Parameter | Description |

Denicotinisation | |

NS→CS (age 20 only) NS→DU (age 20 only) | 90% (SD 5%) of BAU initiation at age 20 by 5 years after implementation (X=beta (32.4, 3.6), median 90.7%, 95% UI: 78.5% to 97.4%.). Implemented as 1–(1–X)^{(t/5)} scalar applied to the BAU initiation rates in years t (1–5) after introduction of the policy, then held at 1–X% thereafter. |

CS→FSFV CS→FSCV DU→FSFV DU→FSCV | Using an expert knowledge elicitation (see online supplemental appendix D), the reduction in smoking prevalence 5 years after the low nicotine policy compared with BAU in 5 years, due to quitting or switching to vaping, was mean 84.4% (SD 7.84%, X=beta (17.78, 3.19), median 85.9%, 95% UI: 67.1% to 96.3%). Implementation was as 1–(1–X)^{(t/5)} scalar applied to BAU CS and DU prevalence, where t is the 1–5 years after intervention. For the sixth and subsequent years, the transition probabilities were twice those in BAU (due to an ongoing higher NCR, given non-addictive levels of nicotine in tobacco). |

NS→NSCV | No change. |

Denicotinisation plus mass media | |

NS→CS (age 20 only) | As above for low nicotine. |

NS→DU (age 20 only) | As above for low nicotine. |

NS→NSCV (age 20 only) | No change. |

CS→FSFV CS→FSCV DU→FSFV DU→FSCV | As above for low nicotine from year 1 to 5+ twice the absolute contribution of the routine media/Quitline campaign added to background net cessation (ie, 1.055%×2=2.1%)30 Subsequent years: transition to quitting or vaping was twice those in BAU. |

Retail outlet restriction to about 300 outlets (about 5% of current outlets; assumed supply of e-cigarettes reduces commensurately)* | |

NS→CS | As per the increase in cessation probabilities (CS→FSFV, etc, below), we reduced the initiation rate by X=beta (23.4, 97.2), median 19.2%, 95% UI: 12.9% to 26.9%. Applies in 2023 onwards (as youth contemplating initiating in the future confront lesser retail availability as well). |

NS→DU | As above for NS→CS. |

CS→FSFV CS→FSCV DU→FSFV DU→FSCV | As a low estimate of one-off quitting, we used that from studies modelling reducing retail outlets in terms of increased travel costs37: a reduction in the prevalence of 15.6% for Māori, and 16.0% for non-Māori—or 15.8% overall. As a high estimate, we used that from the New Zealand - International Tobacco Control study where—in response to a question whether they would quit in response to a 95% reduction in retail outlets—23.0% said they would quit (half quitting→FSFV, half switching to FSCV).38 Placing the mean at 19.4% (average of above 15.8% and 23%) and using 15.8% and 23% as one SD either side of the mean (SD=3.6%), we parameterised the one-off increase in smoking net cessation as X=beta (23.4, 97.2), median 19.2% (ie, percentage point increase), 95% UI: 12.9% to 26.9%. Note this increase was on top of BAU transition probabilities and halved over CS→FSFV and CS→FSCV and halved over DU→FSFV and DU→FSCV. For example, if the CS→FSFV was 5%, the intervention CS→FSFV transition probability was 5%+(1–5%)×0.5×X%. This effect was in the year of intervention only—in years after the retail outlet restriction, the transition probabilities out of CS and DU reverted to BAU. |

NS→NSCV | Unchanged. |

Tobacco-free generation | |

Smoking initiation rate (NS→CS; occurs only at age 20) | For two reasons, a tobacco-free generation proposal will not immediately achieve zero uptake at age 20; (1) our model for parsimony assumes all uptake at age 20, but the minimum legal age of purchasing is 18 years; (2) social supply will allow some young people to keep initiating. We therefore assumed that initiation at age 20 in our model (essentially an average of all initiation by (say) age 25) will asymptote to a mean of X=10% (SD 5%) of BAU in 10 years (beta (3.6, 32.4), median 9.3%, 95% UI: 2.6% to 21.5%), with the scalar of BAU initiation rate of X^{(t/10)} for t=1–10 years after the tobacco-free generation policy is implemented, then X of BAU initiation thereafter. |

NS→DU | As above for NS→CS. |

NS→NSCV | Unchanged.* |

Combined: denicotinisation+retail+tobacco-free | |

NS→CS (age 20 only) | Cumulative impact. If the % reduction in initiation in year t for denicotinisation, retail and tobacco-free was A%, B% and C%, then the reduction in the combined intervention was 1–(1−A)(1–B)(1– C). |

NS→DU (age 20 only) | As above for NS→CS. |

CS→FSFV CS→FSCV DU→FSFV DU→FSCV | Cumulative impact. If the % increase in quitting or switching in year t for denicotinisation, media and retail was A% and B%, then the increase in the combined intervention was 1–(1 – A)(1–B)(1 –C). |

NS→NSCV | Unchanged. |

*If the availability of alternative nicotine delivery systems (ANDS, for example, e-cigarettes) does not reduce commensurately with these policy interventions, one would expect larger switched to ANDS which would reduce smoking prevalence further (but increase DU, FSCV and possibly NSCV state prevalence). We do not model this explicitly but consider it in the Discussion section.

BAU, business as usual; CS, current smoker (but not a dual user); DU, dual user; FSCV, former smoker current vaper; FSFV, former smoker and/or former vaper; NCR, net cessation rate; NS, never smoker; NSCV, never smoker current vaper; NZ, New Zealand; UI, uncertainty interval.