Ninety-three cases of legionnaires’ disease (43
confirmed, 12 presumptive, and 38 possible/clinical) were identified in an
outbreak associated with a trade fair in Kapellen, Belgium in November
1999. Five cases died. Epidemiological investigation showed that the
length of time spent at the fair and exposure to particular areas of the
tent were associated with illness. Polymerase chain reaction tests showed
that a whirlpool and a fountain were contaminated with legionella.
An outbreak of legionnaires’ disease occurred among visitors to the
annual trade fair at Kapellen – a small town in the north of Belgium -
which was held from 29 October to 7 November 1999. The investigation began
on 13 November 1999 after a respiratory physician notified presumptive
cases of legionellosis to the health authorities of the province of
The fair employed 830 people (commercial stand employees or as
technical staff) in a large exhibition tent (9000 m2) that
housed 305 stands and attracted 50 000 visitors. The occurrence of cases
of legionnaires’ disease caused considerable alarm, coming shortly after
a major outbreak of legionella pneumonia in the Netherlands at
Bovenkarspel with 181 cases arose and 21 deaths (1,2). This paper
describes the epidemic, the results of the source tracing, and the control
measures that were taken.
Cases were sought among the stand employees (professionals or
volunteers) (800), the technical staff of the exhibition hall (30), and
the visitors at the fair (50 000). Cases were defined as follows. A
possible or a clinical case was defined as any patient who developed a
radiologically proved pneumonia within two weeks after visiting the fair
which was not caused by another microbiological pathogen. A presumptive
case was a patient with pneumonia and a single high serological legionella
antibody titre. A confirmed case was a clinical case who met at least one
of the following laboratory criteria: isolation of legionella, detection
of the legionella antigen in urine, seroconversion – a fourfold rise in
titre - or a combination of a positive polymerase chain reaction (PCR)
test with a positive legionella antibody titre in serum (3).
Sources of information were notifications of legionnaires’ disease by
physicians, a hospital and a laboratory survey throughout Belgium, and a
questionnaire survey of stand employees and technical staff of the fair.
Cases in the Netherlands were identified in collaboration with the
Landelijke Coördinatie van Infectieziekten Nederland (LCI)
Most of the specimens were examined at the microbiology laboratory of
the University Hospital of Edegem. Clinical specimens were respiratory
secretions or tissues, blood, and urine. Environmental samples were of
water and swabs from the surfaces of devices. Confirmation was organised
in collaboration with the reference laboratory for legionella in the
Netherlands (Tilburg). Techniques applied were culture for legionella,
PCR, urine antigen analysis (Biotest®, Binax®),
and enzyme immunosorbent assay (EIA-IgM and –EIA-IgG (Serion®)).
An exploratory case control study was conducted in order to trace the
source of the epidemic. A questionnaire was mailed to 88 reported cases
and to 350 controls who consulted a hospital after their visit to the fair
and in whom legionella pneumonia had been excluded. Potential controls
with fever and a cough in the two weeks after their visit were excluded
from the analysis. Personal risk factors, days of visit, duration of stay,
and proximity to aerosol producing devices were recorded.
A cohort study was conducted among exhibitors and staff of the fair to
evaluate risk factors such as exposure time and the precise location of
activities. Eight hundred and forty-two stand employees and 3O individuals
of the technical staff who could be identified were contacted by mail and
asked to fill out a questionnaire with personal characteristics, symptoms
and signs, exposure time and working place at the fair. Blood and urine
specimens were collected at the hospital. A map of the exhibition hall,
showing the precise location of all the exhibitors, was divided into six
areas (A to F). A risk rate was calculated for each area.
A list of the stands that demonstrated aerosol-producing devices was
drawn up. Some devices used during the fair were confiscated in the
context of a police inquiry. Samples of water and swabs from surfaces of
these devices were analysed by PCR and culture for legionella.
Patient data were checked and validated by the surveillance team and
analysed using Epi Info version 6.04. Descriptive and associative
parameters were calculated. Logistic regression was performed with LogXact
version 2.1 (Cytel Software Corporation) and with SAS 6.12 using Wald chi
square, likelihood ratio, and odds ratio.
Ninety-three people who met the case definition were identified. Among
these 93 patients 43 could be considered as confirmed cases, 12
presumptive, and 38 possible/clinical cases. Eighty-six of the patients
were visitors and seven were exhibition staff.
The mean incubation period of the disease was nine days (range 3 to
14). Common symptoms were malaise, headache, fever, chills, chest pain,
shortness of breath, and non-productive cough.
Eighty eight patients were admitted to hospital and five died. Four of
the deaths occurred early in the epidemic; one patient died a month later.
The average age of the deceased patients was 64 years (range 52 to 75).
The average age of all the patients was 49 years (range 2 to 84). Both
sexes were equally affected. Twenty-five per cent of cases lived in
Kapellen, 8% in Antwerp, and 5% in the Netherlands. Patients were admitted
to 12 different hospitals, 8% were not admitted to hospital. Fifty-four
(58%) patients were seen in the same hospital.
The outbreak curve is shown in the figure. The index patient reported
ill on 4 November, six days after his visit to the fair. He was notified
as a case on 13 November and a general alert was given the same day. The
attack rate was 8.5 per 1000 for exhibitors and 1.7 per 1000 for visitors.
Twenty-eight (65%) patients had a positive legionella urinary antigen
EIA ( Legionella pneumophila serogroup 1,2,3,4,6,10 Biotest ®), 25
(60%) of the 43 confirmed cases had a positive urine antigen test EIA (
serogroup 1, Binax®) , 22 (51%) was IgM ( SerionÒ
) positive, 22 (51%) had a seroconversion, 20 (46%) were positive with PCR
on respiratory secretions, 21 (23%) had a positive IgG(SerionÒ
) and 10 (23%) had a positive legionella culture.
A total of 64 cases and 162 controls responded to the questionnaire; 48
controls were excluded from further analysis, for reasons of cough and
fever. Cases were significantly older than controls average 52 vs. 34
years (p < 0.001). The sex ratio (M:F) for cases was 1.2:1 and for
controls 0.45:1(p=0.13). Cases were more likely than controls to smoke
(odds ratio (OR) 2.2, p=0.02) but there was no significant difference in
previous respiratory problems between both groups. Some of the aerosol
producing stands were potential risk factors (whirlpool, rainproof roof,
and a fountain) but no significant association was found with the duration
of direct exposure to each of them. There was a significant association
with a specific visit to a steam ironing stand (OR 2.9 (95% confidence
interval (CI) 1.2-6.9) in the univariate analysis and OR 3.5 (95% CI
1.3-9.8) in the logistic regression).
Two hundred and thirty-four (28%) out of 842 identified exhibitors and
staff took part in the cohort study, 180 of whom provided blood and urine
specimens. Twenty-one (9%) fulfilled the case definition (atypical
pneumonia plus a positive urine antigen test or a positive serological
test). The average age of the participants was 38 years and males
outnumbered females by 1.3:1. For 210 out of the 234 participants a
specific area in the tent could be identified. The prevalence of infection
in those exposed for one day or less was 9%; among subjects exposed for
longer it was 13% (table 1). A clustering effect was shown for individuals
who had worked in the central parts of the hall (table 2). Age, smoking,
area and exposure time were not significantly linked with the outcome.
Table 1. Cohort study (univariate analysis), case distribution by
exposure time, Kapellen, 1999