Concept Architecture
Concept
Theoretically, Structural Sensitivity Analysis is a form of sensitivity analysis that evaluates the impact of alternative model structures, structural assumptions and modelling approaches on the results of a health economic evaluation. Unlike parameter sensitivity analysis, which varies numerical inputs, structural sensitivity analysis investigates uncertainty arising from the way the model is formulated. In health economics, it is used to assess whether conclusions remain robust when plausible alternative representations of disease processes, treatment pathways or decision frameworks are considered.
Mathematically, structural sensitivity analysis compares the outputs of alternative mathematical model structures rather than modifying individual parameter values within a single model. Each structural specification represents a different mathematical formulation of the same decision problem, and differences in predicted costs, health outcomes and cost-effectiveness quantify the effect of structural uncertainty.
In practice, structural sensitivity analysis is implemented by developing alternative model structures or modifying key modelling assumptions, such as time horizon, cycle length, health state definitions, survival extrapolation methods or disease progression pathways. Results from each structural specification are compared with the base-case analysis to determine whether policy recommendations remain consistent across plausible model structures.
Purpose
Used to evaluate the influence of alternative model structures and structural assumptions on health economic conclusions and assess the robustness of decision-making under structural uncertainty.
Mathematical Formulae
Primary Formula
There is no universally recognised canonical mathematical formula.
Supporting Formulae
Incremental Cost-Effectiveness Ratio:
ICER = ?C / ?E
Net Monetary Benefit:
NMB = ?E ? C
Incremental difference between structural models:
?Y = Y? ? Y?
where:
Y? = outcome from an alternative model structure
Y? = outcome from the base-case model
Related Mathematical Methods
- Structural Uncertainty
- Scenario Analysis
- Sensitivity Analysis
- Threshold Analysis
- Markov Model
- Partitioned Survival Model
- Decision Tree
Example
A health technology assessment compares a Markov model with a partitioned survival model using identical clinical evidence. The Markov model estimates an ICER of �24,600 per QALY, whereas the partitioned survival model estimates �29,100 per QALY. Structural sensitivity analysis demonstrates that the modelling framework itself influences the estimated cost-effectiveness of the intervention.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| CHOOSE | =CHOOSE(A2,Model1,Model2,Model3) | Select alternative structural model specifications. |
| INDEX | =INDEX(ResultRange,ModelNumber) | Retrieve results from different structural scenarios. |
| MATCH | =MATCH("Partitioned Survival",ModelList,0) | Identify the required model structure for comparison. |
| IF | =IF(B2<=30000,"Cost-effective","Not cost-effective") | Compare cost-effectiveness conclusions across structural models. |
VBA (Optional)
VBA can automate execution of alternative model structures, compare their outputs and generate summary reports describing the impact of structural assumptions on cost-effectiveness.
Sources
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation. Oxford University Press.
- Philips Z, Ginnelly L, Sculpher M, et al. Review of guidelines for good practice in decision-analytic modelling in health technology assessment. Health Technology Assessment. 2004;8(36).
- NICE. NICE Health Technology Evaluations: The Manual.
- ISPOR-SMDM Modeling Good Research Practices Task Force Reports.
Related Concepts (2)
Library
Publications
1
Parameter Estimation and Uncertainty: A Report of the ISPOR-SMDM Modeling Good Research Practices Task Force-6 — Briggs, Weinstein, Fenwick, Karnon, Sculpher & Paltiel, Task Force Report 6 ed., 2012 (Value in Health / Medical Decision Making)
Best-practice guidance on parameter estimation and the characterisation of uncertainty in decision models, covering probabilistic sensitivity analysis, distributional choices, and correlation between parameters.
Journal ArticleView source →
Frequently Asked Questions (6)
What is structural sensitivity analysis?
A sensitivity analysis testing how results change under alternative structural assumptions, such as a different survival extrapolation function, rather than parameter values.
Source: Bojke et al. 2009
Why is structural sensitivity analysis harder than parameter sensitivity analysis?
Testing sensitivity to a parameter means changing a number, but testing sensitivity to structure means building and running a genuinely different model, with alternative health states, a different survival extrapolation, or another way of representing the disease. This is far more work, and there is often no natural range to vary over, only a handful of discrete alternatives to compare. Because whole models must be reconstructed, structural sensitivity is explored less often than parameter sensitivity, though it can matter more. It examines the model's shape, not its numbers. Bojke and colleagues (2009) discuss this.
Source: Bojke et al. 2009
How is structural sensitivity analysis conducted?
Structural sensitivity analysis is conducted by identifying the structural assumptions that are uncertain, such as the extrapolation function, the model type, or which states to include, specifying plausible alternatives, and running the model under each to compare the results. Each alternative structure is implemented coherently with its appropriate inputs. Comparing the outputs reveals how sensitive the conclusions are to the structural choices. Where alternatives are considered jointly, model averaging can combine them, weighting each by its support, rather than relying on a single structure.
Source: Briggs, Claxton & Sculpher 2006
Why is structural sensitivity analysis important?
Structural sensitivity analysis is important because structural choices, such as the survival extrapolation or the states included, can affect results as much as parameter values, yet parameter sensitivity analysis leaves them unexamined, so ignoring structural uncertainty could give a false sense of robustness. Testing alternative structures reveals whether conclusions depend on these choices, informing decision makers of an uncertainty that would otherwise be hidden. This matters particularly where the structure strongly influences results, as in extrapolation beyond the observed data.
Source: Bojke et al. 2009
What structural assumptions are commonly tested?
Structural assumptions commonly tested include the choice of survival extrapolation function, which can strongly affect long-term and mean survival; the set of health states and transitions in the model; the model type, such as a cohort versus an individual-level approach; assumptions about treatment effect duration and waning; and how costs and utilities are attached to states. Each is a structural rather than a parameter choice. Testing plausible alternatives for these shows how much the results depend on the way the model is constructed.
Source: Bojke et al. 2009
What are the limitations of structural sensitivity analysis?
Structural sensitivity analysis examines only the alternative structures considered, so important structural possibilities may be omitted, and it usually attaches no probabilities to the alternatives, giving no likelihood for each. Implementing several coherent alternative structures can be demanding, and the choice of which to test involves judgement. These limitations mean structural sensitivity analysis is used to illustrate the effect of key structural choices, with the alternatives chosen carefully, and model averaging considered where the aim is to combine rather than merely compare competing structures.
Source: Briggs, Claxton & Sculpher 2006
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 30 Oct 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-EM-UA-072
Stable URI · Machine-readable · Resolvable · CC BY 4.0