We have described one of the largest Q fever outbreaks in Germany which, due to its point-source nature, provided the opportunity to assess many epidemiological features of the disease that can be rarely studied otherwise.
In 1954, more than 500 cases of Q fever were, similar to this outbreak, linked to the abortion of an infected cow at a farmers' market. More recently a large outbreak occurred in Jena (Thuringia) in 2005 with 322 reported cases associated with exposure to a herd of sheep kept on a meadow close to the housing area in which the cases occurred.
The first case control study served to confirm the hypothesis of an association between the outbreak and the farmers' market. The fact that only attendance on the second, but not the first day was strongly associated with illness pointed towards the role of the ewe that had given birth in the early morning hours of May 4, 2005. This strong association and the very high attributable fraction among all cases suggested a point source and justified defining cases notified through the reporting system as outbreak cases if they were clinically compatible with Q fever and gave a history of having visited the farmers' market. The point-source nature of the outbreak permitted calculation of the incubation period of cases which averaged 21 days and ranged from 2 to 48 days with an interquartile range of 16 to 24 days. This is compatible with the literature. An additional interview with the two cases with early onset (2 and 4 days after attending the market on May 4, respectively) could not identify any other source of infection. A short incubation period was recently observed in another Q fever outbreak in which the infectious dose was likely very high.
The second case control study among persons who visited the market on May 4 demonstrated that both close proximity to the ewe and duration of exposure were important risk factors. This finding was confirmed by the cohort study on vendors which showed that those who worked in a stand close to (within 6 meters) the sheep pen were at significantly higher risk of acquiring Q fever. The study failed to show a significant role of the location of the stand in reference to the wind direction, although we must take into account that the wind was likely not always and exactly as reported by the weather station. However, if the wind had been important at all more cases might have been expected to have occurred among vendors situated at a greater distance to the sheep.
According to statutory surveillance system data, the proportion of clinical cases hospitalized was 25%, similar to the proportion of 21% found in persons pooled from the other studies conducted. Several publications report lower proportions than that found in this investigation: 4% (8/191), 5% and 10% (4/39)), and there was at least one study with a much higher proportion (63% (10/16)). It is unlikely that hospitals reported cases with Q fever more frequently than private physicians because the proportion hospitalized among Q fever patients identified through random telephone calls in the Soest population or those in the two cohorts was similar to that of notified cases. Thus reporting bias is an unlikely explanation for the relatively high proportion of cases hospitalized. Alternative explanations include overly cautious referral practices on the part of attending physicians or the presumably high infectious dose of the organism in this outbreak, e.g. in those cases that spent time in the sheep pen.
The estimated attack rate among adults in the four studies varied between 16% and 33%. The estimate of 23% based on the random sample of persons visiting the market on the second day would seem most immune to recall bias, even if this cannot be entirely ruled out. The estimation based on information about persons accompanying the cases may be subject to an overestimation because these individuals presumably had a higher probability of being close to the sheep pen, similar to the cases. On the other hand the estimate from the cohort study on vendors might be an underestimate, since the vendors obviously had a different purpose for being at the market and may have been less interested in having a look at the sheep. Nevertheless, all estimates were independent from each other and considering the various possible biases, they were remarkably similar. In comparison, in a different outbreak in Germany, in which inhabitants of a village were exposed to a large herd of sheep (n = 1000–2000)[5, 7] the attack rate was estimated as 16%. In a similar outbreak in Switzerland several villages were exposed to approximately 900 sheep. In the most severely affected village, the clinical attack rate was 16% (estimated from the data provided). It is remarkable that in the outbreak described here, the infectious potential of one pregnant ewe – upon lambing – was comparable to that of entire herds, albeit in different settings.
Our estimate of the proportion of serologically confirmed cases that became symptomatic (50% (3/6)) is based on a very small sample, but consistent with the international literature. In the above mentioned Swiss outbreak, 46% of serologically positive patients developed clinical disease.
Only approximately half of all symptomatic cases were reported to the statutory surveillance system. Patients who did not seek health care due to mild disease as well as underdiagnosis or underreporting may have contributed to the missing other half. Our estimated 3% attack rate among children is based on a number of successive assumptions and must therefore be interpreted with caution. Nevertheless, sensitivity analysis confirmed that adults had a significantly elevated attack rate compared to children. While it has been suggested that children are at lower risk than adults for developing symptomatic illness[7, 8] few data have been published regarding attack rates of children in comparison to adults.
The estimated C. burnetii seroprevalence in the sheep flocks in the area varied from 8% to 24%. The 25% seroprevalence in the flock of the exhibited animals together with a positive polymerase chain reaction in an afterbirth in June 2003 suggested a recent infection of the flock. Seroprevalence among sheep flocks related to human outbreaks tend to be substantially higher than those in flocks not related to human outbreaks. The median seroprevalence in a number of relevant studies performed in the context of human outbreaks[7, 20, 21], was 40% compared to 1% in sheep flocks not linked to human outbreaks.
This outbreak shows the dramatic consequences of putting a large number of susceptible individuals in close contact to a single infected ewe that (in such a setting) can turn into a super-spreader upon lambing. There is always a cultural component in the interaction between people and animals, and these may contribute to outbreaks or changing patterns of incidence. During the past decades urbanization of rural areas and changes in animal husbandry have occurred, with more recent attempts to put a "deprived" urban population "in touch" with farm animals. Petting zoos, family farm vacations or the display of (farm) animals at a market such as this may lead to new avenues for the transmission of zoonotic infectious agents[20, 22–24]. While not all eventualities can be foreseen, it is important to raise awareness in pet and livestock owners as well as to strengthen recommendations where necessary. This outbreak led to the amendment and extension of existing recommendations which now forbid the display of sheep in the latter third of their pregnancy and require regular testing of animals for C. burnetii in petting zoos, where there is close contact between humans and animals.