Classification tree analysis of factors associated with oral cancer exam

Abstract

Objective: In this study, we performed a classification tree analysis (CTA) to identify population subgroups that are less likely to have an oral cancer examination (OCE). Methods: We conducted telephone surveys (N = 2401) of adults residing in north Florida to collect data on OCE status and potential OCE predictors including demographics, medical and dental experience, and psychosocial factors. The CTA algorithm exhaustive chi-square automatic interaction detector (E-CHAID) was employed to determine the relationships between OCE status and the predictors. Results: The overall OCE rate was 46.8% in our sample. Participants’ rating of the dentist was the first level splitting variable, leading to 15 unique participant subgroups: (1) high dentist rating (N = 1269) led to 5 splits and 11 subgroups; (2) low dentist rating (N = 308) led to zero splits; and (3) no regular dentist (N = 824) led to 2 splits and 3 subgroups. Conclusions: The CTA has identified unique population subgroups that could be targeted in future tailored public health interventions. Among underserved populations, it is important to develop and implement community-based interventions that encourage regular dental visits and provide oral cancer self-examination education..