Screening for Cognitive Impairment, Being Cognizant of the Liminal Deities and Demons

Chetan Vekhande, Bei Jiang, Mahesh KateCanadian 2020 · DOI · OpenAlex

Abstract

Screening for Vascular Cognitive Impairment (VCI) after stroke or transient ischemic attack (TIA) often relies on binary "pass/fail" thresholds that fail to account for clinical complexity. This paper discusses a shift toward a tiered, Bayesian framework that integrates clinical history, neuroimaging, and sequential testing. By moving beyond simple cut-off scores, clinicians can better navigate the "liminal" diagnostic zones, improving accuracy and patient management in cerebrovascular disease.

Introduction

Cognitive impairment (CI) is a frequent and debilitating consequence of cerebrovascular disease, affecting up to 60% of patients following a moderate-to-severe stroke. While the Montreal Cognitive Assessment (MoCA) has become the standard screening tool, its clinical utility is often hampered by the lack of uniform threshold values. Relying on a single cut-off score frequently leads to high rates of false positives or negatives, particularly in cases of mild impairment. This editorial argues for a more nuanced approach, treating cognitive screening not as a binary outcome but as a Bayesian process of updating diagnostic probability based on clinical context.

The Pitfalls of Binary Diagnostic Logic

Current diagnostic practices often fall prey to two logical "demons":

  1. Reference Standard Bias: The belief that neuropsychological batteries are a perfect "gold standard." In reality, these tests have their own inherent sensitivities and specificities.
  2. Conditional Independence Fallacy: The assumption that errors in a screening tool (like the MoCA) are independent of errors in a secondary test. Since both often measure overlapping cognitive domains, their errors are frequently correlated.

A Tiered Bayesian Framework

To address these issues, a "tiered" screening approach is proposed. Instead of a single threshold, patients are categorized into three groups based on their MoCA scores:

  • Low Probability (>27): High likelihood of being cognitively intact.
  • High Probability (<24): High likelihood of impairment.
  • Indeterminate (24–27): A "liminal" zone requiring further investigation.

For patients in the indeterminate group, sequential testing—using specific metrics like the Symbol Digit Modalities Test or learning scores—is employed to refine the diagnosis.

Bayesian update of posterior probability based on prior probability. Panel A shows a symmetric relationship, while Panel B illustrates asymmetry where a negative test is less reliable in the presence of high disease burden.

The Role of Prior Probability and Imaging

A core tenet of this approach is that a test score should never be interpreted in isolation. Clinicians must consider "pre-test" factors—such as age, stroke severity, and neuroimaging markers—to establish a prior probability of CI.

Neuroimaging, in particular, serves as a powerful modifier. Markers of small vessel disease, such as white matter hyperintensities (WMH), significantly shift the diagnostic landscape. The presence of WMH can increase the odds of CI by nearly twofold, while their progression increases the odds sixfold.

In patients with significant disease burden, the relationship between prior and posterior probability becomes asymmetrical. As shown in the figure above (Panel B), a "negative" (high) MoCA score in a high-risk patient is less reliable for ruling out impairment than it would be in a low-risk patient (Panel A). This is because the high prior probability "drags" the posterior probability upward, even in the face of a negative test result.

Key Results and Clinical Implications

Implementing this tiered approach has shown significant promise, accurately classifying 79% of patients in a cohort of stroke and TIA survivors. The use of specific likelihood ratios (e.g., 5.4 for high-probability MoCA scores and 0.18 for low-probability scores) allows clinicians to mathematically "update" their diagnostic confidence.

Furthermore, the concept of "brain reserve" is critical. A patient with high baseline cognitive reserve may maintain a "normal" MoCA score despite significant underlying pathology, whereas a low score in a patient with no clinical or imaging markers of disease may be less indicative of true vascular impairment.

Conclusion

Moving forward, the goal of cognitive screening should be to move beyond the "pass/fail" mentality. By adopting a Bayesian perspective that incorporates clinical modifiers and sequential testing, clinicians can more accurately identify and manage patients in the complex gray zones of cognitive decline. Future work must focus on quantifying the impact of various clinical modifiers to make these nuanced assessments more feasible in standard clinical practice.

Summary updated

Abstract

An abstract is not available for this content. As you have access to this content, full HTML content is provided on this page. A PDF of this content is also available in through the ‘Save PDF’ action button.
BibTeX
@inproceedings{vekhande2020screening,
  author = {Chetan Vekhande and Bei Jiang and Mahesh Kate},
  title = {{Screening for Cognitive Impairment, Being Cognizant of the Liminal Deities and Demons}},
  booktitle = {Canadian},
  year = {2020},
  doi = {10.1017/cjn.2020.137},
}
Figure