Threshold Analysis for NMA
This method checks how strong biases would need to be in a network meta-analysis to change the treatment recommendation. It helps decision makers understand if their conclusions are solid or could easily change if the data were slightly biased.
At a glance
Use when
Evaluating confidence in NMA-based treatment recommendations, especially when GRADE ratings are low or moderate
Avoid when
When no clear decision criterion exists or when NMA results are used descriptively rather than for decision making
Inputs
Network meta-analysis results, treatment effect estimates, decision criteria for treatment choice
Outputs
Magnitude of bias required to change a recommendation, alternative treatment recommendations under bias
How it works
Threshold analysis evaluates the robustness of treatment recommendations derived from network meta-analysis (NMA) by determining the magnitude of potential bias required to alter the decision. Applied iteratively to each pairwise contrast, it assesses whether plausible bias could shift the preferred treatment, thereby informing the credibility of NMA-based conclusions beyond GRADE quality ratings.
- HTA domains
- Clinical Effectiveness, Organisational aspects, Aspects Beyond HTA
- Assumptions
- Bias may be present in NMA evidence; treatment recommendations are based on relative effect estimates; changes in effect size can alter decisions
- Strengths
- Provides decision-focused assessment of NMA credibility; identifies fragile recommendations; complements GRADE by focusing on decision impact rather than evidence quality alone
- Limitations
- Requires predefined decision rules; does not quantify actual bias but hypothetical thresholds; interpretation depends on what is considered 'plausible' bias
- Also known as
- Threshold Analysis
Questions this answers
References & sources
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.