Fig. 1
Incidence ratios of fall and fall-related fracture in each tobacco product use status. The incidence ratios (circles) and 95% CIs (bars) of current tobacco use status/types, compared with never users, in different populations were estimated with two-level multilevel Poisson regression with robust variance. Random intercepts for residential areas (47 prefectures, Level 2 variable) were used. We adjusted for confounding factors (sex, age, educational attainment, occupation class, industry sector, equivalent annual household income, workplace size, and Work Functioning Impairment Scale). Additionally, we accounted for lifestyle and behavioral variables (drinking habits, weekly physical activity, hypertension, dyslipidemia, diabetes, body mass index, and habitual use of hypnotics or anxiolytics). The samples used in the analysis were as follows: (1) n = 18,440 (1,338 falls and 17,102 non-fallers), (2) n = 17,610 (508 fractures and 17,102 non-fallers), (3) n = 15,470 excluding previous fallers (226 falls and 15,244 non-fallers), and (4) n = 15,300 excluding previous fallers (56 fractures and 15,244 non-fallers). Abbreviations: CI, confidence interval; HTP, heated tobacco product.
