|
Abstract:
|
Node positivity is a major prognostic factor among oral cancer patients. Hence, knowledge about its associated factors may help in optimal management of the patients. Keeping in view of original scale of measurement of nodal involvement and its distribution, in contrary to earlier models on its nominal form (yes/ no), or ordinal form (grouping of number of positive nodes), modelling its absolute frequency in count form may obviously be a better option to get accuracy in the predictors. This study deals with above modelling using data on 945 oral squamous cell carcinoma patients available under regularly updated data-base, who underwent surgery including neck dissections at our institute. The nodal involvement was among 39.8%. In developing the model, basic principles regarding co-linearity, confounding & effect modifiers were followed. The analytical results reveals that finally out of 19 variables found appropriate for the model exploration, 10 variables including submucous fibrosis, tumour size, oral site and presence of clinical neck-node are major associated factors. This presentation will have comparative appraisal of earlier reported results for its nominal and ordinal forms
|
