Table 2 Bivariate and multivariate regression analyses of trend and determinants of nicotine yield (mg/cig) from 1998 to 2005.
| Nicotine yield in smoke (mg/cig) Bivariate analyses* | Nicotine yield in smoke (mg/cig) multivariate analysis† | Market category and design time trends‡ | ||||
|---|---|---|---|---|---|---|
| ββ§ (95% CI) | p Value | ββ§ (95% CI) | p Value | ββ¶ (95% CI) | p Value | |
| Time trend (years) | ||||||
| All manufacturers | 0.029 (0.010 to 0.044) | 0.003 | 0.009 (0.006 to 0.011) | <0.001 | ||
| Brown & Williamson | 0.059 (0.049 to 0.069) | <0.001 | ||||
| Lorillard | 0.030 (0.187 to 0.041) | <0.001 | ||||
| Philip Morris | 0.011 (0.006 to 0.015) | <0.001 | ||||
| RJ Reynolds | 0.020 (0.016 to 0.024) | <0.001 | ||||
| Market category | ||||||
| Length (⩾100 mm) | 0.097 (0.005 to 0.190) | 0.038 | – | NS | NS | |
| Mentholated vs non‐mentholated | –0.043 (–0.137 to 0.051) | 0.367 | – | NS | NS | |
| Candy‐flavoured vs non‐flavoured | –0.099 (–0.194 to –0.004) | 0.041 | – | NS | NS | |
| Full flavour vs non‐full flavour | 0.348 (0.300 to 0.395) | <0.001 | 0.187 (0.155 to 0.219) | <0.001 | 0.006 (0.004 to 0.009) | <0.001 |
| Filtered vs non‐filtered | –0.092 (–0.221 to 0.036) | 0.160 | – | NS | 0.002 (0.001 to 0.003) | 0.001 |
| Design feature | ||||||
| Nicotine concentration (mg/g) | 0.037 (0.032 to 0.042) | <0.001 | 0.030 (0.025 to 0.034) | <0.001 | 0.177 (0.147 to 0.207) | <0.001 |
| Nicotine (mg) per cigarette | 0.061 (0.055 to 0.067) | <0.001 | NS | 0.135 (0.110 to 0.161) | <0.001 | |
| Puffs per cigarette | 0.091 (0.079 to 0.103) | <0.001 | 0.106 (0.098 to 0.113) | <0.001 | 0.048 (0.035 to 0.060) | <0.001 |
| Percentage filter ventilation | –0.013 (–0.014 to –0.011) | <0.001 | –0.014 (–0.015 to –0.013) | <0.001 | NS | |
| Weight (g) | 1.563 (1.325 to 1.801) | <0.001 | NS | NS | ||
cig, cigarette; NS, not significant.
*Each individual multilevel regression model consists of the Massachusetts (MA) regimen‐measured nicotine yield as the dependent variable, and one market category, design feature or time (in years) as the independent variable.
†Results are given for the multilevel mixed‐effects multiple regression model of MA regimen‐measured nicotine yield in smoke, including all significant parameters.
‡Each individual multilevel regression model consists of one market category or design feature as the dependent variable, and time as the independent variable.
§ β‐coefficients represent mg increase in nicotine yield in smoke per unit change of each parameter per year.
¶ β‐coefficients represent unit change of each parameter per year.