[Association of signs and symptoms with cerebral neoplasms in infancy].

OBJECTIVE: To validate groups of cerebral neoplasia symptoms/signs capable of performing diagnosis of children with neoplasia, with a higher reliability. METHODS: Cross-sectional survey, clinical-epidemiological, and double-blind survey for cerebral neoplasia diagnosis based on data from encephalic and cranial computered tomography scanning (ECCT) and/or magnetic resonance imaging (MRI) services in the state of Bahia, first semester of 1995. A total of 753 children from 1 to 15 years of age, with exams and information regarding symptoms/signs were included. These were classified in 4 groups (1 to 4), according to location of the cerebral tumor. The ECCT and/or MRI were the gold standard and independent evaluation for each group. Basic calculation: sensibility, specificity, negative and positive predictive values (PPV), accuracy, pretest prevalence, and confidence interval (CI) at 95%. Complementary: Bias, odd ratio (OR), prevalence ratio (PR) and Kappa (k) with qualification of validation. RESULTS: 34 cases of tumors were detected, with the loss of one; prevalence of 4,4%. Evaluation results >/= 90% for group 2 symptoms/signs (ataxia and/or cranial nerves palsy), except PPV = 82%; elevated k, and "optimal" evaluation. It was the best tumor predictive group. The prevalence CI was significant, with small amplitudes. Group 3 (endocrine/visual dysfunction), was optimal for k, but with a lower PPV. CONCLUSION: Symptoms/signs of group 2 are important for cerebral tumor diagnosis, followed by group 3. A complete anamneses and neurological examination of children with cephalea and/or seizures are important for screening and reference to bioimaging services. High number of normal exams increases assistance costs, becoming a negative factor for children who really need the exams.