In this article· 13 sectionsClick to collapse
- 11. Common clinic scenarios
- 22. Common patient Q&As
- 33. Key takeaways: 8 things you really need to know
- 44. DR basics: staging and epidemiology
- 55. Dementia 3 main types: all-cause, AD, VD
- 66. What is the TriNetX 770K cohort study?
- 77. The six key findings
- 88. Why is ophthalmology an "early dementia risk marker"?
- 99. Five things diabetics can do
- 1010. Visual red flags for diabetics — when to go in immediately
- 1111. Taiwan context and NHI notes
- 1212. Conclusion: fundus exams are a two-birds-one-stone health investment
- 13References
Enlarge diagram
Scroll horizontally and vertically to read the diagram. Close Enlarge diagram to restore its original size.
1. Common clinic scenarios
'Doctor, my father is 72, has had diabetes for 20 years, and ophthalmology says he has diabetic retinopathy and has had a few laser treatments. Lately he's been forgetting things — even calling us by the wrong names. Is this related to the diabetic eye disease?'
'Doctor, I'm 58, diabetic for 8 years, last year's fundus exam was essentially normal. Recently I heard 'diabetes can cause dementia' — is that true? How much is my risk? Do I need any special workup?'
'Doctor, my mother is 80, diagnosed with mild dementia, also unsteady gait. She's never had a fundus exam — her doctor said «it doesn't matter at this point». But she's had diabetes for years. Do we really not need ophthalmology?'
All three scenarios revolve around the same theme: diabetes ⇄ retina ⇄ dementia. A May 2026 paper in American Journal of Ophthalmology (Khangura et al.) analysed 769,930 individuals ≥65 with an eye-exam record in the TriNetX global research network, stratifying by DR severity (one of the few studies to do so) to answer: does more severe DR really correlate with higher dementia risk? This article distils it in clinic-friendly language.
2. Common patient Q&As
Diabetic retinopathy (DR) is microvascular damage from diabetes manifesting in the retina — microaneurysms, haemorrhages, exudates, neovascularisation. Clinically two stages: non-proliferative (NPDR) — early, with only microvascular abnormalities, and proliferative (PDR) — late, with fragile new vessels prone to vitreous haemorrhage and retinal detachment. DR is a leading cause of blindness in working-age adults.
Dementia is the umbrella term for chronic progressive cognitive decline affecting daily function. The most common types are Alzheimer's disease (AD) and vascular dementia (VD).
The link: diabetes itself elevates dementia risk (some authors even use the term type-3 diabetes), and DR is the most observable proxy for diabetic microvascular damage — if retinal vessels are damaged, cerebral vessels likely are too. Khangura 2026 (TriNetX) quantifies this association.
• Vascular dementia (VD): PDR vs non-diabetic HR 2.08; NPDR 1.92; PDR vs NPDR 1.18. Clearly stepwise.
• All-cause dementia: PDR 1.58; NPDR 1.41; PDR vs NPDR 1.12. Also stepwise.
• Alzheimer's (AD): PDR vs non-diabetic 1.18 (statistically significant but modest); NPDR vs non-diabetic 1.23; but PDR vs NPDR 0.93, PDR vs DM-no-DR 0.97 — neither significant. So AD risk is driven by diabetes itself, not DR severity.
Clinical meaning: DR represents vascular pathology, hence its strongest link to vascular dementia (essentially the same disease in different tissues — small-vessel occlusion). AD's core mechanism is amyloid and tau accumulation rather than microvascular damage, so DR severity has less leverage.
What to do: (1) Glycaemic control — HbA1c < 7% (individualised by your metabolic doctor based on age and comorbidities); (2) Blood pressure — target < 130/80; (3) Smoking cessation, regular exercise, Mediterranean/DASH diet — good for both heart and brain; (4) Regular ophthalmology follow-up at the interval your ophthalmologist recommends; (5) Cognitive monitoring — if memory, judgement, language, or spatial function declines, see neurology / a dementia clinic early.
Treat DR as a warning light, not a verdict — a chance for earlier integrated care, not an unavoidable trajectory.
In Taiwan, the diabetes shared-care network and many hospitals offer routine fundus screening for diabetics — discuss referral with your metabolic / family medicine physician. This article does not make claims about specific NHI programmes or fees; please check with your healthcare facility.
3. Key takeaways: 8 things you really need to know
30-second takeaways
- Large cohort: TriNetX 770K global analysis, one of the few DR–dementia studies stratified by DR severity
- Diabetes itself elevates dementia risk: HR 1.26 even without DR
- DR adds further risk: NPDR HR 1.41, PDR HR 1.58 — stepwise
- Vascular dementia is the most linked: PDR's VD HR = 2.08 — doubled risk
- AD link is weaker: AD risk driven by diabetes itself, no clear gradient by DR severity
- The eye is the brain's window: DR and cerebral small-vessel disease share microvascular pathology — retinal vessels may reflect the state of cerebral microvessels
- Elevated risk ≠ certainty: PDR's actual 6.7-yr incidence was 15.9%; most did not develop dementia
- Fundus exam has dual value: already standard diabetes care, now also an early dementia risk marker
4. DR basics: staging and epidemiology
Diabetic retinopathy is diabetic microvascular damage in the retina, classified by severity:
Scroll horizontally to read the table. With a keyboard, focus the table area and use the arrow keys.
| Stage | Fundus findings | Clinical meaning |
|---|---|---|
| No DR | Normal fundus | Best status for diabetics; with no DR on annual exams and on-target glucose, every 1-2 yr |
| Mild-moderate NPDR | Microaneurysms, dot/blot haemorrhages, hard exudates | Early signal; usually asymptomatic; follow-up every 12 mo (mild) or 6-12 mo (moderate) |
| Severe NPDR | Extensive haemorrhages, venous beading, IRMA | On the verge of PDR; close 3-4 mo monitoring |
| Proliferative PDR | Neovascularisation, vitreous haemorrhage, fibrous proliferation, possible RD | High vision-loss risk; follow-up every 3-4 mo (non-high-risk PDR) or 2-4 mo (high-risk PDR); PRP recommended for high-risk PDR, and intravitreal injection or surgery may be needed |
| With macular edema (DME) | Thickened macula, cystic spaces on OCT | Can occur at any DR stage; direct visual threat; follow-up every 2-6 mo if non-center-involved (depending on DR stage) or about monthly if center-involved, usually with intravitreal anti-VEGF; excluded from this study to cleanly assess DR severity |
Source: follow-up intervals for the NPDR, PDR and macular-edema rows follow the initial management recommendations (Table 5) of the AAO Diabetic Retinopathy Preferred Practice Pattern, 2024 edition; the no-DR row follows the ADA Standards of Care 2026 (the AAO table gives 12 months). Your ophthalmologist sets the actual interval for each patient.
Epidemiology: DR is among the most common diabetic microvascular complications, with prevalence up to 26% in people with diabetes (US 2021), and was the fifth leading cause of blindness worldwide among adults aged 50+ in 2020 (GBD 2019 Vision Loss Expert Group). Critically, fundus changes can develop while the patient feels nothing — which is why scheduled eye exams matter.
5. Dementia 3 main types: all-cause, AD, VD
Dementia denotes chronic, progressive cognitive decline that interferes with daily function. Three common types:
- Alzheimer's disease (AD): 60-70% of dementia, the most common. Core pathology: β-amyloid plaque and tau-tangle accumulation, neuronal death, hippocampal atrophy, progressive cognitive loss.
- Vascular dementia (VD): second most common. Core pathology: cerebral small-vessel occlusion, ischaemia, white-matter lesions, causing stepwise cognitive decline, often with gait disturbance and executive dysfunction. Directly linked to hypertension, diabetes, smoking, dyslipidaemia.
- All-cause dementia: composite epidemiological endpoint capturing every cause of dementia (the two above plus less common Lewy body, frontotemporal, mixed types) — reflects total dementia burden.
Bottom line: the strongest association in this study is DR severity → vascular dementia, fitting the pathological reality that both share small-vessel disease as the core mechanism.
6. What is the TriNetX 770K cohort study?
TriNetX is a global federated health research network that, at the time of the analysis (September 2025), covered 153 healthcare organisations and ~185 million de-identified patient records (diagnoses, procedures, medications, lab values, genomic data). Data refresh every 2-4 weeks; most institutions contribute 7-10 years of history. Khangura et al. 2026 design:
Scroll horizontally to read the table. With a keyboard, focus the table area and use the arrow keys.
| Item | Detail |
|---|---|
| Inclusion | ≥ 65 yr, eye exam or OCT between 1 Jan 2010 and 1 Jan 2020 (later entrants excluded to allow enough follow-up) |
| Exclusion | DME (it can occur at any DR stage and is linked to renal failure, so it was excluded to isolate DR stage); dementia diagnosed before type-2 diabetes |
| Four cohorts | PDR (14,034), NPDR (29,188), T2DM without DR (208,640), non-diabetic (447,054) |
| Matching | 1:1 propensity score matching (greedy nearest neighbour). Variables: age, sex, SES, glaucoma, AMD, hypertension, CKD, cerebral infarction, acute MI, HF, depression |
| Follow-up | Mean 6.73 yr (SD 3.77) |
| Outcomes | Incidence of all-cause dementia, AD, VD; Cox proportional hazards; Kaplan-Meier curves |
This is a large DR–dementia study: 43,222 people with DR (PDR 14,034, NPDR 29,188), and one of the few stratified by DR severity. In their discussion the authors call it the largest such cohort and the first to stratify by DR severity, saying prior studies included roughly 3,000-4,000 patients with DR. Some earlier studies were indeed smaller (e.g., 536 participants in Lee 2023), but larger registry studies exist — the Danish nationwide registry study (Pedersen 2022, also cited by Khangura et al.) included about 786,000 people, 19,221 of them with DR, and already compared Alzheimer's risk with and without DR — so this is not the only large study on this topic.
7. The six key findings
Scroll horizontally to read the table. With a keyboard, focus the table area and use the arrow keys.
| Finding | Key number | Clinical meaning |
|---|---|---|
| 1. Diabetes itself ↑ dementia risk | DM without DR vs non-DM: all-cause HR 1.26, AD 1.12, VD 1.38 (all p<0.0001) | Even with a normal fundus, diabetics need ongoing risk-factor control |
| 2. DR severity → stepwise ↑ dementia | NPDR HR 1.41, PDR HR 1.58 (vs non-DM) | Retinal severity mirrors systemic microvascular burden |
| 3. VD has the strongest link | PDR vs non-DM VD HR 2.08; NPDR 1.92; DM-no-DR 1.38 | Fundus = window to cerebral small-vessel disease — eye findings warn of brain risk |
| 4. AD doesn't track DR severity | PDR vs NPDR AD HR 0.93 (p = 0.31); PDR vs DM-no-DR 0.97 (p = 0.70) | AD is amyloid-driven not microvascular; diabetes itself raises AD, DR severity adds nothing further |
| 5. Mortality also rises stepwise | PDR vs non-DM mortality HR 2.14; NPDR 1.53; DM-no-DR 1.23 | Higher mortality may pre-empt AD onset in severe DR — possible survivorship bias underestimating AD link |
| 6. Methodological rigor | DME excluded; matching included glaucoma and AMD (both linked to dementia) | These controls reduce confounding from other eye diseases |
Limitations (as stated by the authors): (1) a retrospective database study, so diagnoses depend on ICD coding that varies across institutions; (2) education and socioeconomic status were poorly captured (ICD-10 Z55-Z65 codes were present in only 1-3% of the cohort); (3) over 95% of the data came from US institutions, which may limit generalisability to other countries (including Taiwan); (4) mean follow-up of about 6.7 years may be too short for people who had not yet reached the typical age of dementia onset; (5) no data on diabetes duration, long-term HbA1c or medications — DR severity serves as a surrogate for this unmeasured metabolic burden, and the authors stress it cannot establish causality; (6) TriNetX provides no age at death and only Cox (cause-specific) modelling, so the competing risk of death could not be addressed — possibly one reason no association was seen between severe DR and Alzheimer's disease; (7) only 1:1 matching was possible, and using all-cause dementia as a composite endpoint may obscure associations with rarer dementia subtypes.
8. Why is ophthalmology an "early dementia risk marker"?
Anatomically and developmentally, the retina is an extension of the central nervous system — derived from neural ectoderm, containing neurons (retinal ganglion cells, bipolar cells, photoreceptors), myelinated nerve fibres (the optic nerve), and a CNS-like microvascular barrier (blood-retinal barrier). This makes the "the eye is the brain's window" metaphor biologically literal.
When diabetes damages microvasculature systemically, retinal and cerebral microvessels deteriorate in parallel: microaneurysms/haemorrhages/neovascularisation in the fundus correspond to lacunar infarcts, white-matter lesions, and microbleeds in the brain. The fundus is a miniature of the brain.
Khangura et al. also note their prior work showing that DR patients have higher plasma AD biomarkers (β-amyloid, tau, neurofilament light chain, GFAP) — suggesting DR may reflect not only vascular but also broader neurodegenerative processes. Although the present study found no significant DR-severity gradient for AD, this area remains active.
Clinical practical use: for diabetics, scheduled fundus exams are no longer "just eye care" — they are an opportunity for whole-body risk assessment. When ophthalmologists detect moderate-or-worse DR, they can flag systemic risk and prompt joint care with family medicine / metabolic / neurology — a low-cost, low-invasion, scalable model of early dementia risk-flagging.
9. Five things diabetics can do
💡 5 evidence-backed actions to cut both DR and dementia risk
- 1. Glycaemic control: HbA1c < 7% (individualised by your endocrinologist for age, comorbidities, hypoglycaemia risk). Long-term glycaemic variability worsens both DR and dementia risk.
- 2. Blood pressure: target < 130/80. Hypertension accelerates microvascular damage — a major driver of both DR progression and VD.
- 3. Quit smoking: smoking is an independent microvascular damaging factor, worsening DR and small-vessel dementia.
- 4. Regular exercise + Mediterranean / DASH diet: 150 min/wk moderate aerobic + high-fibre, omega-3-rich, low-refined-sugar eating benefits both cardiovascular and cognitive function.
- 5. Regular fundus follow-up + cognitive awareness: per international guidelines (every 1-2 years with no DR and on-target glucose; at least yearly with any DR, more often if worsening); family members should monitor for changes in memory, judgement, spatial sense, and seek early evaluation.
10. Visual red flags for diabetics — when to go in immediately
Even if you follow regular check-ups, see ophthalmology same-day for any of the following:
🚨 Diabetic visual red flags
- Sudden visual decline (one or both eyes, any degree)
- Sudden floaters or dark shadows — possible vitreous haemorrhage (PDR progression)
- Visual field defect (curtain over part of vision) — possible retinal detachment
- Distorted vision (straight lines bent, sizes changing) — possible macular edema progression
- Acute eye pain + blurred vision + nausea — neovascular glaucoma, a serious PDR complication
If any of the above, go to ophthalmology ER or general ER the same day.
11. Taiwan context and NHI notes
⚠️ This section strictly follows "no source — no claim"
- Diabetic fundus exam: a routine ophthalmic exam; Taiwan's diabetes shared-care network and most hospitals offer scheduled fundus screening for diabetics. Specific procedures, intervals, and coverage vary by facility and current NHI policy — check with your provider and the NHIA portal.
- OCT (optical coherence tomography): the key tool for diagnosing diabetic macular edema; available at most facilities. NHI coverage depends on indication and physician judgement.
- DR treatment (intravitreal anti-VEGF, laser, vitrectomy): Under Taiwan NHI drug reimbursement rules (updated 2026-10-22), section 14.9.2, anti-VEGF injections (one of ranibizumab, aflibercept or faricimab) are covered for diabetic macular edema (DME) when central retinal thickness (CRT) is ≥ 300 μm, best-corrected visual acuity in the affected eye is 0.05–0.5, and HbA1c in the past three months is below 10%; pre-approval is required and an ophthalmologist must give the injections. Approvals allow 5, then 5, then 4 injections — at most 14 per eye. To continue, vision must still be below 0.8, OCT must still show edema (CRT ≥ 300 μm) with at least a 10% thickness reduction versus the previous application, and HbA1c must be ≤ 8.0%. If the response is poor or side effects occur, one switch to a drug with a different mechanism may be requested (no switching among the three anti-VEGF drugs). Proliferative DR without macular edema is not a listed indication. The dexamethasone intravitreal implant (e.g. Ozurdex, section 14.9.4) is also covered for DME with pre-approval — first approval up to 3 implants for eyes without prior anti-VEGF, at most 5 per eye, and not for patients with glaucoma. Laser and vitrectomy fall outside the drug rules and are not covered here; discuss details with your doctor.
- Dementia diagnosis and medication: this article makes no claims about cognitive workup, neuroimaging, or pharmacotherapy (e.g., donepezil, memantine). Please see neurology / a dementia clinic. On coverage: under Taiwan NHI drug reimbursement rules (updated 2026-10-22), section 1.3.3, Alzheimer's dementia drugs must be prescribed by a neurologist or psychiatrist, the drugs allowed depend on MMSE or CDR scores, and patients must be reassessed yearly, with the drug stopped if decline reaches the set threshold.
12. Conclusion: fundus exams are a two-birds-one-stone health investment
Khangura et al.'s 2026 TriNetX 770K cohort delivers three key messages:
- Diabetes itself is associated with higher dementia risk: all-cause HR about 1.26 even without retinopathy (a retrospective cohort shows association only; it cannot establish mechanism or causation)
- DR severity → dementia (especially VD) risk rises stepwise; the fundus marks systemic microvascular burden
- Regular fundus exam = dual value: standard diabetes care + early dementia risk marker enabling integrated care
The most important take-home for diabetics: don't wait until your eyes or mind feel «off». The best protection for ophthalmic and cognitive health remains solid glycaemic and BP control + lifestyle + scheduled eye exams — the most cost-effective investment for heart, brain, and eyes alike.
📚 HsiaoEye Related Articles
- 6 Floater Red Flags — Vitreous haemorrhage from DR often presents as new floaters
- Glaucoma — Patient Education — PDR can lead to neovascular glaucoma
- Ophthalmic Trauma — Overlooked Burden — Public-health angle on eye health red flags
- Recurrent SK After HZO — Another chronic eye condition needing ongoing follow-up
References
- Khangura MS, Spratt MA, Gao A, Manhapra A, Siegel NH, Chen X, Poulaki V, Ness S, Stein T, Subramanian ML. The Association Between Diabetic Retinopathy Severity and Dementia Risk: A TriNetX Longitudinal Cohort Study. Am J Ophthalmol. 2026;285:300–309. doi:10.1016/j.ajo.2026.02.014.
- Fang M, Hu J, Weiss J, et al. Lifetime risk and projected burden of dementia. Nat Med. 2025;31(3):772–776.
- Biessels GJ, Despa F. Cognitive decline and dementia in diabetes mellitus: mechanisms and clinical implications. Nat Rev Endocrinol. 2018;14(10):591–604.
- Chai YH, Han YP, Zhang JY, Zhou JB. Diabetic retinopathy and brain structure, cognition function, and dementia: a bidirectional mendelian randomization study. J Alzheimer's Dis. 2024;97(3).
- Lee CS, Krakauer C, Su YR, et al. Diabetic Retinopathy and Dementia Association, beyond diabetes severity. Am J Ophthalmol. 2023;249:90–98.
- Pedersen FN, Stokholm L, Pouwer F, et al. Diabetic retinopathy predicts risk of Alzheimer's Disease: a Danish registry-based nationwide cohort study. J Alzheimer's Dis. 2022;86(1).
- Hwang PH, Longstreth WT, Thielke SM, et al. Ophthalmic conditions associated with dementia risk: the cardiovascular health study. Alzheimers Dement. 2021;17(9).
- Exalto LG, Biessels GJ, Karter AJ, Huang ES, Quesenberry CP, Whitmer RA. Severe diabetic retinal disease and dementia risk in type 2 diabetes. J Alzheimer's Dis. 2014;42 Suppl 3:S109–117.
- Cheng D, Zhao X, Yang S, Wang G, Ning G. Association between diabetic retinopathy and cognitive impairment: a systematic review and meta-analysis. Front Aging Neurosci. 2021;13:692911.
- Lundeen EA, Burke-Conte Z, Rein DB, et al. Prevalence of diabetic retinopathy in the US in 2021. JAMA Ophthalmol. 2023;141(8).
- Flaxman SR, Bourne RRA, Resnikoff S, et al. Global causes of blindness and distance vision impairment 1990-2020. Lancet Glob Health. 2017;5(12):e1221–e1234.
- GBD 2019 Blindness and Vision Impairment Collaborators; Vision Loss Expert Group of the Global Burden of Disease Study. Causes of blindness and vision impairment in 2020 and trends over 30 years, and prevalence of avoidable blindness in relation to VISION 2020: the Right to Sight: an analysis for the Global Burden of Disease Study. Lancet Glob Health. 2021;9(2):e144–e160. doi:10.1016/S2214-109X(20)30489-7.
- Lim JI, Kim SJ, Bailey ST, Kovach JL, Vemulakonda GA, Ying GS, Flaxel CJ; American Academy of Ophthalmology Preferred Practice Pattern Retina/Vitreous Committee. Diabetic Retinopathy Preferred Practice Pattern®. Ophthalmology. 2025;132(4):P75–P162. doi:10.1016/j.ophtha.2024.12.020.
- Mutlu U, Colijn JM, Ikram MA, et al. Association of retinal neurodegeneration on optical coherence tomography with dementia: a population-based study. JAMA Neurol. 2018;75(10):1256–1263.
- Little K, Llorián-Salvador M, Scullion S, et al. Common pathways in dementia and diabetic retinopathy: understanding the mechanisms of diabetes-related cognitive decline. Trends Endocrinol Metab. 2022;33(1):50–71.
- Subramanian ML, Sampani K, Ness S, et al. Biomarkers for Alzheimer's disease are upregulated in patients with diabetic retinopathy. J Alzheimer's Dis. 2025.
- de la Monte SM. The full spectrum of Alzheimer's disease is rooted in metabolic derangements that drive type 3 diabetes. Adv Exp Med Biol. 2019;1128:45–83.
- Tomlinson G, Detsky AS. Composite end points in randomized trials: there is no free lunch. JAMA. 2010;303(3):267–268.