No indication of Coxiella burnetii infection in Norwegian farmed ruminants

Background Infection with Coxiella burnetii, the cause of Q-fever, has never been detected in Norwegian animals. Recognising the increasing prevalence of the infection in neighbouring countries, the aim of the study was to perform a survey of Norwegian farmed ruminants for the prevalence of C. burnetii infection. Results Milk and blood samples from more than 3450 Norwegian dairy cattle herds, 55 beef cattle herds, 348 dairy goat herds and 118 sheep flocks were serologically examined for antibodies against C. burnetii. All samples were negative for antibodies against C. burnetii. The estimated prevalences of infected herds were 0 (95% confidence interval: 0% - 0.12%), 0 (0% - 12%), 0 (0% - 1.2%) and 0 (0% - 10%) for dairy cattle herds, beef cattle herds, goat herds and sheep flocks, respectively. Conclusions The study indicates that the prevalence of C. burnetii infection in farmed Norwegian ruminants is low, and it cannot be excluded that Norway is free of the infection. It would be beneficial if Norway was able to maintain the current situation. Therefore, preventive measures should be continued.

. Input to Freecalc for estimation of the upper limit of the confidence interval of the prevalence estimate and the resulting upper limit of confidence interval.

Population size and number of examined herds
The population size was the number of herds with commercial production as registered in the Register of Production subsidies (Norwegian Agricultural Authority, Oslo). For the cattle population, only herds located in the counties included in the studies were considered when calculating the population size.
The number of examined herds was the number of herds tested in the current studies.
The number of positive herds was the number of herds with positive serological reaction, which was zero for all populations.

Test specificity
The test specificity was not known; hence, a conservative approach was used and the specificity was set to 1.

Test sensitivity for individual samples
According to the manufacturer there were 37 test positive of 37 positive animals examined [1]. The probability of obtaining 37 of 37 positive for different sensitivities is illustrated in Figure A1. The sensitivity was estimated to 0.983 which was the weighted mean for obtaining 37 positive of 37 trials. where Se Ind = Test sensitivity of individual samples as estimated above. P(Lactating) = the probability that the animal is lactating at the time of sampling. For cows the average lactating period is 10 months so that on average 83% of the cows are delivering milk at a time [2]. Likewise, 92% of the goats within a herd would be lactating at the time of sampling [3]. P(Detected) = the probability that a positive sample is not diluted below the detection level.
The test's ability to detect a single positive animal in a bulk milk sample was estimated by examining eight positive individual milk samples that were serially diluted in serologically negative milk. The end-point titres were 1:10 for one sample, 1:20 for two samples, 1:40 for two samples, 1:80 for one sample, 1:320 for one sample, and 1:640 for one sample.
The population structure of dairy cattle and goat herds and the estimated number of antibody positive animals assuming that the within herd prevalence was 5% is given in Table  A2. Based on this, 1 positive animal would not be diluted below detection level for herds up to 10 animals, it would be diluted below detection level in 1 of 8 in herds from 11 to 20 animals. In herds larger than 20 animals and assuming 5% antibody producing animals, the probability of detecting the animals was calculated to 0.75 (0.5 * 7 / 8 + 0.5 * 5 / 8). With the Norwegian population structure, the mean P(Detected) for a herd was estimated by mean P(Detected) = ∑ P(Detected i )* (PopProp i ) where i= Herd size group. The mean P(Detected) was estimated to 0.81 and 0.63 for dairy cattle and goat herds, respectively.
The Se H was estimated to 0.66 and 0.57 for dairy cattle and goat herds, respectively.

Herd sensitivity for beef cattle herds and sheep flocks
For beef cattle herds and sheep flocks, the herd level test sensitivity was set to 1the probability of detecting 0 positive reactors, assuming a hypergeometric distribution. The estimations were performed using the data analysis module in Freecalc (Survey toolbox, © Angus Cameron, Australia, 1998). As input to Freecalc were used the population size, number of examined animals, number of antibody positive animals, specificity, sensitivity (Table A3) and within-herd prevalence of 5%. The mean herd size among the eligible herds was used as the population size of the cattle herds, i.e. 22 adult cattle. Likewise, the mean flock size among the ram circle members was used as the population size of the sheep flocks, i.e. 110 adult sheep.
From each cattle herd 10 individual blood samples were tested, and 5 individual blood samples tested from each sheep.
The herd level sensitivity for beef cattle herds was estimated to 0.45 and the flock level sensitivity for sheep flocks was estimated to 0.25.