Intervention input parameters
| Parameter | Assumptions and data source data | Trend and uncertainty |
|---|---|---|
| Cost of a law | Costs of a new law in NZ to mandate the tobacco retail outlet reduction interventions: NZ$3.54 million. We did not consider ongoing costs around auction operation or retailer licensing since we assumed that this would be self-financing from auction fees. Also we did not assume any costs to businesses since it was assumed that they would replace tobacco sales with other product sales (retail outlets that only sell tobacco are very rare in NZ). | Uncertainty: γ, SD=NZ$1.05 million in 2011 only.20 |
| Travel costs (treated as an incremental increase in the price of a pack of 20 cigarettes) | The incremental travel costs for each intervention year were calculated using: (1) a monetary value on distance travelled (eg, fuel, car maintenance, but excluding insurance and depreciation cost), (2) a monetary value on time spent travelling (ie, personal non-work-related travel time) and (3) an increasing proportional rate for the fraction of the total trip (and hence the travel costs) that would be tobacco-related.15 We used the NZ Ministry of Health mileage reimbursement rate for private vehicles under the National Travel Assistance Scheme: $0.28/km (2011 NZ$).22 We used the car, non-work-related travel time value of $7.18 per hour (2011 NZ$).23 Costs were calculated for each neighbourhood centre and then averaged by rural, semiurban or urban area types. Costs were then weighted by the proportion of Māori and non-Māori in each area type to produce costs by ethnicity for each intervention and each year. We assumed that the progressive reduction in outlets over time would increase the proportion of travel (and thus the total travel costs) that is tobacco-related. Each year, the proportion of tobacco-related travel increased by 5%, up to 50% in year 10. | An uncertainty with a log-normal distribution of ±20% SD around the running cost of a car was assumed. An uncertainty with a log-normal distribution of ±20% SD around the value on personal travel time was assumed. An uncertainty with a β-distribution of ±20% SD around the amount of travel explicitly for tobacco was assumed. Total cost uncertainty: log-normal distribution, ±25% SD. Uncertainties around costs and the amount of travel explicitly for tobacco were mathematically combined in the total travel cost formulae. The formula was run a 1000 times with Monte Carlo simulation for each intervention year in TreeAge software. |
| Tobacco price elasticities used to estimate the impact of increased travel costs on tobacco consumption | Non-Māori: The following price elasticities for smoking prevalence were used: −0.38 (for 15–20 years), −0.29 (for 21–24 years), −0.19 (for 25–34 years) and −0.10 (for 35+ years; as per our previous published work30). These were applied in the year the travel costs increased as a result of reducing the number of tobacco retail outlets under the four different interventions.Māori: For each intervention, we scaled up the non-Māori price elasticity by 20% for Māori given economic theory, the patterns in the international literature for other social groupings and some NZ evidence for increased price sensitivity for Māori for tobacco and for food products.31 32 | No trend. Uncertainty: Non-Māori, SD ±20%, normal, correlated 1.0 across four age groups; Māori absolute scalar of +20% within each age group, ±10% normal (ie, 95% range of absolute scalar of 0.4% to 39.6%). |
| Illicit tobacco market dynamics | The illicit tobacco market was set to start at 1% of the market share in 2011 based on the best data at that time.3 The average tobacco price used in the modelling was a combined function of the legal market price of tobacco, the illicit price of tobacco (75% of the legal price which we estimated from Australian work in the absence of NZ data33) and the size of the illicit market. In the best model, we assumed a stable illicit market (as per our previous work15 and reflecting the absence of commercial tobacco growing in NZ). But in a scenario analysis, we assumed a growing illicit market, whereby every 10% increase in the price of a pack of 20 cigarettes would result in a 1% increase in the size of the illicit market. | Uncertainty of ±20% SD β around the illicit price of tobacco. |
NZ, New Zealand; QALYs, quality-adjusted life years.