Background Neurology clinical trials frequently use clinical outcome assessment instruments (COAs) as outcome measures. We question the extent to which measurement limitations of COAs contribute to clinical trials failures in Alzheimer's disease.

Objectives We conducted a series of literature reviews to: identify the concepts assessed in AD clinical trials, and how these concepts were defined and measured; identify which COAs have been used in AD clinical trials, and on which grounds they were selected. We examined published measurement properties of COAs used in AD clinical trials in accordance with FDA guidance.

Results A set of literature reviews identified >6500 publications from PUBMED/EMBASE inception. In the extracted 850 articles, 984 uniquely named concepts were assessed and 1283 COAs were used. However, few trials provided definitions of the concepts they measured; very different COAs were used to measure the same (named) concepts; the same COAs were used to measure different concepts; COA selection was rarely justified or evidence-based. Further reviews of COA development papers (n=174) indicated most fail to meet recommended criteria.

Conclusions Findings imply substantial measurement confusion in AD clinical trials and suggest they have provided weak evaluations of treatment effectiveness. Findings were replicated in Parkinson's disease (n366 articles).