Association of Covert Cerebrovascular Disease Identified Using Natural Language Processing and Future Dementia

Abstract

Objective: To estimate the risk of dementia associated with incidentally-discovered covert cerebrovascular disease (CCD), including both covert brain infarction (CBI) and white matter disease (WMD). Patients and Methods: We included individuals aged ≥ 50 years enrolled in the Kaiser Permanente Southern California health system receiving a head CT or MRI for a non-stroke indication from January 1, 2009 and December 31, 2019, without prior ischemic stroke, transient ischemic attack, hemipelegia, hemiparesis, dementia/Alzheimer's disease or a visit reason / scan indication suggestive of cognitive decline. Using natural language processing (NLP), we identified the presence of CBI and WMD on the neuroimage report; WMD was characterized as mild, moderate, severe, or undetermined. Results: Among 241,050 qualified individuals, NLP identified 69,931 (29.0%) with WMD and 11,328 (4.7%) with CBI. The dementia incidence rates (per 1,000 person-years) were 23.5 (95% CI 22.90 to 24.0) for patients with WMD; 29.4 (95% CI 27.9 to 31.0) with CBI and 6.0 (5.8 to 6.2) without CCD. The effect of WMD on dementia risk was stronger in younger versus older patients and for CT- versus MRI-discovered lesions. For patients with versus without WMD on CT scan, the adjusted HR (aHR) was 2.87 (2.58 to 3.19) for those < age 70 and 1.87 (1.79 to 1.95) for those ≥ age 70. For patients with versus without WMD on MRI, the aHR for dementia risk was 2.28 (1.99 to 2.62) for patients < age 70 and 1.48 (1.32 to 1.66) for those ≥ age 70. The aHR associated with CBI was 2.02 (1.70 to 2.41) for patients age <70 and 1.22 (1.15 to 1.30) for patients age ≥ 70 for either modality. Dementia risk with WMD was strongly correlated with WMD severity. Conclusion: Incidentally-discovered CCD is common and identifies patients at high risk of dementia, representing an opportunity for prevention.

Stratifying Future Stroke Risk with Incidentally Discovered White Matter Disease Severity and Covert Brain Infarct Site

“…Additionally, the incidence of future stroke risk is dependent on the imaging modality used to identify WMD severity, with mild WMD identified via CT having a similar incidence compared to severe WMD identified via MRI. This is a similar pattern to what we observed for the outcome of dementia [9] and likely reflects the much higher sensitivity of MRI for detecting WMD. Indeed, patients with mild disease on MRI are at similar risk to patients with no WMD on CT.…”

Development of Parkinson Disease and Its Relationship with Incidentally Discovered White Matter Disease and Covert Brain Infarction in a Real‐World Cohort

“…We additionally excluded patients with a visit reason or scan indication suggestive of cognitive symptoms or decline (e.g., confusion, disorientation, altered mental status, or dementia evaluation). The complete list of ICD‐10/CPT codes used to exclude patients can be found in supplements of prior manuscripts 5, 14 …”

“…The complete list of ICD-10/CPT codes used to exclude patients can be found in supplements of prior manuscripts. 5,14 If there were multiple neuroimaging studies for the same individual, the first study was considered the index scan. For neuroimaging evidence of cerebral infarction to be considered "covert," individuals were only included in the study if they did not acquire a new ICD code for a diagnosis of PD, stroke, or dementia within 180 days after the index scan.…”

“…In previous work, we developed and validated natural language processing (NLP) algorithms to identify the presence of id‐CCD and severity of incidentally discovered WMD (id‐WMD) using neuroimaging reports and deployed these algorithms in a large integrated health system, 13 thereby permitting us to examine associations between exposures and rare outcomes in a cohort by using “big data.” We previously found that id‐CCD is strongly predictive of subsequent stroke 5 and dementia, 14 particularly in younger patients and in those with more severe WMD. In this study, we examined the association of id‐CBI and id‐WMD with a subsequent diagnosis of PD, using NLP of routinely obtained neuroimaging reports to identify these lesions.…”