Major Adverse Cognitive Events (MACE-Cog): A new “MACE” framework for cognitive outcomes in cardiovascular diseases and stroke

Abstract

Cardiovascular diseases, particularly stroke, are leading causes of dementia. Several common cardiac interventions such as coronary artery bypass grafting and transcatheter aortic valve implantation are also associated with cognitive decline. However, cognitive outcomes continue to be poorly collected in clinical trials of cardiovascular diseases. In this article, we review the limitations of current approaches to cognitive assessment in cardiovascular disease studies. When assessed, there is wide variation in cognitive tasks used, tasks have limited population-specific validation, are subject to floor and ceiling effects, and may suffer from socio-cultural and linguistic biases. Many tasks are not available in multilingual formats and often rely on face-to-face testing with trained coordinators. All conventional cognitive outcomes in cardiovascular trials are associated with substantial incompletion, especially amongst older and more impaired individuals – the very people most important to capture. This non-random missingness generates survivor and attrition biases, an unacceptable situation for any trial outcome. Finally, the meaning of measured cognitive outcomes is often unclear for patients, carers, clinicians, and regulators. To help address these limitations, we propose a new framework – Major Adverse Cognitive Events (MACE-Cog) – to better capture cognitive outcomes in stroke and other cardiovascular populations. As an inclusive construct reflecting the multi-dimensional nature of cognitive decline, MACE-Cog does not solely rely on performance on cognitive testing, but also considers inability to complete cognitive testing due to cognitive-behavioural factors, reported symptoms of cognitive decline, impairment in activities of daily living as a result of cognitive impairment, new clinical diagnoses of dementia, as well as care home admission. We hope that the proposed composite outcome, and multiple use cases presented, spark progress in cardiovascular research towards more inclusive approaches to the study of cognitive outcomes that move beyond the confines of cognitive tests alone.

Definitions, mechanisms and assessment of cognitive issues in cardiovascular diseases

Cognitive impairment is a leading contributor to the global burden of disease, affecting over 57 million people worldwide.¹ Cognitive impairment refers to deterioration in one or more cognitive abilities, including memory, executive function, language, visuospatial function and social cognition.² These changes can arise from a variety of processes that damage brain tissue, including neurodegenerative disorders (the most common one being Alzheimer’s disease), cerebrovascular disorders, and other pathologies, both chronic and acute (e.g. anesthesia impact in post-surgical patients).³ Cognitive function exists along a spectrum. This spectrum ranges from normal cognition to mild cognitive impairment/mild neurocognitive disorder (characterized by evidence of cognitive decline with preserved independence in activities of daily living), to dementia/major neurocognitive disorder (characterized by cognitive decline accompanied by a loss of independence in activities of daily living (Figure 1).⁴

Figure 1.

The spectrum of neurocognition and cognitive decline ranging from normal cognition to mild cognitive impairment (mild neurocognitive disorder) to dementia (major neurocognitive disorder).

Cardiovascular diseases (CVD), particularly stroke, are a leading cause of cognitive decline and dementia. For example, cognitive impairment is a well-recognized short- and long-term complication of stroke that progresses to dementia in up to a third of patients within 1-year of a moderate to severe stroke.^5–7 Incident stroke is associated with a more than two-fold risk of dementia, suggesting that it is a strong, independent and potentially modifiable risk factor for all-cause dementia.^8–10 Even strokes that do not cause detectable physical deficits ( “covert” strokes) are associated with cognitive decline.¹¹

Cardiac diseases, such as heart failure, atrial fibrillation, and myocardial infarction, are each also associated with cognitive decline³, and up to 50% of cardiac arrest survivors have cognitive impairment at 3-month follow-up.¹² Several common cardiac interventions are also associated with cognitive decline.^{13, 14} For example, cognitive impairment is seen in around one-third of patients undergoing transcatheter aortic valve implantation¹⁵ and up to 40% of patients after coronary artery bypass graft.¹⁶

Mechanisms and Assessment of Cognitive Issues in Cardiovascular Diseases

Overt cerebrovascular events, covert events, brain injuries from cardiovascular diseases and even vascular risk factors such as hypertension and diabetes, can cause cognitive deficits or dementia via multiple mechanisms.³ They can cause direct injury to strategic brain areas causing immediate cognitive consequences. They can undermine fragile brain networks, unmasking or accelerating cognitive changes related to pre-existing neurodegenerative diseases.¹⁷ They can also indirectly accelerate neurodegeneration via multiple mechanisms including neuroinflammation, hypoperfusion with altered cerebral circulation and collateral flow, and white matter injury.¹⁸ Thus, the onset of cognitive decline may become apparent immediately after a cardiovascular event/procedure, or there may be a delay.¹⁹ Regardless, the consequences can be dire; for instance, post-stroke dementia is a major contributor to post-stroke dependency^{20, 21} and mortality.²² Indeed, cognitive impairment and dementia are among the most feared outcomes among persons with cardiovascular disease and their families, as highlighted in research priority-setting exercises.^{23, 24}

It is essential to note that ascertainment of cognitive outcomes in a clinical trial setting (i.e. determining whether someone has developed an adverse cognitive outcome such as cognitive impairment or dementia) is fundamentally different from determining the underlying cause or etiology of the outcome in question. However, a few etiologic considerations bear mentioning. The International Society of Vascular Behavioural and Cognitive Disorders (VasCog) group recently published a second iteration (VasCog-2–World Stroke Organization) of criteria for diagnosing the clinicopathological entity of vascular cognitive impairment and dementia (VCID) as a successor to the original version from 2014, providing operationalization and guidance on potential neuroimaging and fluid biomarkers to facilitate diagnostic consistency among clinicians and researchers.²⁵ Those criteria describe preclinical, mild, and major dementia levels of VCID, reflecting the spectrum of cognitive outcomes as we have outlined above and in Figure 1. Very importantly, that version also acknowledges that there is no standard cognitive testing profile that is diagnostic of vascular disease, a crucial point for CVD trials.^{26, 27} Of course, when a patient presents with abrupt and persistent onset of cognitive impairment in the immediate aftermath of a clinical stroke or cardiovascular event/procedure, there is little controversy about CVD as a main underlying etiology, but cases are often not so clear-cut in practice. Even that situation may have underlying neurodegenerative disease pathology that is unmasked by the event. Ultimately, we must acknowledge that in a CVD trial, an adverse cognitive outcome itself is important to capture – regardless of our ability to specifically attribute it to VCID, neurodegenerative disease, mixed disease or other etiologies. From a participant perspective, the outcome (e.g., disabling cognitive impairment) has immediate impact on their life, while the cause, although potentially relevant for devising future prevention strategies, is of secondary importance.²³ This is similar to how we think about other well-established outcomes such as all-cause mortality, major adverse cardiac events (MACE), physical function (modified Rankin Scale; mRS) or health-related quality of life in CVD trials. In these instances, their importance is universally understood, but for the most part, ascertainment of these outcomes does not presume that the outcome is always due to a particular etiology. In fact, overthinking the etiologic underpinning of outcomes can lead to bias, as has been noted in oncologic trials where all-cause mortality may be less vulnerable to clinician biases which can impact which events are categorized as “cancer-specific” mortality.²⁸ In practice, cognitive impairment is also often multi-factorial in nature, and disentangling VCID from non-VCID contributors can be a challenging endeavour for both clinical research personnel and expert cognitive specialists. Therefore, it is important for ascertainment of cognitive outcomes in CVD trials to be etiologically agnostic, unconstrained by black-and-white categorical diagnoses of CVD-related cognitive impairments.²⁹

Regardless of mechanism, cognitive function has received insufficient attention as a clinically relevant outcome in cardiovascular trials, which have traditionally prioritized cardiovascular events (MACE), physical function (mRS), quality of life, and mortality.^{30, 31} These outcomes are widely used because they are well validated, clinically interpretable, and feasible to collect in a large trials: 1) MACE (with or without mortality) are objective, impactful and can be ascertained from routinely collected clinical data; 2) the mRS is recommended by regulators as a primary endpoint in stroke trials; and 3) quality of life measures support health economic analyses.³² However, these commonly used outcomes do not capture cognitive status. In particular, the mRS, which is heavily relied upon in most acute stroke trials, measures global functional disability and does not assess cognition, particularly when cognitive impairment occurs in the absence of overt physical disability. As a result, individuals with identical mRS scores may differ substantially in independence and quality of life due to cognitive deficits.^{33, 34} Conversely, patients with severe motor impairment may score poorly on the mRS despite intact cognition and good quality of life. Even in people with “excellent” outcomes (mRS 0–1), 54% of people have some degree of cognitive impairment, 32% of people endorse symptoms of depression and 52% are not fully reintegrated to their pre-stroke activities.³⁵ These limitations underscore the need for cognitive outcome measures that are clinically meaningful, grounded in routine clinical data, and feasible to apply at large scale within cardiovascular trials.

In this article, we discuss challenges and limitations of current approaches to cognitive assessment in CVD, discuss how these issues impede progress, and propose a new construct – which we term Major Adverse Cognitive Events (MACE-Cog) – to help better capture cognitive outcomes after stroke, cardiac events (such as cardiac arrest), cardiovascular procedures, and other cardiovascular populations (Figure 2). This work is intended to be conceptual; future work will be required to operationalize and validate these concepts.

Figure 2.

Summary of the problems with current approaches to cognitive outcome evaluation in stroke and other cardiovascular diseases, and of the proposed concept of Major Adverse Cognitive Events (MACE-Cog).

Challenges with current approaches to post-stroke cognitive assessment in cardiovascular trials

Current Approaches

The main approach to assessing cognitive function in CVD trials has focused on cognitive screening tests or neuropsychological assessments.³¹ Screening tests such as the Montreal Cognitive Assessment (MoCA), Mini-Mental State Examination, or Oxford Cognitive Screen (OCS) aim to briefly evaluate global cognition.^{36, 37} Domain-specific tasks focus on individual cognitive functions (e.g. processing speed or verbal memory). Formal neuropsychological assessments and research batteries³⁸ consist of combinations of validated tasks assessing multiple cognitive domains, administered by a trained neuropsychologist. However, they are infrequently used in CVD research,³⁰ and there is substantial variation in the tasks used and the definition of cognitive impairment.³¹ Though online assessments have gained significant attention since COVID-19 pandemic, they have not been routinely used and their incorporation within stroke RCTs have been limited³¹ and may introduce their own biases due to technology access or usability.³⁹ Studies also vary greatly in terms of the setting, and thus the time and staff available for cognitive testing (e.g. emergency treatment vs. prevention vs. recovery), evaluation of pre-event cognitive impairment (and inclusion/exclusion of such patients), whether first-ever and/or recurrent strokes were included, rates of missing cognitive outcomes on follow-up, heterogeneity in timing of assessment of post-event cognitive impairment or dementia and the method by which they were diagnosed.^{31, 40–42}

Challenges with cognitive tasks

Even when used, cognitive assessments after cardiovascular events/procedures have significant limitations as a trial outcome measure. Some commonly-used tests are not validated or well suited for use in multicentre CVD trials, given the higher prevalence of language, physical and cognitive impairments, as well as medical instability in the acute phase, compared with non-CVD survivors.⁴³ In addition, many brief screening assessments are specific, but not sensitive to subtle but important cognitive issues, and suffer from both floor- and ceiling-effects.⁴⁴ Many tools are not available or validated in multiple languages. There have been efforts to develop consensus-based cognitive test batteries for post-stroke cognitive decline, most notably the National Institutes of Neurological Disorders and Stroke–Canadian Stroke Network (NINDS-CSN) Vascular Cognitive Impairment Harmonization Standards which proposed neuropsychological test batteries of various lengths (60-minute, 30-minute, and 5-minute) using various cognitive tests validated for stroke patients, and the consensus recommendations for neurological outcomes in CVD trials by the Neurologic Academic Research Consortium (NeuroARC) (Supplementary Table 1).^{38, 45} The 60- and 30-minute NINDS-CSN batteries are too long for the pressures of typical CVD trial visits, whereas the 5-minute battery consists of components of the Montreal Cognitive Assessment (MoCA), and thus is limited by floor- and ceiling-effects and does not have the responsiveness to change required for a trial outcome.⁴⁶ Commonly used global cognitive tests for identifying cognitive impairment in the setting of stroke and CVD include the full MoCA,⁴⁷ the Addenbrooke’s cognitive examination-revised (ACE-R),⁴⁸ and the Mini-Mental State Examination (MMSE).⁴⁹ The MoCA and ACE-R have been shown to achieve good sensitivity and specificity for detecting cognitive impairment compared to the 60-minute NINDS-CSN battery in patients evaluated ≥1 year following stroke or transient ischemic attack, while the MMSE has been found to be less sensitive for non-memory impairments.^{47, 50} Both detailed testing and, especially, cognitive screens have significant intra- and inter-person variability. This variability, whether person-based (e.g. years of education, baseline intellectual functioning, practice effects, recent sleep quality, distractions etc.) or disease-based (e.g. impacts of acute hospitalizations, anesthetics and recovery over time) dramatically increases the noise in the measure, making identification of meaningful differences challenging.⁴⁶

In multi-cultural settings, or multi-centre studies, administering assessments in a participant’s non-native (even if fluent) language,⁵¹ or using assessments that include culturally specific items, can influence performance and affect test scores.^51–53 Translating cognitive tests can improve their accessibility, but translations must be validated, socio-cultural biases can remain⁵⁴ and application of translated tests are still often limited by availability of trained study personnel able to speak the translated language.

Relevance of cognitive scores

The meaning of cognitive test scores for patients or clinicians is often unclear – for example, if an intervention reduced trails B completion times by 10 seconds, or improved a composite outcome by 0.5 standard deviations, would this be enough to motivate doctors, patients or regulators to change practice? There have been some studies of minimal clinically important differences (MCIDs) to define the smallest amount a test or scale must change to be meaningful to patients but this remains unclear for many cognitive tests.⁵⁵ Though literature have shown that MCIDs for cognitive scales are sensitive to severity in major neurocognitive disorders (dementia), they have not been widely studied in CVD-specific contexts.⁵⁶ For instance, what may be relevant for community-dwelling aging may not apply equally to stroke or cardiac arrest survivors.

Survivor biases and non-random missingness

Many commonly used cognitive measures cannot be completed by patients with severe motor, visual or language impairments. How missing data are handled is thus a major challenge; relying solely on patients who can complete follow-up assessments can grossly distort the cognitive outcomes in CVD studies.⁵⁷ Data are unlikely to be missing completely at random, since the very presence of cognitive decline can interfere with completing the outcome. Bias is inevitable if trial participants have missing outcomes post-randomization for reasons related to the trial intervention (e.g. severity of CVD). Severe functional and behavioural decline (i.e. dementia) from the cognitive impact of CVD can prevent patients from participating meaningfully in testing. Dementia or severe CVD outcomes can result in a participant being admitted to a nursing home or other long-term care facility, or moving to live with a relative, interfering with follow-up and increasing risk of bias. This is particularly problematic when cognitive outcomes rely on face-to-face clinic assessment, since cognitive impairment is strongly associated with increased morbidity and mortality, with a consequent substantial loss to conventional methods of follow-up.⁴¹ Providing telephone, home-based, or web-based methods of cognitive assessment may potentially increase the number of patients who are able to contribute to cognitive outcomes. For example, in the Oxford Vascular Study (OXVASC), the 5-year cumulative incidence of post-event dementia after stroke/TIA was 29% overall but was only 17% in clinic-assessed patients, compared to 45% in patients who were captured using other follow-up methods (home visits and/or telephone visits).⁴¹ However, such alternative methods will not overcome the intrinsic limitations of testing in CVD survivors who are unable to meaningful engage in cognitive tasks.³⁹

CVD trials differ from conventional dementia studies

The biases and limitations for CVD trials differ from trials of dementia, which more routinely use cognitive measures as part of primary or secondary endpoints and are designed to capture cognitive decline over a longer period.⁵⁸ Dementia trials are typically conducted in stable outpatient settings, where patients can be carefully screened and selected for baseline ability to complete testing, informants are often available, and time-intensive assessments can be feasibly administered. In contrast, CVD trials, especially those with a focus on hyperacute interventions (e.g. cardiac arrest, stroke), operate under conditions of extreme time pressure, without opportunity to assess the pre-enrollment cognitive status of participants. In these settings, participants may be incapacitated, and informants may not be immediately available, making the collection of high-quality baseline cognitive data much more difficult. In outpatient secondary prevention trials, a staged approach to cognitive assessment – with pre-enrollment screening followed by more comprehensive assessments for mild cognitive impairment or dementia – may be more feasible.

Why this matters for clinical trials – the need for a new approach

In clinical trial design, any outcome that is missing in a significant portion of participants, especially in a non-random manner, cannot be relied upon as a primary, or even secondary, outcome. The issues of survivor bias, missingness and loss-to-follow-up become far too important. Additionally, to change clinical practice, interventions⁵⁸ must impact cognition in a manner that is meaningful to regulatory agencies, funders, CVD survivors, carers, and clinicians. The fact that there is no agreement on tasks, and that even those recommended by consensus panels are often not included in trials, highlights this profound gap. Therefore, cognitive tasks alone, whether via screening tests, in person batteries, or online assessments, clearly cannot be relied upon as outcome measures for CVD clinical trials.

Rethinking approaches to cognitive outcomes

Cognitive outcomes should emphasize change from baseline and impact on daily activities

The limitations of pen-and-paper cognitive tests are only partially addressed by currently available solutions (as summarized in Table 1). While cognitive tests certainly have a place in this field, it is essential to avoid relying solely on them for evaluating cognitive outcomes in CVD trials. In fact, it must be emphasized that even the Diagnostic and Statistical Manual of Mental Disorders (currently in its Fifth Edition, Text Revision, DSM-5-TR) does not actually require cognitive testing for the diagnosis of major neurocognitive disorder or dementia.⁵⁹ Rather, it requires evidence of cognitive decline and difficulties with activities of daily living, which can be established through patient and carer reports.

Table 1.

Limitations to current approaches to post-stroke cognitive outcome collection, partial solutions in use, and proposed paradigm shifts to better capture these outcomes

Current limitations	Partial solutions in use	Proposed paradigm shift
Variation in cognitive tests used	Use NINDS-CSN VCI battery components – 30- and 60-minute batteries are too long for typical trials, 5-minute protocol just has elements of the MoCA, but may be too insensitive	Rethink reliance on cognitive testing – capture cognitive symptoms and impact on daily life in addition to, or instead of, cognitive tests
Cognitive tests used are often not validated for stroke
Cognitive tests used are insensitive and have both floor and ceiling effects
Reliance on face-to-face testing, with trained coordinators	Offer web or telephone-based testing – still suffers from high rates of incompletion	Consider the use of informant assessments when possible
Substantial incompletion rates, high rates of missing data	- Statistical imputation; however, since data is not missing at random, this approach is statistically limited - Categorize as “impaired” anyone with aphasia (NIHSS item 9 = 2 or 3), or severe neglect (item 11 = 2), or death (as the ultimately poor cognitive outcome)	Consider cognitively relevant reasons (e.g. inattention, memory, aphasia) for incompletion as themselves being adverse cognitive outcomes; non-cognitive reasons for incompletion (e.g. refusal, hemiplegia) would not be considered an adverse cognitive outcome.
Survivor and attrition biases, patients with severe cognitive impairment cannot complete testing	Imputation techniques as above	In addition to above, consider institutionalization as an outcome – although hard to establish that the reason is cognitive decline
Socio-cultural and linguistic biases of tests	Offer translated versions of cognitive tests – still does not overcome sociocultural bias as tests were primarily developed in Western countries	Capture an individual’s pre-stroke cognitively-relevant activities and compare to post-stroke activities to account for sociocultural differences
Unclear meaning of cognitive outcomes for patients, carers, providers and regulators	Studies of minimal clinically important differences to infer relevance of changes of tests or scales	Combined outcome with components that reflect DSM definitions of neurocognitive disorders and incorporate clear change in cognitive symptoms with functional relevance

Questionnaires can incorporate patient and carer reports

Cognitive decline in specific domains can be assessed using standardized questionnaires that can be completed by informants and/or patients. For example, the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE), the short form of which has 16 items, has been validated for identifying both pre-stroke and post-stroke cognitive impairment. The briefer 8-item Interview to Differentiate Aging and Dementia (AD8 ⁶⁰), has also been validated for identifying pre- and post-stroke cognitive impairment, and can be administered either to an informant or to the participant.^61–63 Additionally, evidence of decline on cognitively-relevant daily activities can be provided by completion of instrumental activities of daily living (IADL) questionnaires such as the 20-item Bristol ADL Scale, or the considerably briefer 8-item Lawton IADL Scale, which can also be completed by either informants or patients.^64–66 The Standard Assessment of Global Everyday Activities (SAGEA) collects information on both cognitive symptoms and ADLs.⁶⁷ Having a valid outcome that can be reliably obtained from either an informant or participant reduces missing outcomes and the resulting need to classify missingness as attributable to cognitive or physical limitations. Combining a cognitive symptom assessment with a cognitively-relevant IADL scale may allow greater inference as to whether IADL impairments are reflective of at least some underlying cognitive contributors rather than solely physical impairments.

Seeking data from informants can help mitigate the problem of missing cognitive outcome data. Moreover, cognitive decline can have several important functional and behavioural manifestations in the patient’s daily life that are observed by family members, caregivers, or other individuals close to the patient. Their input has clear value in identifying an adverse cognitive outcome for patients, be it symptoms of cognitive decline^{63, 68, 69} or their impact on IADLs.^{64, 70} However, the availability of informants is imperfect. For example, in one study screening for cognitive impairment using informant-based questions in acute care, only around 50% of participants had a suitable informant within 24 hours.⁷¹ While this may improve with a longer assessment window, it highlights the practical challenges of relying soley on informant data in acute clinical trials. This does not diminish the value of informant-based information but does suggest the need for careful interpretation when incorporating informant data into clinical trials.

Considering the above needs, one consideration that utilizes informants and captures daily functioning, is the Clinical Dementia Rating Scale Sum of Boxes (CDR-SB). The CDR-SB is a widely used outcome measure in dementia intervention trials.It was the main outcome measure in the CLARITY-AD trial evaluating the efficacy of lecanemab for those with early Alzheimer’s Disease related cognitive impairment and mild dementia.⁷² The CDR-SB is a patient- and informant-rated scale reflecting the severity of cognitive and functional impairment in day-to-day life in the domains of memory, orientation, judgement and problem solving, functions in community affairs, in home and with hobbies and with personal care. It does not require cognitive testing to assess the severity of impairment, it captures granularity in increasing degrees of severity and it can be administered by trained clinical personnel over the telephone. Thus, it is less vulnerable to non-completion. However, it is not validated in those with CVD. Increasing CDR-SB severity may not distinguish difficulties caused by physical CVD-related deficits from those due to cognitive impacts. CDR-SB is also a long battery (~30 minutes) which may limit feasibility in large CVD trials.

Administrative and medication data may be informative

A clinical diagnosis of dementia is a relevant outcome that can be obtained from healthcare records or administrative data in many regions, with high specificity, but moderate sensitivity.^{73, 74} In those without a documented diagnosis, a diagnosis can also be ascertained based on prescription records for dementia medications like cholinesterase inhibitors or memantine.¹⁰ Indeed, diagnostic algorithms have been developed and validated for ascertaining a dementia diagnosis from recorded diagnoses, prescriptions, and physician claims; for example, such an algorithm was recently used for quantifying the magnitude and time-course of post-stroke dementia in an entire Canadian province.¹⁰ However, that method is not applicable in many jurisdictions and has lower sensitivity for mild dementia, which is underdiagnosed in the community.⁷⁵ Direct patient or informant reporting of new medications for, or diagnosis of, dementia/mild cognitive impairment can also be a specific, if not fully sensitive outcome.

Institutionalization for cognitive/behavioural changes is a meaningful, but complex, endpoint

Another outcome component to consider is institutionalization from dementia, which can also often be captured well in some jurisdictions using administrative or electronic health records and represents a clinically relevant, “end-stage” indicator of cognitive decline.^{76, 77} However, establishing the reason why a given patient ended up at a long-term care facility can be challenging, particularly in the aftermath of an event like stroke, which can lead to institutionalization because of physical disability alone. Institutionalization derived from routine administrative data may be a valid outcome in European, North American, and Australasian sites, but the definitions of care home/institutional care, and the case-mix/severity of residents, may vary considerably globally.⁷⁸ Indeed, institutionalization is also influenced by numerous socioeconomic and socio-cultural factors like cultural views of institutions, marital status, health of caregivers, the extent of family or other social supports, public/private or insurance supports available to the patient/family and financial resources. With these social constructs in mind, the potential outcome of “institutionalization”, if considered in isolation, would have to be closely adjudicated as many variables – beyond cognitive impairment – may contribute to institutionalization. This is one reason to have multiple contributors to the definition of adverse cognitive outcomes, rather than relying on a single construct. At a group level, particularly after adjusting for confounding comorbidities and baseline stroke severity, institutionalization can still play an important role in distinguishing patients likely to have fared poorly from a neurocognitive standpoint.

Finding meaning in the missingness

In rethinking the approach to cognitive outcomes, it must be acknowledged that being unable to perform a cognitive test – particularly when this is attributable to one or more cognitively-relevant neurological impairments such as aphasia or neglect – is itself an adverse neurocognitive outcome and not missing data. Prospectively capturing the context for incompletion on study case record forms can help improve the ascertainment of adverse cognitive outcomes. If non-completion can be systematically captured and imputed as an adverse outcome, much of the missing data in CVD trials can be reclassified as interpretable and clinically meaningful. For example, assessors could be encouraged to document standardized reasons for test incompletion or factors that limited their assessment. Some factors, like the participant having moved to a new city, may have little to do with cognitive impairment, but others – like severe aphasia or speech disturbance, inappropriate or inconsistent responses, behavioural agitation, neglect, severe apathy or amotivation could indeed be counted towards adverse cognitive outcomes. A similar approach was employed in a secondary analysis of the Endovascular Treatment for Small Core and Anterior Circulation Proximal Occlusion with Emphasis on Minimizing CT to Recanalization Times (ESCAPE) trial of thrombectomy for acute ischemic stroke, which examined cognitive outcomes at 90 days. In that study, when data were missing due to severe aphasia (NIHSS item 9 score 2–3) or severe neglect (NIHSS item 11 score 2), those patients were assigned to the unfavorable performance group in sensitivity analyses.⁷⁹ When doing so, thrombectomy was associated with better cognitive outcome on all five cognitive tests. In sensitivity analyses, which included only patients who completed the cognitive testing, the effect of treatment was significant in only three of the five cognitive tests across all sensitivity analyses with imputed values. This illustrates how reframing missing data can reduce attrition and survivor bias, by moving away from analyzing only the skewed subset of survivors able to complete the tests. It may be challenging for non-clinical research personnel to accurately attribute reasons for non-completion based on neurocognitive context. Clear operational guidance, structured questions or checklists and training would be essential to support consistent and valid classification. Such an approach has been used in the Assessing Post-Stroke Psychology Longitudinal Evaluation (APPLE) study.⁸⁰

In summary, we must consider a different approach to defining cognitive outcomes in CVD trials to support their use as a reliable and valid primary clinical trial endpoint. To do so, the approach to cognitive outcomes must: (a) reflect the multiple different potential manifestations of cognitive impairment; (b) appreciate the value of multiple potential sources of information regarding a patient’s cognitive functioning; (c) permit a more inclusive, data-informed approach to outcome capture that does not disproportionately exclude patients with the most severe cognitive decline; and (d) be inclusive of, and sensitive to, diverse ethnocultural perspectives and languages spoken.

Major Adverse Cognitive Events – MACE-Cog

Major Adverse Cardiovascular Events (MACE) have become very well-accepted as primary outcomes in cardiovascular disease, such that specific guidance has been provided by the Food and Drugs Act (FDA) and the European Medicines Agency (EMA) on utilizing a three-point MACE outcome, which includes acute myocardial infarction (MI), stroke, and cardiovascular mortality, in all trials evaluating the cardiovascular safety of diabetic agents.⁸¹ Applying this concept to the problem of evaluating cognitive outcomes after stroke, cardiac arrest, or other cardiovascular events or procedures – which could also be extrapolated to other cardiovascular or neurological diseases – we propose a composite definition of adverse neurocognitive outcomes which we call “Major Adverse Cognitive Events” (MACE-Cog).

A proposed framework for such a composite outcome, building on the conceptual shifts we have discussed, is presented in Table 2. The components of the MACE-Cog framework are envisioned to be flexible; not all components must be collected in a every study, depending on study design and available resources. Components could include: 1) cognitive screen or task performance below a cutoff or meeting MCID thresholds, 2) symptoms of impairment on validated informant/participant questionnaires, 3) inability to complete cognitive testing due to cognitive/behavioural limitations, 4) impairment in cognitively relevant activities of daily living, 5) new diagnosis of dementia, 6) institutionalization or care home admission related to cognitive impairment. While including more options to define a Major Adverse Cognitive Event will improve sensitivity, it is also important to maintain some specificity – an overarching motivation for these definitions is that they should include cognitively-relevant events that most people would agree are clearly “adverse”.

Table 2.

Proposed framework for Major Adverse Cognitive Events in the setting of stroke or other cardiovascular diseases

Dichotomous framework	Ordinal framework (example)
“No MACE-Cog” or “Normal cognitive outcome” if all components are absent.	“No MACE-Cog” or “Normal cognitive outcome” if all components are absent.
A MACE-Cog event is “positive” if any one of the following is present:  • Impairment (based on standardized cut-off or MCID) on ≥1 cognitive test (or domain), if performed as part of the study protocol, OR  • Incompletion or limitation of cognitive testing due to ≥1 cognitive-behavioural factors (Table 4), OR  • Reported symptoms of cognitive decline on standardized informant (ideal) or participant questionnaires, OR  • Impairment in cognitively-relevant instrumental activities of daily living or basic activities of daily living, OR  • New diagnosis of dementia based on clinic or health records, or prescription of dementia medications, OR  • Institutionalization or care home admission related to cognitive impairment or dementia	“Mild MACE-Cog” or “Mild cognitive impairment” if only impaired cognitive testing and/or symptoms of decline are present, or there are one or more cognitive-behavioural factors identified as limiting/preventing testing   - Single domain or multiple domain impairment could be distinguished based on number of cognitive domains affected on testing or more simply the number of components (1–3) that are positive
	“Severe MACE-Cog” or “Dementia” if there is impairment in cognitively-relevant IADLs or institutionalization, AND either impaired cognitive testing, cognitive symptoms, or cognitively-relevant reason for test incompletion.  - Additional gradations could be added based on extent of IADL loss or by weighting institutionalization higher *

^*Note that additional gradations for cognitive outcomes have been used in the R4VAD and LACI-2 studies.^{87, 88} With respect to gradations for the MACE-Cog outcome, adding more categories may make it harder to standardize implementation and may also require more intensive data acquisition (e.g. multiple cognitive tests, or a trial involving both ADL and IADL outcomes).

In a study that is using objective cognitive testing, one component of MACE-Cog is evidence of impairment on one or more completed cognitive tests (acknowledging expected missingness of this outcome). What counts as impaired performance on the tests can be flexibly considered based on the trial design and defined in the statistical analysis plan. For example, acute intervention trials (e.g. MI, stroke, cardiac arrest) would only have a single time-point of cognitive testing (i.e. at 90-days) in which case impaired performance could be defined based on published normative data for the tests, ideally against age/sex/education-standardized norms. This approach would require large enough trials to ensure that random differences in baseline cognition between treatment groups are unlikely to be attributable to chance. On the other hand, CVD prevention or recovery studies may often have longitudinal participant testing, in which case the change in the patient’s score from baseline (pre-randomization) to follow-up could be considered, with impairment defined based on a validated MCID.^{56, 82}

The second proposed component of MACE-Cog is the presence of informant- or patient-reported symptoms of cognitive decline, as captured using a standardized questionnaire like the IQCODE or AD8. Given that patients with severe deficits, neglect or anosagnosia may have reduced insight into their cognitive decline, it is ideal to have an option for an assessment that can be completed by an informant who knows the patient well. In the absence of such an informant, patient input could also be obtained, as permitted by instruments like the AD8. For trials, and participants, that can support more detailed and time-consuming assessments, the CDR-SB could also be used. Adopting the DSM-5 definition, the presence of either informant or participant concerns should be considered a positive fulfilment of this component. In trials with a single time-point of assessment, published cut-offs could be used, whereas with multiple time-points, one could look for a decline from baseline to follow-up. The latter approach, especially for a prevention/longitudinal trial, is preferable since a substantial minority of CVD patients are expected to have pre-existing cognitive impairment.⁸ Hyperacute trials may be limited to asking about symptoms or changes in function since the CVD event compared to just prior to the event.

The third proposed component of MACE-Cog is the inability to complete cognitive testing due to a relevant cognitive-behavioural factor as noted above. For this component to be considered, a study would need to consistently track the reasons for incompletion or limited completion of cognitive testing, ideally using a standardized case report form with adequate training of study personnel. We have included suggestions regarding what factors might contribute towards a MACE-Cog outcome and what factors would not, in Table 3. For instance, we argue that severe apathy/amotivation is a major neuro-behavioural manifestation that is well justified as an adverse neurocognitive outcome, given that loss of goal-directed behavior or cognitive activity is a key element.⁸³ In contrast, hearing loss that renders a participant unable to answer telephone questions would not count as a MACE-Cog event. This judgement relies on consistent assessment but subjective biases could be introduced. As with other outcome scales with potential for subjectivity (e.g. the mRS), we propose multiple mitigation strategies, including brief assessor training and certification modules, using standardized case record forms with prompts focused on observable context rather than inference, and optional blinded adjudication for ambiguous cases. While there may remain some degree of subjectivity in this rating, as long as it is applied consistently across trial groups, we feel strongly that any modest increase in subjectivity is preferable to the inevitable bias created when impaired cognitive outcomes are classified simply as missingness.

Table 3.

Suggested reasons for incompletion or limited completion of cognitive testing to track in clinical trials, that would and would not count towards Major Adverse Cognitive Events (MACE-Cog)

Reasons that would count towards MACE-Cog	Reasons that would not count towards MACE-Cog
Severe aphasia or speech disturbance	Impaired hearing or vision limiting data collection
Inappropriate or inconsistent responses	Patient refuses to participate for reasons of scheduling, convenience etc. If there is not a clear reason (when in doubt) it would not count towards MACE-Cog.
Patient prematurely terminated testing for reasons of poor performance, frustration with testing or inability to understand instructions	Patient prematurely terminated testing for reasons other than poor performance, test frustration, or inability to learn instructions.
Behavioural agitation	Severe tremor, incoordination, or weakness interfering with a written test
Impaired level of consciousness prohibiting assessment
Severe apathy or lack of motivation

The fourth MACE-Cog component is impairment in IADLs on validated scales like the Lawton or Bristol scales. Such functional impacts of cognition are often prioritized over cognitive task scores by patients, families, and cognitive experts alike.⁸⁴ Furthermore, regulators like the FDA approve treatments based on change in function; change in a neuropsychological test alone is not sufficient for an FDA indication.⁸⁵ Cognitively-relevant IADL impairment could be defined either using cut-offs on a relevant scale if there is a single assessment, or by looking for decline from baseline (pre-event or study start, depending on the trial design). The latter approach is particularly valuable for IADL evaluation because of socio-cultural variations in the extent to which different individuals perform different IADLs – for example, there remain gender disparities in such activities as cooking or driving, and patients should not be penalized for pre-existing differences in function.⁸⁶ By using the participant as their own baseline, socio-cultural or gender role assumptions are minimized and change in the individual’s own function is emphasized. Using a close observer (i.e. informant) as the source of information for the assessment adds further objectivity. Importantly, standardized instruments used to assess symptoms and functioning (such as the Lawton and Bristol scales) are not based on a simple point assessments, instead they ask respondents to think about the past few weeks for context; this is an important way in which they differ from quality-of-life assessments which can vary greatly based on how the patient is feeling at the specific moment of the assessment.

The fifth MACE-Cog component is a new clinical diagnosis of dementia, not present at study enrolment, determined from the patient’s medical records or from reliable administrative health data or prescription records of dementia-specific medications.¹⁰

The final MACE-Cog component can consider institutionalization or care home admission related to cognitive impairment or loss of function in cognitively determined ADLs. Similar mitigation strategies (standardized case record forms, training and certification for adjudicators) can reduce subjectivity, and it may be possible to identify purely physical or social reasons for admission. If assessment of cognitive symptoms and ADL changes is available, this can also support whether institutionalization is related to cognitive challenges. However, purely administratively defined outcomes will not be able to disentangle cognitive from other causes of functional decline.

For trialists, our proposed absolute minimum data that should be collected to get a MACE-Cog outcome would include: 1) cognitive symptoms (participant- or informant-reported), and 2) impact on IADLs (participant- or informant-reported). Where robust data is available, through clinical follow-up, administrative records or other sources, a new diagnosis of dementia and new institutionalization are recommended to be included. Finally, if trial resources permit, cognitive screening or assessment (at minimum a brief, telephone assessment like the Telephone Montreal Cognitive Assessment (t-MoCA) or the modified Telephone Interview for Cognitive Status (TICS-m)) that can be performed in the language of all participants, is recommended, along with reasons for incompletion.

In this manner, the MACE-Cog framework can help ensure that cognitive outcome ascertainment in CVD trials do not rely exclusively on the patient’s ability to complete cognitive testing. Our proposed approach, particularly with its consideration of patient/informant-reported cognitive symptoms and functional decline, is much better aligned with how neurocognition is evaluated in clinical practice, since neurocognitive disorders are never diagnosed solely using cognitive testing, which in turn is only interpreted in the context of reported complaints and changes in daily functioning. Importantly, the additional components of MACE-Cog can be incorporated efficiently into CVD trials with minimal added effort for outcome evaluators, particularly if trials facilitate form completion by informants/patients and linkage or review of administrative or health records, as shown in Table 4.

Table 4.

Examples of efficient methods of evaluating each MACE-Cog component

MACE-Cog components	Exemplar method of evaluation	Anticipated time required in a study visit*
Impairment (based on standardized cut-off or MCID) on ≥1 cognitive test (or domain), if performed as part of the study protocol	Face-to-face or telephone/remote cognitive testing using a validated test or test battery (see Supplementary Table 1)	<10 minutes if brief batteries are used (see Supplementary Table 1). Of course, this time will be reduced if the patient is unable to complete testing.
Incompletion or limitation of cognitive testing due to ≥1 cognitive-behavioural factors	Assessors document reasons for test incompletion or factors that limited their assessment in a standardized manner (see Table 3)	No additional time as it is based on outcome assessor’s observations of patient behaviour during testing above
Reported symptoms of cognitive decline on standardized informant or participant questionnaires	Completion of a structured questionnaire, a brief example being the Eight-item Interview to Differentiate Aging and Dementia (AD8)60	5 minutes if administered by the outcome assessor – but could also be filled out separately by informant
Impairment in cognitively-relevant instrumental activities of daily living or basic activities of daily living	Completion of a structured questionnaire, a brief example being the 8-item Lawton IADL Scale, which can be completed by either informants or patients63	<5 minutes if administered by the outcome assessor – but could also be filled out separately by informant
New diagnosis of dementia based on clinic or health records, or prescription of dementia medications	Review of diagnoses and medications in the patient’s medical records or linkage with any regional administrative records or registries tracking dementia diagnoses	<1 minute to ask about a new dementia diagnosis or related medications. No additional time if passive linkage to health or administrative records is used.
Institutionalization or care home admission related to cognitive impairment and dementia	Review of living situation with the patient/informant or tracking using health or administrative records	<1 minute to ask about current living situation. No additional time if passive linkage to health or administrative records is used.

^*Based on internally conducted timed trials with trained clinical research staff

Statistical analysis considerations for MACE-Cog

The MACE-Cog outcome, as proposed, offers useful versatility for analytical approaches in clinical trials as it can be considered in either dichotomous, ordinal, or hierarchical fashions. Any of the below statistical approaches will benefit from the MACE-Cog framework, by supporting successful outcome ascertainment in most, if not all, participants.

Binary approaches:

Binary approaches would simply consider MACE-Cog as a dichotomous outcome i.e. present or absent. When MACE-Cog is evaluated at a single time-point, such as 90-days post-stroke or 1-year post-cardiac surgery, binary analytical approaches could use binary logistic regressions for multivariable analyses, or the Chi-squared test or Fisher’s exact for univariable analyses (Figure 3A). With this approach, fulfillment of any one of the MACE-Cog components would count as a MACE-Cog event. Much as the original MACE construct does, this equates all components (see below for Hierarchical approaches as alternatives). However, when the outcome is evaluated longitudinally at multiple time points, as might be done in a preventative trial, both the occurrence and timing of attainment of MACE-Cog could be analyzed using survival curves (e.g. Kaplan-Meier curves) and models like Cox or Poisson regressions.

Figure 3.

Examples of how MACE-Cog outcomes can be analyzed using (A) binary and (B) ordinal approaches

Ordinal approaches:

As previously shown with ordinal analyses of measures like the mRS, there can be value in considering a spectrum of cognitive outcomes from both statistical and clinical standpoints.⁸⁷ In this regard, the first three components of the MACE-Cog framework may be seen as helping distinguish normal cognitive outcome (no MACE-Cog) from mild cognitive impairment (or mild MACE-Cog), whereas the last three diagnosis- and function-related components may be seen as distinguishing mild cognitive impairment from dementia (or severe MACE-Cog, Figure 3B).

One might also consider additional gradations, similar to what has been pursued recently by trials. The Rates, Risks and Routes to Reduce Vascular Dementia (R4VaD) and Lacunar Intervention Trial-2 (LACI-2) studies used a 7-point ordinal classification of cognitive decline similar to the mRS (normal, minor: single domain, minor: multi domain, major: mild, major: moderate, major: severe, and death),^{88, 89} designed to reflect DSM-5 criteria for neurocognitive impairment.⁸⁹ Each category was operationalized using a combination of T-MoCA and/or the TICS-m scores, plus mRS or Barthel functional measures. In LACI-2, 86% of people completed either a T-MoCA or TICS-m and a functional outcome and could thus generate an ordinal cognitive outcome. Although the scale minimized attrition (data was missing in 14% of participants), the scale has some limitations – for example, application for individuals with discrepant cognitive and functional status, given the mRS is not focused on cognitively-relevant activities (e.g., mild functional disability with severe cognitive scores or severe functional disability with mild cognitive impairment) may be inconsistent. Further, unlike MACE-Cog, these ordinal approaches could not account for all participants unable to complete testing. Nevertheless, it is an important attempt to create an ordinal scale that captures cognitive function and is available in majority of the participants, and provides a helpful precedent for how ordinal analyses of a MACE-Cog type outcome may be performed. Finally, as mentioned above, trials with both resources and participants able to complete a CDR-SB, could use an ordinal ranking of scores, potentially along with other aspects of MACE-Cog. Ordinal analyses of MACE-Cog could be undertaken using approaches such as ordinal logistic regression (using proportional odds or partial proportional odds models) for multivariable analyses or the Wilcoxon rank-sum or Mann-Whitney U test for univariable analyses.⁹⁰

Hierarchical approaches:

The various MACE-Cog components can also be ordered in a hierarchical manner for analysis. While individual trialists can define the most relevant hierarchy for their trial with appropriate justification, and the choice of ranking may differ by population, an exemplar hierarchy ordered to prioritize diagnostically or functionally worse outcomes could include: (1) Institutionalization or care home admission; (2) New diagnosis of dementia by a physician, or prescription of dementia medications; (3) Impairment in cognitively-relevant IADLs; (4) Reported symptoms of cognitive decline on standardized informant or participant questionnaires; (5) Impairment (based on standardized cut-off or MCID) on ≥1 cognitive test (or domain); and (6) Incompletion or limitation of cognitive testing due to ≥1 cognitive-behavioural factors. Such a hierarchy could be analyzed using an approach like a Win Ratio, which has gained popularity in recent CVD trials (Figure 4).⁹¹ A Win Ratio approach, one method for presenting and analyzing data from a Desirability Of Outcome Ranking (DOOR), would involve comparing the outcome of all possible pairs of participants, one assigned to intervention and one to control, starting at the first hierarchical outcome, and declaring a win for the participant with a better outcome within the pair, or a tie if neither participant in the pair had a better outcome. If tied at the first level of the hierarchy, the pair would proceed to the next level of the hierarchy to decide which participant fared better, and so on. The Win Ratio then ends up being the ratio of wins to losses in the treatment group compared to the control group, ignoring ties. This valuable analytical approach gives trialists the ability to assign priority to more clinically important events within the MACE-Cog framework.

Figure 4.

Example of how MACE-Cog outcomes can be analyzed using hierarchical approaches, specifically in this case using a Win Ratio approach.

The MACE-Cog framework also supports integrating cognitive outcomes into broader cardiovascular outcomes in a larger hierarchical analysis - for example, one that includes being alive, discharged home, physical function as well as cognition. Further work would be required to determine how cardiovascular survivors prioritize these various aspects along with MACE-Cog events to generate meaningful hierarchies.

Consideration of mortality:

Death poses an additional complex challenge when seeking to define and analyze an adverse cognitive outcome. As noted above, conventional cognitive assessments inevitably result in missing data for patients who have died. This problem will be particularly amplified in diseases with high mortality like cardiac arrest and intracranial hemorrhage, for which 50% or more of admitted patients do not survive to discharge.^{92, 93} If deaths inevitably mean missing cognitive data, this can lead to misleading conclusions when the treatment itself (like thrombectomy for stroke) has an effect on mortality and survivors report cognitive symptoms.⁷⁹ Death is, undoubtedly, an adverse cognitive event, as it represents the cessation of all cognition. However, can a patient who died without contributing cognitive test results truly be considered to have had an adverse cognitive event? Perhaps not – but while death does not necessarily mean a MACE-Cog event has occurred, a trial in a condition with substantial expected mortality like stroke or cardiac arrest, or in which there is a potential treatment effect on mortality, should analyze MACE-Cog in combination with mortality as a “MACE-Cog or death” outcome to avoid the bias of penalizing survivors or rewarding mortality. In a hierarchical analysis, death could be included as the most severe event, while in an ordinal analysis it could be included as the most severe outcome on the ordinal scale. With this approach, deaths will no longer equate to missing data, nor be paradoxically classified as a favourable outcome. In keeping with this philosophy, death itself was included in the 7-point cognitive ordinal outcome in R4VaD and LACI-2.^{88, 89} Sensitivity analyses could then separately examine intervention effects on the MACE-Cog component in survivors (i.e. when excluding deaths from the sample).

Future directions

One of the benefits of the MACE definition in cardiovascular disease has been its standardization by the FDA and the EMA. For a construct like MACE-Cog to gain widespread adoption, a well-justified application of the composite outcome across different studies and settings would be the most desirable way forward. We plan to conduct an expert consensus conference to standardize more challenging components of the proposed MACE-Cog definition, such as what factors for cognitive test incompletion should or should not count towards this outcome. Generation of real-world data on the MACE-Cog outcome from completed or ongoing clinical trials would help bring clarity by demonstrating the feasibility and reliability of its different components. It will be particularly illuminating to examine the agreement among MACE-Cog assessments made using various different combinations of these components.

Validation studies of the MACE-Cog outcome should be conducted in parallel across multiple CVD settings; including cardiac arrest, heart failure, stroke, and post-surgical populations. Studies must verify the extent to which our proposed approach mitigates missingness of cognitive outcomes compared to conventional cognitive test-reliant strategies. The concurrent validity of MACE-Cog should be compared to conventional cognitive approaches and to other useful outcome measures such as health-related quality of life, health and social care utilization, and caregiver burden – at least for those participants who are able to complete cognitive tasks and patient-reported outcomes. Stronger observed correlations with MACE-Cog would strengthen our confidence that it is is clinically valuable. Similarly, the predictive validity of MACE-Cog should be verified in relation to robust, long-term clinical and health economic outcomes – for example, examining how MACE-Cog at 3-months predicts quality of life, health care utilization or mortality at 1-year, compared to purely cognitive test-based methods. Such efforts are already underway and collectively, these data will help facilitate the acceptance and adoption of MACE-Cog in CVD trials.

Although the work is motivated by CVD trials, the challenges we have discussed – non-random missingness, test incompletion, and survivor biases – are not unique to these populations, and clearly extend to other conditions such as Alzheimer’s disease, multiple sclerosis, and survivors of critical illness, among others.^94–96 While further validations are required, this work advances a more inclusive, functionally relevant, and scalable framework for cognitive outcome ascertainment. MACE-Cog facilitates outcome capture across the full spectrum of survivors, including those unable to complete formal testing.

We hope that the proposed composite outcome sparks important discussions about more inclusive approaches to improve the measurement of adverse cognitive outcomes in cardiovascular diseases trials and beyond. Ultimately, we hope that this approach can support trials and interventions to reduce the burden of cognitive impairment in multiple populations.

Non-standard Abbreviations and Acronyms

CVD

Cardiovascular diseases

VasCog

Vascular Behavioural and Cognitive Disorders

VCID

Vascular cognitive impairment and dementia

MACE

major adverse cardiac events

MACE-Cog

Major Adverse Cognitive Events

mRS

modified Rankin Scale

MoCA

Montreal Cognitive Assessment

MMSE

Mini-Mental State Examination

OSC

Oxford Cognitive Screen

RCT

Randomized Controlled Trial

NINDS-CSN

National Institutes of Neurological Disorders and Stroke–Canadian Stroke Network

NeuroARC

Neurologic Academic Research Consortium

ACE-R

Addenbrooke’s cognitive examination-revised

MCID

minimal clinically important differences

OXCASC

Oxford Vascular Study

DSM-5

Diagnostic and Statistical Manual of Mental Disorders

IQCODE

Informant Questionnaire on Cognitive Decline in the Elderly

AD8

8-item Interview to Differentiate Aging and Dementia

IADL

instrumental activities of daily living

SAGEA

Standard Assessment of Global Everyday Activities

CDR-SB

Clinical Dementia Rating Scale Sum of Boxes

ESCAPE

Endovascular Treatment for Small Core and Anterior Circulation Proximal Occlusion with Emphasis on Minimizing CT to Recanalization Times

APPLE

Assessing Post-Stroke Psychology Longitudinal Evaluation

FDA

Food and Drugs Act

EMA

European Medicines Agency

MI

myocardial infarction

t-MoCA

Telephone Montreal Cognitive Assessment

TICS-m

Modified Telephone Interview for Cognitive Status

RAVaD

The Rates, Risks and Routes to Reduce Vascular Dementia

LACI-2

Lacunar Intervention Trial-2

DOOR

Desirability Of Outcome Ranking