Aim 3: Determine which of the identified ADRD interventions should be tested in additional trials for the population at-large and by race and ethnicity to inform nondrug ADRD research prioritization using value of information methodology.
- Value of information analysis
- We will use a Gaussian approximation and a meta-modeling approach to conduct the VOI analysis (developed by CO-I Dr. Jalal).
- We will generate five outcomes from our VOI analysis to inform nondrug ADRD research prioritization.
- First, we will calculate the expected value of perfect information (EVPI), which is the maximum amount that should be spent to eliminate uncertainty in the ADRD-MM.
- Second, we will calculate the expected value of population partial perfect information (EVPPI), which quantifies the value of eliminating uncertainty for specific parameters (e.g., effectiveness of an intervention).
- Third, we will calculate the expected value of sample information (EVSI), which represents the amount of uncertainty reduced from the hypothesized data collection effort (e.g., a new RCT of COPE) with a finite sample size (n).
- Fourth and fifth, we will calculate the expected net benefit of sampling (ENBS) and the optimal sample size of the hypothesized data collection efforts, respectively. ENBS provides an estimate of the marginal benefit of acquiring information from the hypothesized research design given population EVSI and the cost of conducting the study. We will engage the expert advisors and consult the literature to identify the cost of conducting research.
- Sensitivity analyses
- We will conduct sensitivity analyses in which we will evaluate the optimal sample size for each proposed study design by varying the current prevalent and future incident cases by ±50%, varying the cost of research by ±50%, and evaluating a 5- and 15-year decision lifetime.