Abstract
INTRODUCTION
Multimorbidity – the coexistence of multiple chronic conditions – is common among older adults with cognitive impairment but is rarely addressed in an integrated manner.
METHODS
We analyzed 17,915 adults aged 55 to 84 years (8997 receiving uMETHOD clinical decision‐support care plans and 8918 from the National Health and Nutrition Examination Survey). The mean age was 69.0 years (SD 7.65; 95% CI: 68.86 to 69.1). Multimorbidity was quantified from diagnoses, labs, medications, vitals, and platform‐inferred conditions. Associations with cognitive impairment were modeled using age‐ and sex‐adjusted logistic regression.
RESULTS
Cognitive impairment was reported in 25.91% of participants (97.5% with ≥2 conditions); each additional condition was associated with higher odds (odds ratio: 1.069; 95% CI: 1.061 to 1.077; p < 0.0001). Prominent clusters included vascular‐metabolic disorders and micronutrient deficiencies.
DISCUSSION
Cognitive vulnerability rises stepwise with disease burden. Precision, multimorbidity‐aware decision support that integrates multidomain data can align cognitive and systemic care in real‐world practice.
Keywords: Alzheimer's disease and related dementias, clinical decision‐support systems, cognitive decline, cognitive impairment, electronic health records, mild cognitive impairment, multimorbidity, personalized care planning
Highlights
Cognitive impairment rose stepwise with multimorbidity burden.
Among participants with cognitive impairment, 97.5% had two or more chronic conditions.
Each added condition raised adjusted odds of cognitive impairment by 6.9%.
Higher‐burden clusters spanned renal, metabolic, immune, endocrine domains.
Clinical decision support can coordinate multimorbidity‐aware cognitive care.
1. INTRODUCTION
Multimorbidity refers to the presence of two or more chronic conditions in an individual. These conditions often span different physiological domains (e.g., cardiovascular, endocrine), and many—such as lipid disorders—directly contribute to the onset and progression of cognitive decline. As chronic diseases accumulate, they can interact in ways that impair compensatory mechanisms, increasing the likelihood of both physical and cognitive deterioration. Cognitive impairment, in turn, reduces an individual's ability to manage their health, creating a self‐reinforcing cycle.
RESEARCH IN CONTEXT
Systematic review: The authors reviewed PubMed and recent dementia literature through September 2025 on multimorbidity, cognitive impairment, and multimorbidity‐oriented clinical decision support. Prior work links accumulating chronic disease burden to dementia risk, but most studies define multimorbidity from diagnosis codes alone and rarely operationalize management using transparent, rule‐based decision‐support software.
Interpretation: In 17,915 adults aged 55 to 84 years from a referred clinical cohort and NHANES, 97.5% of those with cognitive impairment had ≥2 chronic conditions. Each additional condition was associated with higher odds of cognitive impairment (adjusted OR 1.069, 95% CI 1.061 to 1.077). Multimorbidity spanned renal, metabolic, immune, endocrine, and micronutrient domains, underscoring the need for integrated cognitive and systemic care.
Future directions: Prospective, independent validation should test whether multimorbidity‐aware decision support improves prioritization, reduces treatment conflicts and polypharmacy, and benefits cognitive and functional outcomes in routine practice.
Although multimorbidity has been linked to cognitive impairment, it is often measured using diagnosis codes alone, which can miss clinically relevant contributors apparent in objective clinical data such as laboratory results, medication exposure, and vital signs. In this study, we operationalize multimorbidity using diagnoses supplemented by pre‐specified, auditable rules applied to this wider range of medical data. We examine its association with cognitive impairment in both a clinically referred electronic health record (EHR)‐derived cohort and a population survey sample. We also report the most frequent combinations of co‐occurring conditions observed at greater multimorbidity burden to improve clinical interpretability and inform integrated care planning.
To enable this multidomain ascertainment, we leveraged clinical decision‐support software (CDSS; uMETHOD Health) that harmonizes EHR diagnoses with additional structured and extracted clinical signals, including laboratory results, medications, allergies, and vital signs, to generate an analyzable chronic‐condition profile. Condition identification is anchored in deterministic clinical logic, with limited AI‐based components used for language‐based extraction, anomaly detection, and test‐case validation.
Originally developed for Alzheimer's disease (AD), the CDSS platform now addresses a broad spectrum of chronic conditions and multimorbidity patterns. Designed for transparency, auditability, and scalability, it functions as a unified inference engine for generating personalized care plans and conducting population‐level analyses of unmet needs.
The platform integrates biospecimen data, medication and supplement histories, comorbidities, vital signs, genomic variants, cognitive assessments, allergy records, and demographic information. Inputs are standardized using RxNorm for clinical drugs; Logical Observation Identifiers Names and Codes (LOINC) for laboratory and clinical observations; the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD‐10‐CM) for diagnoses; HUGO Gene Nomenclature Committee nomenclature (HGNC) for human genes; and the Centers for Disease Control and Prevention vaccine‐administered code set (CVX) for immunizations. Rather than relying on static thresholds, the software applies sophisticated inference logic that evaluates longitudinal medication exposure, biomarker trajectories, and interrelated clinical findings. This enables identification of disease drivers, treatment gaps, adverse drug effects, and modifiable risks, producing a high‐fidelity clinical portrait that informs individualized intervention.
2. MATERIALS AND METHODS
We used clinical decision‐support software (CDSS; uMETHOD Health) to manage chronic disease, including neurological conditions, operationalizing multidomain intervention principles from the Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER), the Weill Cornell Alzheimer's Prevention Clinic, the 2017, 2020, and 2024 Lancet Commission reports, and the U.S. Study to Protect Brain Health Through Lifestyle Intervention to Reduce Risk (US POINTER). 1 , 2 , 3 , 4 , 5 , 6 , 7 For each individual, the platform generated tailored care plans from their clinical data, including problem lists, medical history, current medications, genomics, and biospecimen and cognitive test results, with recommendations spanning lifestyle, medication, and diagnostic domains.
Supplementary Methods Sections S1–S4 provide more details.
2.1. Cohorts and data sources
We analyzed two sources: a clinical cohort of 8997 adults (55 to 84 years old) who received uMETHOD care plans and 8918 adults (55 to 84 years old) from the National Health and Nutrition Examination Survey (NHANES); the combined dataset comprised 17,915 individuals. 8 Across the combined dataset, the mean age was 69.0 years (standard deviation [SD] 7.65; 95% confidence interval [CI]: 68.86 to 69.14). Cognitive impairment prevalence was higher in the clinical cohort (as expected for a referred population) and very low in NHANES. The combined prevalence was 25.91%; because this dataset pools a clinically referred cohort with a population survey sample, this pooled prevalence should be interpreted as descriptive of the analytic dataset rather than as a national prevalence estimate. For context, national studies of US adults aged 55 to 84 report prevalence estimates in a comparable range. 9 , 10
Case identification used three sources of evidence available within the data: (1) diagnostic coding consistent with mild cognitive impairment and dementia; (2) recorded results of standard cognitive assessments, indicating impairment; and (3) use of medications prescribed specifically for cognitive impairment or AD (donepezil, rivastigmine, galantamine, memantine, lecanemab, and donanemab). Unlike certain chronic comorbidities, cognitive impairment status was not derived from biomarker‐based inference rules. For this analysis, mild cognitive impairment and dementia were combined into a single binary outcome rather than modeled separately.
2.2. Exposure, conditions, and covariates
The primary exposure was multimorbidity, measured as the count of chronic conditions. Conditions were assembled from documented ICD‐10‐CM diagnoses and from platform‐inferred conditions using biomarkers, medication lists, and vitals (e.g., chronic kidney disease [CKD] stages 3–5 inferred from persistently low estimated glomerular filtration rate or creatinine clearance). To standardize granularity, families of specific codes were collapsed to a single parent diagnosis (e.g., multiple E11* codes → E11 for type 2 diabetes; osteoarthritis M15–M19 → M15).
Within each cohort, candidates for analysis were ICD‐10‐CM conditions flagged as “chronic” by ICDList 11 and observed in ≥50 people.
ICD‐10‐CM code E72.12, methylenetetrahydrofolate reductase (MTHFR) deficiency, was not reported in the NHANES dataset. In the uMETHOD dataset, it was determined from genomes. Since genomes were not available for all uMETHOD individuals, the presence of this condition could not be determined reliably, and it was excluded from this analysis.
ICD‐10 Code E88.810, Metabolic syndrome, was also excluded, since it is an umbrella condition that can include obesity, hypertension, dyslipidemia, and type 2 diabetes mellitus. This avoids double counting.
Candidates that showed a positive, statistically significant association with cognitive impairment after adjusting for age and sex were included. The primary pooled models adjusted for age and sex because these covariates were consistently available and comparable across both cohorts. Additional behavioral and socioeconomic covariates (e.g., physical activity, education, and income proxies) were available in NHANES but were not captured in sufficiently complete, standardized form in the uMETHOD EHR‐derived dataset to support harmonized adjustment across cohorts.
2.3. Statistical analysis
We summarized demographics, condition prevalence, and cognitive status using standard descriptive statistics (means ± SD; counts and percentages). Associations between condition count, individual conditions, and cognitive impairment were estimated using age‐ and sex‐adjusted logistic regression and reported as odds ratios (ORs) with 95% CI and p values.
We conducted stratified analyses by cohort (uMETHOD and NHANES), including testing of cohort–multimorbidity interactions, given the marked differences in the prevalence of cognitive impairment between the cohorts.
Although longitudinal data are not available for NHANES (and sporadically available for uMETHOD), we provided a stratified analysis by age group (55 to 64, 65 to 74, and 75 to 84), and tested for age group–condition interactions.
3. RESULTS
Table 1 shows the ORs, 95% CI, and p values for the combined cohort. Figure 1 shows the odds ratios for the chronic conditions versus cognitive impairment. Only chronic conditions that had an OR greater than 1 and were statistically significant (p value < 0.05) are included in the figure.
TABLE 1.
Odds ratios for chronic conditions versus cognitive impairment.
| Chronic Condition | Odds Ratio | Lower CI | Upper CI | p‐value |
|---|---|---|---|---|
| Vitamin B12 deficiency | 3.121 | 2.497 | 3.900 | <0.0001 |
| Atherosclerosis of arteries | 1.952 | 1.633 | 2.334 | <0.0001 |
| Hypertensive CKD stage 1–4 | 1.934 | 1.570 | 2.383 | <0.0001 |
| Parkinson's disease | 1.925 | 1.372 | 2.700 | 0.0001 |
| Vitamin D deficiency | 1.811 | 1.650 | 1.987 | <0.0001 |
| Chronic pain | 1.761 | 1.479 | 2.095 | <0.0001 |
| Essential tremor | 1.587 | 1.086 | 2.320 | 0.0171 |
| Hypothyroidism | 1.585 | 1.411 | 1.780 | <0.0001 |
| Anxiety | 1.575 | 1.411 | 1.759 | <0.0001 |
| Cataract | 1.560 | 1.228 | 1.983 | 0.0003 |
| Atrial fibrillation | 1.511 | 1.248 | 1.829 | <0.0001 |
| ADHD | 1.506 | 1.030 | 2.201 | 0.0349 |
| Disorders of zinc metabolism | 1.502 | 1.274 | 1.771 | <0.0001 |
| Irritable bowel syndrome | 1.491 | 1.144 | 1.944 | 0.0031 |
| Nicotine dependence | 1.463 | 1.168 | 1.832 | 0.0009 |
| Achlorhydria | 1.436 | 1.211 | 1.704 | <0.0001 |
| Primary insomnia | 1.419 | 1.048 | 1.923 | 0.0238 |
| BPH | 1.369 | 1.166 | 1.606 | 0.0001 |
| Peripheral vascular disease | 1.337 | 1.106 | 1.617 | 0.0027 |
| Adrenocortical insufficiency | 1.332 | 1.159 | 1.532 | 0.0001 |
| Osteoporosis | 1.272 | 1.097 | 1.474 | 0.0014 |
| Polyneuropathy | 1.251 | 1.025 | 1.525 | 0.0274 |
| Incontinence | 1.245 | 1.023 | 1.515 | 0.0291 |
| CKD stages 3‐5 | 1.223 | 1.112 | 1.344 | <0.0001 |
| Iron deficiency anemia | 1.221 | 1.057 | 1.409 | 0.0065 |
| Anemia | 1.182 | 1.001 | 1.397 | 0.0489 |
| Hearing loss | 1.182 | 1.001 | 1.395 | 0.0486 |
| Immunodeficiency | 1.178 | 1.065 | 1.302 | 0.0014 |
| Dyslipidemia | 1.156 | 1.052 | 1.270 | 0.0025 |
FIGURE 1.
Odds ratios for chronic conditions versus cognitive impairment.
3.1. Cohort heterogeneity
We observed nominally significant effect modification by cohort for six of the 137 comorbidities (unadjusted p interaction < 0.05). Stratum‐specific odds are presented in Table S1 for these comorbidities. For the remaining 131 comorbidities, no significant heterogeneity was detected (all p interaction > 0.05), supporting similar associations across the cohorts.
While six comorbidities showed nominally significant cohort interactions in unadjusted analyses (raw p < 0.05), none remained statistically significant after Benjamini‐Hochberg false discovery rate (FDR) correction (all q > 0.05). These are shown in the rightmost column of Table S1. These findings indicate no strong evidence of effect modification by cohort after accounting for multiple comparisons.
3.2. Cognitive impairment and multimorbidity
The set of chronic conditions that had ORs greater than 1 and were statistically significant was used to determine the association between the number of chronic conditions and cognitive impairment.
Of the 17,915 individuals in this cohort, 4641 (25.91%) had cognitive impairment. Of those with cognitive impairment, 4523 (97.5%) had two or more chronic conditions of any type in the full 137‐condition candidate set. Among those without cognitive impairment (n = 13,274), 12,568 (94.7%) had ≥2 chronic conditions. Mean condition burden was higher among those with cognitive impairment (9.26 conditions) than those without cognitive impairment (7.04 conditions). A top 10 list of the individual comorbid conditions is shown in Table S2.
Table 2 illustrates that as the number of chronic conditions an individual has increases, the likelihood of that individual having cognitive impairment increases. Adjusted for age and sex, the OR of the number of chronic conditions versus cognitive impairment was 1.069 (95% CI: 1.061 to 1.077; p < 0.0001). Sex was not significantly correlated with cognitive impairment. Counts in Table 2 are restricted to the 29 conditions that showed a positive, statistically significant association with cognitive impairment.
TABLE 2.
Association between number of chronic conditions and cognitive impairment.
| Number of chronic conditions | Number of Individuals | Number with cognitive impairment | Percentage with cognitive impairment |
|---|---|---|---|
| 0 | 1843 | 126 | 6.84% |
| 1 | 3698 | 324 | 8.76% |
| 2 | 3619 | 568 | 15.69% |
| 3 | 2707 | 677 | 25.01% |
| 4 | 1978 | 721 | 36.45% |
| 5 | 1430 | 679 | 47.48% |
| 6 | 979 | 515 | 52.60% |
| 7 | 705 | 427 | 60.57% |
| 8 | 451 | 278 | 61.64% |
| 9 | 245 | 145 | 59.18% |
| 10 | 128 | 87 | 67.97% |
| 11 | 69 | 46 | 66.67% |
| 12 | 38 | 30 | 78.95% |
| 13 | 14 | 10 | 71.43% |
| 14 | 7 | 5 | 71.43% |
| 15 | 3 | 2 | 66.67% |
| 16 | 1 | 1 | 100.00% |
With smaller numbers, percentages can easily fluctuate. To avoid relying on percentages with smaller numbers, the number of chronic conditions is limited to eight in Figure 2.
FIGURE 2.
Number of chronic conditions versus percentage of people with cognitive impairment. Numbers above bars indicate sample size in each chronic‐condition category; categories ≥8 are aggregated due to small cell sizes.
These findings are visually represented in Figure 2. As multimorbidity increases, so does the likelihood of cognitive impairment, suggesting a strong dose–response relationship that underscores the importance of care models that account for chronic disease complexity in cognitive health management.
Table 3 summarizes the mode (most frequent) condition cluster observed at each multimorbidity count. Across levels, a vascular–metabolic core (CKD stages 3–5, dyslipidemia) recurs across most levels, while endocrine and micronutrient disorders (hypothyroidism, iron deficiency anemia, vitamin B12 and D deficiencies) accrete in higher‐order clusters. An “immunodeficiency” flag co‐occurs in several high‐burden clusters. These patterns are coherent with shared pathophysiology and polypharmacy risks, and they align with known vascular and metabolic contributors to cognitive decline.
TABLE 3.
Most frequently occurring combinations of chronic conditions.
| Number of chronic conditions | Most frequent combination |
|---|---|
| 1 | Dyslipidemia |
| 2 | Dyslipidemia, immunodeficiency |
| 3 | CKD stages 3–5, dyslipidemia, immunodeficiency |
| 4 | CKD stages 3–5, dyslipidemia, immunodeficiency, vitamin D deficiency |
| 5 | CKD stages 3–5, dyslipidemia, hypothyroidism, immunodeficiency, vitamin D deficiency |
| 6 | Dyslipidemia, hypothyroidism, immunodeficiency, iron deficiency anemia, vitamin B12 deficiency, vitamin D deficiency |
| 7 | CKD stages 3–5, dyslipidemia, hypothyroidism, immunodeficiency, iron deficiency anemia, vitamin B12 deficiency, vitamin D deficiency |
As the number of conditions increases, so does the number of ways diseases and their treatments can affect one another. These may contribute to diagnostic uncertainty, fragment care, and raise the risk of adverse events.
Understanding these patterns is central to personalized care planning. By highlighting the most common and consequential groupings, the CDSS care plans help clinicians prioritize both individual conditions and the points where they intersect, targeting the layered complexity that is associated with much of the functional decline in older adults.
3.3. Age and multimorbidity
Figure 3 illustrates that the average number of chronic conditions increases with age. In the adjusted model, older age (per 1‐year increase) was associated with higher odds of cognitive impairment (OR 1.114; 95% CI: 1.108 to 1.120; p < 0.0001).
FIGURE 3.
Average number of chronic conditions and percentage of people with cognitive impairment by age. The red solid line shows the average number of chronic conditions using the left vertical axis; the black dashed line shows the percentage of people with cognitive impairment using the right vertical axis.
This confirms a well‐documented trajectory: Multimorbidity accumulates with age, spanning cardiovascular, metabolic, endocrine, and neuropsychiatric domains. Many of these age‐associated conditions, especially vascular–metabolic clusters, are positively associated with cognitive decline.
These findings emphasize the need for age‐sensitive, multimorbidity‐aware models of care. Disease‐specific guidelines are poorly suited to older adults with multiple interacting conditions. Clinical decision support that evaluates cumulative disease burden offers a scalable path for tailoring management to aging patients at cognitive risk.
We explored analyses stratified by age group, 55 to 64, 65 to 74, and 75 to 84, to identify comorbidities that are more strongly associated with cognitive impairment at earlier ages. Table S3 shows 10 statistically significant interactions.
However, after correction for multiple comparisons using the Benjamini‐Hochberg FDR procedure (137 tests), three remained statistically significant at q ≤ 0.05. These are again shown in the rightmost column of Table S3.
Vitamin D deficiency had the strongest association with cognitive impairment in the 55 to 64 age group. Adrenocortical insufficiency had the strongest association in the 65 to 74 age group. Peripheral vascular disease had the strongest association in the 75 to 84 age group.
3.4. Cognitive impairment, multimorbidity, and age
Cognitive impairment correlates with multimorbidity, and multimorbidity becomes more common with age. To show how these two signals move across the study age range (55 to 84 years), Figure 3 plots, by age, the average number of chronic conditions and the percentage of people with cognitive impairment.
Both curves rise with age, consistent with disease accumulation and increasing cognitive vulnerability. This visualization does not imply causation – age and disease burden travel together – but it motivates analyzing them separately (as in Section 3.2) and planning care that accounts for both.
4. DISCUSSION
Here, we synthesize the findings of this study and their practical meaning. We quantify the dose–response interaction between multimorbidity and cognitive impairment; situate the results in the context of prior literature; describe the condition constellations that plausibly drive risk; explain detection gaps outside problem lists and how the model links labs, medications, vitals, and diagnoses; translate these into clinical and implementation steps; and end with strengths, limitations, and concise conclusions for practice and validation at scale.
4.1. Principal findings
Across the combined cohort, cognitive impairment rose stepwise with multimorbidity. Using the study's age‐ and sex‐adjusted model, each additional chronic condition increased the odds of cognitive impairment (OR 1.069; 95% CI: 1.061 to 1.077). Prevalence moved from 6.84% among those with 0 chronic conditions to 63.2% among those with ≥8 conditions (from Table 2, aggregating 8 to 16), demonstrating a strong dose–response gradient. Composition mattered alongside count: vascular–metabolic constellations were over‐represented in higher‐burden strata.
We measure multimorbidity using coded diagnoses supplemented by rule‐based signals from laboratory results, medication exposure, and vital signs, and we summarize the most frequent co‐occurring condition patterns at higher burden. Results are shown for both cohorts, with cohort‐specific and sensitivity analyses in the Supplementary Material.
4.2. How the pattern fits prior evidence
Our gradient mirrors external literature showing that monotonic risk increases with the accumulation of chronic diseases, with disproportionate hazards when cardiometabolic and mental‐health conditions co‐occur. 12 , 13 , 14 , 15 The dominant clusters in this cohort – CKD, dyslipidemia, hypothyroidism, and micronutrient deficiencies – are consistent with known pathways linking vascular injury, metabolic dysfunction, and inflammation to cognitive vulnerability. 16 , 17 , 18
4.3. Which constellations drive risk (and why)
Repeated clusters in Table 3 converge on shared mechanisms:
CKD stages 3–5 + dyslipidemia + immunodeficiency (3–5 conditions): Hormonal drive that raises pressure and retains salt, endothelial dysfunction, and small‐vessel injury connect kidney and brain phenotypes, while chronic immune activation adds an inflammatory component; tight blood pressure control, albuminuria reduction, and lipid management are key priorities to mitigate cerebrovascular risk.
Dyslipidemia + hypothyroidism + immunodeficiency + iron deficiency anemia + vitamin B12 deficiency + vitamin D deficiency (6 conditions): This constellation reflects a convergence of metabolic, endocrine, immune, and hematologic dysfunction. Dyslipidemia and hypothyroidism jointly impair lipid handling and energy metabolism, while iron and B12 deficiencies reduce oxygen delivery and disrupt myelin integrity. Concurrent vitamin D deficiency and immunodeficiency impair immune resilience and signal chronic inflammation. Together, these deficits produce diffuse cognitive slowing, fatigue, and reduced physiologic reserve, driven by systemic insufficiency rather than a single dominant pathology.
CKD stages 3–5 + dyslipidemia + hypothyroidism + immunodeficiency + iron deficiency anemia + vitamin B12 deficiency + vitamin D deficiency (7 conditions): Parallel hits on oxygen delivery, myelin integrity, lipid metabolism, and inflammatory tone produce diffuse cognitive slowing and fatigue rather than a single etiologic lesion.
These patterns highlight leverage points with cross‐condition benefit (e.g., blood pressure and albuminuria targets in CKD; correcting B12/iron/vitamin D deficiencies; deprescribing or dose‐rationalizing long‐term PPI use when appropriate).
4.4. Detection gap and data model implications
Structured EHR problem lists undercount cognitively relevant burden. Signals often reside in laboratory trajectories, medication histories (including anticholinergic load and drug–drug interactions), vitals, and genomics. By linking these inputs, the platform is designed to surface under‐recorded conditions of potential cognitive relevance (e.g., metabolic acidosis, hyponatremia, thrombocytopenia) that are not reported among the analyses above, while maintaining specificity via confidence scoring and review, an approach consistent with scalable, person‐centered multimorbidity assessment frameworks. 19 , 20
4.5. Clinical and implementation implications
Taken together, counts matter and composition matters. The idea of treating to target for a single disease, in isolation, risks exacerbating competing conditions, polypharmacy, and monitoring burden. Multimorbidity‐aware care planning – supported by interoperable decision support – offers a practical route to coordinate actions across cardiometabolic, renal, endocrine, inflammatory, and neuropsychiatric domains. Reports from a large randomized lifestyle program and a platform that transforms clinical guidelines into implementable clinical decision‐support services underscore the feasibility and generalizability of structured implementation at scale. 7 , 21
4.6. Strengths and limitations
Strengths include integration of heterogeneous data sources at scale; terminology standardization using ICD‐10‐CM, RxNorm, LOINC, and Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT); and explicit, auditable inference rules linking inputs to derived conditions. The study's limitations include differential selection and ascertainment across constituent datasets (including differences between the clinically referred and population‐based cohorts), age‐censoring conventions in NHANES, and residual confounding typical of observational designs. Residual confounding by education, physical activity, and other social or behavioral factors remains possible because these variables could not be harmonized across the two cohorts.
This study is cross‐sectional and therefore cannot establish temporal directionality. Cognitive impairment may contribute to worsening disease self‐management, medication adherence challenges, reduced activity, and accelerated accumulation or detection of chronic conditions; thus, part of the observed association between multimorbidity burden and cognitive impairment may reflect downstream consequences of cognitive impairment rather than upstream risk. Longitudinal analyses with disease onset timing and cognitive trajectories are needed to disentangle bidirectional effects.
5. CONCLUSIONS
Cumulative multimorbidity burden is strongly associated with cognitive impairment, and across the full range of observed burden its aggregate signal exceeds that of any single diagnosis in this dataset. In this cohort, 97.5% of individuals with cognitive impairment had at least two chronic conditions, and risk rose stepwise with each added condition (OR 1.069; 95% CI: 1.061 to 1.077). Age was also independently associated with higher odds of cognitive impairment.
We observe that disease‐specific guidelines are poorly suited to older adults with multiple interacting conditions. Older adults typically present with interacting cardiometabolic, renal, endocrine, and neuropsychiatric conditions whose combined effects outweigh any one diagnosis. Disease‐specific guidance seldom reconciles conflicting recommendations, offers little help in cross‐condition prioritization, and can drive polypharmacy, therapeutic competition, and monitoring burden. Multimorbidity‐aware guidance must explicitly resolve conflicts, sequence work, and favor actions with cross‐condition benefit.
A practical response starts with measurement. Problem lists undercount relevant burden; integrating laboratory patterns, medication histories, vitals, genomics, and recorded diagnoses yields a more complete clinical picture that clarifies who is at risk and why. In our implementation, the uMETHOD platform performs this integration – identifying both documented and rules‐based inferred conditions – and can support consistent plan construction as part of routine workflows. However, the present analysis is observational and cross‐sectional; prospective and independently replicated evaluations are needed to determine whether multimorbidity‐aware decision support improves cognitive or functional outcomes.
CONFLICT OF INTEREST STATEMENT
Authors are employed by and/or hold stock in uMETHOD Health. Because all authors are affiliated with the platform developer, we have aimed to separate empirical findings from implementation implications and to specify inference logic transparently to enable independent replication. Author disclosures are available in the Supporting Information.
ETHICS APPROVAL AND CONSENT TO PARTICIPATE
This secondary analysis used only pre‐existing, de‐identified data from the uMETHOD cohort and publicly available NHANES files. All uMETHOD participants had previously provided written informed consent at program entry authorizing de‐identified research use of their data, and NHANES participants provided consent under the NCHS Research Ethics Review Board. Because no identifiable private information was obtained or used and no interaction or intervention with participants occurred for this analysis, the analysis did not involve human subjects as defined by 45 CFR 46.102(e)(1); therefore, IRB review and additional consent were not required.
CLINICAL TRIAL NUMBER
Not applicable.
Supporting information
Supporting Information: dad270395‐sup‐0001‐SuppMat.docx
Supporting Information: dad270395‐sup‐0002‐Figure S1.docx
Supporting Information: dad270395‐sup‐0004‐Table S2.docx
Supporting Information: dad270395‐sup‐0005‐Table S3.docx
Supporting Information: dad270395‐sup‐0006‐ICMJE.pdf
ACKNOWLEDGMENTS
uMETHOD Health Inc. (Raleigh, NC, USA) provided internal support for data processing/analysis and manuscript preparation. No external grant funding was received.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from uMETHOD Health, but restrictions apply to the availability of these data, which are proprietary company information and are not publicly available. Data are, however, available from the authors upon reasonable request and with the permission of uMETHOD Health. An earlier version of this paper was published at the July 2020 Alzheimer's Association International Conference. 22
REFERENCES
- 1. Keine D, Walker JQ, Kennedy BK, Sabbagh MN. Development, application, and results from a precision‐medicine platform that personalizes multi‐modal treatment plans for mild Alzheimer's disease and at‐risk individuals. Curr Aging Sci. 2018;11(3):173‐181. doi: 10.2174/1874609811666181019101430 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Ngandu T, Lehtisalo J, Solomon A, et al. A 2‐year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at‐risk elderly people (FINGER): a randomized controlled trial. Lancet. 2015;385(9984):2255‐2263. doi: 10.1016/S0140-6736(15)60461-5 [DOI] [PubMed] [Google Scholar]
- 3. Isaacson RS, Ganzer CA, Hristov H, et al. The clinical practice of risk reduction for Alzheimer's disease: a precision medicine approach. Alzheimers Dement. 2018;14(12):1663‐1673. doi: 10.1016/j.jalz.2018.08.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Livingston G, Sommerlad A, Orgeta V, et al. Dementia prevention, intervention, and care. Lancet. 2017;390(10113):2673‐2734. doi: 10.1016/S0140-6736(17)31363-6 [DOI] [PubMed] [Google Scholar]
- 5. Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396(10248):413‐446. doi: 10.1016/S0140-6736(20)30367-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Livingston G, Huntley J, Liu KY, et al. Dementia prevention, intervention, and care: 2024 report of the Lancet Standing Commission. Lancet. 2024;404(10452):572‐628. doi: 10.1016/S0140-6736(24)01296-0 [DOI] [PubMed] [Google Scholar]
- 7. Baker LD, Espeland MA, Whitmer RA, et al. Structured vs self‐guided multidomain lifestyle interventions for global cognitive function: the US POINTER randomized clinical trial. JAMA. 2025;334(8):681‐691. doi: 10.1001/jama.2025.12923 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. National Center for Health Statistics (NCHS) , Centers for Disease Control and Prevention (CDC). National Health and Nutrition Examination Survey (NHANES). NCHS; [cited 2025 Sep 4]. Available from: https://wwwn.cdc.gov/nchs/nhanes/Default.aspx [Google Scholar]
- 9. Taylor CA, Bouldin ED, McGuire LC. Subjective cognitive decline among adults aged ≥45 years—United States, 2015‐2016. MMWR Morb Mortal Wkly Rep. 2018;67(27):753‐757. doi: 10.15585/mmwr.mm6727a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Manly JJ, Jones RN, Langa KM, et al. Estimating the prevalence of dementia and mild cognitive impairment in the US: the 2016 Health and Retirement Study Harmonized Cognitive Assessment Protocol Project. JAMA Neurol. 2022;79(12):1242‐1249. doi: 10.1001/jamaneurol.2022.3543 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. ICD List – Trusted 2025 ICD‐10‐CM & PCS code lookup. ICD List; 2025. [cited 2025 Sep 7]. Available from: https://icdlist.com/ [Google Scholar]
- 12. Zhou Y, You Y, Zhang Y, Zhang Y, Yuan C, Xu X. Multimorbidity and risk of dementia: a systematic review and meta‐analysis of longitudinal cohort studies. J Prev Alzheimers Dis. 2025;12(7):100164. doi: 10.1016/j.tjpad.2025.100164 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Xin B, Zhang D, Fu H, Jiang W. Association between multimorbidity and the risk of dementia: a systematic review and meta‐analysis. Arch Gerontol Geriatr. 2025;131:105760. doi: 10.1016/j.archger.2025.105760 [DOI] [PubMed] [Google Scholar]
- 14. Du M, Liu M, Liu J. Effects of physical and psychological multimorbidity on the risk of dementia: multinational prospective cohorts and a meta‐analysis. BMC Med. 2024;22:423. doi: 10.1186/s12916-024-03642-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Calvin CM, Conroy MC, Moore SF, Kuzma E, Littlejohns TJ. Association of multimorbidity, disease clusters, and modification by genetic factors with risk of dementia. JAMA Netw Open. 2022;5(9):e2232124. doi: 10.1001/jamanetworkopen.2022.32124 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Tanaka S. Targeting inflammation in perivascular cells and neuroimmune interactions for treating kidney disease. Clin Exp Nephrol. 2024;28(6):505‐512. doi: 10.1007/s10157-024-02494-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Zhou X‐W, Zhang H, Zhang J‐H, et al. Role of ALOX5AP in the pathogenesis of cerebral small vessel disease: impact of inflammation and lipid metabolism dysregulation. Brain Res. 2025;1865:149814. doi: 10.1016/j.brainres.2025.149814 [DOI] [PubMed] [Google Scholar]
- 18. Varra F‐N, Varras M, Varra V‐K. Theodosis‐Nobelos P. Molecular and pathophysiological relationship between obesity and chronic inflammation in the manifestation of metabolic dysfunctions and their inflammation‐mediating treatment options. Mol Med Rep. 2024;29(6):95. doi: 10.3892/mmr.2024.13219 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. World Health Organization . Integrated care for older people (ICOPE): guidance for person‐centred assessment and pathways in primary care. 2nd ed.: World Health Organization; 2024. [cited 2025 Sep 7]. Available from: https://www.who.int/publications/i/item/9789240103726 [Google Scholar]
- 20. Bierman AS, Burke BT, Comfort LN, Gerstein M, Mueller NM, Umscheid CA. From precision medicine to precision care: choosing and using precision medicine in the context of multimorbidity. Camb Prism Precis Med. 2023;1:e19. doi: 10.1017/pcm.2023.8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Gencturk M, Laleci Erturkmen GB, Akpinar AE, et al. Transforming evidence‐based clinical guidelines into implementable clinical decision‐support services: the CAREPATH study for multimorbidity management. Front Med (Lausanne). 2024;11:1386689. doi: 10.3389/fmed.2024.1386689 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Zelek MC, Walker JQ, Sabbagh MN. Severe multimorbidity is prevalent in those with cognitive decline, which challenges neurologists and family physicians: a clinical decision‐support software (CDSS) platform provides guidance. Alzheimers Dement. 2020;16(S7):e042243. doi: 10.1002/alz.042243 [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information: dad270395‐sup‐0001‐SuppMat.docx
Supporting Information: dad270395‐sup‐0002‐Figure S1.docx
Supporting Information: dad270395‐sup‐0004‐Table S2.docx
Supporting Information: dad270395‐sup‐0005‐Table S3.docx
Supporting Information: dad270395‐sup‐0006‐ICMJE.pdf
Data Availability Statement
The data that support the findings of this study are available from uMETHOD Health, but restrictions apply to the availability of these data, which are proprietary company information and are not publicly available. Data are, however, available from the authors upon reasonable request and with the permission of uMETHOD Health. An earlier version of this paper was published at the July 2020 Alzheimer's Association International Conference. 22