Healthcare decisions are complex and involve trade-offs between multiple, often conflicting objectives. The aim of the study is to describe the inductive (empirical) and deductive (theoretical) methods applied to develop a comprehensive theoretical framework to measure the latent construct of value in a unidimensional measure.

Based on the literature on value assessment frameworks (VAF), multi-criteria decision analysis (MCDA), composite indices (CI), a generic framework was developed and used to discuss the basis steps for constructing a value assessment framework. Methodological principles of VAF, MCDA and CI were applied: (1) defining value in VAF currently discussed; (2) describing the properties and scope of MCDA and CI, (3) identifying basic steps for constructing a value assessment framework and (4) guiding decisions on its components.

A VAF was constructed to assist decision makers with a composite value indicator (CVI). CVIs assess the performance of complex phenomena that are not directly measurable. The concept of the framework consists of 7 sequential steps: 1) Definition of the decision context (of the phenomenon of value), 2) Identification of performance indicators, 3) Performance measurement of all relevant indicators, 4) Scoring of indicators, 5) Weighting of normalized effects, 6) Aggregation of weighted effects, 7) Interpretation of results. The proposed generic model describes a general, structured approach of value assessment based on MCDA procedures which can be adapted to specific circumstances of a concrete choice situation, to specific objectives or stakeholders.

Methods of value measurement derived from MCDA and CI literature offer frameworks to assess the value of health care products and services. However, especially the description and application of scoring and weighting showed large gaps in existing value assessment frameworks. In addition, it remains unclear how patient preferences and costs are included and interpreted. A uniform concept of a value assessment framework and a universal definition is needed.