Concept Architecture
Concept
Theoretically, Decision Curve Analysis (DCA) is a decision-analytic method for evaluating the clinical usefulness of prediction models, diagnostic tests or biomarkers by quantifying their net benefit across a range of decision thresholds. It was developed to determine whether using a predictive model improves decision-making compared with treating all individuals or treating none. In health economics, Decision Curve Analysis supports the assessment of diagnostic strategies and risk prediction tools by incorporating the consequences of false-positive and false-negative decisions into treatment recommendations.
Mathematically, Decision Curve Analysis is based on the calculation of net benefit, which combines the proportion of true positives with a weighted penalty for false positives determined by the selected threshold probability. The threshold probability reflects the relative value placed on the benefits of treatment compared with its potential harms. Net benefit is calculated across multiple threshold probabilities to produce a decision curve that enables comparison of alternative decision strategies.
In practice, Decision Curve Analysis is performed after developing or validating a prediction model. For each threshold probability, the net benefit is calculated and plotted against competing strategies such as treat-all, treat-none or alternative prediction models. In health economics, these analyses are frequently used alongside cost-effectiveness studies to determine whether implementing a prediction model leads to better healthcare decisions and more efficient allocation of healthcare resources.
Purpose
Used to evaluate the clinical utility of prediction models and diagnostic strategies by quantifying the net benefit of treatment decisions across a range of clinically relevant decision thresholds.
Mathematical Formulae
Primary Formula
NB = (TP / N) ? (FP / N) ? (p? / (1 ? p?))
where:
- NB = net benefit
- TP = number of true positives
- FP = number of false positives
- N = total sample size
- p? = threshold probability
Supporting Formulae
Weight = p? / (1 ? p?)
Net Reduction = (NBmodel ? NBtreat-all) ? ((1 ? p?) / p?)
Related Mathematical Methods
- Decision Curve Analysis
- Net benefit analysis
- Receiver operating characteristic (ROC) analysis
- Calibration analysis
- Logistic regression
- Prediction modelling
Example
A cardiovascular risk model is evaluated using a treatment threshold probability of 20%. Among 1,000 patients, the model identifies 180 true positives and 40 false positives.
NB = (180 / 1000) ? (40 / 1000) ? (0.20 / 0.80)
NB = 0.180 ? 0.010
NB = 0.170
The prediction model therefore provides a net benefit of 0.170, indicating that its use results in more beneficial treatment decisions than treating all patients or treating none at the chosen threshold.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| COUNTIFS | =COUNTIFS(Predicted,1,Observed,1) | Counts true positives |
| COUNTIFS | =COUNTIFS(Predicted,1,Observed,0) | Counts false positives |
| COUNTA | =COUNTA(Observed) | Calculates total sample size |
| Formula | =(TP/N)-((FP/N)*(Threshold/(1-Threshold))) | Calculates net benefit for Decision Curve Analysis |
| Scatter Chart | Plot Threshold Probability vs Net Benefit | Produces the decision curve for comparing prediction strategies |
VBA (Optional)
VBA can automate calculation of net benefit across multiple threshold probabilities and generate decision curves for competing prediction models.
Sources
- Vickers AJ, Elkin EB. Decision Curve Analysis: A Novel Method for Evaluating Prediction Models. Medical Decision Making. 2006.
- Vickers AJ, van Calster B, Steyerberg EW. Net Benefit Approaches to the Evaluation of Prediction Models, Molecular Markers and Diagnostic Tests. BMJ.
- Steyerberg EW. Clinical Prediction Models.
- Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
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 decision curve analysis?
A method evaluating a diagnostic or prognostic model's clinical value by calculating its net benefit across a range of possible intervention thresholds.
Source: Vickers & Elkin 2006
What does decision curve analysis add beyond accuracy measures?
Standard measures of a prediction model, such as how well it discriminates, say how accurate it is but not whether using it does more good than harm in practice. Decision curve analysis fills this gap by computing the model's net benefit, weighing correct detections against the cost of unnecessary interventions, across the range of risk thresholds at which a clinician might act. This shows whether, and over what range of thresholds, using the model beats simply treating everyone or no one. It judges usefulness, not just accuracy. Vickers and Elkin (2006) developed this method.
Source: Vickers & Elkin 2006
How does decision curve analysis work?
Decision curve analysis works by calculating, across a range of threshold probabilities at which one would opt to intervene, the net benefit of using a model to guide treatment, where net benefit weighs the true positives gained against the false positives incurred, using the threshold to set their relative value. The net benefit of the model is compared with that of default strategies, treating all or treating none. Plotting net benefit against the threshold probability produces the decision curve, showing over what range of thresholds using the model is beneficial. This links the model to clinical decisions.
Source: Vickers & Elkin 2006
What is net benefit in decision curve analysis?
Net benefit in decision curve analysis is a measure combining the benefits of correctly identifying and treating those who need it, the true positives, with the harms of unnecessarily treating those who do not, the false positives, weighted according to the threshold probability at which one would choose to intervene. The threshold reflects how the decision maker values avoiding a missed case relative to avoiding an unnecessary treatment. Net benefit thus places benefits and harms on a common scale defined by the decision context, allowing strategies to be compared by their clinical value rather than statistical accuracy alone.
Source: Weinstein & Fineberg 1980
Why is decision curve analysis useful?
Decision curve analysis is useful because it evaluates a model by its practical value for clinical decisions, showing whether using it to guide treatment yields greater net benefit than simple default strategies of treating everyone or no one, across the range of thresholds reflecting different preferences. Unlike measures of discrimination or calibration alone, it incorporates the consequences of decisions, so it directly addresses whether a model helps in practice. This makes decision curve analysis valuable for judging the clinical usefulness of diagnostic and prognostic models, complementing statistical performance measures with a decision-relevant assessment.
Source: Vickers & Elkin 2006
What are the limitations of decision curve analysis?
The limitations of decision curve analysis include that it relies on the threshold probability to represent the trade-off between benefits and harms, which may vary between patients and decision makers and may not be known precisely; that it summarises net benefit without capturing all aspects of a decision, such as costs beyond the benefit-harm trade-off; and that its results depend on the data and model used. It also does not by itself indicate the magnitude of clinical impact in absolute terms for all contexts. These limitations mean decision curve analysis is interpreted alongside other evidence and with attention to the relevant thresholds.
Source: Vickers & Elkin 2006
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-007
Stable URI · Machine-readable · Resolvable · CC BY 4.0