Premium
P4‐001: COMPOSITE ENDPOINTS – PRO PERSPECTIVE: THE KEY TO SUCCESSFUL CLINICAL DEVELOPMENT IN ALZHEIMER'S DISEASE
Author(s) -
Hendrix Suzanne B.,
Dickson Samuel P.,
Ellison Noel
Publication year - 2019
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2019.06.3660
Subject(s) - overfitting , perspective (graphical) , sample size determination , disease , clinical trial , population , statistical power , medicine , computer science , statistics , artificial intelligence , mathematics , artificial neural network , environmental health
Background: Progressive diseases such as Alzheimer’s disease are devastating precisely because they are degenerative, suggesting that the passage of time can be used as a “gold standard” for assessing disease. Symptoms that change the most and have minimal variability best reflect the disease symptoms. A perfect measure of disease has the most progression and least variability over time. Methods: Composite scores optimally combine item scores and can be derived from existing historical datasets. Composite scores should be validated in new dataset or split sample to reduce overfitting, and to adjust performance estimates for bias. Composite scores should progress relative to a normal population, allowing the most room for improvement. Similarly performing composite scores can be compared with theoretical considerations such as domain representation, time and ease of administration. Global Statistical Tests (GSTs) were used to combine primary and key secondary outcomes from historical studies to estimate study results if an optimized composite had been used reflecting overall disease effects. Results: Using properly derived composite scores in clinical trials substantially improves the power for detecting treatment effects. Use of Global Statistical Tests demonstrate that historical studies are often inconclusive due to inconsistent outcome measures potentially missing significance with effective treatments or proceeding to phase 3 when phase 2 should have failed. Conclusions: Composite scores are not all a single type. Theoretical considerations combined with empirical derivations result in robustly performing composite scores across multiple data sets offering the best chance for successful clinical study outcomes by improving power.