Keratinized lesions have been a conceivable false-negative (FN) factor in oral exfoliative cytology (OEC); however, other factors are poorly analyzed. In this study, we aimed to identify the factors influencing the accuracy of OEC and FNs focusing on the lesion characteristics, patient background, and surgeon factors in oral potentially malignant disorders (OPMD).

We retrospectively studied 44 patients who underwent both OEC and histopathological diagnosis. Sensitivity, specificity, FN rate, false-positive (FP) rate, and prevalence of both methods were compared. Similarly, accuracy indices were compared among clinical diagnosis groups (leukoplakia vs. other diagnosis). The association between patient and surgeon-related factors influencing FN OEC results were investigated using Fisher's exact test and a multiple logistic regression analysis.

Overall, the sensitivity; specificity; and FN, FP, and prevalence rates of OEC were 31.8%, 82.1%, and 68.8%, 17.9%, and 36.4%, respectively. Leukoplakia was significantly more common in clinical diagnosis (P = 0.007) with sensitivity, specificity, and FN rates of 20.0%, 95.2%, and 80.0%, respectively. Contrarily, non-keratinized lesions had sensitivity, specificity, and FN of 83.3%, 85.7%, and 16.7%, respectively. In the prevalent group, leukoplakia and anucleate squamous cells were significantly associated with FN cases (P = 0.013, P = 0.050). On multivariate analysis in OEC negative patients, age ≤64 (P = 0.050) and location on the tongue (P = 0.047) was independently associated with FNs.

FN of OEC was conceivable to be due to poor deep-seated cell sampling, which was associated with leukoplakia, age, and location. Therefore, these factors may be considered in the evaluation of OEC results.