Probabilistic Sensitivity Analysis (PSA)
At TreeAge Software, we design powerful analytical tools that help researchers, health economists, analysts, and decision scientists make better decisions when faced with uncertainty. One of the core methods we provide for understanding uncertainty in models is Probabilistic Sensitivity Analysis (PSA).
What is Probabilistic Sensitivity Analysis?
Probabilistic Sensitivity Analysis is a structured technique for evaluating how combined uncertainty in your model’s inputs affects its outcomes and conclusions. Instead of analyzing each uncertain parameter one at a time, PSA assesses all relevant uncertainties simultaneously by representing them as probability distributions and repeatedly sampling from them.
In practical terms:
- Input uncertainty is expressed using statistical distributions (e.g., beta, normal, lognormal) for model parameters such as probabilities, costs, utilities, or transition rates.
- The model is then recalculated many times, each time with a different set of randomly sampled values from those distributions.
- The result is a Monte Carlo simulation of your model across a plausible range of parameter combinations, generating a distribution of outcomes rather than a single point estimate.
This approach to sensitivity analysis gives you a clearer picture of how uncertainty influences both model outputs and your confidence in base-case conclusions.
Why PSA Matters in Decision Modeling
In real-world decision analysis (whether in health economics, clinical decision making, or business strategy), input parameters are seldom known with certainty. Probabilistic Sensitivity Analysis helps you answer critical questions such as:
- How robust are my conclusions to uncertainty in key assumptions?
- What is the probability that one strategy is optimal given the uncertainty in inputs?
- What level of confidence can I place in cost-effectiveness results or other decision outcomes?
By examining outcomes across many sets of sampled parameters, PSA quantifies overall uncertainty and supports more informed decisions under uncertainty.
How PSA Works in TreeAge Pro
TreeAge Pro implements PSA as part of its Monte Carlo simulation capabilities:
- Define Parameter Distributions
You assign probability distributions to any uncertain inputs (e.g., transition probabilities, costs, utilities). These distributions reflect the best available evidence about uncertainty. - Run PSA (Monte Carlo Sampling)
TreeAge Pro automatically draws samples from all defined distributions and recalculates your model for each sampled set. You typically specify how many iterations (samples) to run—for example, 1,000 or more—to ensure stable results. - Analyze Results
After the simulation, a suite of reports and visualizations helps you interpret PSA results, including:- Cost-effectiveness scatterplots
- Acceptability curves showing the probability each strategy is cost-effective across willingness-to-pay thresholds
- Incremental cost-effectiveness histograms
- Probability distributions for outcomes and strategy selection frequencies
These outputs show not just average results but the range and likelihood of possible outcomes given uncertainty in your model parameters.
PSA Across Model Types
TreeAge Pro supports PSA across a range of decision models, including:
- Decision trees
- Markov models
- Partitioned survival models
- Microsimulation and patient-level simulation models
- Discrete event simulation models
For models requiring individual patient simulation, PSA can be combined with microsimulation so that both patient-level and parameter uncertainty are captured concurrently, allowing a comprehensive uncertainty analysis.
Interpreting PSA Results
Once the simulation is complete, TreeAge Pro’s outputs help you make transparent, uncertainty-aware decisions:
- The percent of iterations confirming base case conclusions gives a measure of confidence in your chosen strategy.
- Acceptability curves show the probability each strategy is optimal across different cost-effectiveness thresholds.
- Scatterplots and histograms provide visual insights into the variability and correlation of costs and outcomes.
Together, these outputs help stakeholders understand not just what decision is preferred in the average case, but how confident that preferred decision is given uncertainty in model inputs.
Try It Now
Probabilistic Sensitivity Analysis is essential for sound decision science. It enables modelers to:
- Quantify the impact of input uncertainty on results
- Assess confidence in decision recommendations
- Communicate uncertainty clearly through visual and statistical outputs
- Support evidence-based policy, clinical, and business decisions
PSA is a cornerstone of robust decision modeling and is fully integrated into TreeAge Pro’s analytical ecosystem to give users a rigorous and practical approach to uncertainty analysis. Discover a free trial of our software now and implement Probabilistic Sensitivity Analysis in your workspace.
Contact Us
For Sales:
- Send an email to sales@treeage.com.
- Call 413-458-0104, then 1 for Sales.
For Technical Support:
- Submit a support ticket to our help desk system.
- Call our support line at 413-458-0104, then 2 for Support.
- Search our Knowledge Base.
For Training:
- Send an email to training@treeage.com.
- Call 413-458-0104 x802.