Vascular cognitive impairment in China: Epidemiology, risk factors, healthcare system, management challenges, therapeutic modalities, and prevention strategies

Highlights

• •VCI burden in China is substantial but varies across phenotypes and settings.
• •PSCI offers a practical entry point for detection and secondary prevention.
• •Hypertension control provides the strongest indirect support for VCI prevention.
• •Direct treatment evidence for established VCI syndromes remains limited.
• •Scalable care in China requires tiered screening, referral, and follow-up.

Abstract

Vascular cognitive impairment (VCI) is an increasingly important contributor to cognitive decline in China, where rapid population aging intersects with a substantial burden of stroke, cerebral small vessel disease, and cardiometabolic risk. This narrative review synthesizes Chinese epidemiological, clinical, health-services, and policy literature on VCI, supplemented by key international consensus and interventional evidence when directly relevant. Available evidence indicates that the vascular contribution to cognitive disorders in China is substantial but heterogeneously measured because of differences in case definition, screening tools, study setting, and follow-up design. Post-stroke cognitive impairment is especially important because it is common, clinically recognizable, and reachable through existing stroke follow-up pathways. Across studies, hypertension and cardiometabolic multimorbidity emerge as the most actionable modifiable drivers. China has strengthened its policy response through updated clinical guidance and a national dementia action plan, yet implementation remains constrained by inconsistent screening, uneven specialist capacity, fragmented referral pathways, and limited longitudinal follow-up. Therapeutically, direct evidence for interventions targeting established VCI syndromes remains limited. The strongest available support is indirect and comes from vascular risk reduction, particularly blood pressure lowering for broader dementia or cognitive decline endpoints, while multidomain and symptomatic interventions show phenotype-specific but more heterogeneous signals. A practical China-adapted response is therefore a tiered life-course model linking chronic disease management, post-stroke surveillance, specialist diagnosis, multidomain intervention, and caregiver support. Future priorities include harmonized diagnostics, clearer evidence characterization, biomarker-enabled longitudinal cohorts, and pragmatic implementation-science studies focused on adaptation, cost, and sustainability in community and rural settings.

1. Introduction

Vascular cognitive impairment (VCI) is best understood as a spectrum ranging from mild vascular cognitive disorder to major vascular cognitive disorder or vascular dementia (VaD), and it frequently overlaps with neurodegenerative pathology rather than appearing as a purely vascular syndrome [[1], [2], [3], [4], [5], [6], [7], [8]]. The 2024 Chinese guideline identifies VCI as the second leading cause of cognitive impairment in people aged 60 years or older in China and contextualizes Jia et al.'s national survey estimate that dementia prevalence in this age group is 6.0%, including 1.6% with vascular dementia [[9], [10]]. This mixed-pathology reality matters clinically because it complicates diagnosis, prognosis, and treatment planning. It matters especially in China, where rapid population aging is occurring alongside a sustained burden of stroke, hypertension, diabetes, and cerebral small vessel disease [9,[11], [12], [13], [14], [15], [16]]. Globally, dementia prevalence is also projected to increase substantially by 2050, reinforcing the need for prevention-oriented vascular-cognitive strategies in high-burden countries such as China [17].

In the Chinese healthcare system, VCI is not only a memory-clinic problem. Many patients first appear in hypertension clinics, diabetes follow-up programs, cardiology services, stroke units, rehabilitation settings, or community primary care rather than in specialist cognitive clinics [10,[18], [19], [20], [21], [22]]. As a result, outcomes depend not only on the biology of vascular brain injury, but also on whether the health system can identify high-risk individuals early, perform cognitive surveillance in routine care, and connect patients to appropriate staged interventions. A China-focused review therefore needs to integrate epidemiology, risk factors, service capacity, treatment options, and implementation barriers rather than discussing pathology alone.

Recent harmonization efforts from NINDS-CSN, VASCOG, and later updates have improved conceptual clarity and cross-study comparability [1,3,[6], [7]]. However, operational heterogeneity remains substantial in Chinese real-world settings, particularly in cognitive test selection, case thresholds, neuroimaging use, and follow-up practices. Unlike the 2024 Chinese guideline, which is recommendation-focused [10], the national dementia action plan, which is policy-focused [22], or the recent China-specific prevalence meta-analysis, which is burden-focused [12], the purpose of this review is to compare study-level sources of heterogeneity, distinguish direct syndrome-level treatment evidence from indirect vascular-risk-reduction evidence, and translate those findings into a pragmatic China-adapted implementation pathway. The emphasis is therefore less on restating broad burden claims and more on showing where diagnostic heterogeneity, endpoint directness, and service-delivery realities most affect interpretation for China.

2. Methods

This article is a narrative review rather than a formal systematic review. It prioritizes Chinese epidemiological studies, national guidelines, health-services surveys, registries, and interventional studies relevant to VCI, PSCI, cerebral small vessel disease (CSVD), and dementia prevention, while incorporating key international consensus statements and trials when needed to interpret endpoint directness or implementation relevance. Relevant English- and Chinese-language literature was identified through targeted searches of PubMed, Web of Science, and CNKI, with emphasis on studies published from January 2000 through January 2026. Search terms included combinations of 'vascular cognitive impairment', 'vascular dementia', 'post-stroke cognitive impairment', 'cerebral small vessel disease', 'China' or 'Chinese', 'stroke', 'hypertension', 'dementia prevention', 'multidomain intervention', and 'implementation'; backward citation tracking of key guidelines, consensus statements, and review articles was also used to identify additional relevant sources. Studies were preferentially included when they reported Chinese epidemiological, risk-factor, health-system, or interventional data with interpretable phenotype definitions, settings, and endpoints, or when non-Chinese consensus or trial data were needed to contextualize directness of evidence. Case reports, highly repetitive reports without additional methodological detail, and basic mechanistic papers without clear relevance to the China-focused clinical or implementation questions were not emphasized. Because this review was designed as a narrative synthesis rather than a formal systematic review, a PRISMA-style screening log and study-flow count were not prospectively maintained. Conclusions are therefore framed as a structured narrative synthesis organized by directness, phenotype specificity, and implementation relevance rather than as a formal graded evidence review or pooled causal estimate.

For terminological clarity, VCI is used here as an umbrella construct spanning mild vascular cognitive disorder or vascular cognitive impairment no dementia (VCIND) to major vascular cognitive disorder, including VaD. PSCI is defined by its temporal relationship to stroke and may occur across the mild-to-major VCI spectrum; it is therefore not treated as a separate category parallel to VCIND or VaD. Mixed vascular–neurodegenerative pathology may overlap with these vascular-cognitive phenotypes, whereas cognitive and executive dysfunction, gait and functional decline, loss of independence, and caregiver burden are presented as downstream clinical consequences rather than separate diagnostic phenotypes. The review is therefore organized along a burden–risk–evidence–implementation pathway that moves from epidemiology to practical action. Fig. 1 summarizes the conceptual pathway from vascular and social exposures through vascular brain injury to the VCI spectrum and its downstream clinical consequences, while mapping the major healthcare entry points for detection and management in China.

Fig. 1.

Vascular cognitive impairment in China: from risk to care. Abbreviations: BP, blood pressure; CSVD, cerebral small vessel disease; TIA, transient ischemic attack; VCI, vascular cognitive impairment; VCIND, vascular cognitive impairment no dementia; PSCI, post-stroke cognitive impairment; VaD, vascular dementia.

2.1. Epidemiology in China

Chinese epidemiological evidence on vascular-cognitive disorders should be interpreted at three levels: national dementia context, China-specific VCI meta-analytic estimates, and phenotype-specific stroke cohorts (Table 1). Two national surveys provide the broadest context but are not direct VCI prevalence studies. Jia et al. examined 46,011 adults aged 60 years or older across 96 sites using a two-stage evaluation with MMSE, CDR, neuropsychological testing, and mandatory neuroimaging for dementia diagnosis and reported dementia prevalence of 6.0%, VaD prevalence of 1.6%, and MCI prevalence of 15.5% [9]. Gan et al. later studied 20,438 adults aged 65 years or older from 14 centers, 28 communities, and 56 villages using MMSE-triggered second-stage assessment and reported crude dementia prevalence of 9.1% and VaD prevalence of 1.4% [11].

Table 1.

Chinese epidemiological studies of VCI, PSCI, and VaD.

Study	Setting / region	Sample size	Phenotype	Diagnostic criteria / case definition	Screening / cognitive tool	Cutoff / threshold	Time window	Main estimate	Likely source(s) of heterogeneity / interpretive note
Jia et al., 2020 [9]	National cross-sectional survey; 96 sites in 12 provinces/municipalities	46,011 adults aged ≥60 years	Dementia, VaD, MCI (context-setting, not VCI-specific)	Cognitively normal if global CDR=0; dementia by DSM-IV-TR; AD by NINCDS-ADRDA or NIA-AA; VaD by NINDS-AIREN; MCI by NIA-AA workgroup criteria	MMSE, WHO-UCLA Auditory Verbal Learning Test, ADL, Hachinski Ischemic Score, physician-rated CDR; MRI/CT compulsory for dementia diagnosis	No single screening cutoff; classification based on CDR plus syndrome-specific criteria after battery and imaging	Cross-sectional; recruitment 2015–2018	Dementia 6.0%; VaD 1.6%; MCI 15.5%	National context-setting study, not a direct VCI prevalence study.
Gan et al., 2024 [11]	National cross-sectional study; 14 centers, 28 communities, 56 villages	20,438 adults aged ≥65 years	Dementia and VaD prevalence (not VCI-specific)	Dementia by DSM-5; AD by NINCDS-ADRDA; VaD by NINDS-AIREN	MMSE, ADL, second-stage examination, medical record review, daily living questionnaires	MMSE ≤17 illiterate, ≤20 for 1–6 years education, ≤24 for ≥7 years education triggered second-stage assessment	Cross-sectional; May-Dec 2019	Crude dementia 9.1%; VaD 1.4%; standardized dementia 8.8%	Older sampling frame and later survey period complicate direct comparison with Jia et al.
Liang et al., 2025 [12]	Systematic review and meta-analysis of Chinese observational studies	81 studies; 784,846 participants in prevalence dataset	VCI prevalence and incidence	Mixed across included studies	Mixed across included studies	Mixed across included studies	Studies from 1980–2023	Pooled prevalence 1.54%; pooled incidence 0.29 per 100 person-years	Useful for burden framing, but source studies are highly heterogeneous by design.
Zhou et al., 2005 [23]	Hospital-based ischemic stroke cohort; Chongqing	434 consecutive ischemic stroke patients	Post-stroke cognitive impairment / stroke-related cognitive impairment	Pre-stroke cognition from IQCODE; post-stroke cognitive impairment defined with stroke-concurrent or stroke-related cognitive impairment	IQCODE on admission; MMSE and RVR at 3 months post-stroke	MMSE 17/20/24 by education; RVR ≤15; prestroke impairment screened by IQCODE	3 months after ischemic stroke	Post-stroke cognitive impairment 37.1%; stroke-related cognitive impairment 32.2%; after first-ever stroke 29.6%	Ischemic-stroke hospital cohort with education-adjusted MMSE thresholds; not interchangeable with community VCI prevalence.
Zhang et al., 2012 [24]	Hospital-based multicenter study across 5 geographic areas of China	633 recruited; 577 completed cognitive tests	PSCI after first-ever stroke	PSCI diagnosed according to VCI criteria; impairment defined below 10th percentile on at least one domain	MoCA, Chinese MMSE, then 30-min domain battery: Fuld Object Memory Test, Rapid Verbal Retrieve, Block Design, Digit Span	MoCA <26; C-MMSE <20 illiterate, <23 elementary education, <27 middle/higher education; excluded if s-IQCODE ≥55 before stroke	3 months after first-ever stroke	PSCI incidence 30.7%	Two-stage screening plus domain testing likely yields lower estimates than ultrabrief high-sensitivity screens.
Qu et al., 2015 [18]	Community-based cross-sectional stroke-survivor study; Fangshan (rural Beijing) and Huangpu (urban Shanghai)	607 evaluated; 599 included (48% of registered stroke survivors in 4 communities)	PSCI, PSCI-ND, and post-stroke dementia	Neurologist diagnosis based on MoCA/MMSE/HIS plus clinical information; CT/MRI-confirmed prior stroke required	MoCA, MMSE, HIS	MoCA <26; MMSE ≤17 illiterate, ≤20 primary school, ≤24 middle school+; HIS ≥7 for post-stroke vascular dementia	Mixed chronicity after stroke; mean 4.5 years since last stroke	PSCI 80.97%; PSCI-ND 48.91%; post-stroke dementia 32.05%	Community registry sampling, rural inclusion, long time since stroke, and MoCA-based ascertainment likely increased case capture.
Chen et al., 2016 [26]	Hospital-based post-stroke cohort; Nanjing	56 analyzed post-stroke patients aged ≥50 years	Post-stroke VCI	VCI if impairment present in ≥1 domain on 60-min Chinese NINDS-CSN battery and no pre-stroke impairment	60-min Chinese NINDS-CSN battery; MMSE and MoCA also recorded; IQCODE informant screen for prestroke status	Domain impairment if >1.5 SD below age- and education-matched controls; excluded if IQCODE ≥3.4	3–12 months after stroke; mean 7.1 months	VCI 55.4%	Small but deeply phenotyped cohort; demonstrates that brief global screens may miss domain-specific post-stroke impairment.
He et al., 2023 [25]	Multicenter hospital-based stroke-center network; 563 centers in 30 provinces	24,055 first-ever ischemic stroke patients	PSCI after first-ever ischemic stroke	Clinical first-ever ischemic stroke with overt symptoms confirmed by MRI/CT; prestroke cognitive impairment excluded	5-min NINDS-CSN	Cutoff <9	3–6 months after indexed stroke	PSCI incidence 78.7%	Large hospitalized cohort using a brief high-sensitivity screen; best interpreted as screen-positive PSCI burden, not general VCI prevalence.

Abbreviations: ADL, activities of daily living; CDR, Clinical Dementia Rating; C-MMSE, Chinese Mini-Mental State Examination; CT, computed tomography; DSM, Diagnostic and Statistical Manual of Mental Disorders; HIS, Hachinski Ischemic Score; IQCODE, Informant Questionnaire on Cognitive Decline in the Elderly; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; MRI, magnetic resonance imaging; NIA-AA, National Institute on Aging-Alzheimer's Association; NINCDS-ADRDA, National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association; NINDS-AIREN, National Institute of Neurological Disorders and Stroke-Association Internationale pour la Recherche et l'Enseignement en Neurosciences; NINDS-CSN, National Institute of Neurological Disorders and Stroke-Canadian Stroke Network; PSCI, post-stroke cognitive impairment; PSCI-ND, post-stroke cognitive impairment no dementia; RVR, Rapid Verbal Retrieve; s-IQCODE, short-form IQCODE; VaD, vascular dementia; VCI, vascular cognitive impairment; WHO-UCLA, World Health Organization-University of California Los Angeles.

These national datasets are valuable because they quantify the cognitive-health landscape in which vascular syndromes occur, but they should not be treated as direct estimates of VCI. The more recent China-specific meta-analysis by Liang et al. pooled 81 observational studies and reported a prevalence of 1.54% and an incidence of 0.29 per 100 person-years for VCI [12]. However, that synthesis combined studies published from 1980 to 2023 that used mixed case definitions, mixed settings, and inconsistent cognitive tools, so the pooled figures are better interpreted as burden signals than as operational benchmarks for service planning.

The clearest illustration of heterogeneity comes from post-stroke cohorts, which capture a time-anchored PSCI phenotype rather than population VCI. In the Chongqing Stroke Study, Zhou et al. followed 434 ischemic stroke patients and defined stroke-related cognitive impairment at three months using admission IQCODE and post-stroke MMSE plus rapid verbal retrieval testing; the corresponding estimate was 32.2% [23]. In the five-region first-ever stroke study by Zhang et al., 633 patients were recruited and 577 completed testing, and PSCI at three months was 30.7% using a two-stage screen (MoCA <26 or education-adjusted MMSE) followed by a domain battery with impairment defined below the 10th percentile in at least one domain [24].

By contrast, Qu et al. reported PSCI in 80.97% of 599 community-registered stroke survivors and post-stroke dementia in 32.05% [18]. That study differed fundamentally from the two hospital cohorts: stroke survivors were identified through community health service center registries, mean time since the last stroke was 4.5 years, rural participants were included, and case definition relied on MoCA, MMSE, and Hachinski scores rather than a uniform neuropsychological protocol. The resulting estimate therefore reflects a chronic mixed-survivor community phenotype, not a directly comparable 3-month post-stroke incidence.

A different source of variation is screening sensitivity. He et al. analyzed 24,055 first-ever ischemic stroke patients from 563 stroke centers in 30 provinces and found a PSCI incidence of 78.7% at 3–6 months using the 5-min NINDS-CSN protocol with a cutoff of <9 [25]. This instrument is brief and operationally attractive for large-scale stroke follow-up, but it is intentionally sensitive, and the study excluded patients unable to complete assessment. The very high estimate should therefore be interpreted as an upper-range screen-positive burden within hospitalized first-ever ischemic stroke survivors rather than as a directly interchangeable VCI prevalence figure.

Smaller, more deeply phenotyped cohorts further show that cognitive case yield depends on how intensively cognition is measured. Chen et al. assessed 56 Chinese post-stroke patients a mean of 7.1 months after onset using the 60-min Chinese NINDS-CSN battery and diagnosed VCI when impairment was present in at least one domain at >1.5 SD below matched controls, yielding a prevalence of 55.4% [26]. In the same cohort, conventional MMSE cutoffs would have identified only a minority of cases, underscoring that executive and visuoconstructive deficits can be missed by simpler screens [26].

For this reason, VCI, PSCI, and VaD should not be used interchangeably in the interpretation of Chinese studies. VCI is an umbrella construct, PSCI is a post-event phenotype with explicit time anchoring, and VaD represents the major syndrome at the severe end of the spectrum [1,3,[6], [7], [8],10]. Differences in education-adjusted thresholds, use or non-use of informant screening for pre-stroke decline, and variable access to neuroimaging further shift observed rates. Table 1 is intended to make these methodological contrasts visible rather than leaving them implicit.

Even with these definitional differences, the public health implication is consistent. China has a very large exposed population because stroke prevalence and incidence remain high, and vascular risk exposure accumulates long before patients reach memory clinics [10,[15], [16]]. Epidemiological interpretation should therefore be longitudinal: midlife hypertension, diabetes, and small-vessel injury feed into later stroke, chronic disability, and cognitive decline. From a service perspective, this means that stroke follow-up clinics, chronic disease management platforms, and community primary care are all epidemiologically relevant entry points into the vascular-cognitive pathway.

The most robust epidemiological conclusion is therefore not a single national percentage, but a pattern. National surveys show a substantial dementia and MCI burden with a meaningful vascular component [[9], [10], [11]]; meta-analysis confirms that China already has a measurable VCI burden [12]; and stroke cohorts show that PSCI is common enough to justify routine surveillance after cerebrovascular events [18,[23], [24], [25], [26]]. The remaining uncertainty is chiefly about where on the spectrum a given study is sampling, which instrument it is using, and how aggressively it defines impairment.

2.2. Risk factors

Most Chinese data on VCI risk factors are observational rather than interventional, and key Chinese risk-factor studies are summarized in Table 2. These studies are highly valuable for risk stratification and prevention planning, but they do not provide equally strong evidence that modifying every associated factor will change VCI-specific outcomes. This distinction matters when association signals are translated into clinical or policy recommendations.

Table 2.

Risk factors for vascular cognitive outcomes in Chinese studies.

Factor	Study / cohort	Population	Outcome	Main finding	Study type	Main limitation
Age / lower education	Jia et al. 2020 [9]; Gan et al. 2024 [11]	National older-adult samples	Dementia, VaD, MCI	Older age and lower education consistently associated with higher cognitive risk and dementia burden	National cross-sectional surveys	Broad cognitive outcomes, not VCI-specific incidence
Earlier hypertension onset / longer exposure	Ding et al. 2023 [28]	Chinese adults in longitudinal analysis	Cognitive decline trajectories	Earlier hypertension diagnosis associated with less favorable cognitive trajectories	Longitudinal observational study	Observational exposure-outcome association, not intervention evidence
Antihypertensive adherence	Li et al. 2025 [29]	Chinese adults using antihypertensive therapy	Cognitive aging	Better medication adherence associated with slower cognitive decline	Observational cohort analysis	Adherence may proxy broader health behavior
Cardiometabolic multimorbidity	Liu et al. 2024 [30]	Rural older adults in China	Dementia risk	Cardiometabolic multimorbidity associated with higher dementia risk	Observational cohort	Dementia endpoint broader than VCI
Prior / recurrent stroke	Qu et al. 2015 [18]; Chen et al. 2016 [26]	Community and hospital post-stroke cohorts	PSCI / post-stroke VCI	Recurrent stroke associated with higher odds of cognitive impairment	Cross-sectional and cohort studies	Phenotype restricted to stroke survivors
Acute-phase stroke complications	Qu et al. 2015 [18]	Community stroke survivors	PSCI	Complications during acute phase independently associated with PSCI	Community cross-sectional study	Cross-sectional design and mixed chronicity after stroke
Lower education / female sex / recurrent stroke / cortical atrophy	Chen et al. 2016 [26]	Chinese post-stroke hospital cohort	Post-stroke VCI	Independent predictors included lower education, female sex, recurrent stroke, and global cortical atrophy	Hospital cohort with multivariable regression	Small sample size
CSVD and microvascular markers	Yang et al. 2023 [35]; Jiang et al. 2023 [36]; Zhou et al. 2025 [37]	Community and cohort studies	Cognitive impairment / MCI	WMH, lacunes, and microvascular dysfunction associated with worse cognitive outcomes	Observational imaging studies	Marker-outcome associations not equivalent to validated clinical staging
Social engagement / social vulnerability	Zhou et al. 2018 [31]	Chinese older adults	Dementia risk	Lower social engagement associated with higher dementia risk	Longitudinal observational study	Not specific to vascular cognitive phenotypes

Abbreviations: CSVD, cerebral small vessel disease; MCI, mild cognitive impairment; PSCI, post-stroke cognitive impairment; VaD, vascular dementia; VCI, vascular cognitive impairment; WMH, white matter hyperintensities.

Chinese evidence is consistent with global literature in showing that advanced age and lower educational attainment are major correlates of cognitive decline and dementia [9,11,[27], [28], [29], [30], [31], [32], [33], [34]]. Social determinants also matter. Lower socioeconomic status, reduced social engagement, sensory loss, depression, air pollution, physical inactivity, smoking, harmful alcohol use, obesity, diabetes, hypertension, and broader social vulnerability are among the modifiable or partially modifiable dementia-risk domains emphasized by the Lancet Commission reports [[33], [34]]. In practice, these factors influence not only disease incidence, but also the stage at presentation and the feasibility of sustained management.

Among modifiable factors, hypertension is the most important and actionable target in China. Longitudinal Chinese analyses show that earlier hypertension onset and longer exposure are associated with less favorable cognitive trajectories, while better antihypertensive adherence is associated with slower cognitive decline [[28], [29]]. This combination of epidemiological association, biological plausibility, and intervention feasibility makes blood pressure control the clearest and best-supported prevention anchor for Chinese VCI strategy, even though much of the intervention literature uses broader dementia or cognitive decline endpoints rather than uniform VCI-specific outcomes.

Mechanistic interpretation is most useful when tied back to Chinese longitudinal and imaging data rather than left at a textbook level. In Chinese cohorts, earlier hypertension onset and poorer antihypertensive adherence are associated with less favorable cognitive trajectories [[28], [29]], while CSVD markers such as white matter hyperintensities and lacunes are linked to worse vascular-cognitive outcomes [[35], [36], [37]]. These observations fit a pathway in which long-standing blood pressure dysregulation promotes arterial stiffness, endothelial dysfunction, impaired autoregulation, and small vessel injury, which then appear clinically as non-amnestic slowing, executive dysfunction, and later disability. Because hypertension control remains suboptimal in many Chinese regions, the mechanistic argument is not abstract: it helps explain why chronic disease clinics and stroke follow-up services are such important prevention platforms.

Diabetes, dyslipidemia, and cardiometabolic multimorbidity are repeatedly associated with higher cognitive risk in Chinese cohorts [30,38]. Multimorbidity likely reflects cumulative vascular stress, metabolic dysregulation, and inflammatory pathways rather than merely the sum of independent disease effects. This supports integrated risk-management models in which blood pressure, glycemia, lipid profile, kidney function, weight, and lifestyle factors are managed together with cognitive preservation as an explicit goal.

Prior stroke and recurrent ischemic injury are strong predictors of subsequent cognitive decline [18,[23], [24], [25], [26]]. Cognitive outcomes after stroke reflect lesion burden, recurrence risk, premorbid reserve, comorbidity, and rehabilitation quality. Secondary stroke prevention programs should therefore include cognitive follow-up as a core outcome rather than focusing only on recurrent vascular events or mortality.

Chinese cohort studies increasingly highlight CSVD markers as useful tools for risk stratification [[35], [36], [37]]. White matter hyperintensities, lacunes, and related imaging findings are associated with worse cognitive outcomes and may help identify high-risk individuals before overt dementia develops. As imaging access and biomarker infrastructure improve in China, CSVD-related markers may become more useful for both clinical staging and trial enrichment.

2.3. System capacity and access in China

The implementation challenge in China has two layers: whether sufficient capacity exists, and whether patients move through that capacity in a coherent longitudinal pathway. This section focuses on the structural layer, including policy momentum, specialist access, primary care reach, and economic pressure. The next section then focuses on care-delivery failures that occur along the patient pathway even when some capacity is already present.

China has recently strengthened the policy framework relevant to VCI. The 2024 Chinese guideline for vascular cognitive impairment provides updated recommendations for diagnosis and management [10], and the national dementia action plan for 2024–2030 sets strategic goals for awareness, screening, referral, workforce development, and long-term care coordination [22]. This policy momentum is important because it moves VCI from a loosely defined specialist issue toward a recognized national aging and chronic disease priority.

Policy progress does not eliminate capacity constraints. National survey data still show limited availability of specialized memory clinics and dementia inpatient units in many hospitals, with marked regional imbalance in specialist distribution [19]. In many settings, patients are assessed late or are managed in non-specialized services where cognitive evaluation is variable in depth and consistency.

Capacity problems are not only about the number of clinics. They also include uneven access to formal neuropsychological testing, MRI-based CSVD characterization, caregiver counseling, and longitudinal follow-up infrastructure. Because VCI diagnosis often depends on combining vascular history, cognitive change, imaging evidence, and time course, fragmented access to these elements increases under-recognition, especially in mild or mixed cases [10,19].

Primary care should be the first contact point for high-risk adults with hypertension, diabetes, prior stroke, gait decline, or functional complaints. In practice, however, cognitive screening and referral workflows remain inconsistent across Chinese frontline settings [19,27]. A feasible model in overloaded community clinics is to separate case finding from diagnostic work-up: a brief trigger step embedded in routine chronic disease review, followed by a short-structured screen only for patients with concern, and specialist escalation only when abnormalities persist. This sequencing is essential because chronic disease platforms already reach large exposed populations, whereas specialist cognitive resources remain unevenly distributed.

A candidate stepped workflow is outlined in Table 3. In hypertension or diabetes follow-up, trigger questions on new forgetfulness, slowed thinking, medication mismanagement, gait-functional decline, or caregiver concern can be asked by a community physician or nurse at least annually. If concern is present, a brief first-line tool such as the AD8 when an informant is available, or an education-aware cognitive screen such as MoCA-Basic/MoCA or MMSE using locally validated Chinese education-adjusted thresholds, can be completed during the same visit or within 2–4 weeks. In stroke follow-up, the 5-min NINDS-CSN protocol is a practical first-line option at 3–6 months and again around 12 months because it is brief, domain-sensitive, and already used in large Chinese cohorts [25]. Patients with abnormal screening results, rapid functional decline, or diagnostic uncertainty should then be referred for formal neuropsychological assessment and phenotype clarification.

Table 3.

Candidate stepped screening and referral workflow for Chinese settings.

Setting	Trigger item(s)	First-line tool	Second-stage tool	Who administers	Suggested interval	Referral trigger	Feasibility notes
Hypertension / diabetes follow-up in community primary care	New forgetfulness, slowed thinking, medication mismanagement, gait-functional decline, caregiver concern	AD8 when informant available; if no informant, brief cognitive screen in same visit	MoCA-Basic / MoCA or MMSE with education-adjusted thresholds plus ADL	Community physician or nurse	At least annually; earlier if new symptoms emerge	Persistent trigger-positive screen, AD8 ≥2, abnormal cognitive screen, or rapid functional decline	Separates case finding from full work-up; keeps burden low in busy chronic-disease clinics.
Post-stroke clinic / rehabilitation follow-up	Any cognitive complaint after stroke or routine surveillance after first-ever ischemic stroke	5-min NINDS-CSN	MoCA or formal neuropsychological battery if available	Stroke nurse, rehabilitation therapist, or clinic physician	3–6 months after stroke and again around 12 months	5-min NINDS-CSN <9, decline in function, or diagnostic uncertainty	Best aligned with existing large Chinese PSCI data; operationally brief and domain-sensitive.
County hospital neurology / stroke clinic	Abnormal first-line screen or complex mixed symptoms	MoCA / MoCA-Basic, functional staging, mood screen	Imaging review and syndrome clarification	Neurologist with trained nurse support	At first abnormal referral and again in 6–12 months	Persistent cognitive-functional decline, mixed pathology, behavior symptoms, recurrent stroke	Useful bridge between primary care and tertiary memory services.
Memory clinic / tertiary center	Referral from community or stroke services for confirmation	Comprehensive syndrome review	Formal neuropsychology, imaging integration, caregiver counseling, longitudinal plan	Neurologist / geriatrician / neuropsychology team	Individualized; commonly every 6–12 months	Confirmed VCIND / PSCI / VaD or unresolved diagnostic complexity	High specificity but limited capacity; should be reserved for confirmation and complex cases.

Note. This table is a candidate implementation workflow intended for pragmatic adaptation across Chinese care settings; it is not a validated national protocol. Abbreviations: AD8, Eight-item Informant Interview to Differentiate Aging and Dementia; ADL, activities of daily living; CSVD, cerebral small vessel disease; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; NINDS-CSN, National Institute of Neurological Disorders and Stroke-Canadian Stroke Network; PSCI, post-stroke cognitive impairment; VaD, vascular dementia; VCIND, vascular cognitive impairment no dementia; VCI, vascular cognitive impairment.

Chinese studies consistently show that dementia imposes high direct and indirect economic burden, with costs rising as impairment becomes more severe [[20], [21]]. This cost pressure is highly relevant to VCI because many patients also live with stroke-related disability, gait problems, depression, or broader multimorbidity. Delayed diagnosis can therefore increase repeated hospital use, informal caregiving burden, and inefficient medication management. Earlier identification of vascular cognitive syndromes should thus be viewed not only as a diagnostic objective, but also as a cost-containment and disability-reduction strategy.

Taken together, the Chinese healthcare context supports a layered service model. Primary care should identify risk and triage patients; stroke units and neurology clinics should implement routine post-event cognitive surveillance; memory clinics and tertiary centers should confirm complex or mixed diagnoses; and community or long-term care services should provide continuity, caregiver support, and disability prevention. This architecture is broadly compatible with current policy direction, but it requires clearer referral standards, stronger data linkage, and explicit assignment of responsibility for longitudinal follow-up [10,19,22].

2.4. Care delivery gaps along the patient pathway

Many patients with cognitive symptoms still enter specialist care at moderate or advanced stages [19]. This delays preventive intervention and makes etiologic differentiation between vascular, mixed, and neurodegenerative processes more difficult. In VCI, late presentation is particularly costly because the largest gains are likely to come from modifying vascular exposures before functional decline becomes entrenched. Time to first cognitive evaluation should therefore be treated as a service-quality metric, not merely a descriptive feature of current care.

Variation in screening tools, cutoff thresholds, and neuropsychological depth remains common across institutions. Although guideline-level recommendations now exist [10], implementation fidelity is uneven, which creates uncertainty in staging, follow-up intensity, and multicenter research comparability. Diagnostic inconsistency also affects communication with patients and families, because similar clinical pictures may be labeled differently across hospitals. A minimum national dataset for suspected VCI, including cognitive screening, functional assessment, vascular history, and core neuroimaging descriptors aligned with updated STRIVE-2 standards for small-vessel disease markers, would improve both care consistency and research comparability [[5], [6], [7],10].

VCI management commonly spans neurology, geriatrics, rehabilitation, psychiatry, and primary care, yet communication across these services is often weak. Some patients leave stroke care without structured cognitive reassessment; others reach memory clinics without systematic optimization of vascular risk factors. Primary care records may track blood pressure and diabetes but not cognitive trajectory. Without clear handoffs, no service owns the longitudinal picture, and patients experience a sequence of disconnected encounters rather than an integrated vascular-cognitive pathway.

Shortages of dementia-trained clinicians and allied health professionals remain a major implementation barrier [19]. However, scaling cannot rely only on producing more specialists. A more realistic strategy is tiered task sharing: community physicians, nurses, and rehabilitation staff can conduct trigger questioning, AD8 collection, or first-line brief screening; stroke or neurology clinics can perform second-stage confirmation with MoCA-based or domain-sensitive testing; and memory clinics or tertiary neurology services can adjudicate mixed pathology, behaviorally complex cases, or refractory management. For this model to be workable in China, frontline teams will need short competency-based training modules, standardized scoring sheets, referral criteria, tele-mentoring from higher-level centers, and audit-feedback systems that monitor both completion burden and referral yield rather than screening volume alone.

Informal caregiving burden in China is substantial and often underestimated [[20], [21]]. In VCI, burden may be amplified by executive dysfunction, fluctuating insight, emotional lability, gait impairment, and recurrent vascular events. Families are often expected to coordinate medications, appointments, finances, and safety supervision across multiple institutions. A realistic management model should therefore treat the caregiver as part of the care unit and routinely include education on vascular risk adherence, cognitive symptom monitoring, and crisis planning.

Another major management problem is the limited use of standardized longitudinal outcomes. Many services document a diagnosis, but fewer consistently track change in cognition, function, mood, caregiver burden, and vascular control over time. Without repeated structured measures, it is difficult to judge treatment response, identify deterioration early, or compare service quality across regions. Implementation research in China should therefore test not only whether stepped screening increases case detection, but also whether it is affordable, sustainable, and operationally acceptable across tertiary hospitals, county hospitals, urban community health centers, and rural township clinics [[39], [40], [41]]. Relevant metrics include reach, false-positive burden, referral completion, staff training time, supervision intensity, and cost per high-risk case detected.

2.5. Therapeutic evidence and clinical management

The therapeutic literature is more heterogeneous than the epidemiological or policy literature, and Table 4 and Table S1 make that heterogeneity explicit. Table 4 provides a concise main-text summary of intervention evidence relevant to VCI in China, whereas Supplementary Table S1 provides the expanded design-level details, including target phenotype, sample size, endpoint, follow-up window, directness to VCI, and interpretation notes. For interpretive clarity, the studies are grouped heuristically by endpoint directness and clinical purpose into upstream vascular-risk reduction in exposed or high-risk populations, prevention of broader cognitive decline or all-cause dementia, and direct syndrome-level treatment for established VCIND, PSCI, or VaD. This is not a formal evidence-quality grading system. Rather, it is a narrative way of separating what is best supported for implementation from what has been most directly tested in established vascular-cognitive syndromes. The therapeutic argument for China is therefore strongest for prevention-first management and weaker for disease-modifying treatment of established syndromes.

Table 4.

Intervention evidence relevant to VCI in China.

Study	Intervention	Population / phenotype	Design / sample	Endpoint / follow-up	Main result	Directness to VCI	Main interpretation
He et al., 2025 [42]	Intensive blood pressure reduction via non-physician-led stepped-care protocol	Adults ≥40 years with uncontrolled hypertension in 326 rural Chinese villages; Prevention of all-cause dementia in vascular-risk population	Open-label, blinded-endpoint cluster RCT; n = 33,995	Adjudicated all-cause dementia; follow-up: 48 months	Dementia 4.59% vs 5.40%; RR 0.85; CIND and serious adverse events also reduced	Indirect but highly relevant	Primary endpoint was all-cause dementia, not incident VCI; baseline cognition not formally characterized.
SPRINT MIND, 2019 [43]	Intensive vs standard systolic blood pressure control	Ambulatory adults with hypertension, no diabetes or prior stroke; US/Puerto Rico; Prevention of probable dementia / MCI	Randomized clinical trial; n = 9361	Adjudicated probable dementia; secondary MCI and composite MCI/probable dementia; follow-up: Median intervention 3.34 years; median follow-up 5.11 years	No significant reduction in probable dementia alone; reduced MCI and MCI/dementia composite	Indirect	Not China-specific and not VCI-specific; probable dementia endpoint may have been underpowered.
Hughes et al., 2020 [44]	Antihypertensive blood pressure lowering	Randomized trial populations across 14 RCTs; Prevention of dementia or cognitive impairment	Systematic review and meta-analysis; n = 96,158	Dementia or cognitive impairment; follow-up: Mean 49.2 months	OR 0.93 for dementia/cognitive impairment; OR 0.93 for cognitive decline	Indirect	Aggregates mixed non-VCI endpoints and does not isolate vascular-cognitive syndromes.
Peters et al., 2022 [45]	Blood pressure lowering treatment	Five placebo-controlled RCTs from 20 countries; Prevention of incident dementia	Individual participant data meta-analysis; n = 28,008	Incident dementia; follow-up: Median 4.3 years	Adjusted OR 0.87 favoring BP lowering	Indirect	Supports upstream vascular-risk reduction, not syndrome-specific VCI treatment.
Liu et al. (COMBAT), 2023 [46]	Meditation, cognitive training, exercise, and nutrition counseling	Community-dwelling older adults at risk of cognitive decline; 10 community hospitals in Beijing; Prevention of cognitive decline in at-risk older adults	Two-arm cluster-randomized controlled trial; n = 209	Change in cognition measured by composite Z score of 7 tests; follow-up: 9-month intervention plus 1-year post-intervention follow-up	Between-group difference in cognition change 0.20 (95% CI 0.053 to 0.35; Hedges g 0.40) at intervention end; not significant at 1 year	Indirect	At-risk population rather than established VCI; benefit not durable at 1-year follow-up.
Meng et al., 2024 [47]	Life-course-model multidomain intervention delivered through WeChat and community activities	Community-dwelling at-risk Chinese adults aged ≥60 years; Prevention of dementia risk / cognitive decline	Proof-of-concept randomized controlled trial; n = 96 (48 vs 48)	ANU-ADRI-SF dementia risk score and cognitive composite Z score; follow-up: 6 months	Intervention lowered dementia risk score and improved cognitive composite Z score versus control	Indirect	Short duration, modest sample size, and proof-of-concept design; not a trial in established VCI syndromes.
Jin et al., 2020 [48]	Qigong vs stretching exercise	Community-dwelling Shanghai older adults aged 60–80 at risk of cognitive decline; Prevention of cognitive decline progression	Single-blind cluster randomized controlled trial; n = 66 randomized; 51 completers	MoCA global cognitive score; follow-up: 1 year	Lower odds of progression to MCI (OR 0.314, 95% CI 0.103 to 0.961) and better MoCA/RBANS scores in qigong group	Indirect	Small sample, completer-dependent interpretation, and non-VCI population.
Lam et al., 2012 [49]	Tai Chi vs stretching and toning exercise	Older Chinese adults in Hong Kong with CDR 0.5 or amnestic MCI; Prevention of cognitive decline / progression to dementia	Single-blind cluster randomized controlled trial; n = 389	Progression to DSM-IV dementia and change in cognitive/functional scores; follow-up: 1 year with assessments at baseline, 5, 9, and 12 months	Lower conversion-to-dementia signal and better CDR sum-of-boxes preservation in Tai Chi arm, especially in completer analyses	Indirect	High dropout, especially in intervention arm; at-risk population rather than established VCI.
Han et al., 2025 [59]	Butylphthalide treatment	Community-recruited ischemic stroke survivors 1–6 months after stroke onset in Tianjin; Prevention of post-stroke cognitive decline	Prospective follow-up study; n = 605 analyzed (309 butylphthalide, 296 control)	Change in MMSE and incidence of cognitive decline; follow-up: 12 months	Lower MMSE decline and lower incidence of cognitive decline; treatment-group incidence 16.5%	Direct to PSCI-related phenotype, not full VCI spectrum	Non-randomized staged recruitment and phenotype-specific applicability.
Yang et al., 2019 [60]	Acupuncture vs oral citicoline	Patients with VCIND in three Beijing hospitals; Treatment of established VCIND	Open-label randomized multicenter trial with blinded assessors/statisticians; n = 216 randomized; ITT 108 vs 105	Change in ADAS-cog at 3 months; follow-up: 3-month treatment with 6-month follow-up	ADAS-cog favored acupuncture at 3 and 6 months; no CDT difference; ADL advantage mainly in per-protocol analyses	Direct	Active comparator not guideline-standard; open-label design and limited functional benefit warrant caution.
Shi et al., 2020 [58]	Tianzhi granule vs donepezil vs placebo	Patients with mild to moderate probable/possible VaD in 16 centers in China; Treatment of established VaD	Phase III randomized double-blind placebo-controlled three-arm trial; n = 543 randomized; 520 ITT	VADAS-cog and CIBIC-plus; follow-up: 24 weeks after 2-week placebo run-in	Both Tianzhi and donepezil improved cognition/global impression versus placebo; no clear difference between active groups; AE rates similar	Direct to VaD	Symptomatic benefit only; does not establish disease modification or broad VCI efficacy.
Wilkinson et al., 2003 [50]	Donepezil 5 or 10 mg/day vs placebo	Probable or possible VaD; Treatment of established VaD	Randomized placebo-controlled trial; n = 616	ADAS-cog and global impression outcomes; follow-up: 24 weeks	Modest cognitive and global function improvement vs placebo	Direct to VaD	International trial; symptomatic effect size modest and not disease-modifying.
Orgogozo et al., 2002 [55]	Memantine 20 mg/day vs placebo	Mild to moderate probable VaD; Treatment of established VaD	Randomized placebo-controlled trial; n = 321 randomized; 288 ITT	ADAS-cog and CIBIC-plus; follow-up: 28 weeks	Cognitive improvement relative to placebo; global response less robust	Direct to VaD	Symptomatic treatment evidence only; international, not China-specific.
Battle et al., 2021 [57]	Cholinesterase inhibitors compared with placebo or each other	Adults with vascular dementia or other VCI; Treatment of established VCI / VaD	Cochrane network meta-analysis; n=Multiple RCTs	Cognition, global impression, function, adverse events; follow-up: Varied across included RCTs	Small cognitive benefits may occur, but strong functional evidence is lacking	Direct but methodologically mixed	Benefits are modest, certainty varies, and no recommended pharmacologic treatment has robust disease-modifying evidence.
Wang et al., 2025 [61]	Tongxinluo capsule for CSVD (TOPS-CSVD)	Chinese patients aged 50–80 years with MRI-confirmed CSVD and mild cognitive impairment; Emerging CSVD-targeted intervention	Randomized, double-blind, placebo-controlled multicenter protocol; n = 1052 planned; 894 evaluable needed	Change in VADAS-cog from baseline to 12 months; follow-up: 12 months with 3-month interval assessments	No efficacy result yet; protocol only	Potentially relevant but currently indirect	Protocol only; should not be discussed as efficacy evidence.

Note. Directness to VCI indicates whether the intervention was tested in an established vascular-cognitive syndrome or whether the evidence is indirect because the endpoint was all-cause dementia, cognitive decline, cognitive impairment no dementia, or an at-risk population. Table 4 provides the concise intervention-evidence summary for the main submission file; Supplementary Table S1 provides the expanded design-level details for the same intervention evidence, including target phenotype, sample size, primary endpoint, follow-up window, directness to VCI, and interpretation notes. Abbreviations: ADAS-cog, Alzheimer's Disease Assessment Scale-cognitive subscale; ADL, activities of daily living; AE, adverse event; ANU-ADRI-SF, Australian National University Alzheimer's Disease Risk Index-short form; BP, blood pressure; CDR, Clinical Dementia Rating; CDT, Clock Drawing Test; CIBIC-plus, Clinician's Interview-Based Impression of Change plus caregiver input; CIND, cognitive impairment no dementia; COMBAT, Cognitive and Multidomain Intervention for Brain Health in Beijing Aging Trial; CSVD, cerebral small vessel disease; ITT, intention-to-treat; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; OR, odds ratio; PSCI, post-stroke cognitive impairment; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; RCT, randomized controlled trial; RR, risk ratio; SPRINT MIND, Systolic Blood Pressure Intervention Trial Memory and Cognition in Decreased Hypertension; TOPS-CSVD, Tongxinluo capsule in the treatment of cerebral small vessel disease trial; VaD, vascular dementia; VADAS-cog, Vascular Dementia Assessment Scale-cognitive subscale; VCI, vascular cognitive impairment; VCIND, vascular cognitive impairment no dementia.

The clearest therapeutic principle for VCI in China is prevention-first management: reduce ongoing vascular injury early, then add symptom-oriented and function-oriented interventions once impairment is established. For most patients, the realistic therapeutic goal is not reversal of accumulated brain injury, but slowing further injury, preserving function, and delaying dependency. This framing is also the safest way to align the literature with clinical practice, because it distinguishes the high-certainty need for vascular-risk control from the lower-certainty evidence for syndrome-specific therapies.

Among currently available interventions, blood pressure control has the strongest interventional relevance for VCI prevention, but the evidence remains indirect. The CRHCP-3 cluster-randomized trial in rural China assigned 33,995 adults with uncontrolled hypertension in 326 villages to a non-physician-led stepped-care program or usual care for 48 months and used adjudicated all-cause dementia, not VCI, as the primary endpoint [42]. Dementia occurred in 4.59% of the intervention group and 5.40% of the usual-care group (risk ratio 0.85), with parallel reductions in cognitive impairment no dementia and serious adverse events [42]. International trials and meta-analyses point in the same direction [[43], [44], [45]], but these studies likewise target broader dementia or cognitive decline outcomes rather than incident VCI diagnosed by vascular-specific criteria. They should therefore be interpreted as the clearest support for upstream vascular-risk reduction, not as direct proof of efficacy for every VCI phenotype.

Chinese multidomain prevention trials also sit mainly in the indirect tier. COMBAT randomized 209 community-dwelling at-risk older adults from 10 community hospitals in Beijing to meditation, cognitive training, exercise, and nutrition counseling versus usual care, delivered weekly for 9 months; cognition improved at the end of the intervention, with a between-group difference of 0.20 in composite Z-score change, but the benefit was no longer significant at 1-year follow-up [46]. In Northeast China, Meng et al. randomized 96 at-risk community-dwelling older adults to a 6-month life-course-based multidomain program delivered through WeChat messages, cognitive training, physical activity, and social/community activities; the intervention lowered dementia risk scores and improved cognitive composite Z scores, but it remained a proof-of-concept study with short follow-up and modest sample size [47]. These trials support the plausibility of multidomain brain-health interventions in Chinese communities, yet they remain prevention studies in at-risk populations rather than direct treatment trials for established VCI syndromes.

Exercise-based mind-body interventions are similarly promising but heterogeneous. Jin et al. randomized 66 older Shanghai adults at risk of cognitive decline to qigong or stretching for 1 year and found lower odds of progression to mild cognitive impairment and better MoCA and RBANS performance in the qigong group among 51 completers [48]. Lam et al. studied 389 older Hong Kong adults with CDR 0.5 or amnestic MCI in a cluster-randomized Tai Chi trial and reported better preservation of CDR sum-of-boxes and a lower 1-year conversion-to-dementia signal in completer analyses, but dropout was substantial, especially in the Tai Chi arm [49]. These studies are clinically interesting because they combine physical and cognitive stimulation, but they remain indirect to established VCI, and their real-world value depends heavily on adherence, dose, and delivery logistics.

Direct syndrome-level evidence in Chinese or international cohorts is thinner and more phenotype-specific. For VaD, randomized placebo-controlled trials and reviews of cholinesterase inhibitors and memantine show, at most, modest symptomatic cognitive benefit without convincing disease modification: donepezil evidence includes multiple trials and a Cochrane review [[50], [51], [52], [53]], rivastigmine has been tested in probable VaD [54], memantine has randomized trial evidence [55], and broader meta-analytic syntheses report small or uncertain clinical effects [[56], [57]]. The Chinese phase III Tianzhi granule trial enrolled 543 patients with mild-to-moderate probable or possible VaD across 16 centers, randomized them to Tianzhi granule, donepezil, or placebo for 24 weeks, and found better VADAS-cog and CIBIC-plus outcomes for both active treatments versus placebo but no clear difference between the two active groups [58]. In PSCI-related populations, Han et al. followed 605 ischemic stroke survivors and reported less MMSE decline with butylphthalide, but the study used nonrandomized staged recruitment and addressed a post-stroke phenotype rather than the full VCI spectrum [59]. These data justify discussing phenotype-specific symptomatic options, but not claiming a general disease-modifying treatment for VCI.

Evidence for VCIND-specific adjunctive therapy should also be read cautiously. Yang et al. randomized 216 patients with VCIND in three Beijing hospitals to acupuncture or citicoline for 3 months and found greater improvement in ADAS-cog at 3 and 6 months in the acupuncture group, while CDT differences were absent and ADL advantages were limited to per-protocol analyses [60]. However, the study was open-label, lacked a sham control, used an active comparator not endorsed in contemporary guidelines, and produced limited functional signal. Its findings are therefore hypothesis-generating within a narrow VCIND phenotype rather than definitive evidence for broader VCI treatment. Meanwhile, TOPS-CSVD remains a protocol rather than an efficacy study [61]. In routine care, treatment should therefore be staged: optimize vascular risk first, characterize cognition and function second, and then select phenotype-specific adjunctive pharmacologic, rehabilitation, or multidomain interventions with follow-up that jointly tracks cognition, function, vascular control, and caregiver strain [10,22,42,[46], [47], [48], [49],[58], [59], [60], [61]].

2.6. A China-adapted prevention and care pathway

The clinical implementation of VCI management requires integration across multiple healthcare settings. Unlike disorders identified primarily through specialist cognitive clinics, many individuals with vascular cognitive risk first present in chronic disease clinics, stroke services, rehabilitation units, or community healthcare settings. Therefore, a stepped pathway that connects risk identification, cognitive screening, specialist clarification, and longitudinal management may better fit the Chinese healthcare context (Fig. 2). This pathway is intended as a pragmatic implementation framework rather than a validated national care algorithm. Its feasibility will depend on locally appropriate screening thresholds, workforce capacity, referral infrastructure, and future implementation research.

Fig. 2.

A stepped screening and longitudinal management pathway for vascular cognitive impairment in China. Abbreviations: AD8, Ascertain Dementia 8-item Informant Questionnaire; BP, blood pressure; CSVD, cerebral small vessel disease; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; NINDS-CSN, National Institute of Neurological Disorders and Stroke–Canadian Stroke Network; TIA, transient ischemic attack; VCI, vascular cognitive impairment.

Primary prevention should focus on sustained vascular risk control, particularly blood pressure management, together with cardiometabolic optimization and lifestyle intervention [[28], [29], [30],42]. In China, prevention messaging may be more effective when framed as brain-health preservation rather than only dementia avoidance. Linking hypertension control to maintenance of independence, mobility, and cognitive function could make prevention more concrete for middle-aged and younger-old adults.

Secondary prevention should target groups at clearly elevated risk, especially stroke survivors and adults with poorly controlled hypertension. In stroke services, a practical pathway is bedside or clinic-based 5-min NINDS-CSN screening at 3–6 months and again at about 12 months, administered by a stroke nurse, rehabilitation therapist, or clinic physician, with MoCA or fuller neuropsychological testing reserved for those with scores <9 or persistent concern [25]. In chronic disease clinics, annual trigger questioning plus informant-based AD8 or a brief cognitive screen is more realistic than universal full-battery testing. Older adults with recurrent stroke, CSVD markers, falls, medication mismanagement, or new functional decline should move more quickly to second-stage assessment because vascular presentations are often non-amnestic [10,18,[23], [24], [25], [26], [27]].

For patients with established impairment, tertiary prevention should combine cognitive rehabilitation, functional support, vascular risk reinforcement, and caregiver-focused planning to delay progression and reduce disability [10,[19], [20], [21], [22]]. An effective model also requires anticipatory guidance on medication supervision, mobility safety, financial decision support, fall prevention, and recognition of recurrent stroke symptoms. Because VCI often coexists with gait slowing, mood symptoms, and instrumental dependency, follow-up should not be limited to a cognitive score alone.

Four near-term priorities appear both practical and scalable: integrate cognitive case finding into chronic disease clinics; standardize brief screening and referral algorithms; expand regional vascular-cognition and memory-service capacity; and use registry-supported quality indicators to track outcomes [10,19,22,[39], [40], [41]]. To make these priorities workable in China, implementation design matters as much as policy endorsement. Frontline teams will need concise training packages, referral scripts, documentation templates, access to MoCA-Basic/MoCA or education-adjusted MMSE scoring aids, and dependable back-referral or tele-support from higher-level centers if screening is to be added without simply increasing workload and false-positive referrals.

Implementation should be judged by outcomes rather than policy issuance alone. Useful indicators include blood pressure control among cognitively at-risk adults, the proportion of stroke survivors screened at 3–6 months and 12 months, time from trigger-positive screen to specialist assessment, documentation of functional staging, caregiver-burden measures, and avoidable hospitalization rates. Future Chinese implementation studies should also capture reach, adherence, cost-effectiveness, referral completion, staff burden, and sustainability after pilot support is withdrawn, because scalable VCI care will depend as much on delivery design as on biological efficacy.

2.7. Limitations and unresolved questions

Several limitations should temper interpretation of this review. First, VCI, PSCI, vascular dementia, and related vascular-cognitive phenotypes are not operationalized uniformly across studies, so burden estimates and treatment findings are not fully interchangeable. This diagnostic heterogeneity is itself an important result, but it also limits clean cross-study comparison [1,3,[6], [7], [8],10,12].

Second, the intervention evidence base is uneven. Some of the strongest data cited here concern all-cause dementia, cognitive decline, or broader at-risk older-adult outcomes rather than VCI-specific endpoints. These studies are highly relevant to a prevention-first vascular-cognitive strategy, but they should not be overinterpreted as direct proof for every VCI phenotype [[42], [43], [44], [45], [46], [47]]. The same caution applies when phenotype-specific findings from PSCI, VCIND, or VaD cohorts are extrapolated to the broader VCI umbrella.

Third, Chinese data are not uniformly representative of all care contexts. Urban tertiary centers, selected stroke cohorts, and well-characterized research registries often contribute a disproportionate share of detailed evidence, whereas rural settings and lower-resource pathways remain less well characterized [19,25,27,[39], [40], [41]].

Finally, the tiered life-course model proposed in this review is a synthesis of epidemiological evidence, guideline direction, and implementation logic rather than a care pathway already validated in large pragmatic trials. It is intended as a practical framework for action, but its comparative effectiveness and scalability still need direct testing in Chinese real-world settings [10,22,[39], [40], [41]].

2.8. Future directions

China is well positioned to contribute substantially to VCI translational research because it now has expanding longitudinal cohorts, stroke registries, and biomarker-capable platforms [[39], [40], [41],62]. The next phase of research should prioritize harmonized diagnostics [[6], [7], [8]], CSVD-informed risk stratification [[35], [36], [37]], and pragmatic implementation trials in community and rural settings where unmet need is highest.

Future work should also move beyond efficacy-only designs. Studies should increasingly include implementation feasibility, cost-effectiveness, equity, and patient-centered functional outcomes [63]. Hybrid effectiveness-implementation designs are especially needed for China, because the field now has to answer not only whether a strategy can work, but also where, by whom, at what cost, and with what ongoing supervision burden. A particularly important agenda is to test how stepped screening and referral models should be adapted across tertiary hospitals, county hospitals, urban community health centers, and rural township clinics. Relevant outcomes should include reach, false-positive burden, staff training time, supervision intensity, referral completion, cost per high-risk case detected, and sustainability after initial pilot support.

Large Chinese cohorts are now capable of supporting stratified analyses by CSVD burden, multimorbidity pattern, and care context, but these data will have greatest policy value if they are linked to outcomes such as dependence, institutionalization, recurrent vascular events, and caregiver burden. In other words, the field should connect mechanistic precision to service design rather than treating them as separate agendas.

Available evidence for VCI management in China spans multiple domains, but these domains differ in their clinical objectives, endpoint relationships, and implementation requirements. Evidence from vascular risk reduction primarily informs prevention and risk modification, whereas multidomain and service-delivery approaches address detection, functional preservation, and healthcare delivery. In contrast, phenotype-specific interventions aim to address established cognitive syndromes but remain limited by heterogeneous populations, variable definitions, and insufficient disease-modifying evidence. These evidence domains should therefore be interpreted as complementary components rather than a hierarchical treatment ranking (Fig. 3).

Fig. 3.

Evidence domains and implementation considerations for vascular cognitive impairment management in China. Abbreviations: BP, blood pressure; VaD, vascular dementia; VCIND, vascular cognitive impairment no dementia.

3. Conclusions

VCI in China is a high-burden, systems-sensitive condition at the intersection of cerebrovascular medicine, cognitive neurology, and public health. The available evidence most strongly supports a prevention-first approach that prioritizes vascular risk control, especially blood pressure management, while acknowledging that these data provide indirect but clinically important support for VCI prevention because the strongest interventional endpoints are often dementia or broader cognitive outcomes rather than uniform VCI-specific measures. Direct evidence for interventions targeting established syndromes such as VCIND, PSCI, or VaD is more limited and more phenotype-specific. China has already strengthened its policy and guideline foundations, but the main barriers now lie in implementation: inconsistent screening, uneven specialist capacity, fragmented referral pathways, and weak longitudinal follow-up. A tiered life-course model linking chronic disease management, post-stroke surveillance, specialist assessment, multidomain intervention, and caregiver support is therefore best understood as a practical synthesis of current evidence and service logic rather than a fully trial-validated national pathway. Strengthening evidence grading, review transparency, and pragmatic implementation research will be essential if this framework is to reduce future vascular-cognitive burden in China.

Funding

This work was supported by the National Natural Science Foundation of China (82571515), the Sichuan Provincial Science and Technology Department (2026YFHZ0202, 2024ZYD0136), Chengdu Medical Project (2025044).

Declarations

Fig. 1, Fig. 2, Fig. 3 were created by the authors with the assistance of the image-generation function in ChatGPT (OpenAI). All graphical elements are original schematic illustrations generated specifically for this manuscript and were subsequently reviewed and edited by the authors for scientific accuracy. No third-party copyrighted figures, commercial templates, patient images, or identifiable clinical imaging data were reproduced or adapted.

CRediT authorship contribution statement

Liang Gong: Writing – review & editing, Writing – original draft, Methodology, Funding acquisition, Conceptualization. Hongmei Lang: Writing – review & editing, Validation, Methodology, Funding acquisition. Jie Yang: Writing – review & editing, Supervision, Funding acquisition, Conceptualization.

Declaration of competing interest

There is no conflicts of interest.