Value of Information Application
Conducting a Study is Not Recommended if Decision Makers Plan to Implement the Intervention for 5 Years or Less.
As it would take more than 5 years to accrue enough health benefits to outweigh the cost of the study, conducting further research is not recommended if decision makers plan to implement the intervention for 5 years or less.
Check 10-year and 20-year results.
- Optimal sample size without budget contraints
- Optimal sample size for your study budget
Conducting a Study is Not Recommended Based on Current Assumptions
Based on the assumptions you entered, the expected net benefit regardless of sample size is zero or negative (blue curve) indicating that the cost of conducting the study outweighs the benefits of the additional information produced by the study. The benefits are small because the added information would not change the current decision to adopt or not adopt the intervention and therefore further investments in additional research is not worthwhile. This is reflected in the flat “expected value of sample information” curve.
You can find the circumstances in which conducting additional research could have added value by changing your assumptions:
- Willingness to pay threshold: Select a value within the recommended range. VOI is sensitive to the decision maker’s willingness to pay for 1 additional QALY as it is the comparator against which the probability of a strategy’s “cost-effectiveness” is determined.
- Cost of sampling: Reduce the cost of sampling. Entertain the possibility of using less costly approaches.
- Target population: Increase target population. In general, the intervention has greater value when expanded to include more eligible people (i.e., # of eligible patients nationally).
Optimal sample Size for Your Study Budget =
Optimal Sample Size without Budget Constraints =
The optimal sample size for your study without budgetary constraints is , and is the size at which maximum health gains can be achieved for your investment. For your study budget of thousand per year the ideal sample size is participants. As shown by the “expected net benefit for sample size” blue curve, conducting a study with participants is considered beneficial as the health benefit potentially gained is greater than the cost of the study. However, increasing your sample size toward is recommended as additional gains in health may be realized. For example, an increase in sample size of to would translate to an incremental increase in expected net benefit of (). Net monetary benefit is the net gain in health expressed in monetary units and is calculated by multiplying quality adjusted lifer years (QALY) by the decision maker’s willingness to pay per additional QALY.
To provide some context, a power calculation estimates a required sample size of 4710 participants for a one tailed logistic regression to detect an odds ratio of 1.49 (increase in abstinence) assuming an error probability of 5%, an event probability in the control arm of 5%, and no covariates. With a sample size of 4710 the trial would still be considered beneficial, but increasing the sample size toward the optimal sample size would increase net health benefit (gain more health).
To provide some context, a power calculation also estimates a required sample size of 4710 participants for a one tailed logistic regression to detect an odds ratio of 1.49 (increase in abstinence) assuming an error probability of 5%, an event probability in the control arm of 5%, and no covariates. The calculated sample size using both methods are equal and represents an ideal situation.
To provide some context, a power calculation estimates a required sample size of 4710 participants for a one tailed logistic regression to detect an odds ratio of 1.49 (increase in abstinence) assuming an error probability of 5%, an event probability in the control arm of 5%, and no covariates. If ENBS is a positive value at a sample size of 4710, we expect the maximum achievable health benefit could be attained and pursuing the study is of value. If ENBS is zero or a negative at a sample size of 4710, conducting research is not recommended as the return on investment would not be realized.
To provide some context, a power calculation estimates a required sample size of 2647 participants for a one tailed robust Poisson regression analysis to detect a relative risk of 0.78 (decrease in loss to follow up) assuming an error probability of 5%, a baseline event probability of 30%, and no covariates. With a sample size of 2647 the trial would still be considered beneficial, but increasing the sample size toward the optimal sample size would increase net health benefit (gain more health).
To provide some context, a power calculation also estimates a required sample size of 2647 participants for a one tailed robust Poisson regression analysis to detect a relative risk of 0.78 (decrease in loss to follow up) assuming an error probability of 5%, a baseline event probability of 30%, and no covariates. The calculated sample size using both methods are equal and represents an ideal situation.
To provide some context, a power calculation estimates a required sample size of 2647 participants for a one tailed robust Poisson regression analysis to detect a relative risk of 0.78 (decrease in loss to follow up) assuming an error probability of 5%, a baseline event probability of 30%, and no covariates. If ENBS is a positive value at a sample size of 2647, we expect the maximum achievable health benefit could be attained and pursuing the study is of value. If ENBS is zero or a negative at a sample size of 2647, conducting research is not recommended as the return on investment would not be realized.
To provide some context, a power calculation estimates a required sample size of 1915 participants for a one tailed logistic regression analysis to detect an odds ratio of 1.8 (increase in successful tracing) assuming an error probability of 5%, an event probability of 5%, and no covariates. With a sample size of 1915 the trial would still be considered beneficial, but increasing the sample size toward the optimal sample size would increase net health benefit (gain more health).
To provide some context, a power calculation also estimates a required sample size of 1915 participants for a one tailed logistic regression analysis to detect an odds ratio of 1.8 (increase in successful tracing) assuming an error probability of 5%, an event probability of 5%, and no covariates. The calculated sample size using both methods are equal and represents an ideal situation.
To provide some context, a power calculation estimates a required sample size of 1915 participants for a one tailed logistic regression analysis to detect an odds ratio of 1.8 (increase in successful tracing) assuming an error probability of 5%, an event probability of 5%, and no covariates. If ENBS is a positive value at a sample size of 1915, we expect the maximum achievable health benefit could be attained and pursuing the study is of value. If ENBS is zero or a negative at a sample size of 1915, conducting research is not recommended as the return on investment would not be realized.