Rapid communications A community outbreak of Legionnaires ’ disease in

During August and September 2010, an outbreak comprising 22 cases of Legionnaires' disease was identified by the public health service in Wales. The cases are distributed over a wide geographical area in South East Wales. There are two space-time clusters centred on the upper Rhymney Valley and the lower Cynon Valley respectively. Epidemiological investigations are compatible with cooling towers in each location as the potential source, but environmental inspections were satisfactory and microbiological investigations are inconclusive.

During August and September 2010, an outbreak comprising 22 cases of Legionnaires' disease was identified by the public health service in Wales.The cases are distributed over a wide geographical area in South East Wales.There are two space-time clusters centred on the upper Rhymney Valley and the lower Cynon Valley respectively.Epidemiological investigations are compatible with cooling towers in each location as the potential source, but environmental inspections were satisfactory and microbiological investigations are inconclusive.

Outbreak description
In mid and late August 2010, six cases of Legionnaires' disease, with no history of recent travel abroad, were reported to the public health service.All the patients tested positive for urinary antigen for Legionella pneumophila serogroup 1 (mAb2 positive), which is the most common cause of Legionnaires' disease in the United Kingdom (UK).There were 24 cases of Legionnaires' disease in Wales in 2009 and an average of 13 cases per year over the past 10 years.A multidisciplinary Outbreak Control Team was convened on 3 September 2010 and an outbreak of Legionnaires' disease was declared.

Epidemiological investigation
Active case finding was undertaken by alerting clinicians throughout Wales and by alerting public health professionals throughout the UK.All cases of Legionnaires' disease reported in Wales from 1 July 2010 to 30 September 2010 were reviewed.A probable outbreak-associated case was defined as a person with a positive urine antigen test for L. pneumophila and onset of symptoms after 1 July 2010, who lived in, or had visited, the outbreak area during the 14 days before onset of symptoms.The outbreak case definition was based on the European Union case definitions for Legionnaires' disease [1].The outbreak area was defined as the 12 km corridor on either side of the Heads of the Valleys Road (A465).This is a major road that links South West Wales with South East Wales and the English Midlands.Over the next two weeks a further 16 cases of Legionnaires' disease were identified.
Environmental health officers from 10 county or city councils interviewed all cases as soon as possible after notification.Information on demographic factors and recent movements within and outside the outbreak area was collected for the 14-day period before the onset of symptoms.Patients' residence and movements, as well as the locations of cooling towers in the area, were mapped using a geographical information system in order to help generate hypotheses about potential sources of exposure.

Environmental and microbiological investigations
By law, all cooling towers and evaporative condensers in the UK are required to be registered with the local council [2].Owners should also follow the Approved Code of Practice (ACOP) on their operation and maintenance [3].The Health and Safety Executive (HSE) inspected registered premises in the Merthyr Tydfil, Blaenau Gwent, Rhondda Cynon Taff and Caerphilly county council areas to identify any operating deficiencies.A search was also undertaken for unregistered premises.In addition, other potential sources within the outbreak area that might generate aerosols such as car wash and jet wash facilities were visited and inspected by local authority environmental health officers.Water samples were taken from a wide variety of sources at all sites that were found to have operating deficiencies or that were epidemiologically linked to the outbreak and analysed for legionella by PCR and culture.
Environmental samples were sent to the Severn Trent Water Company laboratory for testing and to the Respiratory and Systemic Infection Laboratory (RSIL) of the Health Protection Agency for further typing.Patient samples were also collected and sent to RSIL for testing and typing, in order to identify a match with the potential environmental source.

Results
Thirty-one patients with Legionnaires' disease with onset since 1 July 2010 were identified, 22 of whom met the outbreak case definition [1].Dates of onset of symptoms ranged from 4 August to 10 September 2010 (Figure 1) and none had travelled abroad in the two weeks beforehand.
Cases had a median age of 65 years (range  and most had underlying medical risk factors that are known to be associated with Legionnaires' disease.There were 15 males and seven females.All 22 cases were admitted to hospital and two died (case fatality rate 7%).There were two distinct spatio-temporal clusters (Figure 2): a cluster of seven people in the upper Rhymney Valley (cluster A) and a cluster of six people in the lower Cynon Valley (cluster B) including one case linked to both.The clusters are located around 15km apart but both are within a 5 km radius of a cooling tower.Of the remaining 10 cases, two were epidemiologically linked to a retail premises outside the outbreak area and one was microbiologically linked to another premises outside the outbreak area.The source for the remaining cases is unknown.

Laboratory results
All cases tested positive for urinary antigen for L. pneumophila serogroup 1 (mAb2 positive).Respiratory samples were available for typing from 11 patients.L. pneumophila has so far been cultured/ typed from four patients and typed directly from sputum in a further three.Six strains are different subtypes and/or genotypes and neither cluster A nor B can be clearly characterised (Table ).

Environmental samples
In total, 28 registered premises were visited by the HSE.Another three unregistered premises were identified and visited.A total of 26 environmental samples were collected and tested and all but one (linked to a single case not associated with either cluster) were negative on culture.

Control measures
In the vicinity of cluster A, there is a cooling tower and an air scrubber.Both were voluntarily closed down after inspection and the cooling tower was cleaned and

Figure 1
Patients with Legionnaire's disease by date of symptom onset, South Wales, July-September 2010 (n=31) disinfected.Both have resumed normal operation following microbiological clearance.A cooling tower in the vicinity of cluster B was also closed, disinfected and re-opened following microbiological clearance.None of the sites has been definitively identified as the source of the outbreak.
In addition, a Prohibition Notice was served by the HSE at a site in Merthyr Tydfil.The notice was served as the cooling towers were not being operated in accordance with the Approved Code of Practice.Improvement Notices were also served on six further companies that were found to have minor deficiencies in their training, risk management policies, or maintenance procedures requiring them to improve the operation of their systems.None of these companies were located in the vicinity of Cluster A or B.

Discussion and conclusion
The outbreak investigation has so far identified two time-space clusters compatible with cooling towers in each location as the potential source.The outbreak has proved a particular challenge to investigate by virtue of the wide geographical distribution of cases, the identification of two distinct spatio-temporal clusters, the existence of different strains of L. pneumophila in cases, and the absence of L. pneumophila in environmental samples.
Previous outbreak investigations have identified geographical spread of Legionella up to 10 km from an industrial source [4,5].However, even this would not explain the wide geographical distribution in this outbreak.Although the two clusters are only 15 km apart in a straight line, the topography comprises a series of hills and valleys and the road distance is considerably greater.
When microbiological results do not confirm a single source, or are contradictory, it can be difficult to decide if an outbreak is actually taking place [6].The number of cases in this outbreak is clearly in excess of what would normally be observed in South Wales at this time of year.Some of this may be the consequence of heightened awareness and active case finding after declaration of the outbreak, but this does not explain the clustering of cases.
Investigations are continuing and some typing results are still awaited.So far, there has been only one successful match between an environmental and human sample.This highlights the importance of isolating and typing Legionella from as many clinical and environmental samples as possible to help identify the source [7,8].The fact that no further cases have been detected since mid September 2010 indicates that the control measures taken appear to have been successful.Since its emergence in early 2009 [1] the pandemic influenza A(H1N1) virus has remained closely related to one of the earliest viruses detected, A/California/7/2009, with little change in the viruses' genetic makeup in even the most variable genes, haemagglutinin (HA) and neuraminidase (NA).This lack of drift was reflected in the World Health Organization's (WHO) Vaccine formulation decisions which recommended an A/California/7/2009-like pandemic influenza A(H1N1) virus for both the southern hemisphere 2010 and the northern hemisphere 2010-11 influenza vaccines [2].While some genetic variants have been reported such as the D222G (D239G numbering if starting at the first methionine) HA mutation which was linked with more severe outcomes following pandemic influenza virus infection [3] and a more commonly seen E391K change in the HA gene [4] during late 2009, no clear variant has predominated in a country or region and no vaccine update has been forthcoming.This report, however, describes the recent emergence in Singapore and subsequent spread of a genetic variant of the pandemic influenza A(H1N1) virus to Australia and New Zealand during their 2010 winter influenza season, where it now predominates and has been detected in some vaccine breakthroughs and fatal cases.

Genetic characterisation of the pandemic influenza A(H1N1) variant
We sequenced the HA, NA and other genes of 2010 pandemic influenza A(H1N1) viruses from Singapore, Australia, New Zealand and elsewhere using conventional Sanger sequencing.Viruses early in 2010 (January to April) from Singapore and Australia showed the E391K (numbering beginning at the first methionine in HA; equivalent to E374K if starting after the signal peptide sequence in HA at DTLC) change in the HA but were scattered throughout the phylogenetic trees for HA (Figure 1) and the whole genome (Figure 2).
On 13 April 2010 an influenza A(H1N1) strain, A/ Singapore/CC01/2010, was detected in Singapore that had further changes in HA (N142D; numbering beginning at the first M in HA; equivalent to N125D if starting after the signal peptide sequence in HA at DTLC) and in NA (M15I, N189S).Viruses with these changes then increased in frequency during May and June 2010 in Singapore and became the predominant viruses by mid-2010.Of the pandemic influenza viruses sequenced in Australia in 2010, those sampled in January and February mostly had the E391K change.
Viruses with the dual HA mutation (E391K and N142D) were first detected in late April 2010 (e.g.A/ Brisbane/10/2010, sample date 29 April 2010), and by June 2010, viruses with these HA (and NA) changes predominated.In the North and South Islands of New Zealand, viruses that were collected in July and August 2010 also showed this dual change in the HA along with the NA changes.Viruses with these genetic characteristics in the HA protein have only been detected sporadically in some other countries (e.g.Guam; Figure 1, Table 1), and the complete set of changes in HA and NA has not yet been reported in the northern hemisphere to date in 2010.
These variant viruses have also been associated with several vaccine breakthroughs and a number of fatalities in both Singapore and Australia (labelled 'dec' in Figure 1).Examination of other gene segments of several pandemic influenza A(H1N1) variants showed that the other six segments were all very similar to the A/California/7/2009 strain (nucleotide identity ranged from >99% to 100%) with no evidence of gene reassorting between the pandemic influenza (H1N1) virus and seasonal influenza A(H1N1) or H3N2 viruses or another influenza A subtype.Nevertheless, as marked in the whole genome phylogenetic tree (Figure 2), some additional mutations in the other gene segments (PB2, PB1, NP, NS1) appeared commonly among the recent variant strains, but the significance of these changes remains to be determined.
To further investigate the importance of these surface antigen mutations, we built a structural homology model of HA from the A/Brisbane/10/2010(H1N1) virus based on the template structure of A/ California/04/2009(H1N1) (PDB:3LZG) [5] using MODELLER with loop refinement [6] and ProQ [7] for model quality control.In Figure 3, we superimpose our model with the complex of the antigenically similar HA of the 1918 influenza A(H1) virus bound to an antibody that recognises the classical Sa epitope (PDB:3LZF) [5].We show that N142D is centrally located in this epitope, which led us to further investigate the effect of the mutation on antigenic properties with haemagglutination inhibition assays.
Adding to the possibility of the N142D mutation affecting antigenicity, the equivalent mutation N129D(H3)/N124D(H5) in influenza A/Mallard/ Pennsylvania/10218/84(H5N2) virus was previously reported to cause antigenic drift as an escape mutant [8].However, the findings in the context of avian H5 may not be easily transferable to the swine-origin H1.
Generally, a single mutation will only partially affect antigenicity as typically several mutations in the same epitope are needed to seriously alter vaccine efficacy.
An additional mutation in the HA sequence, D111N, was common among samples from New Zealand, and an equivalent mutation in avian influenza has been reported to be related to a shift in host specificity (from avian towards human) which could hypothetically mean a small fitness advantage in the human host [9].The equivalent mutation (referred to as D94N in [9]) enhanced binding of HA to the human-type SA-α-2,6-Gal receptor and decreased binding to the aviantype SA-α-2,3-Gal receptor.It was also observed that the mutation was able to enhance HA-mediated membrane fusion in mammalian cells.Structurally, D111N is located on the outside of the bottom of the sialic acid binding pocket with the side chain pointing to the outside (Figure 3) and the mechanism that causes the reported effect is not fully clear.
HA D111N is almost exclusively found in combination with another mutation, HA V267A, which is located at an internal beta sheet below the receptor binding pocket facing the Sa epitope (Figure 2).Exchanging valine for the smaller alanine at this position creates a small cavity which may slightly alter the surface of the epitope on top and could add to the effects of N142D.However, so far the HA D111N and V267A mutations have only occurred in a close temporal and geographic context (four in New Zealand and three in eastern Australia in July and August 2010, see Figure 1) and their increased local occurrence may simply be due to founder effects.
Two additional mutations occurred in the NA sequence, M15I and N189S, which were predominant in viruses from Singapore and Australia by mid-2010.NA M15I is located in the signal peptide region.The signal peptide is the motif required for cell surface expression of the viral protein and its existence and quality can be predicted with the programme SignalP 3.0 [10].For the NA M15I mutation, the prediction score (D-score) increases from 0.326 for M15 to 0.404 for I15.This could hypothetically indicate that the mutated version represents a better signal peptide with potentially increasing secretion and surface expression efficiency, but this needs to be further tested experimentally.NA N189S on the other hand is located at the bottom side of the NA structure, far away from the sialic acid-and drug-binding pocket and any phenotypic change cannot easily be predicted.

Antigenic analysis of variant viruses
While genetic differences were apparent in this variant group of pandemic influenza A(H1N1) viruses, when they were assessed for antigenic variation in haemagglutination inhibition assays (HI) using ferret antisera raised to A/California/7/2009-like viruses and viruses from the new variant group (e.g.A/ Singapore/548/2010, A/Brisbane10/2010), no differences in titres were apparent, indicating that these viruses were not antigenically distinguishable from the reference and vaccine virus A/California/7/2009 (Table 2).
Further antigenic analysis was performed using a small human serum panel (n=48) containing pre-and postvaccination sera from Australian adults (  ncbi.nlm.nih.gov/genomes/FLU/FLU.html).Phylogenetic analysis was conducted on 1,877 strains with full-length nucleotide sequences available for all eight segments.The protein-coding nucleotide sequences for these strains were concatenated such that a single sequence representing a single strain contained nucleotide sequences encoding all 10 proteins.The vaccine strain A/California/07/2009 and recent Singaporean, Thai, New Zealand and Australian strains were set aside and redundancy was removed within the remaining strains with Cd-hit [13] by allowing a maximal sequence identity of 80% to reduce the set to 20 non-redundant strains.This representative set was aligned with the vaccine strain and 16 other strains from Singapore, Thailand, New Zealand and Australia using MAFFT [14] with the FFT-NS-2 option.Sequences flanking the coding region were removed from the alignment.A maximum likelihood tree was created using PhyML [12] with the approximate likelihood ratio test, the HKY85 substitution model and parameters such as for the shape of the gamma distribution (0.258) were estimated by the programme.Substitutions discussed in this analysis were identified and marked in the resulting phylogenetic tree using the MEGA software package [15].
Amino acid changes in the various genes are indicated by name of protein and mutation e.g.PB1 A652V.
tested against an egg-grown A/Brisbane/10/2010 virus (one of the genetically variant viruses) compared to the GMT obtained against egg-grown wildtype A/ California/7/2009 virus.Despite some reduction in HI titres with human post-vaccination sera, there were no clear differences with ferret sera, suggesting that there are no major antigenic differences in these variant viruses at this stage in their evolution and that they still share most of their antigenic properties with the early pandemic influenza A(H1N1) viruses.

Discussion
While the 2009 pandemic has recently been downgraded by the WHO [11], the pandemic influenza A(H1N1) virus still remains the predominant influenza virus in most countries including those in the southern hemisphere that recently experienced their winter influenza season (with the exception of South Africa where influenza B and A(H3N2) viruses have predominated in 2010) [11].To date there has been little change detected in either the genetic or antigenic characteristics of the pandemic H1N1 influenza virus in the nearly 18 months that it has infected humans.clear variant has appeared apart from minor changes occurring in the HA, NA and other viral genes during this time.
Recently however a genetically distinct variant containing several signature amino acid changes in both the HA and NA genes has emerged in Singapore, Australia

Conclusions
A new genetic variant of the pandemic influenza (H1N1) virus has emerged in Singapore, Australia and New Zealand in the second and third quarters of 2010 that does not appear at this stage to represent a significant antigenic change for the virus.However, it may represent the start of more dramatic antigenic drift of the pandemic influenza A(H1N1) viruses that may require a vaccine update sooner than might have been expected, with a new human influenza virus.

Introduction
Campylobacter infections resulting in gastroenteritis are recognised as an emerging problem worldwide.With 47 cases per 100,000 population campylobacteriosis is the most commonly reported gastrointestinal disease in the European Union (data from 2007) [1].Notification rates differ markedly, ranging from zero per 100,000 population in Romania to 95 per 100,000 population in the United Kingdom in 2007 [1].In Germany, the annual number of reported cases rose from 47,937 in 2003 (58 per 100,000 population) to 62,807 in 2009 (79 per 100,000 population) [2].A number of case-control studies identified travelling abroad, eating poultry, pork and sausages, drinking untreated water or unpasteurised milk, barbecuing and having contact with domestic animals as risk factors for infection [3][4][5], but most infections are believed to result from the ingestion of contaminated food [6].The primary source of food contamination is believed to be animal faeces.This is consistent with high Campylobacter carriage rates in poultry, pigs and cattle and the fact that similar Campylobacter genotypes have been identified in farm animals and humans [7][8][9].Contamination of the environment by faeces of domestic and wild animals presents an alternative exposure pathway for human infection, for example, via contamination of drinking and recreational water sources [10].Humans may also be exposed to animal faeces in the environment through other outdoor activities such as playing, camping, walking and picnicking.
Since 2001, when the country's disease reporting system was reorganised, specific notification data on human campylobacteriosis have been available in Germany.Using national case definitions [11], local health authorities verify locally identified notifiable diseases and send case reports electronically via state health departments to the national surveillance unit at the Robert Koch Institute in Berlin [12].For campylobacteriosis, data collected in this system include demographic characteristics, dates of illness, county and -for internationally imported cases -country of infection, diagnostic procedure used (bacterial culture or enzyme-linked immunosorbent assay (ELISA)), and if performed, results of species differentiation, but not the techniques used) and association with outbreaks.
Hesse is one of the 16 German Laender, with a population of 6.In addition, Hessian local health authorities provided information on the municipality of the cases' residence (postal code and/or name of municipality) for all cases included in the study.Population data were provided by the Hesse Statistical Office.The 426 Hessian municipalities were grouped into six categories according to their degree of urbanicity, as defined by the Federal Office of Building and Regional Planning based on population density (urban, intermediate and rural) and accessibility of centres (inner and outer) [13].The spatial distribution of categories of urbanisation in Hesse is shown in Figure 1.In Hesse, the total inner urban area (IUA) has a population density of 1,441 inhabitants per km 2 .Some 35% of the Hessian population live in municipalities of the IUA (Table 1).
Age-specific campylobacteriosis incidences for the six categories of urbanisation and five age groups (under 5 years, 5-14 years, 15-44 years, 45-64 years, more than 64 years) were calculated.In our study, C. coli and C. jejuni together represent over 96% of campylobacteriosis reports with species information.Due to the small numbers of C. coli cases, species-specific campylobacteriosis incidences for C. jejuni and C. coli were calculated only for three categories of urbanisation (urban, intermediate and rural areas).We also calculated 95% confidence intervals (CI) for incidence rates and incidence rate ratios (IRRs) and the attributable fraction among the exposed and the population attributable fraction for living in non-urban areas according to Boice and Monson, using the incidence in urban areas as reference for calculation [14].
As species information was not available for all cases and in order to take into account variation in the frequency of species differentiation according to degree of urbanicity, the number of species-specific cases was estimated using the formula: corrected number of species-specific cases = (number culture confirmed/ number differentiated to species level) x (reported number of species-specific cases).Corrected incidence rates and IRRs were then calculated.For corrected incidences, 95% CI were not calculated, due to the additional uncertainty resulting from incompleteness of species differentiation.Data were analysed with Stata version 10.0.3) was significantly associated with a higher campylobacteriosis incidence.For children aged under five years and those aged 5-14 years, the association between living in the ORAs and higher campylobacteriosis incidence was not statistically significant (children under five years IRR: 1.8, 95% CI: 0.8-3.6;children aged 5-14 years IRR: 1.6, 95% CI: 0.8-3.0)(Figures 1 and 2).

From
We then calculated the attributable risk of living in non-urban areas for children aged 0-14 years In the exposed children, the risk was 46%; the population attributable risk was 25%.
While urban-rural differences were most pronounced in children 14 years of age and younger, they were also seen in the three older age groups.In these age groups there was a tendency towards lower campylobacteriosis incidences for persons living in more rural areas (Figure 2).However, only for those aged 65 years and above living in ORAs did this difference reach statistical significance (IRR: 0.

Discussion
In this analysis, degree of urbanisation was found to be associated with campylobacteriosis in children under 15 years of age.Calculation of the attributable risk indicated that 25% of all reported cases of campylobacteriosis aged under 15 years were associated with living in non-urban areas.These attributable risk calculations reflect the degree to which the true, unknown sources of infection are more abundant in non-urban areas than in urban areas.
Recent studies investigated urban-rural differences in campylobacteriosis incidence in Canada [15], the Netherlands [16], Scotland [17], Denmark [18] and Sweden [19].Four of these studies found higher incidences in rural environments [15,[17][18][19] and one in urban and urbanised environments [16].The authors of the last study suggested the higher incidence of campylobacteriosis in urban and urbanised areas could be related to higher consumption of ready-to-eat foods.Of the two studies that presented age-specific data, one reported the greatest urban-rural differences for children 0-4 years-old [15] and one that urban-rural differences were limited to children 0-14 years of age [18].Authors of both studies suggested that contact with farm animals and the environment were the source of a substantial proportion of sporadic Campylobacter infections.
When interpreting age-specific differences in urbanrural gradients of campylobacteriosis incidence, two main factors need to be considered: age-specific risk factors for infection and immunity acquired during childhood towards local, i.e. rural sources of Campylobacter infection.Few studies reported on agespecific risk factors for campylobacteriosis [20][21][22][23][24][25].However, in case-control studies on risk factors for the disease in infants and young children food exposures explained less than 40% of the infections [20,22,23].
Campylobacter infections in these age groups have been associated with contacts with diarrhoeic pets [20,23,24] and live chickens [24,25], drinking water from a well, lake or river [20,23], riding in a shopping trolley next to meat or poultry [23], visiting or living on a farm [23], ownership of farm animals or visiting farm animals outside the household [22] as well as different food exposures.Among these food exposures were the consumption of fruits and vegetables prepared at home [23], mayonnaise [24], butter [25], porridge [25], undercooked meat [22], products containing raw eggs [22] and grilled meat [20,22].In contrast to many other published case-control studies [4][5], these studies did not find an association between eating chicken and Campylobacter infection.In the light of these findings we believe that environmental exposure accounts for a considerable part of Campylobacter infections in children and that children living in non-urban areas have more opportunities for direct or indirect contact with animals or their excrement.In addition, children living in urban and rural environments may differ in their eating and drinking habits.C. jejuni outbreaks, for example, have been repeatedly related to the consumption of raw milk [26]; children living in non-urban areas may drink raw milk more frequently.
The absence of urban-rural differences in campylobacteriosis incidence in persons aged 15 years and above may be related to differences in behaviour and/ or a higher level of immunity from previous exposures [27].In developing countries, clinical disease due to C. jejuni is common among children younger than two years, but rare among individuals later in life [28].This relative absence of disease is thought to be related to acquired immunity [29].If a higher proportion of inhabitants of Hessian rural areas aged 15 years and above are immune to Campylobacter infection, then the association between living in rural areas and campylobacteriosis may be decreased or even reversed.These questions should be addressed in Campylobacter seroprevalence studies or the inclusion of only non-immune controls in future case-control studies for the identification of risk factors.
In our study, 15% of all campylobacteriosis cases with species information were due to C. coli.Germany is one of the European Union Member States with the highest proportion of cases due to C. coli [30].Within Germany, the proportion of cases due to C. coli differs widely between States and is higher in the former East German or new Laender (in 2006, 14% of all cases (n=7,494) with species information) than in the former West German or old Laender (in 2006, 6% (n=26,205).For whole of Germany, the proportion in 2006 was 8% (n=33,699) [2].The new Länder are more rural, i.e. the population density is lower [31], a greater proportion of the total area is agricultural [32] and a smaller proportion of the total area is inhabited [33].
When analysing species-specific differences in campylobacteriosis incidence in urban, intermediate and rural areas and correcting for differences in frequency of species differentiation, we found a higher incidence for C. coli in non-urban areas.Microbiological data show that the prevalence of different Campylobacter species varies between different potential sources of infection, including animal species, food and water [10,34,35].Poultry has been recognised as the primary reservoir of C. jejuni, while pigs are mostly implicated as reservoirs of C. coli [36][37][38].Differences in food-borne exposures between C. coli and C. jejuni have been shown to exist [22,39] and differences in consumption habits between persons living in urban and rural areas may contribute to the observed difference in species distribution.However, it has been suggested that C. coli may survive in the environment better than C. jejuni [10] and people living in non-urban areas may be exposed more frequently to environmental sources of C. coli.
This study is limited by constraints inherent to all ecological analyses: a sample size limiting detailed subgroup analysis and a limited availability of further data on municipality level (for example, information on water supply, animal density or consultation of health services and diagnostic practices).However, our analysis suggests that differences in risk factors by Campylobacter species, cases' age and degree of urbanicity do exist.

Figure 2
Figure 2Location of clusters of Legionnaires' disease cases South Wales, August-September 2010

Figure 1
Figure 1 Phylogenetic analysis of haemagglutinin sequences from recent pandemic influenza A(H1N1) viruses

Figure 2
Figure 2Full genome maximum likelihood phylogenetic analysis of pandemic influenza A(H1N1) variants from Singapore, Australia and New Zealand and other non-redundant (<80% identity) strains

Figure 1
Figure 1 Geographical distribution of Campylobacter infections, by municipality, showing (A) degree of urbanicity, (B) incidence in children aged 0-14 years and (C) incidence in people aged 15 years and above, Hesse, Germany, July 2005 -June 2006

Table
Microbiological results from clinical samples, Legionnaires' disease outbreak, South Wales, August-

September 2010 Case Cluster Sero-group mAb a Monoclonal antibody subgroup Sequence-based type (SBT)
b By direct-nested PCR typing.c Data obtained were insufficient to distinguish this strain from Case b or Case g. d Unique strain.

Table 1
Population density and percentage population, by degree of urbanicity, Hesse, Germany, July 2005 -June 2006 Methods Data on age, sex and Campylobacter species of all campylobacteriosis cases with disease onset (or if missing, date of diagnosis) from July 2005 to June 2006 were extracted from the state surveillance database.

Table 3
Association between campylobacteriosis cases and degree of urbanicity, by Campylobacter species, Hesse, Germany, July 2005 -June 2006 a Number of cases corrected for incomplete differentiation to species level, which differed by level of urbanicity.
urban, intermediate and rural areas, respectively.However, a higher proportion of isolates from patients in non-urban areas were differentiated to species level: 53.7%, 79.4% and 71.3% of culture-confirmed cases were differentiated to species level in urban, intermediate and rural areas, respectively (Pearson's chi-square test p<0.001).When compared with urban areas, species-specific incidences for C. coli and C. jejuni were higher in rural and intermediate areas.However, when the number of C. coli and C. jejuni cases was corrected for incomplete differentiation to species level, only incidence of C. coli infections differed by degree of urbanicity (Table3).In addition, a relatively higher proportion of C. coli cases lived in non-urban areas: the ratio of the C. coli to C. jejuni cases (corrected) was 0.13 in urban areas, 0.26 in intermediate areas and 0.28 in rural areas.