Modelling count outcomes from dental caries in adults: a comparison of competing statistical models
Abstract
The objectives of this research are to identify the key risk factors contributing
to dental caries in adults and to determine an appropriate model suitable for
the analysis of Decayed, Missing and Filled Teeth Data (DMFTDATA). The
study was conducted between June, 2015 and January, 2016 in Sefwi Wiawso
District Hospital, Sefwi Wiawso, Ghana, and it involved a systematic random
samples of 1158. The average age of participants was about 40 years with 54%
of them being females and the rest 46% being males. Count data models such
as Poisson, negative binomial models, zero inflated Poisson (ZIP), zero inflated
negative binomial (ZINB) and Poisson hurdle (PHURDLE) models were fitted to
the DMFTDATA and compared using the Vuong test statistic (V) and Akaike
information criterion (AIC). ZINB was found to be the appropriate model for
analysing DMFTDATA characterised with overdispersion and contains 28.8%
of zeros. It was found that age (in years), jaw accident, frequency in sweet
consumption, rinsing habit after meal with water and frequency of brushing were
significant risk factors of dental caries in adults. It is therefore recommended that
adults should endeavour to brush their teeth at least twice daily and rinse their
mouth with water after meal. Also proper dental care should be considered for
adults with jaw accident and adults should reduce intake of sugar foods.
Description
A thesis submitted to the Department of Mathematics,
Kwame Nkrumah University of Science and Technology In partial fulfillment of the requirement for the Degree of M.Phil Mathematical Statistics,