Concept Architecture
Concept
Theoretically, Meta-Regression Transfer is a benefit transfer method that uses a meta-regression model to predict economic values for a policy or healthcare setting using evidence synthesised from multiple primary valuation studies. It is founded on meta-analysis, regression modelling and benefit transfer theory, and exists to improve the transferability of economic values by explicitly accounting for differences in study characteristics, populations and settings.
Mathematically, Meta-Regression Transfer is represented by a regression equation relating observed economic values to explanatory variables describing study design, population characteristics, intervention attributes and contextual factors. The estimated regression coefficients are used to predict values for a new policy or healthcare context by substituting the characteristics of the target setting into the model.
In practice, Meta-Regression Transfer is implemented by systematically identifying relevant valuation studies, estimating a meta-regression model using pooled study data, and applying the resulting prediction equation to the target population or intervention. It is commonly used in health economics and environmental economics when primary valuation studies are unavailable or impractical.
Purpose
Used to transfer economic values between settings, improve the external validity of benefit transfer, estimate values for new populations, and support health economic evaluation when primary valuation data are unavailable.
Mathematical Formulae
Primary Formula
V? = ?? + ????? ??X?
where:
- V? = predicted transferred value
- ?? = intercept
- ?? = estimated regression coefficients
- X? = characteristics of the target study or population
Supporting Formulae
There is no universally recognised canonical mathematical formula.
Related Mathematical Methods
- Meta-analysis
- Meta-regression
- Benefit transfer
- Ordinary least squares regression
- Weighted least squares regression
- Mixed-effects meta-regression
Example
A meta-regression model is estimated using 120 published willingness-to-pay studies for health interventions.
V? = 45.2 + 0.85(Income) ? 12.4(Age) + 18.7(Disease Severity)
For a target population with the specified characteristics, the regression equation predicts the monetary value used in the economic evaluation without conducting a new primary valuation study.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| LINEST | =LINEST(B2:B121,C2:E121,TRUE,TRUE) | Estimates meta-regression coefficients. |
| SUMPRODUCT | =SUMPRODUCT(B1:D1,B2:D2)+A1 | Predicts transferred values using estimated coefficients. |
| TREND | =TREND(B2:B121,C2:E121,C122:E122) | Predicts values for a target setting. |
| RSQ | =RSQ(B2:B121,F2:F121) | Assesses model fit. |
VBA (Optional)
Automate estimation of meta-regression models and generation of transferred economic values for multiple target populations.
Sources
- Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. Oxford University Press.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation. Oxford University Press.
- Johnston RJ, Boyle KJ, Adamowicz W, et al. Contemporary Guidance for Stated Preference Studies. Journal of the Association of Environmental and Resource Economists.
- ISPOR Good Practices for Outcomes Research Task Force Reports.
- NICE. Health Technology Evaluation Manual.
Related Concepts (2)
Library
Publications
1
NICE DSU Technical Support Document 4: Inconsistency in Networks of Evidence Based on Randomised Controlled Trials — Dias, Welton, Sutton, Caldwell, Lu & Ades, TSD 4 ed., 2011 (NICE Decision Support Unit (University of Sheffield))
Guidance on assessing and handling inconsistency — conflict between direct and indirect evidence — in network meta-analysis, a key validity check for mixed treatment comparisons.
Frequently Asked Questions (6)
What is meta-regression transfer?
A form of benefit transfer in which a regression model estimated across many original studies predicts a value for a new context based on its characteristics.
Source: Nelson & Kennedy 2009
How does meta-regression transfer work?
Values reported by many original studies become the observations in a regression, with the characteristics of each study and its context as explanatory variables. Applying the characteristics of the new context to the estimated relation predicts a value for it. The approach draws on the whole literature rather than a single study. Because each observation is a published result rather than an individual response, the model is estimating a relation across studies rather than across people.
Source: Nelson & Kennedy 2009
What advantage does meta-regression transfer have?
It uses far more evidence than transferring from one study, and it adjusts explicitly for the ways the new context differs on the variables included. Where the underlying studies are numerous and reasonably consistent, the prediction is more stable than any single estimate would be. It also produces an estimate for contexts where no primary study exists, which is the situation the method was developed to address.
Source: Johnston & Rosenberger 2010
What weakens meta-regression transfer?
The model must control for differences in method between the contributing studies as well as differences in context, and it rarely controls for them fully. Published estimates are also a selected sample, since studies finding low or implausible values are less likely to appear, which biases the relation the regression estimates. Heterogeneity in how the original studies were conducted therefore enters the prediction as unexplained variation, and the reported precision of the model understates the real uncertainty.
Source: Nelson & Kennedy 2009
How accurate is meta-regression transfer?
Tests transferring to contexts where a primary estimate also exists report errors that are frequently substantial, and the approach does not consistently outperform simpler transfers. Accuracy improves where the new context lies inside the range of the contributing studies and deteriorates where it lies outside. Reporting the range of predictions across plausible model specifications is more informative than the confidence interval from any single one.
Source: Johnston & Rosenberger 2010
What should a meta-regression transfer report?
The studies included and the criteria used to select them, the model specification with its coefficients, the values of the explanatory variables at the new site, and a sensitivity range reflecting transfer error rather than the standard errors of the regression alone. The studies excluded and the reasons for excluding them should be recorded as well, since selection into the sample affects the estimated relation directly. Where the new context lies outside the range of the contributing studies, that should be stated plainly, since the prediction is then an extrapolation rather than an interpolation.
Source: Boyle & Bergstrom 1992
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 1 Aug 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-EE-CBA-033
Stable URI · Machine-readable · Resolvable · CC BY 4.0