Combining risk reduction strategies
The previous section reviewed the effectiveness, costs and cost-effectiveness of a series of interventions aimed at reducing specific risks to health. That analysis allows decision-makers with an interest in reducing the burden related to a specific cause -- for example, cardiovascular disease or child undernutrition -- to assess what types of interventions would be cost-effective in that area for the resources that are available. This section takes the broader perspective of a government as the steward of the entire health system. As argued earlier, one of the intrinsic goals of a health system is to improve population health, and information about how best to achieve this for the available resources is of vital importance. This requires not only deciding which combinations of interventions are cost-effective ways of reducing the risks associated with unsafe sex, for example, but also deciding which of the myriad of risks to health that could be targeted should be given priority.
The information considered in the previous section is used again to illustrate how cost-effectiveness analysis can make an important contribution to this debate. Figures 5.2 and 5.3 report the results for interventions considered in the previous section, for two of the 14 subregions, AFR-D and AMR-B.11 Interventions that are both more costly and less effective than alternative ways of achieving the same goal (for example, reducing the impact of unsafe sex) are not shown on the graphs so that the more cost-effective interventions can be identified more easily. That is why most of the interventions that are shown appear to be cost-effective. (The key to the interventions is found in Table 5.3.) The vertical axis depicts the annualized discounted costs of the intervention. All costs are included regardless of who pays.12 The horizontal axis shows the yearly DALYs gained from this action.
The two rays drawn from the origin represent the cut-off points used to denote interventions as cost-effective and very cost-effective. All points on the lower ray (closer to the south-east corner) have a cost-effectiveness exactly equal to GDP per capita in the region. Interventions appearing to the right of it are defined as very cost-effective -- most of the preventive interventions aimed at reducing unsafe sexual practices and improving child undernutrition fall in this category in both regions. All points on the upper ray (closer to the north-west corner) have a cost-effectiveness equal to three times GDP per capita, the cut point used to distinguish between cost-effective and cost-ineffective interventions. Points to the left of this ray would not be cost-effective in that region.
In AFR-D, preventive interventions to reduce the health effects of unsafe sex and the combined approach of population-wide and individual-based interventions for cardiovascular disease are among those in the most cost-effective category. On the other hand, treatment of people based purely on observed levels of blood pressure and cholesterol would not be cost-effective. In AMR-B, high rates of taxation to reduce smoking would be very cost-effective, but the combination of all the possible smoking-reduction interventions would not be in the most cost-effective category.
The figures show which interventions are in the the most cost-effective category. They also illustrate that it is possible for an intervention to be cost-effective but at the same time have a relatively small impact on population health. In AFR-D, for example, iron supplementation at 50% coverage (intervention 165) is cost-effective by itself. So is the combination of case management for pneumonia, ORT, vitamin A and zinc supplementation (intervention 80). The former would gain 1.28 million DALYs while the latter would gain 11.6 million. Despite the fact that both are very cost-effective, policy-makers need to have information about which one will have the greatest total impact on population health, and the total cost of achieving these health gains.13
In both figures, however, interventions cluster close to the origin and it is difficult to identify all of them clearly. Accordingly, the figures are redrawn with the axes on a logarithmic scale, enabling the individual interventions to be identified. In this case, the lines drawn obliquely across the figures represent lines of equal cost-effectiveness. All points on the line at the south-east extreme have a cost-effectiveness ratio (CER) of I$1 per DALY gained. Because of the logarithmic scale, each subsequent line moving in a north-easterly direction represents a one order of magnitude increase in the CER, so all points on the next line have a CER of I$ 10, and the subsequent line represents a CER of I$ 100.
These figures illustrate more clearly that the variation in CERs across interventions within each region is substantial. In both subregions, some interventions (for example, preventive interventions aimed at reducing the incidence of HIV, and interventions to improve unsafe injection practices) gain each DALY at a cost of less than I$ 10. On the other hand, adding nicotine replacement therapy to the cost-effective population-wide set of anti-smoking activities would cost more than I$ 10000 per additional DALY gained (intervention 20). A similar range of cost-effectiveness ratios is observed in AMR-B.
The information on costs and effectiveness of a set of interventions targeting different risk factors can help to identify which interventions would be selected for given levels of resource availability in the different regions if the goal were to maximize population health.14 In AFR-D, for example, a very severe restriction of resources would see most attention paid to preventive interventions to reduce the impact of unsafe sexual behaviours, unsafe injection use and micronutrient supplementation or fortification.
If the substantial increase in resources for health in Africa that is now becoming available allows all interventions costing less than three times GDP per capita to be funded, the optimal mix would include HIV prevention interventions combined with ARV treatment. It would include supplementation or fortification of vitamin A, iron and zinc in combination with treatment for diarrhoea and pneumonia in children. Disinfection at point of use would be combined with provision of improved sanitation facilities, and interventions designed to reduce the overuse of injections and unsafe injection practices would be introduced. Population-wide interventions to reduce the risks of cardiovascular disease would be combined with treatment of individuals with an absolute risk of an event in the next 10 years estimated to be above 25% (possibly even 15%), and high rates of taxation on cigarettes would be introduced and maintained.
These interventions are not exhaustive because not all risk factors were included, nor were all possible interventions analysed. However, they show that an annual expenditure of approximately I$ 6.8 billion would gain over 140 million DALYs in that region alone.