• Open Access

A serosurvey of Coxiella burnetii infection in children and young adults in South West Queensland


Correspondence to:
Dr Neil Parker, Darling Downs Public Health Unit, PO Box 1775, Toowoomba Queensland 4350. Fax (07) 4639 4772; e-mail: neil_parker@health.qld.gov.au


Objective: To describe the seroepidemiology of Coxiella burnetii, the causative agent of Q fever, in those under 25 years of age in South West Queensland.

Methods: A convenience sample of residual sera from a diagnostic laboratory was tested for C. burnetii antibodies by immunofluorescence at 1:10 dilution. Prevalence and annual incidence were calculated from the results.

Results: Twenty-nine of 447 (6.5%, 95% CI 4.5%-9.2%) samples were positive. Seropositivity increased from 2.5% in those <15 (95% CI 1.0%-5.5%) to 11.0% in those 15-24 years old (95% CI 7.4%-16.0%). The estimated annual incidence for the latter age group was 7.7 per 1,000.

Conclusions: Q fever is a relatively common infection in South West Queensland, even in those aged <15 years for whom the vaccine is not recommended.

Implications: Vaccination programs, such as the federally funded National Q fever Management Program, are needed in this and similar high risk rural areas.

Q fever is a worldwide zoonosis causing a wide spectrum of disease in adults, and is increasingly recognised as an issue for children.1,2 The vaccine licensed in Australia is not recommended for children under 15 years of age and requires prevaccination screening, including history, serology and skin testing.3 Notifications of disease underestimate incidence, and this underestimate is probably higher in rural areas where doctors are few and rural workers travel considerable distances for medical services.1 The underestimate may be higher in Q fever as serology is often negative at first presentation, and rural people may not return for repeat serology when their health improves.1 A serosurvey provides an alternative means of assessing the need for population based vaccination programs.

The Australian Federal Government funded a National Q Fever Management Program from 2001 to 2006.4 Phase I funded medical consultations and laboratory tests, and its main beneficiaries were abattoir workers. Phase II included graziers. It began several months after Phase I, and only provided free vaccines and skin test solution.


Sera collected by a private laboratory for clinical purposes were tested using a pre-immunisation serological screen. The study area was defined by postcodes >4416 and <4498, which includes all of the Australian Bureau of Statistics (ABS) South West Queensland Statistical Division and has the highest rate of Q fever in Australia at 107 per 100,000 population in 2006.4 Some adjacent areas in Central West Queensland and the Darling Downs fall into this postcode range.

Sullivan Nicolaides Pathology was the only private provider for most of the study area. Sera from ambulatory patients were tested for IgG antibodies to phase 2 Coxiella burnetii by the indirect immunofluorescent-antibody (IFA) assay. Sera from patients known to be febrile were excluded. Sera were diluted 1:10 in 3% chick yolk sac in phosphate buffered saline and spotted onto slides coated with C.burnetii phase 2 organisms prepared by the Institute of Veterinary and Medical Research (Adelaide, South Australia). Results were classified as positive, negative, equivocal or weakly positive. In final data analysis equivocal results were combined with negative results, and weakly positive results were combined with positive results.

Results were de-identified before analysis, with age, sex, postcode and Q fever serology result available to researchers. With no known studies in similar populations a sample size could not be calculated. The plan was to obtain 100 samples from each five year cohort between five and 25. As significant time elapsed without obtaining 100 samples in younger age groups, children <5 were also included.

The study received approval from Toowoomba Health Service District Human Research Ethics Committee. Chi was estimated squared values were performed with EpiInfo version 6 Statcalc, November 1993. Agresti–Coull5 95% confidence intervals (95% CI) were calculated with the utility for calculating binomial confidence limits for a proportion (AusVet Animal Health Services site http://www.ausvet.com.au/epitools/content.php?page=CIProportion accessed 21/1/2009).

The incidence for the 15-24 age group using the following formula, where page(15-24) and page(0-14) are the average prevalences for ages 15-24 and 0-14 respectively, and age difference is average age for the 15-24 group minus the average age for the 0-14 group.6



A total of 466 sera were collected from 19 January 2001 to 4 November 2002. Nineteen samples were excluded, leaving 447 for analysis. Three people had dual collections, with the same result both times (two negative, one positive). Three negative samples had incorrect collection dates recorded. Thirteen samples (all negative) collected within the study area had postcodes outside this area.

Of the 447 samples, 25 were positive, four weakly positive, four equivocal and 414 negative (Table 1).

Table 1.  Seropositivity for Q fever in South West Queensland by age, sex and health district.
 Total samplesPositiveWeakly PositiveTotal PositiveRate95% CI
Age (years)
  0 to <5520111.9%0.6%-11.1%
  5 to <10782022.6%0.2%-9.4%
  10 to <151073032.8%0.6%-8.3%
  15 to <201017298.9%4.6%-16.3%
  20 to <251091311412.8%7.7%-20.5%
  0 to <152375162.5%1.0%-5.5%
  15 to <252102032311.0%7.4%-16.0%
Sex and age ≥15 years
Sex and age <15 years
District and age ≥15 years

Trend with age

The prevalence for the total sample was 6.5% (29 of 447, 95% CI 4.5%-9.2%). Seropositivity in the originally planned age group of 5-24 years was 7.1% (28 of 395, 95% CI 4.9%-10.1%). As expected the rate increased with age, from 2.5% in those <15 to 11.0% in those 15-24 years old (see Table 1 for confidence intervals). The trend by five-year cohort was highly significant with a chi squared for linear trend of 12.24 (p<0.0005).

Association with sex

In those >14 years, 15.6% of males and 8.9% of females were positive. This difference is not statistically significant (p=0.15, 95% CI 8.5%-26.6% vs 5.2%-14.8%). In younger children there were too few positive tests for meaningful comparison.

Comparisons between districts

South-west Queensland ABS Statistical Division (SD) had two health service districts, Roma (300 samples) with six Statistical Local Areas (SLAs) and Charleville (128 samples) with the four most westerly SLAs. Charleville district had a higher rate (7.8% vs 6.0%) that was more pronounced in those 15-24 (15.8% vs 9.3%), but the difference was not statistically significant. There were 19 samples from the study's postcode range outside this SD.

There were 9,793 people under 25 in the SD in 2001/02, so the 428 samples from this area represented 4.4% of the population.


The calculated average annual incidence for those aged 15-24 was 7.7 per 1,000. Average incidence should be used with caution as outbreaks are a feature of Q fever.1,6


Limitations of this study and serosurveys

Laboratory tests

The most important uncertainties are a test's sensitivity and specificity for past Q fever infection, and these vary with the cut-off titre used.7 Specificity is reduced by cross reactions with antibodies induced by other bacteria such as Legionella and Bartonella species.1 Sensitivity is reduced when higher titres are used. This study employed a titre of 1:10, which is routinely used as a prevaccination screen. This titre prioritises sensitivity over specificity, as severe side-effects may occur following vaccination of immune people. 3 Serosurveys have used titres ranging from 1:87 to 1:64.8 Different cut-offs may be justified by different background rates of seropositivity,7 but make comparisons between serosurveys difficult.

Sensitivity decreases over time. The IFA test remains positive for 10 years or more, so titre decline is unlikely to be an important factor in this study.9

The sample population

The survey used a convenience rather than a random sample. To reduce bias towards inclusion of persons infected with Q fever, those with a known febrile illness were excluded. A bias against including positives is possible as children from remote properties are more likely to have Q fever, but have less access to laboratory testing. More samples were collected from females than males (ratio 1.3:1), which could lead to an underestimate.

The influence of vaccination

Vaccination may increase seropositivity, but an IgG response is usually seen only in those who already have low positive titres (IFA 1:10 or 1:20).10 The observed seroprevalence in Roma district was not higher than Charleville, despite the fact Roma had a large Q fever vaccination campaign in 2000, while one in Charleville was cancelled because of a vaccine shortage.

While the vaccination rate in the study area is unknown, the federally funded vaccination program occurred predominantly after the survey.3,4 In Queensland Phase I began in November 2001, and Phase II in July 2002. Phase I was predominately for meatworkers, of whom there were few in South West Queensland. Phase II extended the program to farmers, but our last sample from a person >14 years was taken on 28 March 2002, before Phase II began.

Infection and disease

About 50% of Q fever infections are asymptomatic, but this rate is higher in children.1 Serosurveys cannot provide an estimate of disease burden unless a reliable estimate of the age-specific rates of symptomatic disease in infected persons is known. These rates may vary from place to place, and also with the type of contact with animals, as different activities result in different ranges of inhaled numbers of organisms.

Australian serosurveys and estimates of annual risk

All the published Australian Q fever serosurveys we found were in occupational risk groups with one exception; a 1953 serosurvey in Northern Territory Aboriginal people.11 In 1998, Casolin listed eight published seroprevalence studies, mainly in abattoirs.12 One listed paper briefly mentions that only one in 577 opportunistically tested sera from New South Wales was positive.13 No demographic details were given. Casolin's study, and four published since then are listed in Table 2. These are all pre-vaccination studies, so are in high risk populations excluding children <15. The seroprevalence in these studies varies between 4.2% and 21.2%, comparable to the rates in this study.

Table 2.  Comparison of skin test positive and blood test positive results in previous Australian serosurveys.
AuthorYearType of blood TestTotal testedPositive serologyPositive skin testSkin test positive in those seropositiveSeropositive in those in skin test positive
  1. Note: ns = not stated

  2. All the above surveys were in occupationally at-risk groups, and did not include children under 15 years of age.

Hutson142000EIA / CF97920821.2%34535.2%16579.3%16547.8%

The annual notified incidence of Q fever in Australia peaked at 0.049 per 1,000 in 1993.17 Estimates of Q fever disease in Australian abattoir workers vary from 3.318 to 10 per 1,000 per year, the latter being the average rate in one Queensland abattoir between 1968 and 1977, excluding one ‘epidemic’ year.19 An economic evaluation of Q fever vaccination used an estimate 30 infections per 1,000 per year.20 A similar rate (45.0 per 1,000) was found in Victorian abattoir workers.21 The seropositivity in this study was 11.7% (1076 of 9196), similar to our finding of 11.0% in those aged 15-24.

Overseas studies in children

There are numerous published serosurveys from other countries, some of which include children.1 Eight of 1,200 hospitalised children in Greece had sera positive for Q fever.22 A Spanish survey found 12% of those <15 years were positive.23 Comparisons between surveys are difficult for the reasons outlined above.

Conclusion and recommendations

The National Q fever Management Program began with abattoir workers (Phase I) and only after seven months extended to graziers (Phase II). This study shows that unselected young people in South West Queensland have seropositive rates approaching those of abattoir workers, so calling into question the prioritisation of abattoir workers over rural workers.

High seropositive rates in young people in South West Queensland demonstrate the need to focus vaccination strategies on this population. As the risk in these younger people approaches the risk in abattoir workers, the life-time risk for the population as a whole will be greater. We therefore recommend that all children in South West Queensland be vaccinated at age 15. Rural areas with a similar risk could be easily determined through an analysis of routine notification data by SLA. Further research is needed before recommendations for younger children are formulated.2 In the meantime, clinicians should make individual decisions for children <15 who engage in high-risk farm activities, particularly butchering and birthing.