The ProVac initiative and evolving decision support

Abstract

Many of the papers in this supplement give results of costffectiveness studies. One of the lessons learned from ProVac has een that the most pressing need, particularly in middle and lowerncome countries, is not so much for new methods of capturing ever ore elusive forms of cost or benefit, but for building and strengthning the decision support systems (DSS) within which economic nalyses are, or should be, embedded. The underlying principle is hat supporting decision-making groups with the tools to do their wn analyses is better than global or regional one-size-fits-all guidnce, or contracting out to consultants. The hope is that in the longer erm this approach will benefit public health decision-making more enerally. This is very much in line with WHO initiatives following he Mexico Statement in 2005 [1], which included the following: health policy, public health, and service delivery should be based n reliable evidence derived from high quality research’. One such nitiative is the Evidence Informed Policy Network (EVIPNet [2]) nd the tools linked to it such as SUPPORT [3]. As well as tools, ountry-level structures such as NITAGs [4], IICCs [5] and NRAs 6] have emerged which provide the fora for collaborative and vidence-based decision-making. The ProVac contribution to this has many of the characteristics f a decision support system (DSS) [7]. It provides a structure for athering and evaluating data; and it provides models or analytic echniques which are interactive and easy to use by non-computer eople, and can respond flexibly to changes in the decision-making nvironment. The database element provides country-specific data, much of t quality graded, from a variety of public sources on demograhy, disease burden, GDP per capita, etc.; data on health services osts, health service utilisation, vaccine coverage, vaccine effectiveess, and vaccine programme costs; a facility for producing tables nd graphs; online training materials, guidelines and methods for ata collection; and model documentation. Users are encouraged o review and understand the shortcomings of the data and correct rrors before using them for modelling. They can search interacively and create tables by filtering on the basis of data quality and eography, but cannot specify and run completely new analyses of aw data.