila in hospital artificial water environment should be strengthened.