Purpose: To investigate the interpretation of the expected value of partial perfect information (EVPPI) and the research decisions it can be used to address.

Methods: Value of information (VOI) analysis can be used to quantify the expected gain in net benefit that can be made by obtaining further information to inform a decision. The first necessary condition is that the expected value of perfect information (EVPI) for the whole decision problem exceed the costs of performing additional research. Subsequently EVPPI can be used to identify: (i) parameters (or groups of parameters) that contribute most to the EVPI for the whole decision problem and that might be worthy of further research; (ii) parameters with no EVPPI that may be disregarded as potential targets for further research. Recently it has been noted that parameters with zero EVPPI for a one-off research design may be associated with high EVPPI when considered as part of a sequential research design. This paper examines the characteristics and role of conditional and sequential EVPPI in VOI analysis.

Results: A simple two parameter decision problem is used to illustrate the calculation of EVPI, EVPPI for single parameters, joint EVPPI for groups of parameters, conditional EVPPI and sequential EVPPI. Conditional EVPPI is defined as the expected value of perfect information about parameter two, conditional on having obtained perfect information about parameter one. Sequential EVPPI is defined as the expected value of perfect information for a sequential research design to investigate first parameter one, then parameter two. Conditional EVPPI for a parameter will not equal the EVPPI for that parameter alone due to differences in the current evidence position. The components of sequential EVPPI are taken from the joint EVPPI for the parameters (equal to the EVPI for a two parameter model) and the EVPPI associated with the first parameter to be investigated. A benefit of a sequential design is the option to forgo research on the second parameter on the basis of additional information obtained on the first parameter.

Conclusions: Published VOI analyses have focussed on a limited research decision space about one-off research designs to inform single parameters or groups of parameters. The research decision space can be widened by incorporating sequential EVPPI into future VOI analyses.