Sensitivity analysis in economic evaluation: an audit of NICE current practice and a review of its use and value in decision-making
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
OBJECTIVES: To determine how we define good practice in sensitivity analysis in general and probabilistic sensitivity analysis (PSA) in particular, and to what extent it has been adhered to in the independent economic evaluations undertaken for the National Institute for Health and Clinical Excellence (NICE) over recent years; to establish what policy impact sensitivity analysis has in the context of NICE, and policy-makers' views on sensitivity analysis and uncertainty, and what use is made of sensitivity analysis in policy decision-making. DATA SOURCES: Three major electronic databases, MEDLINE, EMBASE and the NHS Economic Evaluation Database, were searched from inception to February 2008. REVIEW METHODS: The meaning of 'good practice' in the broad area of sensitivity analysis was explored through a review of the literature. An audit was undertaken of the 15 most recent NICE multiple technology appraisal judgements and their related reports to assess how sensitivity analysis has been undertaken by independent academic teams for NICE. A review of the policy and guidance documents issued by NICE aimed to assess the policy impact of the sensitivity analysis and the PSA in particular. Qualitative interview data from NICE Technology Appraisal Committee members, collected as part of an earlier study, were also analysed to assess the value attached to the sensitivity analysis components of the economic analyses conducted for NICE. RESULTS: All forms of sensitivity analysis, notably both deterministic and probabilistic approaches, have their supporters and their detractors. Practice in relation to univariate sensitivity analysis is highly variable, with considerable lack of clarity in relation to the methods used and the basis of the ranges employed. In relation to PSA, there is a high level of variability in the form of distribution used for similar parameters, and the justification for such choices is rarely given. Virtually all analyses failed to consider correlations within the PSA, and this is an area of concern. Uncertainty is considered explicitly in the process of arriving at a decision by the NICE Technology Appraisal Committee, and a correlation between high levels of uncertainty and negative decisions was indicated. The findings suggest considerable value in deterministic sensitivity analysis. Such analyses serve to highlight which model parameters are critical to driving a decision. Strong support was expressed for PSA, principally because it provides an indication of the parameter uncertainty around the incremental cost-effectiveness ratio. CONCLUSIONS: The review and the policy impact assessment focused exclusively on documentary evidence, excluding other sources that might have revealed further insights on this issue. In seeking to address parameter uncertainty, both deterministic and probabilistic sensitivity analyses should be used. It is evident that some cost-effectiveness work, especially around the sensitivity analysis components, represents a challenge in making it accessible to those making decisions. This speaks to the training agenda for those sitting on such decision-making bodies, and to the importance of clear presentation of analyses by the academic community.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.068 | 0.011 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.006 | 0.000 |
| Bibliometrics | 0.004 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it