Date Published: April 18, 2019
Publisher: Public Library of Science
Author(s): Charlene Soobiah, Mina Tadrous, Sandra Knowles, Erik Blondal, Huda M. Ashoor, Marco Ghassemi, Paul A. Khan, Joanne Ho, Andrea C. Tricco, Sharon E. Straus, Stephen D. Ginsberg.
Selection of optimal outcome measures is a critical step in a systematic review; inclusion of uncommon or non-validated outcome measures can impact the uptake of systematic review findings. Our goals were to identify the validity and reliability of outcome measures used in primary studies to assess cognition, function, behaviour and global status; and, to use these data to select outcomes for a systematic review (SR) on treatment efficacy of cognitive enhancers for Alzheimer’s Dementia (AD).
Articles fulfilling the eligibility criteria of the SR were included in a charting exercise to catalogue outcome measures reported. Outcome measures were then assessed for validity and reliability. Two independent reviewers abstracted data on outcome measures and validity and reliability reported for cognition, function, behaviour and global status.
129 studies were included in the charting exercise; 57 outcome measures were identified for cognition, 21 for function, 13 for behaviour and 10 for global status. A total of 35 (61%) cognition measures, 10 (48%) functional measures, 8 (61%) behavioural measures and four (40%) of global status measures were only used once in the literature. Validity and reliability information was found for 51% of cognition measures, 90% of function and global status measures and 100% of behavioural measures.
While a large number of outcome measures were used in primary studies, many of these were used only once. Reporting of validity and reliability varied in AD studies of cognitive enhancers. Core outcome sets should be used when available; when they are not available researchers need to balance frequency of reported outcome measures, their respective validity and reliability, and preferences of knowledge users.
Outcome measures are tools, instruments or scales used to assess an outcome. For example, the Activities of Daily Living (ADL) [1, 2] is to assess function in older adults. Selection of appropriate outcome measures for inclusion in a systematic review is imperative, to ensure research relevance for knowledge users . Knowledge users are individuals who may use research findings to make a decision and can include patients, clinicians, policymakers, or researchers [4, 5]. Inclusion of non-validated or uncommon measures in a systematic review can make it difficult for knowledge users to interpret and utilize findings to make informed decisions [6, 7]. Multiple measures exist for a particular outcome and their selection for use in systematic reviews can be challenging; a researcher must identify measures that are valid, reliable, and clinically relevant .
As our systematic review methods and results were previously published , the focus of this paper is on how outcome measures were identified for inclusion in the review. Our systematic review was registered (CRD#42012001948)  and included experimental and observational studies that reported on cognitive enhancers approved for use in Canada (i.e., donepezil, galantamine, rivastigmine and memantine) for patients with AD. Studies had to report on at least one pre-specified outcome (i.e., cognition, function, behaviour, and global status) to be considered eligible for inclusion in the systematic review. Several electronic databases were searched from inception to December 31, 2011 to identify studies. Two independent reviewers assessed each citation and full text article against eligibility criteria. Studies fulfilling eligibility criteria were included in the present study.
Data from the charting exercise and validity and reliability assessments were analyzed descriptively using frequencies in Excel.
Overall, 129 studies were included in the systematic review on treatment efficacy of cognitive enhancers for AD patients and from these articles, we identified 101 measures for our outcomes of interest (57 cognition measures, 21 function measures, 13 behaviour measures and 10 global status measures). We identified validity and reliability data for 51% of cognition measures, 100% of behaviour measures and 90% of function measures and global status measures. Studies in which validity or reliability were not supported by a citation were supported by funding from industry or a mix of industry and non-industry sources.
Our paper highlights the variability in the reporting of outcome measures used in AD studies. We identified multiple outcome measures reported in the primary studies; many of these were used only once in the primary studies that were included in our systematic review. The large number of measures used in studies makes it difficult to synthesize the evidence. Cataloguing and assessing validity and reliability of each outcome measure in studies can be resource intensive; using core outcome sets (i.e., agreed upon outcomes and measures used in a particular discipline) in systematic reviews is recommended as this may streamline outcome selection . In lieu of core outcome sets, researchers need to balance frequency of reported outcome measures, their respective validity and reliability, and preferences of knowledge users.