Concept Architecture
Concept
Theoretically, Simulated Treatment Comparison (STC) is a population-adjusted indirect comparison method used to estimate comparative treatment effects when head-to-head clinical trials are unavailable and individual patient data (IPD) are available for one treatment while only aggregate data are available for the comparator. It is based on regression modelling and causal inference, using patient-level outcome models to predict treatment effects for a target population defined by published aggregate characteristics. The method exists to reduce bias arising from differences in baseline characteristics between independent clinical trials.
Mathematically, STC is represented using regression models that relate patient outcomes to baseline covariates, treatment and treatment-covariate interactions. The fitted model is then used to predict outcomes for a population with covariate distributions matching those reported in the comparator trial. Comparative treatment effects are obtained by contrasting the predicted outcomes with the published aggregate results for the comparator.
In practice, STC is implemented by identifying prognostic variables and treatment effect modifiers, estimating an outcome regression model using individual patient data and simulating outcomes for the target population. In health economics it is widely applied in health technology assessment, comparative effectiveness research and reimbursement submissions when direct comparative evidence is unavailable.
Purpose
Used to adjust for cross-trial differences in patient characteristics and estimate comparative treatment effects for economic evaluation and health technology assessment when direct comparative trials are unavailable.
Mathematical Formulae
Primary Formula
E(Y|X) = ?? + ??T + ??X + ??(T ? X)
where:
Y = clinical outcome
T = treatment indicator
X = vector of baseline covariates
? = estimated regression coefficients
Supporting Formulae
Predicted outcome for the target population:
? = E[E(Y|X)]
Treatment effect:
? = ?? ? ??
For time-to-event outcomes:
h(t|X) = h?(t)exp(??X)
Related Mathematical Methods
Outcome Regression
Generalised Linear Models
Cox Proportional Hazards Model
Matching-Adjusted Indirect Comparison
Network Meta-Analysis
Anchored Indirect Comparison
Unanchored Indirect Comparison
Example
A clinical trial of Treatment A includes individual patient data for 420 patients, whereas only published aggregate results are available for Treatment B. A regression model estimates overall survival as a function of age, disease severity, performance status and treatment. The fitted model is used to predict survival for patients with baseline characteristics matching those reported in the Treatment B trial. The simulated population yields an estimated hazard ratio of 0.82 for Treatment A versus Treatment B, which is subsequently incorporated into a cost-effectiveness model.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| LINEST | =LINEST(B2:B201,C2:F201,TRUE,TRUE) | Estimate regression coefficients for continuous outcomes |
| LOGEST | =LOGEST(B2:B201,C2:F201,TRUE,TRUE) | Estimate coefficients for exponential outcome models |
| SUMPRODUCT | =SUMPRODUCT(coefficients,covariates) | Calculate predicted outcomes for target populations |
| EXP | =EXP(A2) | Estimate predicted hazards or probabilities |
| AVERAGE | =AVERAGE(predicted_range) | Calculate simulated population outcomes |
VBA (Optional)
VBA can automate regression estimation, population simulation, outcome prediction and comparative treatment analyses across multiple target populations.
Sources
- Phillippo DM, Ades AE, Dias S, et al. Methods for Population-Adjusted Indirect Comparisons in Health Technology Appraisal.
- NICE. Health Technology Evaluations: The Manual.
- ISPOR Good Practice Reports for Population-Adjusted Indirect Comparisons.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
- Signorovitch JE, et al. Matching-Adjusted Indirect Comparisons and Related Population Adjustment Methods.
Related Concepts (2)
Library
Publications
1
Interpreting Indirect Treatment Comparisons and Network Meta-Analysis for Health-Care Decision Making: ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices, Part 1 — Jansen, Fleurence, Devine, Itzler, Barrett, Hawkins, Lee, Boersma, Annemans & Cappelleri, Vol. 14, No. 4 ed., 2011 (Value in Health)
The ISPOR good-practice guidance on interpreting indirect treatment comparisons, network and mixed treatment comparisons for decision making — terminology, assumptions, validity and how to critically appraise an ITC/NMA when head-to-head trial evidence is unavailable.
Journal ArticleView source →
Frequently Asked Questions (6)
What is a simulated treatment comparison?
A method comparing treatments from separate trials by fitting a regression model to one trial's data to predict outcomes for the other's population.
Source: Caro & Ishak 2010
How does a simulated treatment comparison predict outcomes for another trial's population?
A simulated treatment comparison uses individual patient data from one trial to fit a regression model relating patient characteristics to outcome, then applies that model to the characteristics of the other trial's population to predict how its patients would have fared on the first treatment. Comparing this prediction against the other trial's observed result estimates the relative effect, adjusting for differences in the populations. Unlike reweighting, it works by outcome prediction, which can extend beyond the range of the observed data at some risk. Modelling links the two trials. Ishak and colleagues (2015) describe this.
Source: Ishak et al. 2015
How does a simulated treatment comparison work?
A simulated treatment comparison works by fitting a regression model relating outcomes to patient characteristics using the individual patient data from one trial, then using this model to predict the outcome for the characteristics of the comparator trial's population, effectively simulating how the first treatment would perform in that population. The predicted outcome is compared with the comparator trial's reported result to estimate the relative effect adjusted for population differences. This model-based prediction adjusts for the differences in characteristics between the trials, aiming for a more comparable indirect comparison than an unadjusted one.
Source: Caro & Ishak 2010
How does a simulated treatment comparison differ from a matching-adjusted indirect comparison?
A simulated treatment comparison uses a regression model fitted to one trial's individual patient data to predict outcomes for the other trial's population, whereas a matching-adjusted indirect comparison reweights one trial's patients to match the other's characteristics without an outcome model. So STC relies on correctly specifying the outcome model, while MAIC relies on reweighting to balance characteristics. Both adjust for population differences in indirect comparisons using individual patient data for one treatment, but they differ in approach, model-based prediction versus reweighting, and each rests on its own assumptions about the adjustment.
Source: Signorovitch et al. 2010
When is a simulated treatment comparison used?
A simulated treatment comparison is used when treatments must be compared indirectly, direct head-to-head evidence is unavailable, and the trials differ in their populations, provided individual patient data are available for one treatment and summary results for the other. It is applied in health technology assessment to adjust for population differences when comparing a new treatment with a competitor. So STC, like MAIC, is used to improve the comparability of indirect comparisons through population adjustment, chosen when a model-based prediction approach is preferred and the necessary individual patient data are available for one of the treatments.
Source: Caro & Ishak 2010
What are the limitations of a simulated treatment comparison?
The limitations of a simulated treatment comparison include its reliance on correctly specifying the outcome regression model, since a misspecified model biases the predictions; its ability to adjust only for measured characteristics, leaving unobserved differences as a source of bias; and the assumptions needed to extrapolate the model to the comparator population. Its validity depends on including the relevant prognostic and effect-modifying variables. So STC improves on unadjusted indirect comparisons but remains subject to model dependence and residual bias from unmeasured differences, and its results are interpreted with these limitations in mind.
Source: Caro & Ishak 2010
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 20 Nov 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-ES-CER-025
Stable URI · Machine-readable · Resolvable · CC BY 4.0