Integrated Decision Support to Tobacco Control Policy with Applications to Singapore

Ong Lishan

Abstract

The tobacco epidemic is a serious problem plaguing the world. Although Singapore has a low smoking rate, the government is still planning to implement more policies to reduce the rate further. Given the many effective tobacco control policies available, and given the conflicting criteria of public health improvement and cost, there is a need to compare policies for prioritization by governments. Governments will also want to know how to select multiple policies to maximize public health benefits given a certain budget constraint.

However, studies on tobacco control policies do not always use standardized metrics for comparing policies. Sometimes, only health metrics are tracked. Existing studies on cost-effectiveness cover health and cost outcomes, but they do not do an incremental analysis and do not incorporate subjective stakeholder preference to allow for a policy to be recommended.

Therefore, this thesis proposes an integrated Systems Dynamics-Decision Analysis (SD-DA) and Optimization decision support approach to compare and select policies, based on public health benefit and cost. It leverages on the strengths of three complementary components to rank policies by cost-effectiveness – SD, cost analysis and Analytic Hierarchy Process (AHP). 1) An SD model is built to simulate smoking behavior and policy effects, enabling policy makers to understand complex relationships between system components and feedback loops. Multiple health-related metrics can be obtained from the model outputs. 2) The cost analysis uses a probabilistic discounted cashflow model, considering uncertain cashflows and policy makers’ risk tolerance. 3) Combining the SD and cost analysis results, AHP allows for multiple health and cost metrics and criteria to be weighed, using subjective stakeholder preference. This produces a cost-effectiveness ranking, allowing for single policy comparison and prioritization. Tradeoffs between health and cost metrics can be evaluated. Finally, this thesis uses optimization to select multiple policies under budget constraints.

When the integrated decision support approach is applied to Singapore, the cost-effectiveness ranking is as follows: 1) Cessation Treatment; 2) Tax Increase; 3) Mass Media; 4) School Program; 5) Health Labels and 6) Status Quo. Ranking results are largely insensitive to changes in SD model parameters, risk tolerance, and birth rate scenarios. Assuming that policy makers prioritize public health more than cost, and do not face much cost constraints, the ranking is largely robust, as seen from the SD, cost and AHP analysis. Under almost all situations, Cessation Treatment and Tax Increase are the top two favorable options, due to their strong performance in health and cost. Under tight cost constraints, the Status Quo may be more favorable than some alternatives, though such austerity is unlikely. Where multiple policies can be selected, Cessation Treatment and Tax Increase are the most cost-effective. Mass Media, School Program and Health Labels are still useful, despite their lower cost-effectiveness score, as they can maximize public health benefit and make fuller use of the budget. Policy makers should exploit instances where public health benefit increases greatly with just a slight budget increase.

Overall, this thesis highlights the usefulness of the integrated SD-DA and Optimization approach in policy comparison and selection. By applying the framework to Singapore, useful recommendations about policy prioritization and selection can be proposed. The thesis also lays the groundwork for future work to be done, possibly in other healthcare policies (such as obesity reduction) or in other fields of policy making (such as transport and water management).