Expected Value of Partial Perfect Information

Concept

---

Theoretically, Expected Value of Partial Perfect Information (EVPPI) is a value of information measure that quantifies the expected benefit of eliminating uncertainty in a selected subset of model parameters while uncertainty in all remaining parameters persists. It is founded on Bayesian decision theory and extends the Expected Value of Perfect Information by identifying the value of resolving uncertainty in specific parameters rather than all uncertainty. In health economics, EVPPI is used to prioritise research by identifying which uncertain parameters contribute most to decision uncertainty.

Mathematically, EVPPI is calculated as the difference between the expected net benefit obtained when perfect information is available for a subset of parameters and the expected net benefit obtained under current information. The calculation involves nested expectations because uncertainty is removed only for the parameter subset of interest while all other parameters remain uncertain.

In practice, EVPPI is estimated using probabilistic sensitivity analysis together with nested Monte Carlo simulation or approximation methods such as Gaussian process regression, regression-based approaches or integrated nested methods. Population EVPPI is obtained by multiplying the per-person EVPPI by the expected number of patients affected during the decision relevance period and is widely used to prioritise future research and optimise study design.

Purpose

---

Used to quantify the value of eliminating uncertainty in specific model parameters, supporting research prioritisation and efficient allocation of research resources in health economic evaluations.

Mathematical Formulae

Primary Formula

EVPPI = E?[max(E�|?(NMB))] ? max(E(NMB))

where:

? = parameter subset of interest

� = remaining uncertain parameters

NMB = Net Monetary Benefit

Supporting Formulae

Net Monetary Benefit:

NMB = ?E ? C

Population EVPPI:

Population EVPPI = EVPPI ? N

where:

N = affected population over the decision relevance period

Relationship:

0 � EVPPI � EVPI

Related Mathematical Methods

• Expected Value of Perfect Information
• Expected Value of Sample Information
• Expected Net Benefit of Sampling
• Bayesian Decision Theory
• Probabilistic Sensitivity Analysis
• Monte Carlo Simulation
• Gaussian Process Regression

Example

---

A health economic model identifies treatment efficacy, utility values and treatment costs as uncertain parameters. The EVPPI for treatment efficacy is estimated as �420 per patient, whereas the EVPPI for treatment costs is �35 per patient. With an expected affected population of 20,000 patients, the population EVPPI for treatment efficacy is �8.4 million, indicating that further research should primarily focus on reducing uncertainty surrounding treatment effectiveness.

Excel Implementation

Function	Example Formula	Health Economics Application
MAX	=MAX(B2:D2)	Identify the maximum net monetary benefit for each simulation.
AVERAGE	=AVERAGE(ResultRange)	Estimate expected net monetary benefit across simulations.
SUMPRODUCT	=EVPPI_per_patient*Population	Calculate the population EVPPI.
IF	=IF(B2>C2,"Research priority","Lower priority")	Compare EVPPI estimates across parameter groups.

VBA (Optional)

---

VBA can automate nested simulation outputs, calculate population EVPPI and compare research priorities across alternative parameter groups.

Sources

• Claxton K. The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies. Journal of Health Economics. 1999;18(3):341?364.
• Strong M, Oakley JE, Brennan A. Estimating multiparameter partial expected value of perfect information from a probabilistic sensitivity analysis sample. Medical Decision Making. 2014;34(3):311?326.
• Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation. Oxford University Press.
• ISPOR Value of Information Good Practice Reports.
• NICE. NICE Health Technology Evaluations: The Manual.