Background: The clinical features that may be associated with malignancy in
parotid tumours are well known. Classical teaching dictates caution in the
ir presence hut this raises a false alarm in many cases. Formal studies loo
king at these features are few. The aim of this article was to study quanti
tatively those features that provide a better prediction of malignancy. Met
hods: Clinical records of 186 consecutive patients treated for parotid tumo
urs over a 12-year period were reviewed. Presence of suspicious clinical fe
atures and the final histology in each patient were noted. Results: The ove
rall pick-up rate for malignancy, based on clinical features alone, was aro
und 30%. Palpable cervical lymph nodes, facial nerve palsy, deep fixation a
nd rapid enlargement of the tumour were significant parameters indicative o
f malignancy (p= 0.000 for all 4 parameters, chi-squared test). The risk of
malignancy increased when multiple parameters were present together at the
same time. Conclusion: Clinical features remained the most important singl
e modality identifying malignancy in patients with parotid tumours. The log
istic regression model allowed for simple clinical prediction of malignancy
with improved sensitivity and much better specificity.