Pre-test probability and likelihood ratios for clinical findings in canine leishmaniasis

Canine leishmaniasis is a parasitic zoonosis mainly caused by L. infantum; an obligate intracellular protozoan transmitted by haematophagous insects of the genus Phlebotomus, which affects dogs and wild canids. The clinical implications of this disease are highly variable, since infected animals may remain asymptomatic (absence of observable clinical signs) or present a wide spectrum of clinical alterations and degrees of severity, including the death of the animal. Symptoms such as lymphadenomegaly, alopecia, weight loss, keratoconjunctivitis and onychogryphosis are usually the first diagnostic reference available. The objectives of this study are to evaluate the validity (sensitivity, specificity and likelihood ratios) and diagnostic utility (pre-test probability) of the clinical signs commonly associated with canine leishmaniasis based on the prevalence in the area and to explore the combination of symptoms that best predicts the diagnosis of canine leishmaniasis. It is a matched case-control study in the canine population of southern Spain based on the comparison of the findings collected in the clinical history and the results of the LeisSCAN quantitative ELISA. A total of 39 cases and 78 controls were analysed. Approximately 80% of the infected animals showed signs compatible with the disease. The most frequent alterations were cutaneous (64.1%), systemic (51.3%) and oculo-nasal (30.7%). The most useful signs to support this diagnosis were alopecia and epistaxis (LR+ 6.69 and 6.0, respectively) (pre-test leishmaniasis probability is ≥70% for prevalence ≥28% when alopecia or epistaxis is present), followed by lameness (LR+ 5.0). The combinations of signs that showed greater validity were alopecia with hyperkeratosis of the snout and alopecia with onychogryphosis (LR+ > 10). None of the observed signs or their combinations resulted useful to rule out the diagnosis (LR- 0.55 to 1.15). The results found show notable differences in the diagnostic value of the clinical signs, individually and in combination, so we believe that medical decisions should be based on their diagnostic validity (LR+) and the estimation of the pre-test and post-test probability.