Introduction
Legionnaires’ disease is an atypical pneumonic illness caused by the Legionella bacteria. These bacteria can be found naturally in environmental water sources such as rivers, lakes and reservoirs, usually in low numbers. The organism favours warm, stagnant waters, and becomes infective when aerosolised. Poorly maintained aerosol-generating devices can act as a source of the disease, and have been responsible for outbreaks affecting up to 400 cases [1]. Wet cooling systems, water systems and spa pools are all well documented sources of Legionnaires’ disease [1].
The identification of Legionnaires’ disease in 1977 led to the establishment of the European Working Group for Legionella Infections (EWGLI) in 1986. The group aimed to share knowledge and to monitor trends in legionella infections across Europe. Currently, 35 countries are members of EWGLI.
Every year the EWGLI participating countries are asked to submit data on the cases of Legionnaires’ disease that have been diagnosed in their residents during the preceding year to the co-ordinating centre in London. This allows for analysis of this disease on a European level and for comparison of trends between countries. Data from the years 1996 to 2004 have been published previously [2-7]. This paper presents the dataset for the years 2005-2006.
Methods
Participating countries submit an aggregated epidemiological and microbiological dataset using standardised reporting forms. The following data is collected: the number of confirmed and presumptive cases diagnosed in the reporting country during the preceding year, how many died, and the population base covered by the reporters; the method of diagnosis and the species and serogroup of any isolates obtained; age group and sex of the cases (age standardised rates were introduced into the dataset for the first time in 2005); category of exposure (nosocomial [hospital-acquired], travel or community); countries of travel for the travel-associated cases; outbreaks by type, size and suspected source.
Cases are classified as confirmed or presumptive, based upon the EWGLI case definitions [8]. If the method of diagnosis is not known, the case is classified as ‘diagnosis not known’.
Each case is further categorised by exposure history into ‘travel’, ‘nosocomial’ and ‘community’ cases. This is determined for each case by the country of report according to national definitions. In instances where there is insufficient evidence to allocate a case to one of the existing categories (e.g. cases that spent part of their incubation period both in hospital and travelling), the case is categorised as ‘other’, and if there is no exposure information available, the case is allocated to the ‘not known’ category.
Where incidence rates per million population are calculated, they are based upon the reported population size. Regional rather than national incidence rates were obtained for six countries in 2005 (Bulgaria, Croatia, Czech Republic, Lithuania, Romania, Russia) and four in 2006 (Bulgaria, Lithuania, Russia, Romania), and it should be noted that these data may not be representative of the entire country.
In this report, the term ‘outbreak’ is used to describe outbreaks in hospitals or community settings, whereas the term ‘cluster’ is used to describe this type of incident when associated with hotels or other tourist accommodation sites. Every country defines its outbreaks independently, whilst clusters are defined as ‘two cases associated with the same accommodation site within two years’, based upon EWGLINET’s definition.
Results
In 1993, only 19 countries reported a dataset to EWGLI. The response rate has risen significantly since then to 35 countries in both 2005 and 2006. The number of cases reported was 5,700 in 2005 and 6,280 in 2006. In the fourteen years for which this dataset has been collected, 41,627 cases have been reported (Table 1). [Note that in previous publications, the number of cases for 1993 was erroneously reported as 242 instead of 1,242]
Incidence per million population
In both years the highest incidence rates were reported by Spain (28.4/1,000,000 population in 2005 and 30.0/1,000,000 in 2006), followed in 2005 by France (24.8/1,000,000 in 2005 and 23.0/1,000,000 in 2006) and in 2006 by the Netherlands (16.7/1,000,000 in 2005 and 26.9/1,000,000 in 2006). Five countries reported incidences of less than one case per million population in 2005 (Latvia, Malta, Poland, Slovakia and Turkey), in comparison with four countries in 2006 (Latvia, Romania, Slovakia and Turkey). Table 2 shows rates of Legionnaires’ disease per million population for 10 countries, selection based upon their consistent rates, and in order to allow comparison with previous papers.
The overall incidence for Europe was 10.3/1,000,000 in 2005 (based on a denominator population of 550.8 million) and 11.2/1,000,000 in 2006 (based on a denominator of 562.7 million) (Table 1).
Category of cases
For the two years 2005-2006, 629 cases were reported as nosocomial, 7,041 as community cases, 1,395 as associated with travel abroad, 1,227 as associated with travel within the country of residence, 126 as ‘other’ and 1,562 as ‘not known’ category of infection (Table 3).
Age of cases
A breakdown of cases by age group was available in both years for all countries except Czech Republic, Former Yugoslav Republic of Macedonia, Germany, Iceland and Israel*. National demographic data on population size by age group were also provided in order to calculate age standardised rates.
The peak age group of cases was 50-59 in both years; 1,164 cases in this age group were reported in 2005 and 1,289 cases in 2006. Whilst in both years the number of reported cases in the older age groups decreased with age (60-69 years: 1,076 in 2005, 1,170 in 2006; 70-79 years: 914 in 2005, 1,026 in 2006; 80+ years: 639 in 2005, 726 in 2006), the overall age standardised incidence rates increased with increasing age (60-69 years: 2.6 cases per 100,000 in 2005, 2.86 in 2006; 70-79 years: 2.91 in 2005, 3.32 in 2006; 80+ years: 3.83 in 2005, 4.32 in 2006). This increase in incidence rate with age was seen for some individual countries (e.g. France, Italy), but did not hold for all (e.g. England and Wales, The Netherlands) (Table 4).
Outbreaks
During the two years, there were a total of 214 outbreaks or clusters, detected by 18 countries and involving 1,028 cases, 8.6% of the total dataset (Table 5). Countries reported 408 cases associated with 107 outbreaks in 2005, and 620 cases associated with 107 outbreaks in 2006. The outbreaks ranged in size from two to 146 cases. The largest outbreaks in both years occurred in Spain and were attributed to wet cooling systems; involving 50 cases in 2005 [9] and 146 in 2006 [10]. The number of deaths associated with these outbreaks could not be determined from the information collected in this dataset.
Nineteen outbreaks (8.9%) involving 66 cases were linked to hospitals or healthcare facilities and occurred in Austria, Denmark, England and Wales, France, Germany, Ireland, The Netherlands, Poland, Portugal and Spain. Fifteen of these were attributed to contaminated hot or cold water systems, two to wet cooling systems, and two to an unknown source. These sources are as reported by collaborators, and the standard of investigation may vary between countries.
Forty-four outbreaks (20.6%) were linked to community settings, and involved 522 cases. They occurred in Austria, England and Wales, France, Germany, The Netherlands, Northern Ireland, Poland, Portugal, Scotland and Spain. Wet cooling systems were identified as the source in 19 outbreaks, five were attributed to hot or cold water systems, four to whirlpool spas, 15 to an unknown source and one to the sediment at the base of a pressurised water tank [11].
One hundred and forty-three clusters (66.8%) were associated with travel; 94 (44%) with travel outside the country of residence, and 49 (23%) with travel within the country of residence. Hot or cold water systems were responsible for 52 of the clusters, a wet cooling system in one cluster, and whirlpool spas in three. No source was identified for the remaining clusters.
Two outbreaks (one in each year) were linked to prisons, and in both the source of infection was identified as the hot or cold water system. One 2006 outbreak was associated with a nursing home, for which the source was not identified, and the remaining five outbreaks (one in 2005, the rest in 2006) were associated with private homes or buildings. In two of these latter outbreaks, the hot water systems were identified as the source; for the remaining three no source was identified.
Travel-related legionella infection
Altogether in 2005-2006, 26 countries reported a total of 2,622 travel-associated cases; 1,395 were classified as ‘travel abroad’ and 1,227 were associated with travel in the patient’s country of residence (Table 3). Nine countries in 2005 and five countries in 2006 reported no travel-associated cases. Travel within Europe accounted for 89.2% of the travel-associated cases in 2005 (1142 cases) and 90.0% in 2006 (1208 cases). Travel on cruise ships was associated with two cases in 2005 and 11 in 2006.
Spain was associated with the most travel-related cases over this two-year period (545 cases), followed by France (497 cases) and Italy (450 cases). These countries may be disproportionately represented as countries of infection because they reported the highest number of cases in general, and the majority of the travel-associated cases in these countries (59.6%, 77.3% and 53.3%) occurred as a result of domestic travel.
A more detailed analysis of travel-associated cases of Legionnaires’ disease is published each year from EWGLI’s surveillance scheme (EWGLINET) [12]. EWGLINET operates a strict case definition for travel-associated infections, and so not all cases reported as travel in this dataset are reported to the EWGLINET travel dataset. EWGLINET’s case definition excludes patients who had stayed in private accommodation, patients for whom travel information was incomplete, or those for whom travel did not fall within the strict 2-10 day incubation period. EWGLINET does not include these cases because it would not be possible to investigate them further or to link them to other cases who shared the same accommodation site.
Main method of diagnosis
EWGLI collaborators allocate a main method of diagnosis to each reported case, taking culture as the ‘gold-standard’ test. The majority of cases in 2005 and 2006 were primarily diagnosed by urinary antigen detection (9,100 cases, 76.0%), followed by isolation/culture for 1,067 cases (8.9%), single high antibody titres in 716 cases (6.0%), and a fourfold rise in antibody detection levels for 274 cases (2.3%). The remaining cases were diagnosed by respiratory antigen detection, PCR, other methods or the method was unknown (Table 6).
In 2006 compared with 2005, the percentage of cases diagnosed primarily by culture fell from 9.3% to 8.6%, whilst the number of cases with urinary antigen detection as the main method of diagnosis increased from 71.3% to 80.2%. The proportion of cases diagnosed serologically (including both seroconversions and single high titres) fell from 8.8% to 7.8%.
‘Legionella pneumophila serogroup 1’ accounted for 9,219 cases (77.0%) over the two years, ‘L. pneumophila other serogroup or serogroup not determined’ accounted for 1,862 cases (15.5%), and 899 (7.5%) were reported as ‘other Legionella species’ or ‘species not known’.
Of the 1,067 isolates obtained, 862 (80.8%) were identified as L. pneumophila serogroup 1, 94 (8.8%) were L. pneumophila serogroups 2-16 and 74 (6.9%) were L. pneumophila serogroup unknown. Seventeen isolates were diagnosed as other species of Legionella. These were reported as L. anisa (1), L. bozemanii (2), L. brunensis (1), L. cincinnatiensis (1), L. feeleii (1), L. jordanis (1), L. longbeachae (4), and L. micdadei (6). For 20 isolates, the Legionella species was not known.
Deaths
There were 377 deaths reported in 2005 (case fatality rate of 6.6%) compared with 387 deaths in 2006 (case fatality rate of 6.2%). In some countries it is not compulsory to report deaths, and so these figures may underestimate the true mortality attributable to Legionnaires’ disease.
Discussion
The number of cases reported each year to the scheme continues to increase. This rise in case numbers can be partly attributed to increasing ascertainment as national surveillance schemes strengthen. It is especially notable that awareness of Legionnaires’ disease is rising in the newer European Union member states. Invitation to submit annual datasets of cases each year appears to be helping in raising the profile of the disease in these countries, whereas comparison of the rates between countries can highlight the extent of the under-ascertainment. EWGLI hopes that the number of cases reported by these countries in the future annual datasets will increase to better reflect the true number of cases.
Every year, the number of nosocomial cases reported to the dataset remains relatively static (between 300 and 350 since 2003 [7]). In the context of increasing overall case numbers this stability is an encouraging trend, especially since case fatality rates are higher amongst nosocomial cases than amongst other categories [13]. Also, the number of large community outbreaks has been decreasing in recent years. Community outbreaks are unpredictable, so it is difficult to determine whether this decrease is real or artifactual. However, more extensive legislation has been introduced across Europe in recent years for the control and prevention of Legionnaires’ disease, which is probably having a beneficial effect [14-16]. Authorities should be encouraged to ensure that national or WHO guidelines are being utilised in national health care systems [1].
EWGLI has repeatedly raised the problems associated with a decrease in the number of clinical isolates being obtained but, despite this, the number fell again during 2005-2006 to 8.9% (in comparison with 10.0% in 2003-2004 [6]). Lack of clinical isolates can cause difficulties for public health authorities when investigating clusters; with no clinical isolate to compare with any environmental isolates obtained, the suspected source of the outbreak cannot be microbiologically confirmed. As Legionella is a relatively ubiquitous organism in the environment, microbiological confirmation is an important step in determining the source of infection. As the urinary antigen test becomes ever more widespread, this problem is likely to become exacerbated. The increasing use of the urinary antigen test probably also accounts for the high proportion of L. pneumophila serogroup 1 reported to EWGLI, since the test almost exclusively detects these organisms.
The overall incidence rates recorded in this dataset show an increasing rate with increasing age. This is a new variable collected in the dataset and the resulting figures have important implications for Europe’s aging population. Countries should expect an increase of case numbers and a greater demand for health services due to Legionnaires’ disease in the future. There are some countries that do not show this increasing rate with age, however, it is difficult to determine whether this is due to different testing policies or to true trends in the incidence rates.
In 2010, EWGLINET and the collection of this annual dataset will transfer to the European Centre for Disease Prevention and Control (ECDC). This annual dataset has now been collected for fourteen years, to comprise the largest dataset of cases of Legionnaires’ disease in the world and should be continued. EWGLINET itself has been a valuable tool in raising awareness of the disease among members and national surveillance structures, and has contributed to the introduction of regulations and guidelines across Europe. This type of European surveillance is especially important for preventable diseases with environmental sources, as prompt action can tackle these sources as they emerge. With the ageing of populations across Europe, and therefore more people at risk from Legionnaires’ disease, EWGLINET’s importance will only increase.
EWGLI hopes that the ECDC will encourage countries, especially the new EU Member States, to develop their national surveillance schemes and submit their annual data so that surveillance of the Legionnaires’ disease across Europe can continue unabated.
Acknowledgements
EWGLI is partially funded by the European Centre for Disease Prevention and Control. We would like to thank our collaborators for completing their annual datasets and Mrs Rekha Yadav for collecting the data.
*Authors' correction:
1) In Results, the sentence: "A breakdown of cases by age group was available in both years for all countries except Czech Republic, Former Yugoslav Republic of Macedonia, Germany, Iceland, Israel and Portugal." was corrected to: "A breakdown of cases by age group was available in both years for all countries except Czech Republic, Former Yugoslav Republic of Macedonia, Germany, Iceland, and Israel."
2) Data for Portugal were added in Table 4.
These corrections were made on 21 May 2008.