13 Research methods, statistics, critical review and evidence-based practice13.6 Economic analysis

13.6.9 Sensitivity analysis in economic evaluation

1 min read1 figure

Written by Resident Doctor Committee · Published

What sensitivity analysis does

  • Sensitivity analysis tests how the result of an economic model, such as the incremental cost-effectiveness ratio (ICER), changes when the model's inputs or assumptions vary.

  • A utility-elicitation method such as the standard gamble produces an input to the model. It is not a sensitivity analysis.

Types of analysis

Type

What it does

One-way

Varies one input while holding the others at base-case values; a tornado diagram ranks inputs by their effect

Multi-way

Varies several inputs together

Scenario

Tests coherent alternatives, such as lower effectiveness with earlier waning

Threshold

Finds the input value (switching value) at which the preferred decision changes

Probabilistic

Assigns distributions to inputs, draws values repeatedly and calculates costs and effects for each draw

In this tornado diagram from one-way sensitivity analysis, each input is changed by twenty per cent while the others stay at base case, and the width of each bar shows how far the modelled outcome swings, so the widest bar marks the input the result depends on most. Phernandez2000, CC BY-SA 4.0, via Wikimedia Commons

Cost-effectiveness acceptability curve

  • A cost-effectiveness acceptability curve (CEAC) comes from probabilistic analysis and shows the uncertainty around the ICER across a range of willingness-to-pay thresholds.

  • The horizontal axis is the willingness to pay per unit of health gain, or ceiling ratio.

  • The vertical axis is the probability that the intervention is cost-effective, from 0 to 1, and it rises as the threshold increases.

  • Each point is the proportion of draws in which the option is cost-effective at that threshold.

Reading a CEAC

  1. Find the stated willingness-to-pay ceiling on the horizontal axis.

  2. Read the height of the curve on the vertical axis: 0.75 means a 75% probability of being cost-effective.

  3. Compare curves at that same ceiling.

What a CEAC does not show

  • It does not give the ICER, its confidence limits or the cost-effectiveness plane.

  • Its probability is not clinical effectiveness, a proportion of patients or the chance that the treatment works. It depends on the model and its input distributions.

Revision material for the MRCPsych examinations. Not clinical guidance, and not for use in patient care.