Do Subjects Not Meeting Screening Inclusion Criteria at Baseline Differ in Screening Performance in Diagnostic Tests in Early Dementia Clinical Trails?

Abstract

AbstractBackgroundThe success of any clinical trial begins with ensuring that participants entering the trial meet the inclusion criteria and disease severity. Score manipulation to meet eligibility criteria may contribute to loss of signal in clinical trials. Based on our prior analyses of early Alzheimer’s disease (AD) clinical trials, over 10% of randomized participants likely had inflated MMSE scores at inclusion determination. In addition to ‘traditional’ instruments such as the MMSE or CDR, other ‘diagnostic’ instruments such as the RBANS, FCSRT and Wechsler Memory Scale are often used to select the study population. In the current analysis we explore the relationship between the performance on the diagnostic tests and the likely inflation in MMSE.MethodBaseline MMSE scores and Screening diagnostic tests results were pooled from 6,387 subjects from multiple clinical trials in early AD. Within each trial, participants’ performance on the diagnostic tests was broken into quartiles – best performers, second best, second worst and worst performers. Participants were also classified by their Baseline MMSE score as either meeting or being below the Screening MMSE inclusion criteria. A logistic regression was used to explore the association of the diagnostic test performance quartiles and Baseline MMSE score categories correcting for Screening MMSE score.Result1223 (19.2%) participants comprised the best performing quartile on the diagnostic tests. Comparatively, with each subsequent quartile, the odds of not meeting the Screening inclusion criteria at Baseline increased significantly ‐ by 1.4x in the 1666 participants (26.1%) in the second best, by 2.0x in the 1696 (26.6%) in the second worst, and by 2.4x in the 1800 (28.2%) in the worst quartile.ConclusionThis post‐hoc analysis identified increasingly significant odds of not meeting Screening MMSE related inclusion criteria at Baseline for each progressive quartile of worsening diagnostic test performance, when correcting for the Screening MMSE score. We suggest that despite the limitations of this post‐hoc analysis, outlying poor performance on the diagnostic tests should be considered a risk marker for possible score inflation and assessed in the whole context of the available data. Further work is necessary.