Introduction
Q fever is a zoonotic infection caused by the rickettsial organism Coxiella
burnetii. In the United Kingdom it is most commonly carried, often
asymptomatically, in sheep, cattle and goats, and is transmitted to humans
by inhalation of aerosols. High concentrations of the organism are found
in the placenta/placental fluids. Coxiellae can remain viable for months
in the environment. The disease occurs most frequently in humans exposed
to farm animals or in areas where animal products are handled (1]. Retrospective
serological studies have shown evidence of high rates of past infection
in farm workers, which suggests that many cases are often not identified
at the time of illness [2].
The major clinical manifestations of Q fever are respiratory, cardiac and
hepatic, although symptoms are often non-specific. C. burnetii is thought
to account for 1% (700 cases) of community-acquired pneumonia in the UK
each year, and although more serious complications such as endocarditis
are rare, they do represent a significant burden of disease [3].
Although outbreaks have frequently been reported worldwide, the causes
have often not been identified [4] and we have only been able to find one
previous case-control study in the literature determining risk factors
in sporadic cases [5]. The highest incidence of cases in England is consistently
reported from the southwest and in an epidemiological review this rural
region reported one third of all cases in England and Wales [3]. Northern
Ireland reports even higher rates of Q fever per 100 000 population, with
between 21 and 75 cases per year since 1990 [6].
Methods
We collaborated with laboratories in southwest England and Northern Ireland
to identify cases of Q fever for a matched case-control study to determine
risk factors for sporadic infection. A required sample size of 43 cases
was estimated using Epi Info. This size was based on a case-control
ratio of 1:3, with 95% confidence and 80% power to detect an OR of
3.
Cases in patients resident in southwest England and Northern Ireland
aged 16 years and over between 1 January 2002 and 31 December 2004 were
identified by local laboratories and confirmed as acute by the Health
Protection Agency Regional Laboratory in Bristol on the basis of a history
of acute illness and the detection of specific immunoglobulins to C.
burnetii phase 2 antigens in human sera (Coxiella burneti-Spot IF,
bioMerieux? sa, France, using sheep anti-human IgG and IgM conjugates
supplied by The Binding Site Ltd,UK), to detect either a fourfold rise
in IgM and/or IgG on paired sera, or IgM and IgG titres >= 640.
Initially, three controls of the nearest age, same sex and registered
with the same general practice were selected for each case (general practices
in the UK cover an average population of 6000 people in the same geographical
area). In 2003, the study duration was extended from two to three years
and the number of controls per case increased to five, because case numbers
had been lower than expected and there had been poor response rates,
especially from controls.
Postal questionnaires, including questions about contact with animals,
consumption of pasteurised/unpasteurised milk, and leisure and work activities
within the four weeks before illness (past four weeks for controls),
were sent to cases and controls. Non-responders were sent one reminder
after four weeks. Data were entered onto a Microsoft Access database.
Where responses were not received and there was evidence of individuals
only responding where the answer was ‘yes’, a ‘no’ response
was entered for data that were missing. ‘Don’t know’ responses
were excluded from the analysis. Single variable conditional logistic
regression was carried out using Stata (v8.2). Variables with P<0.2
in the single variable analysis were then included in a multivariable
conditional logistic regression analysis. The study received approval
from the appropriate local ethics committees.
Results
Questionnaires were returned by 39/50 (78%) of the cases identified with
acute Q fever and 90/180 (50%) of the controls. After excluding records
without case or control matches, data from 34 cases and 77 controls
were available for analysis, a ratio of 1:2.3. The age range for both
cases and controls was 20-73 years (mean 47 and 48 years respectively).
Twenty five (73.5%) of the case patients were men, and 9 (26.5%) were
women. Over the three year study period, the majority of cases (63.6%)
were reported between the months of March and June and were from a
rural location (29/34 cases lived on a farm or within 3 miles of farmland).
There was a clustering of four cases within a 10 mile (16 km) radius
in one rural area. Further investigation did not identify any specific
exposure common to these cases.
All cases reported sweating and/or a fever, 28 (82.4%) had a headache,
27 (79.4%) had respiratory symptoms (shortness of breath and/or cough),
27 (79.4%) experienced weight loss, 23 (67.7%) had joint pain and 20
(58.8%) had chest pain. Three (8.8%) had jaundice and 8 (28.6%) patients
experienced other symptoms including vomiting, blurred vision, dizziness,
extreme thirst, ‘sore kidneys’ and increased sensitivity
of senses (taste and smell). The median duration of illness was 21 days.
Twelve patients (35.2%) said they were still unwell at the time of completing
the questionnaire.
In the single variable analysis, occupational exposure to animals or
animal products was the only risk factor associated with cases at the
5% level (P=0.05, OR 3.4, 95%CI 1.0 to 11.8) [TABLE 1]. Long term illness
appeared to be significantly protective (P=0.03, OR 0.3, CI 0.1 to 0.9).
In the multivariable analysis, long term illness remained significantly
protective, and smoking emerged as a possible risk factor [TABLE 2].
Although the P value increased from 0.05 to 0.06 when added to the multivariable
model, the strength of association between occupational exposure and
illness remained high (OR 3.6, 95% CI 0.9 to 14.8).
Discussion
Occupational exposure has been documented as a risk for Q fever in case
series and outbreaks since the organism was first discovered in 1937
[7]. As far as we are aware, this is the first case-control study to
identify it as the most likely route of exposure in sporadic cases.
The temporal distribution of Q fever cases between March and June is
similar to that seen in other studies in the UK and Spain, consistent
with increased exposure to C. burnetii after animal births in spring
[3, 8].
As expected, the majority of cases reported non-specific symptoms such
as fever and sweating. However, cough and shortness of breath were consistent
with respiratory tract involvement, the most common manifestation of
Q fever in the UK. The low proportion of cases with jaundice supports
the observation that hepatitis is not a common presentation in the UK
[3], although patients with mild or granulomatous hepatitis would not
necessarily have been jaundiced. Other countries have reported a higher
proportion of cases with hepatitis, up to 40% of acute cases in one study
in France [9].
The incidence of Q fever in the study regions fell almost as soon as
the study started. It is possible that this was due to the effects of
foot and mouth disease that occurred in England in 2001, just before
the study commenced. Also, a low response rate, especially among controls,
resulted in some variables being dropped from the analysis, and misclassification
bias may have been introduced into the analysis by assigning missing
values to ‘no’. It is also possible that other risk factors
were not included in the study, such as exposure to rats, which have
been identified as an important reservoir for C. burnetii in
the UK [10].
The apparent protective effect of long term illness was surprising, but
could reflect lower outdoor exposure to rural environments in people
with long term illness. Apart from occupational exposure and a possible
link with smoking, other risk factors studied did not reach statistical
significance at the 5% level. Occupational exposure could explain at
most a quarter of cases, but we did not expect to have sufficient statistical
power to identify risk factors below an odds ratio of 3. Further studies
to elucidate risk factors for sporadic Q fever should plan for a larger
sample size. In the meantime, prevention and control measures should
be directed at reducing the risk of occupational exposure [11].
* Members of the South West Q Fever Project Group: David
Dance (chairman),
David Carrington, John Hartley, Simon Hill, Graham Lloyd, Conall McCaughey,
Marina Morgan, Isabel Oliver, Hilary Orr, Mike Smith, Robert Smith, Brian
Smyth, James Stuart
|