Future smoking prevalence under business as usual (BAU) and projected smoking prevalence under interventions (95% uncertainty intervals in parentheses for BAU 2016)*, and ratio of smoking prevalence under tax intervention compared with prevalence under BAU
| Smoking prevalence | Prevalence ratio comparing tax intervention to BAU | ||||||||
| BAU | Tax intervention | Tobacco-free generation | |||||||
| Sec and age† | 2016 | 2026 | 2036 | 2026 | 2036 | 2026 | 2036 | 2026 | 2036 |
| Males | |||||||||
| 20 years | 34.1% (19.1 to 53.1) | 24.4% | 22.9% | 15.4% | 14.2% | 0.0% | 0.0% | 0.63 | 0.62 |
| 30 years | 42.1% (24.9 to 59.1) | 34.7% | 24.9% | 26.8% | 15.7% | 34.7% | 0.0% | 0.77 | 0.63 |
| 40 years | 35.7% (20.2 to 52.4) | 39.7% | 32.5% | 34.7% | 25.2% | 39.7% | 32.5% | 0.87 | 0.78 |
| 50 years | 34.9% (19.7 to 51.6) | 30.7% | 34.6% | 27.2% | 30.2% | 30.7% | 34.6% | 0.89 | 0.87 |
| 60 years | 28.2% (15.2 to 43.8) | 27.3% | 23.9% | 24.2% | 21.2% | 27.3% | 23.9% | 0.89 | 0.89 |
| 70 years | 19.1% (9.7 to 31.9) | 19.4% | 18.9% | 17.3% | 16.7% | 19.4% | 18.9% | 0.89 | 0.88 |
| Females | |||||||||
| 20 years | 10.1% (5.4 to 18.2) | 7.8% | 7.4% | 5.0% | 4.6% | 0.0% | 0.0% | 0.64 | 0.62 |
| 30 years | 12.5% (6.7 to 22.1) | 11.6% | 9.0% | 9.0% | 5.7% | 11.6% | 0.0% | 0.78 | 0.63 |
| 40 years | 12.2% (6.5 to 21.6) | 12.7% | 11.9% | 11.1% | 9.2% | 12.7% | 11.9% | 0.87 | 0.77 |
| 50 years | 13.7% (7.4 to 24.0) | 10.7% | 11.3% | 9.9% | 9.8% | 10.7% | 11.3% | 0.93 | 0.87 |
| 60 years | 12.0% (6.4 to 21.3) | 10.9% | 8.5% | 9.9% | 7.8% | 10.9% | 8.5% | 0.91 | 0.92 |
| 70 years | 7.5% (3.9 to 13.9) | 7.2% | 6.5% | 7.1% | 5.9% | 7.2% | 6.5% | 0.99 | 0.91 |
Uncertainty intervals are provided for the baseline year, 2016, according to that provided from IHME. There will be uncertainty in 2026 and 2036 predicted smoking prevalence as well, but it was impracticable to extract from the model. As a guide, for the uncertainty in BAU predictions the uncertainty will be slightly wider in ratio terms (ie, upper divided by lower limits) than 2016 due to uncertainty in the APC parameter used in smoking prevalence predictions, much the same for smoke-free generation (SFG) (assuming no uncertainty in the implementation; except for the 0% estimates among younger cohorts where it is structurally assumed to be zero with no uncertainty), and widest for the tobacco tax intervention (perhaps half as wide again) due to added uncertainty in price elasticities.
*Future BAU smoking prevalence was estimated with separate logistic regression models by sex on GBD smoking prevalence data, with terms for age and year. Models were fitted with generalised linear models in STATA V.15 with family binomial and logit link. Prevalence data only for ages 20–80 were used for predictions. For more details, see online supplementary additional file 9.
†Age in future year (eg, age in 2026 and 2036).
APC, annual percentage change; BAU, business as usual; GBD, Global Burden of Disease; IHME, Institute for Health Metrics and Evaluation.