A new cognitive safety monitoring plan for interventional studies involving patients with advanced dementia: Insights from the ECT-AD Study

Abstract

<h3>Introduction</h3> Many pharmacological and some non-pharmacological treatments (e.g., electroconvulsive therapy, ECT) are associated with adverse cognitive effects in older adults, and this is particularly true for those with advanced dementia. Thus, it is imperative that studies investigating new treatments with potential cognitive side effects include a robust cognitive safety monitoring plan when the target population includes older adults with advanced dementia. However, detecting adverse cognitive events (ACEs) in this population poses unique challenges for two reasons: (1) the extant literature offers no recommendations or guidelines for defining what level of decline constitutes an ACE in this population, and (2) these patients are more likely to show floor effects on baseline cognitive testing, making it difficult to measure treatment-related adverse cognitive effects. To address these challenges, we created a new, empirically derived, and individually tailored cognitive safety monitoring plan that accounts for potential floor effects on pre-treatment cognitive testing. This plan was developed in the context of an ongoing, randomized controlled trial (the NIA supported "ECT-AD" study) to investigate the safety and efficacy of ECT for refractory agitation and aggression in patients with severe Alzheimer's disease (AD) <h3>Methods</h3> An extensive literature review was conducted to identify cognitive measures that have been found to be psychometrically sound for use in patients with advanced dementia, as well as psychometric data from those scales that could be used to define floor effects. Additionally, we inquired informant-rated scales of daily functioning that could serve as a proxy measure of cognitive status when patients exhibit baseline floor effects on cognitive testing. The team specifically focused on daily functioning scales for this purpose, given the robust association between cognition and activities of daily living in AD – an association that becomes stronger with disease progression (Liu-Seifert et al., 2016). Lastly, we reviewed the literature on statistical approaches to measuring reliable change on these kinds of cognitive/functional scales to define what constitutes statistically meaningful decline. <h3>Results</h3> The Severe Impairment Battery-8 (SIB-8) was chosen as the primary cognitive measure. The SIB-8 is a brief, clinician-administered test that has strong psychometric properties in patients with advanced dementia (Schmitt et al., 2009; Schmitt et al., 2013). Baseline floor effects on the SIB-8 were defined as a total score of ≤ 5/16. This value was chosen because it is approximately two standard deviations below the modal SIB-8 score in a large sample (N > 1300) of patients with moderate-to-severe dementia (Schmitt et al., 2013). For patients who <i>do not</i> exhibit baseline floor effects on the SIB-8, a post-treatment decline of ≥ 6 points would be considered an ACE. This threshold was empirically derived from the review of evidence-based approaches to measuring reliable change. Specifically, we used the Standard Deviation Index (SDI) method that uses the normative psychometric properties of a given test, along with an individual's test scores at two time points to derive a reliable change score. Using normative SIB-8 data from a large sample of comparable patients with AD (Schmitt et al., 2013), the SDI indicates that a decline of ≥ 5.6 points from baseline represents a statistically meaningful change. For patients who <i>do</i> exhibit baseline floor effects on the SIB-8, we will use the Barthel Index (BI) as a proxy measure of cognitive function (Mahoney & Barthel, 1965). The BI is an informant-rated scale (score range: 0-100) that assesses performance in basic activities of daily living (e.g., feeding, bathing). An ACE will be defined as a decline of ≥ 30 points on the BI from baseline, which was derived from the SDI method using normative BI data from a large sample (N=341) of comparable patients with AD (Kamiya et al., 2014). If an ACE is detected (using either the SIB-8 or BI) on the morning of, but prior to, a patient's next scheduled ECT treatment, then treatment will be held until the next treatment day. If a patient exhibits persistent decline prior to treatment on three consecutive days, then ECT will be discontinued. <h3>Conclusions</h3> By accounting for practical and statistical challenges in assessing cognitive function in patients with advanced dementia, we have designed a new, empirically derived, and conditional approach to monitoring for ACEs in real-time in clinical trials targeting this patient population. While the proposed cognitive safety monitoring plan was developed in the context of an inpatient ECT study, the procedures are agnostic to the specific type of treatment and can be readily adapted to a variety of interventional research contexts involving patients with severe dementia. <h3>This research was funded by</h3> National Institute of Health/National Institute on Aging (R01AG061100-01, NCT03926520)