Objective: To determine sensitivity and specificity of cytologic examination used in a clinical setting.
Design: Retrospective study.
Animals: 216 dogs, 44 cats, 4 horses, 2 ferrets, 1 llama, 1 rat, and 1 mouse.
Procedure: Records were reviewed of cases in which a cytologic diagnosis was followed by a surgical biopsy or postmortem examination within 3 days with subsequent histopathologic diagnosis. Diagnoses were compared for agreement at various levels, including complete agreement, partial agreement, no agreement, or no comparison possible because of insufficient or incorrect cytologic specimen. Levels of agreement were compared for different categories of lesions, including neoplastic, inflammatory, dysplastic-hyperplastic-other, and normal tissue. Additionally, levels of agreement for neoplastic lesions were categorized with regard to cell type, degree of malignancy, and location. Sensitivity and specificity of cytologic examination were calculated.
Results: At the level of general agreement (complete and partial agreement), the sensitivity of cytologic examination ranged from 33.3 to 66.1%, depending on the location of the lesion. Cytologic examination was most accurate when used to diagnose cutaneous and subcutaneous lesions and least accurate for diagnosis of liver lesions. Cytologic examination was most effective in diagnosis of neoplastic disease and least effective in diagnosis of dysplastic or hyperplastic conditions.
Conclusions and clinical relevance: Cytologic examination is a valuable diagnostic tool, although our results indicated lower accuracy than previously reported. False-negative results (missing a diagnosis) were far more common than false-positive results (categorizing a healthy animal as diseased): therefore, if the clinical index of suspicion is high, cytologic examination should be repeated or another technique should be selected to rule out the suspected condition.