Cost-effectiveness of interventions to prevent lifestyle-related disease and injury in Australia

Background The costs of healthcare are on the rise. With an ageing population, growing demand for health services and expensive new technologies, Australia’s annual health care bill has more than doubled in the last ten years, and is projected to increase a further 127% by 2033. As third-party provider of health care, the Government must make difficult decisions about how best to allocate limited resources to the many new and existing drugs, technologies and health services available for prevention and treatment of disease. Cost-effectiveness analysis of interventions can help identify those that should be given funding priority in order to maximise population health, but its use in allocating resources to prevention has been limited. There have been few cost-effectiveness analyses of preventive interventions, particularly for the potentially more effective strategies targeting the whole population, such as taxation, regulation and community campaigns. Current methods are poorly designed for capturing the change in population distribution of risk that can occur with these types of interventions. Use of cost-effectiveness analysis has also been limited by the use of many different (and often simplistic) modelling methods and assumptions that prevent league table comparison of results to help identify most cost-effective strategies. This thesis presents new methods for evaluating cost-effectiveness of preventive interventions, with application to interventions promoting physical activity, preventing alcohol misuse, reducing body mass, promoting fruits and vegetables and reducing dietary salt intake, from an Australian health sector perspective. Methods Proportional multi-state life table models were developed for each risk factor and for risk factor combinations, using population impact fraction (PIF) functions to quantify the potential intervention impact of a change in individual or population risk factor exposure on disease and injury. The models were used to simulate population health in disability-adjusted life years (DALYs) and costs of disease treatment, over the lifetime of the Australian population in 2003, for a range of individual- and populationtargeted interventions using intervention costs and effects derived from Australian cost data and published evaluation studies. Monte-Carlo analysis was used to derive uncertainty around all outcome measures, and sensitivity of results to key modelling choices and assumptions was also evaluated. Cost-effectiveness of six physical activity interventions, ten alcohol interventions, two body mass interventions, 23 fruit and vegetable interventions and four dietary salt interventions was evaluated in comparison to current Australian practice. In addition, where multiple mutually-exclusive interventions were evaluated, a partial null (‘no current practice intervention’) scenario was calculated and cost-effectiveness of incrementally adding each intervention to a package was evaluated, to determine the optimal intervention mix and to compare optimal outcomes with the current practice. Findings For physical activity, a package of six individual- and population-targeted interventions is cost-effective and could avert a third of disease burden attributable to physical inactivity. For reducing alcohol misuse, a package of eight individual- and population-targeted interventions could avert a third of disease burden attributable to hazardous and heavy levels of drinking. Although the current practice of random breath testing is cost-effective, if the expenditure on random breath testing had been distributed to more cost-effective interventions, around ten times the improvement in population health could have been achieved. The individually-targeted interventions for body mass, fruits and vegetables and salt intake are not cost-effective. Providing incentives for food industry to reduce salt in processed foods, on the other hand, is far more effective in improving population health and can lead to cost-saving for the health sector in the long term. if (moderate) reductions in salt were made mandatory for food manufacturers, around 20 times the health gains achieved by the current voluntary program could be achieved. Overall, eight interventions are potentially cost-saving for the health sector: voluntary and mandatory limits on salt in processed foods; mass media- and pedometer-based community campaigns to promote physical activity; a community program to promote fruits and vegetables; and volumetric taxation, advertising bans and an increase of the minimum legal drinking age to 21 years to address alcohol misuse. A further 12 interventions for reducing alcohol misuse, and promoting physical activity and fruit and vegetable consumption, are under a $50,000 per DALY threshold of costeffectiveness, and are also recommended for health sector investment. Implications The integration of a proportional multi-state life table model with PIF function in this research has proved to have a number of advantages over previous modelling methods. The PIF function enables better simulation of the true continuous distribution of risk in the population, and facilitates analysis of population-targeted interventions that shift the whole distribution of risk. It also substantially simplifies the integration of multiple risk factors into the one model, which was previously constrained by the need to create separate states for every risk factor category (e.g. active and inactive) and risk factor categories in combination (e.g. obese and active, obese and inactive, etc.). This not only makes for easier and more accurate analyses of interventions targeting multiple risk factors, but enables evaluation of packages of many different risk factor interventions. Further investment in preventive interventions is highly recommended. The population-targeted approaches hold most promise for improving population health. Although there is potential for opposition from industry stakeholders and public concern around Government paternalism, these interventions have most potential to generate cost-savings for the health sector in the long-term. Given the weaker evidence around effectiveness of these interventions, however, it is recommended that programs are implemented with sufficient funding for monitoring and evaluation of outcomes