OHS Defined:

Using Statistical Analysis System version 9.4, we combined the binary variable of brushing daily 1/0 and flossing daily 1/0, and categorised the cohort into four mutually exclusive groups: group 0 consisted of those who never brushed or flossed; group 1 consisted of those who flossed but not brushed; group 2 consisted of those who brushed but not flossed; and group 3 consisted of those who brushed daily and flossed. There were only four people in group 1 and we merged them into group 0. Thus, OHS group 0, 1 and 2 were created: 0 being the poorest oral hygiene group (poor OHS and reference); 1 being those who brushed daily but not flossed (good OHS); and 2 being those who brushed and flossed (better OHS). Because the sample size in Cox regression is event rate (that is, CVD mortality), the sample size in this study is very small between 40-50. Thus, to save the degree of freedom (statistical power), we used linear trend models whenever possible. Combining people who never brushed and brushed seldom led to the reference group (n = 41). Next, we generated two levels of OHS, namely, who only brushed daily (n = 261) (OHS level 1) and the 57 people who brushed and flossed became OHS level 2. The baseline characteristics were stratified by OHS categories and compared by non-parametric three-group comparison by Kruskal-Wallis test or chi-squared test.