The European Legionnaires’ disease Surveillance Network, which comprises Iceland, Norway, and all 28 EU member states, including the United Kingdom, operates under the European Centre for Disease Prevention and Control (Stockholm, Sweden). Each state annually reports its LD cases to The European Surveillance System database hosted by the European Centre for Disease Prevention and Control. Countries report their LD cases with variables such as patient age, patient sex, date of disease onset, probable setting of infection (e.g. travel-associated), whether the case-patient is part of a cluster, laboratory method used for diagnosis, causative pathogen, and clinical outcome. LD patients who travelled (abroad or domestically) 2–10 days before disease onset are considered to have travel-associated LD. Many EU–EEA countries define HCA LD on the basis of whether the patient was in a hospital or healthcare facility <10 days before disease onset (16–18). Community-acquired LD is a diagnosis of exclusion (i.e., non-HCA, non–travel-associated). We defined a locally acquired case as any case not associated with travel.

Our analysis included all locally acquired cases reported during 2008–2017 that met the 2012 EU–EEA case definition of confirmed and probable cases of LD (19). We excluded travel-associated cases because they encompass heterogeneous exposures. We reclassified LD cases reported before 2012 according to the 2012 EU–EEA case definition. We defined hospital-acquired cases as those reported from a hospital, whereas HCA LD cases comprised hospital-acquired cases and cases reported from other healthcare facilities (e.g., nursing homes). We made this distinction for 2 main reasons. First, hospital patients, independent of age, might be immunocompromised and therefore more susceptible to LD. Second, the duration of Legionella exposure might be shorter for patients admitted to the hospital for acute care than for residents of long-term care facilities.

We compared the characteristics of HCA LD patients and community-acquired LD patients. We sorted patients into 8 groups by age at diagnosis (2). We compared their characteristics by using the χ² test with a 2-sided p value of <0.05. In addition, we used logistic regression to analyze the odds of acquiring HCA or community-acquired LD, the odds of death, and the confounding effects of age and sex.

In a subanalysis of culture-confirmed cases (i.e., cases ascertained by isolation of Legionella spp. from respiratory secretions or any normally sterile site), we compared the causative pathogens of HCA LD patients with community-acquired LD patients. We grouped Legionella isolates by species and monoclonal subtypes; we further classified monoclonal subtypes by the virulence-associated epitope recognized by monoclonal antibody (mAb) 3/1 of the Dresden Panel (10). We further explored the factors associated with outcome in a subset analysis of culture-confirmed cases with information about the causative strain. We used Stata 14 (StataCorp LLC, https://www.stata.com) for all statistical analyses.

Of the 4,310 HCA LD patients for whom sex was known, 2,499 (58.0%) were male, resulting in a male:female ratio of 1.4:1. However, the proportion of HCA LD was higher for female LD patients than for male LD patients (14.2% vs. 7.8%; p<0.01). The male:female ratio was lower (0.9:1) for both younger (<20 years of age) and older (>80 years of age) patients; the ratio peaked at 2.2:1 for patients 40–49 years of age. Of the 4,313 HCA LD patients for whom age was known, 2,650 (61.4%) were >70 years of age. The proportion of HCA LD patients >70 years of age was higher for patients linked to other healthcare facilities than for those linked to hospitals (80.0% vs. 52.8%; p<0.01). After adjustment for age, year, and reporting country, women were more likely than men to have acquired their infection in a healthcare facility (odds ratio [OR] 1.60, 95% CI 1.49–1.71) (Table 2). Patients <20 years of age of were twice as likely as patients 50–59 years of age to have HCA LD (OR 2.04, 95% CI 1.25–3.33). The risk for an HCA LD diagnosis increased with age for patients >60 years of age, peaking for patients >80 years of age (OR 4.58, 95% CI 4.11–5.12).

The monthly distribution of onset peaked in August and September for both community-acquired and HCA LD (Figure 2). The proportion of community-acquired LD cases that developed during June–November was greater than that of HCA LD (66.9% vs. 55.8%; p<0.01).

During 2008–2017, The European Surveillance System recorded 48,197 laboratory test results for the 44,726 LD patients included in this analysis. LD diagnosis by urinary antigen test (UAT) was more common for community-acquired than HCA LD cases (88.7% vs. 84.3%; p<0.01). On the other hand, culture confirmation of LD was more common for HCA LD than community-acquired cases (15.9% vs. 10.4%; p<0.01) (Table 2). Of the 4,884 culture-confirmed cases, 2,205 (45.1%) were also ascertained by UAT.

Among HCA LD cases, the proportion of culture-confirmed cases was higher for hospital-acquired cases than cases linked to other healthcare settings (18.8% vs. 9.7%; p<0.01). PCR diagnosis was more likely for HCA LD than for community-acquired LD (6.8% vs. 5.0%; p<0.01). For both community-acquired and HCA LD, the proportion of cases ascertained on the basis of a single high titer of a specific serum antibody was similar (»2.5%). The proportion of cases diagnosed by a 4-fold rise in titer or by direct immunofluorescence assay also was similar for both groups (<1% for both tests).

Of the 4,859 culture-confirmed cases reported with a known causative pathogen species, 4,739 (97.5%) were caused by Legionella pneumophila. This proportion was similar for community-acquired and HCA LD cases (97.4% vs. 98.1%; p = 0.31) (Table 3). Of the 4,533 laboratory-confirmed cases of L. pneumophila reported with a known serogroup, 4,137 (91.3%) were caused by L. pneumophila serogroup 1. This proportion was higher for community-acquired cases than HCA LD cases (92.3% vs. 85.1%; p<0.01). Following L. pneumophila serogroup 1 (537/684 culture-confirmed HCA LD), the most common serogroups associated with culture-confirmed HCA LD were L. pneumophila serogroups 3 (33 cases), 6 (15 cases), and 5 (11 cases). Of the 107 community-acquired cases with culture confirmation of other Legionella species, 48 (44.9%) were caused by L. longbeachae. The European Surveillance System did not record any HCA cases of L. longbeachae.

Of the 856 culture-confirmed cases caused by L. pneumophila serogroup 1 for which isolates were subtyped using mAbs, 679 (79.3%) tested positive for mAb 3/1 (Table 4). This proportion was higher for community-acquired than HCA LD (83.6% vs. 43.3%; p<0.01).

Of the 32,379 case-patients with known outcomes, 3,448 (10.6%) died (Table 5). The proportion of patients who died was higher for those with HCA than community-acquired LD (28.8% vs. 8.6%; p<0.01). This proportion was similar for patients with hospital-acquired LD and LD linked to other healthcare facilities (29.2% vs. 28.1%; p = 0.52). After adjustment for age, sex, year, and reporting country, the death rate was higher for HCA than community-acquired LD (OR 3.02, 95% CI 2.75–3.32). The death rate was higher for hospital-acquired LD than for LD linked to other healthcare facilities (OR 3.50, 95% CI 3.14–3.91) (Table 5). After we restricted our analysis to the 4,121 culture-confirmed cases for which information was available about causative species and serogroups, the death rate for HCA LD remained higher than for community-acquired LD (OR 2.60, 95% CI 2.11–3.22). Patients infected by L. pneumophila nonserogroup 1 had a higher risk for death than those infected by L. pneumophila serogroup 1 (OR 2.17, 95% CI 1.61–2.92). Infection with other species was not associated with a higher death rate. Of the 690 culture-confirmed cases caused by L. pneumophila serogroup 1 for which information about monoclonal subtype was available, patients with HCA LD still had a higher death rate than those with community-acquired LD (OR 1.93, 95% CI 1.04–3.58); cases caused by mAb 3/1–negative strains were 4 times more likely to be fatal than those caused by mAb 3/1–positive strains (OR 4.20, 95% CI 2.32–7.61).

Of the 31,973 LD patients with known cluster status, 2,039 (6.4%) were part of a cluster. This proportion was higher for HCA LD patients than community-acquired LD patients (10.6% vs. 6.0%; p<0.01).