Short abstract

Analysis of simultaneous interventions

Our method of analysis used two innovations to ensure the results had more relevance to practical policy decisions than traditional cost effectiveness analysis.^3,9-12 Firstly, the cost effectiveness of the existing use of resources could be evaluated at the same time as the cost effectiveness of possible future courses of action should new resources become available. Traditional cost effectiveness analysis has usually considered only future use of resources. Secondly, we incorporated interactions between costs and effects of interventions that are undertaken simultaneously, as they would be in practice. Previous studies have generally assumed, mostly implicitly, that every intervention is implemented in isolation from related activities.

Here we have analysed more synergies between concurrent interventions than were included in the analyses for separate health goals.^4-8 For example, different interventions that would be delivered as part of a basic obstetric package, often by the same person during the same visit, had been analysed separately in the maternal and neonatal health (tetanus toxoid),⁵ HIV and AIDS (prevention of mother to child transmission),⁶ and malaria (intermittent presumptive treatment with sulfadoxine-pyrimethamine in pregnancy)⁷ analyses. Cost synergies between tetanus toxoid and other interventions aimed at maternal and neonatal health were included in that paper, but here we add synergies resulting from common delivery platforms across all the health goals.

The individual papers eliminated several interventions from further consideration because they proved to be more costly, with lower health benefits, than others (see table B on bmj.com) The remaining interventions were classified in a way that is useful for setting priorities across multiple health conditions. We earlier argued that the uncertainty around estimates of costs and health gains, especially when information must be taken from a limited number of data points, precludes basing policy advice on the point estimates of cost effectiveness.³ For policy purposes, interventions should be compared in terms of order of magnitude cost effectiveness bands. Within any band, individual decision makers have a menu of interventions to choose from. We deemed interventions to be highly cost effective if they cost less than the gross domestic product per capita to avert each disability adjusted life years (DALY) and cost effective if each DALY could be averted at a cost of between one and three times the gross domestic product per capita. Other interventions are not cost effective.¹³ This incorporates an element of affordability as regions and countries with lower national income will have lower cut-off points.

Recommended strategy changes for each goal

In some cases, we found current strategies and plans to be essentially appropriate, while more opportunities to reallocate resources existed in others (box. Significant reductions in maternal and neonatal mortality require, for example, increased access to clinic based services providing basic and emergency obstetric and neonatal care, but also increased community based prevention, including the encouragement of breastfeeding, support of low birthweight babies, treatment of neonatal pneumonia, and wider provision of tetanus toxoid. If no new resources are forthcoming and substantial resources currently support relatively high cost, low effect interventions (such as antibiotics for premature rupture of membranes) policy makers could consider reallocating current spending to the more cost effective interventions.

Priority setting across health goals

Tables ​Tables11 and ​and22 classify interventions into the cost effectiveness bands described above for the two regions Afr-E (countries in sub-Saharan Africa with very high adult mortality and high child mortality) and Sear-D (countries in South East Asia with high adult and child mortality). Tables C and D on bmj.com gives details of costs, effects, and cost effectiveness ratios.

Application of results

In practice, resources are never allocated according to formulaic cost effectiveness rules described in text books—for example, by choosing the most cost effective intervention, then the next most cost effective, until all resources are used. This can sometimes suggests that only prevention should take place, or only treatment, but in reality mixes of interventions are generally found. Our analysis suggests this is appropriate. The highly cost effective group of interventions reported above includes a selection from each of the five goals in both regions, as well as mixes of curative and preventive actions and of population and individually focused activities. This is true even if the threshold for highly cost effective interventions is reduced to $Int100 per DALY averted.

Both regions have so much unmet need and so many under-used interventions that the opportunities for reallocating resources are limited. Purely on cost effectiveness grounds, however, priority should clearly be given to highly cost effective interventions rather than activities such as second line antiretroviral therapy for AIDS and provision of supplementary food for children in Afr-E. More could be achieved if these resources were reallocated to any of the under-used, highly cost effective group. A similar picture unfolds in Sear-D. Attention should be focused on scaling up interventions that are highly cost effective rather than expanding second line antiretroviral therapy (which is just over the threshold for cost effective interventions), antenatal steroids for preterm births, and provision of supplementary food for children.

Both regions have a relatively large set of highly cost effective interventions, offering considerable flexibility to adapt packages to particular contexts. The relative size of the highly cost effective group reflects the unmet needs but also the fact that the millennium development goals were well chosen and need to be better funded. Many more interventions would fall outside the highly cost effective group had our analysis included conditions outside the goals, and it is here that greater potential to reallocate resources toward the goals may be found.

Validity of cost effectiveness

We accept that in practice, considerations other than cost effectiveness do, and should, influence decisions on resource allocation. Important debate continues about the appropriateness of using cost effectiveness analysis to drive decisions in health. For example, the technique focuses only on the health gains associated with different uses of resources and does not incorporate other effects of concern to society. This may be particularly relevant to antiretroviral treatment for HIV and AIDS, which keeps health workers and school teachers in their posts and could, at the limit, prevent a possible break down of society.^14,15 These benefits cannot be captured in terms of DALYs. Use of cost effectiveness analysis also raises several ethical issues, particularly the fact that equity is not explicitly incorporated.^15,16

Policy makers, however, cannot escape from the unfortunate fact that the resources available are insufficient even to implement all the interventions designated in this paper as highly cost effective, and it is not yet clear that the additional resources required to reach the millennium development goals will be found. In such cases, informed decisions about how to allocate the available resources require knowledge of the likely effect on population health of different courses of action. Without this knowledge, decisions could be made to improve the health of a few people by a small amount at the expense of improving the health of more people by a larger amount, something that neither the proponents nor opponents of cost effectiveness analysis would want.

Many grounds may exist to justify implementing the interventions we have identified as less cost effective. For example interventions, such as feeding malnourished infants or management of obstructed labour, target a group in society with particularly poor health. Although this is perfectly legitimate, we argue that decision makers cannot make an informed decision without information on the opportunities to improve population health that are forgone elsewhere. Our results represent the best evidence currently available and show difficult trade-offs may need to be made. Another equally important message from our results is the need to redouble efforts to raise additional funds for health in poor countries. Our experience with economists in ministries other than health is that it is much easier to convince them of the need for funds if both additional and existing funds are well spent. We hope that this series contributes to not only improving population health with the available resources but to raising more funds for health as well.

Supplementary Material

Notes

Members of the WHO-CHOICE millennium development goals team and further details are on bmj.com

We thank Megha Mukim, Jason Lee and Marilyn Vogel for help with referencing.

Contributors and sources: All authors contributed to the development of the methods used in the paper and helped to decide on the implications of the individual results. DBE wrote the first draft. TA, TTTE and SSL made substantial comments and modifications, and DBE prepared the final version and is the guarantor. Members of the WHO-CHOICE MDG group commented on the initial outline and results, as well as providing guidance on the implications of their papers for the final summary. SSL put the results together for tables and figures. The opinions expressed in the paper are those of the authors and do not necessarily represent the views of the organisations they represent

Competing interests: None declared.

References