Signs and symptoms do not predict, but may help rule out acute Q fever in favour of other respiratory tract infections, and reduce antibiotics overuse in primary care

Background: From early 2009, the Dutch region of South Limburg experienced a massive outbreak of Q fever, overlapping with the influenza A(H1N1)pdm09 pandemic during the second half of the year and affecting approximately 2.9% of a 300,000 population. Acute Q fever shares clinical features with other respiratory conditions. Most symptomatic acute infections are characterized by mild symptoms, or an isolated febrile syndrome. Pneumonia was present in a majority of hospitalized patients during the Dutch 2007-2010 Q fever epidemic. Early empiric doxycycline, guided by signs and symptoms and patient history, should not be delayed awaiting laboratory confirmation, as it may shorten disease and prevent progression to focalized persistent Q fever. We assessed signs' and symptoms' association with acute Q fever to guide early empiric treatment in primary care patients.

Methods: In response to the outbreak, regional primary care physicians and hospital-based medical specialists tested a total of 1218 subjects for Q fever. Testing activity was bimodal, a first "wave" lasting from March to December 2009, followed by a second "wave" which lasted into 2010 and coincided with peak pandemic influenza activity. We approached all 253 notified acute Q fever cases and a random sample of 457 Q fever negative individuals for signs and symptoms of disease. Using data from 140/229(61.1%) Q fever positive and 194/391(49.6%) Q fever negative respondents from wave 1, we built symptom-based models predictive of Q-fever outcome, validated against subsets of data from wave 1 and wave 2.

Results: Our models had poor to moderate AUC scores (0.68 to 0.72%), with low positive (4.6-8.3%), but high negative predictive values (91.7-99.5%). Male sex, fever, and pneumonia were strong positive predictors, while cough was a strong negative predictor of acute Q fever in these models.

Conclusion: Whereas signs and symptoms of disease do not appear to predict acute Q fever, they may help rule it out in favour of other respiratory conditions, prompting a delayed or non-prescribing approach instead of early empiric doxycycline in primary care patients with non-severe presentations. Signs and symptoms thus may help reduce the overuse of antibiotics in primary care during and following outbreaks of Q fever.

Keywords: Antibiotics; Coxiella burnetii; Prediction; Primary care; Q fever; Respiratory tract infection.