[Logistic model in diagnosis of lung cancer]

Specificity of the diagnosis of lung cancer (l.c.) based upon clinical and radiological criteria was investigated in patients who died in Hospital of Lung Diseases in Lódź between in 1985 and 1989. Autopsy revealed diagnostic errors in 37% cases. Twenty most common disease features were taken into consideration. Logistic analysis helped to create clinical-mathematical model which allows for interpretation of disease features in two distinct categories: l.c. (non-neoplasmatic lung disease). Eight features proved to be significant (changes in RBC, weight loss, chest pain, increased sputum production, haemoptisis, hilar tumor, pleural effusion, peripheral round shadow on chest X-ray). Diagnostic usefulness of the model was checked in prospective study a group of patients who died from properly diagnosed l.c. (499 cases). Our model allowed for proper diagnosis of neoplasmatic cause of the disease with higher probability than morphological methods (including both intravital and autopsy examination).