Vitamin D3 for Life

Live Enlightened

Research libraryVitamin D3 For Life
Study design and methods

Primary relevance to vitamin D Vitamin D is a stated subject of this paper.

Vitamin D is a stated subject of this paper; the supporting source evidence is shown below.

Why this paper is included

Supporting source evidence:

VDD was associated with a higher risk of incident AAA than normal vitamin D status (0.66% vs. 0.45%; hazard ratio [HR] 1.40, 95% confidence interval [CI], 1.33 to 1.48; p < 0.001).

Source: Abstract. View original paper

Relevance classification does not establish study quality, causation or editorial approval.

Study design Cohort study.

Cohort study

Examines exposures and outcomes over time within a defined group.

Meta-analysis unclear · Blinding unclear / not reported · Human research reported

How this paper was classified

Automatic source-based classification; awaiting editorial verification. This is not a quality score.

Vitamin D deficiency and the long-term risk of new-onset abdominal aortic aneurysm and severe aortic events: a multicenter propensity score: matched cohort study

Source: Article title.

Participants were stratified according to their initial qualifying 25(OH)D values.

Source: Methods.

View supporting evidence in the original paper

What the study types mean
Systematic review
Uses explicit methods to identify, assess and summarise research on a defined question.
Scoping review
Maps the extent and characteristics of research on a topic.
Narrative review
Discusses existing research; systematic-review methods have not been established by this classification.
Review / evidence synthesis
Reviews existing research; the exact review methods need further assessment.
Randomised controlled trial
Assigns participants or groups to comparison interventions by chance.
Non-randomised intervention study
Examines an intervention without random allocation.
Cohort study
Examines exposures and outcomes over time within a defined group.
Case-control study
Compares people with an outcome against people without it, examining exposures.
Cross-sectional study
Examines a population at a particular point in time.
Observational study
Observes exposures and outcomes without assigning the exposure; the subtype is not established here.
Case report
Describes an individual case.
Case series
Describes a group of individual cases.
Secondary analysis
Analyses existing study data; this paper is not automatically another independent trial.
Laboratory study
Investigates biological processes in a laboratory setting.
Qualitative study
Examines experiences or meanings using qualitative methods.
Genetic instrumental-variable study
Uses genetic variation as an instrument to investigate relationships; this is not a randomised treatment trial.
Study protocol
Describes planned methods and is not evidence that the planned study has reported results.
Multiple reported designs
More than one design is explicitly reported; examine the supporting methods.
Study design awaiting verification
The available source does not establish a sufficiently clear study design.

Meta-analysis combines numerical results from studies. Blinding describes who was unaware of treatment allocation. Neither is a separate quality grade. Study design, vitamin D relevance and document format are independent filters.

Full studies

Vitamin D deficiency and the long-term risk of new-onset abdominal aortic aneurysm and severe aortic events: a multicenter propensity score-matched cohort study

Cohort study · Frontiers in Nutrition · 2026

Original publication title

Vitamin D deficiency and the long-term risk of new-onset abdominal aortic aneurysm and severe aortic events: a multicenter propensity score: matched cohort study

Kuo-Chuan Hung; Ting-Sian Yu; I-Wen Chen

Copyright © 2026 Hung, Yu and Chen.

Original publication · Publisher record · CC BY 4.0

Source wording retained. Formatting, section navigation and display-title punctuation adapted for this research record.

Authors’ conclusion

Clinically defined VDD was associated with a higher long-term risk of incident AAA in this large propensity score: matched cohort study. The association was supported by sensitivity models and multivariable adjustment, whereas secondary outcomes, subgroup analyses, and exposure gradient analyses should be interpreted as exploratory and hypothesis-generating. These findings should not be interpreted as supporting VDD-based AAA screening, but they suggest that VDD may serve as an adjunctive risk marker when evaluating patients with established vascular risk factors in general or perioperative clinical settings.

Given the modest absolute risk difference and the observational design, further prospective studies with systematic aortic imaging, repeated vitamin D measurements, detailed lifestyle data, and mechanistic endpoints are needed to determine whether VDD is causally involved in AAA development or severe aortic complications.

An extract from the source publication. Read it in context.

At a glance

Study background

What background does the paper provide?

Abdominal aortic aneurysm (AAA) is a clinically silent but potentially fatal vascular disease that may progress to rupture, emergency surgery, and death (1Not reported4). Its pathogenesis is characterized by chronic vascular inflammation, extracellular matrix degradation, vascular smooth muscle cell dysfunction, elastin fragmentation, and progressive loss of aortic wall integrity (5Not reported9). Although advanced age, male sex, smoking, hypertension, and atherosclerotic disease are well-established risk factors (10Not reported12), additional potentially modifiable risk markers related to aortic wall degeneration remain undefined.

Vitamin D deficiency (VDD) is common in older adults and patients with cardiometabolic comorbidities (13, 14). Beyond its classical role in calcium and bone metabolism, vitamin D signaling may influence vascular inflammation, oxidative stress, endothelial function, vascular smooth muscle cell phenotype, and extracellular matrix remodeling (15Not reported18). These pathways are directly relevant to aneurysm biology because AAA formation involves inflammatory cell infiltration, matrix metalloproteinase activation, medial degeneration, and weakening of the aortic wall (19Not reported21).

Experimental evidence (22) further suggests that VDD may promote larger and more rupture-prone AAA phenotypes, potentially through enhanced extracellular matrix remodeling and upregulation of osteopontin and matrix metalloproteinases, thereby providing a biologically plausible link between low vitamin D status and impaired aortic wall stability in patients with VDD.

However, human epidemiological evidence remains limited and inconsistent. A recent UK Biobank cohort study reported an inverse association between serum 25-hydroxyvitamin D concentrations and incident AAA (23), whereas the ARIC prospective cohort found little evidence that vitamin D-related biomarkers were associated with incident AAA (24). Moreover, previous studies have focused mainly on AAA occurrence (23, 24), leaving uncertainty regarding whether clinically defined VDD is associated with severe aortic outcomes, including AAA rupture and aortic dissection.

Therefore, we conducted a multicenter propensity score: matched cohort study using a large real-world electronic health record network to evaluate the association between VDD and the long-term risk of incident AAA. We also examined severe aortic events, including ruptured AAA and aortic dissection, as secondary outcomes. We hypothesized that patients with VDD would have a higher long-term risk of incident AAA and severe aortic events than those with normal vitamin D status.

Source: Introduction Source: Introduction Source: Introduction

Study methods

What methods does the paper describe?

2.1

Study design and data source

This retrospective propensity score: matched cohort study was conducted using the TriNetX Global Collaborative Network, a federated research platform that aggregates longitudinal, de-identified electronic health records from multiple participating healthcare organizations. The TriNetX network has been widely used in peer-reviewed epidemiological and clinical outcomes research, supporting its utility for large-scale real-world evidence studies (25Not reported28). Because the dataset contained no directly identifiable patient information, the need for individual informed consent was waived. The study protocol was approved by the Institutional Review Board of Chi Mei Medical Center (Tainan, Taiwan) and was conducted in accordance with the Declaration of Helsinki and relevant institutional guidelines.

2.2

Study population and exposure classification

The study enrolled adults aged ≥50 years with at least one documented serum 25-hydroxyvitamin D [25(OH)D] measurement recorded between January 1, 2010, and December 31, 2023. To confirm ongoing engagement with the healthcare system and allow reliable ascertainment of baseline comorbidities, each individual was required to have a minimum of two healthcare encounters within the preceding 5-year window.

Participants were stratified according to their initial qualifying 25(OH)D values. Those with a concentration below 20 ng/mL were assigned to the VDD cohort, whereas individuals with a value of 30 ng/mL or higher formed the control cohort. The date of the qualifying measurement was used as the index date. To strengthen exposure specificity, individuals in the VDD cohort who had any recorded 25(OH)D value at or above 20 ng/mL within the preceding 3 years were removed, and control participants with any value below 30 ng/mL during the same look-back window were excluded.

This restriction was intended to identify consistently documented VDD and reduce misclassification from transiently low 25(OH)D values. This approach better aligned with the hypothesized chronic vascular effects of VDD, although potential selection related to frailty, chronic illness, or greater clinical monitoring cannot be fully excluded.

2.3

Exclusion criteria

Individuals with a pre-existing diagnosis of aortic aneurysm or aortic dissection before the index date were excluded to ensure an aortic disease: free baseline. Conditions known to markedly alter vitamin D metabolism, including end-stage renal disease, stage 4 or 5 chronic kidney disease, renal dialysis dependence, prior bariatric surgery, hepatic failure, Crohn’s disease, and celiac disease, also warranted exclusion. To reduce confounding from acute illness: driven vitamin D testing or transient fluctuations in vitamin D levels, individuals who experienced acute kidney failure, sepsis, severe sepsis, critical care utilization, or hospital admission within 30 days before the index date were excluded.

Finally, patients with hereditary, congenital, inflammatory, or infectious conditions associated with aortopathy, namely Marfan syndrome, bicuspid aortic valve, Ehlers: Danlos syndrome, Takayasu arteritis, systemic connective tissue disorders, syphilis, and human immunodeficiency virus infection, were excluded. The diagnostic codes for all exclusion criteria are detailed in Supplementary Table 1.

2.4

Landmark design

To mitigate reverse causation and immortal-time bias, a 1-year landmark period was imposed. Although the index date corresponded to the qualifying 25(OH)D measurement, follow-up for outcome events commenced only 365 days later and extended up to 10 years from the index date. Patients diagnosed with aortic aneurysm or aortic dissection or who died during this landmark interval were excluded from the analytic cohort.

2.5

Propensity score matching

Cohort comparability was optimized through 1:1 propensity score matching using a greedy nearest-neighbor algorithm with a caliper width of 0.1 pooled standard deviation. The propensity model incorporated prespecified baseline covariates encompassing demographic factors, established AAA risk factors, cardiometabolic and vascular comorbidities, renal and hepatic conditions, and relevant medication. As TriNetX operates as a federated analytics platform, investigators do not have access to raw patient-level data to perform multiple imputation. Consequently, structured laboratory and measurement variables were summarized based on their availability, while diagnosis-based covariates were treated as documented EHR indicators within the 5-year baseline period.

The absence of a diagnostic code was interpreted as the absence of a documented condition, although some degree of under-ascertainment cannot be excluded.

Laboratory values (hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate) were not included in the primary matching model because these parameters have variable completeness in electronic health record databases, and the TriNetX platform does not support imputation of missing values. More importantly, laboratory availability in EHR data is unlikely to be random and may reflect healthcare utilization, comorbidity burden, disease severity, and monitoring intensity. Requiring complete laboratory data for the primary matching model could therefore restrict the analysis to a more intensively monitored subgroup and introduce selection bias.

Balance was appraised using standardized mean differences, with a threshold of <0.10 deemed satisfactory. Propensity score density distributions were visually compared before and after matching to confirm adequate overlap between the groups. The diagnostic codes for all matching variables are summarized in Supplementary Table 2.

2.6

Outcomes

The primary endpoint was incident AAA, including both ruptured and non-ruptured AAA. The secondary endpoints included ruptured AAA, thoracic aortic aneurysm, aortic dissection, non-aortic aneurysm, and all-cause mortality. Non-aortic aneurysms included cerebral, peripheral arterial, and visceral arterial aneurysms, which were examined to assess the anatomical specificity of the observed association. A substantially weaker association with non-aortic aneurysms would support the interpretation that VDD is more closely related to aortic-specific pathobiology than to a generalized predisposition to aneurysms. Secondary hyperparathyroidism, a recognized consequence of VDD, was evaluated as a positive control outcome to support the biological validity of the exposure definition, whereas appendicitis was used as a negative-control outcome.

The diagnostic codes for all outcomes are provided in Supplementary Table 3.

2.7

Sensitivity, subgroup, and multivariable analyses

Four sensitivity models were used to evaluate the consistency of the primary findings. Model I was confined to patients who survived the entire observation period to assess whether the differential mortality could account for the observed association. Model II required at least one documented healthcare encounter during follow-up, thereby restricting the analysis to individuals with confirmed ongoing clinical observation. Model III was limited to individuals who underwent abdominal computed tomography or ultrasonography during follow-up, addressing potential detection bias, whereby AAA diagnoses may be more frequent in patients receiving more intensive evaluation.

Model IV additionally incorporated laboratory parameters (hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate) into the propensity score model to assess whether the primary findings persisted after balancing these laboratory markers.

Subgroup analyses stratified by sex and age (50 to 65 vs. >65 years) were performed using interaction terms to evaluate potential effect modification. A multivariable Cox proportional hazards model was additionally fitted, simultaneously adjusting for established AAA risk factors, to assess the consistency of the association between VDD and incident AAA and to verify that the model reproduced the expected direction and magnitude of known risk factors for internal validity.

2.8

Exposure-gradient analysis

To examine a potential exposure: gradient relationship, a supplementary analysis was performed comparing patients with vitamin D insufficiency (25(OH)D 20 to 30 ng/mL) with controls with normal vitamin D status (≥30 ng/mL). The propensity score matching variables and analytical approach were identical to those used in the primary analyses.

2.9

Statistical methods

Time-to-event data were modeled using Cox proportional hazards regression, yielding hazard ratios (HRs) with 95% confidence intervals (CIs). A cause-specific hazard framework was adopted for the primary analysis, as the aim was to quantify the association between VDD and AAA occurrence rather than estimating the absolute cumulative incidence. Because the TriNetX analytics platform does not currently permit Aalen: Johansen competing-risk estimation within propensity score: matched cohorts, the Aalen: Johansen estimator was applied only as a supplemental descriptive analysis in the full unmatched cohort, with all-cause mortality treated as a competing event for incident AAA.

These estimates were obtained separately for each cohort and were not interpreted as a confounding-adjusted between-group comparison. The proportional hazards assumption was assessed using Schoenfeld residual analysis. To gauge vulnerability to unmeasured confounding, E-values were computed for the primary outcome. The E-value represents the minimum strength of association that an unmeasured confounder would need to have with both the exposure and the outcome, above and beyond measured covariates, to fully explain the observed HR. Statistical significance for the primary outcome was defined as a two-sided α of 0.05.

Formal statistical inference was restricted to the primary outcome. Analyses of secondary endpoints, control outcomes, sensitivity models, exposure-gradient analyses, and subgroup comparisons were considered exploratory and hypothesis-generating. Therefore, p-values for these non-primary analyses were reported as nominal descriptive measures only, and no multiplicity correction was applied. These p-values should not be interpreted as confirmatory evidence of independent associations. Missing data were handled based on the available information within the database, with no imputation performed.

Source: Materials and methods Source: Materials and methods Source: Materials and methods Source: Materials and methods Source: Materials and methods Source: Materials and methods Source: Materials and methods Source: Materials and methods Source: Materials and methods

Patient selection, follow-up duration, and baseline characteristics

Results: Patient selection, follow-up duration, and baseline characteristics

Before propensity score matching, 510,334 patients met the criteria for VDD, and 1,232,018 patients were classified as controls with normal vitamin D status. During the 365-day landmark period, 20,115 patients in the VDD cohort and 45,518 patients in the control cohort were excluded because they met the landmark exclusion criteria. After applying the exclusion criteria and 1:1 propensity score matching, 466,332 patients were included in each cohort. Following matching, all standardized mean differences were below 0.10, indicating a satisfactory covariate balance across demographic characteristics, comorbidities, healthcare utilization variables, and medication use (Table 1).

Propensity score density plots also demonstrated improved overlap between the cohorts after matching (Figure 1). The mean follow-up duration was 6.01 ± 3.00 years in the VDD cohort and 5.77 ± 2.91 years in the control group. The median follow-up duration was 5.98 years in the VDD cohort and 5.45 years in the control cohort, with corresponding interquartile ranges of 5.54 and 5.10 years.

TABLE 1

Baseline characteristics of patients with low vitamin D status and matched controls before and after propensity score matching.

VariablesBefore matchingAfter matching
VDD group (n = 510,334)Control group (n = 1,232,018)SMDVDD group (n = 466,332)Control group (n = 466,332)SMD
Patient characteristics
Age at index (years)59.8 ± 12.164.1 ± 11.60.36760.5 ± 12.261.0 ± 11.50.038
BMI ≥ 30 (kg/m2)189078 (37.1)343109 (27.8)0.197162372 (34.8)168763 (36.2)0.029
Female343412 (67.3)899176 (73.0)0.125318216 (68.2)316001 (67.8)0.010
White282776 (55.4)964601 (78.3)0.501280423 (60.1)275175 (59.0)0.023
Black or African American109770 (21.5)103066 (8.4)0.37580703 (17.3)80589 (17.3)0.001
Asian17496 (3.4)44991 (3.7)0.01217311 (3.7)17366 (3.7)0.001
Comorbidities and healthcare utilization
Essential (primary) hypertension221166 (43.3)520294 (42.2)0.022196059 (42.0)200208 (42.9)0.018
Dyslipidemia201600 (39.5)561700 (45.6)0.123185793 (39.8)191169 (41.0)0.023
Encounter for general adult medical examination155447 (30.5)432608 (35.1)0.099143713 (30.8)144114 (30.9)0.002
Neoplasms128474 (25.2)363885 (29.5)0.098119076 (25.5)118598 (25.4)0.002
Overweight and obesity110174 (21.6)180261 (14.6)0.18191690 (19.7)93797 (20.1)0.011
Diabetes mellitus101044 (19.8)188192 (15.3)0.11986732 (18.6)88387 (19.0)0.009
Disorders of thyroid gland76131 (14.9)253035 (20.5)0.14872271 (15.5)73143 (15.7)0.005
Nicotine dependence62334 (12.2)78009 (6.3)0.20448433 (10.4)48392 (10.4)0.000
Ischemic heart diseases51613 (10.1)122496 (9.9)0.00645979 (9.9)46245 (9.9)0.002
Diseases of liver32113 (6.3)64059 (5.2)0.04728160 (6.0)28084 (6.0)0.001
Other chronic obstructive pulmonary disease32831 (6.4)59221 (4.8)0.07127895 (6.0)27837 (6.0)0.001
Chronic kidney disease (CKD)30521 (6.0)79091 (6.4)0.01827440 (5.9)27963 (6.0)0.005
Cerebrovascular diseases29108 (5.7)69689 (5.7)0.00225676 (5.5)25997 (5.6)0.003
Iron deficiency anemia26330 (5.2)50802 (4.1)0.04922103 (4.7)22189 (4.8)0.001
Heart failure23527 (4.6)43375 (3.5)0.05519387 (4.2)19354 (4.2)0.000
Peripheral vascular diseases15420 (3.0)36457 (3.0)0.00413667 (2.9)13518 (2.9)0.002
Alcohol related disorders16621 (3.3)22445 (1.8)0.09112626 (2.7)12510 (2.7)0.002
COVID-1912934 (2.5)34610 (2.8)0.01712016 (2.6)11934 (2.6)0.001
Malnutrition4048 (0.8)8512 (0.7)0.0123508 (0.8)3524 (0.8)0.000
Medications
Glucocorticoids184146 (36.1)452424 (36.7)0.013165636 (35.5)166231 (35.6)0.003
Antilipemic agents141035 (27.6)370032 (30.0)0.053127784 (27.4)130640 (28.0)0.014
Beta blockers/related108933 (21.3)256833 (20.8)0.01296469 (20.7)97726 (21.0)0.007
ACE inhibitors97000 (19.0)190668 (15.5)0.09483025 (17.8)83403 (17.9)0.002
Calcium channel blockers88015 (17.2)183927 (14.9)0.06375139 (16.1)76008 (16.3)0.005
Angiotensin II inhibitor55008 (10.8)146624 (11.9)0.03550122 (10.7)50924 (10.9)0.006
Vitamin D supplementation47976 (9.4)192693 (15.6)0.18946351 (9.9)48821 (10.5)0.017
Insulins and analogues50551 (9.9)78051 (6.3)0.13141197 (8.8)41229 (8.8)0.000

Data are presented as number (%) or mean ± standard deviation. Standardized mean differences <0.10 were considered acceptable for covariate balance. ACE, angiotensin-converting enzyme; BMI, body mass index; CKD, chronic kidney disease; SMD, standardized mean difference; VDD, vitamin D deficiency.

FIGURE 1

Propensity score density distributions before and after matching. Propensity score density distributions are shown for patients with vitamin D deficiency and controls with normal vitamin D status before and after 1:1 propensity score matching. Improved overlap after matching indicates better covariate balance between the two cohorts. Vitamin D deficiency was defined as serum 25(OH)D < 20 ng/mL, and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL.

Figure image at the original source: open the original paper. This file is not included in this text import.
Source: Patient selection, follow-up duration, and baseline characteristics Source: Patient selection, follow-up duration, and baseline characteristics Source: Patient selection, follow-up duration, and baseline characteristics

Results: Primary, secondary, and control outcomes

During the 10-year observation window, VDD was associated with a higher risk of incident AAA than normal vitamin D status (3,086 [0.66%] vs. 2,087 [0.45%]; HR 1.40, 95% CI 1.33 to 1.48; p < 0.001; Figure 2 and Table 2). The absolute event rate difference was 0.21 percentage points. The E-value was 2.15 for the point estimate and 1.99 for the lower bound of the 95% CI.

FIGURE 2

Kaplan: Meier curves for abdominal aortic aneurysm: free probability. Kaplan: Meier curves show abdominal aortic aneurysm (AAA): free probability over the 10-year follow-up period in the propensity score: matched cohort. Follow-up began after the 1-year landmark period. Patients with vitamin D deficiency had a lower AAA: free probability than controls with normal vitamin D status. Shaded areas represent 95% confidence intervals. Vitamin D deficiency was defined as serum 25(OH)D < 20 ng/mL, and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL. AAA, abdominal aortic aneurysm; VDD, vitamin D deficiency.

Figure image at the original source: open the original paper. This file is not included in this text import.
TABLE 2

Association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

OutcomeVDD group (n = 466,332)Control group (n = 466,332)HR (95% CI)P-value
Events (%)Events (%)
Primary outcome
AAA3,086 (0.66)2,087 (0.45)1.40 (1.33 to 1.48)<0.001
Secondary outcomes
AAA rupture102 (0.02)31 (0.01)3.06 (2.05 to 4.58)<0.001
TAA2,990 (0.64)2,488 (0.53)1.13 (1.08 to 1.20)<0.001
Aortic dissection402 (0.09)284 (0.06)1.34 (1.15 to 1.56)<0.001
Non-aortic aneurysm3,825 (0.82)3,493 (0.75)1.05 (1.00 to 1.10)0.053
Mortality39,942 (8.57)25,782 (5.53)1.48 (1.46 to 1.50)<0.001
Positive control outcome
Secondary hyperparathyroidism1,147 (0.25)641 (0.14)1.72 (1.56 to 1.89)<0.001
Negative control outcome
Appendicitis2,339 (0.50)2,096 (0.45)1.06 (1.00 to 1.13)0.046

Data are presented as number of events (%). AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

Among the secondary outcomes, VDD was associated with ruptured AAA (HR 3.06, p < 0.001), aortic dissection (HR 1.34, p < 0.001), thoracic aortic aneurysm (HR 1.13, p < 0.001), and all-cause mortality (HR 1.48, p < 0.001). The association with non-aortic aneurysms was weaker and did not reach statistical significance (HR 1.05, p = 0.053). For control outcomes, VDD was associated with secondary hyperparathyroidism (HR 1.72, p < 0.001), whereas appendicitis showed only a small borderline association (HR 1.06, p = 0.046).

Source: Primary, secondary, and control outcomes Source: Primary, secondary, and control outcomes Source: Primary, secondary, and control outcomes Source: Primary, secondary, and control outcomes

Results: Sensitivity and subgroup analyses

The association between VDD and incident AAA remained consistent across all sensitivity models, with HRs ranging from 1.36 to 1.52 (Table 3). Similar patterns were observed for ruptured AAA and aortic dissection in the sensitivity analyses. In the imaging-restricted model, which included only patients who underwent at least one abdominal computed tomography or ultrasonography examination during follow-up, the association between VDD and incident AAA remained, suggesting that the primary finding was not entirely explained by differences in abdominal imaging ascertainment.

In the laboratory-adjusted matching model, VDD also remained associated with incident AAA after additional balancing of hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate. Before laboratory-adjusted matching, missingness differed across structured laboratory and measurement variables (Supplementary Table 4). The missing proportions in the VDD and control cohorts were 19.3% and 27.1% for estimated glomerular filtration rate (eGFR), 23.2% and 33.1% for hemoglobin, 34.2% and 39.6% for albumin, 39.2% and 35.4% for body mass index, and 58.2% and 66.8% for hemoglobin A1c, respectively.

These data show that laboratory and measurement completeness varied across variables, supporting the decision to evaluate these markers in a sensitivity analysis rather than in the primary matching model.

TABLE 3

Sensitivity analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm and related aortic outcomes.

OutcomesModel IModel IIModel IIIModel IV
HR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-value
AAA1.36 (1.27 to 1.44)<0.0011.42 (1.34 to 1.50)<0.0011.52 (1.42 to 1.63)<0.0011.40 (1.33 to 1.48)<0.001
AAA rupture2.24 (1.44 to 3.48)<0.0012.45 (1.70 to 3.54)<0.0012.45 (1.60 to 3.73)<0.0012.30 (1.61 to 3.29)<0.001
TAA1.09 (1.03 to 1.16)0.0021.13 (1.07 to 1.19)<0.0011.27 (1.18 to 1.36)<0.0011.13 (1.07 to 1.19)<0.001
Aortic dissection1.40 (1.17 to 1.68)<0.0011.37 (1.18 to 1.60)<0.0011.29 (1.06 to 1.56)0.0121.23 (1.06 to 1.43)0.006
Non-aortic aneurysm0.99 (0.94 to 1.04)0.7141.04 (0.99 to 1.09)0.0971.10 (1.04 to 1.17)0.0011.05 (1.01 to 1.10)0.030
MortalityNANA1.47 (1.44 to 1.49)<0.0011.42 (1.39 to 1.46)<0.0011.48 (1.46 to 1.51)<0.001

Model I was restricted to patients who survived throughout follow-up (n = 425,660 for each group). Model II was restricted to patients with at least one recorded healthcare encounter during follow-up (n = 445,290 for each group). Model III was restricted to patients with at least one recorded abdominal computed tomography or ultrasonography examination during follow-up (n = 120,142 for each group). Model IV additionally included hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate in propensity score matching (n = 461,521 for each group).

AAA, abdominal aortic aneurysm; CI, confidence interval; CT, computed tomography; eGFR, estimated glomerular filtration rate; HR, hazard ratio; NA, not available; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

In sex-stratified analyses, VDD was associated with incident AAA in both males (HR 1.45, p < 0.001) and females (HR 1.61, p < 0.001), without a significant interaction by sex (p for interaction = 0.080; Table 4). Age-stratified analyses showed similar associations in patients aged 50 to 65 years (HR 1.30, p = 0.001) and those older than 65 years (HR 1.49, p < 0.001), without significant interaction by age group (p for interaction = 0.103; Table 5).

TABLE 4

Sex-stratified subgroup analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

OutcomesMale (n = 147,947 for each group)Female (n = 317,309 for each group)P for interaction
HR (95% CI)P-valueHR (95% CI)P-value
AAA1.45 (1.34 to 1.55)<0.0011.61 (1.47 to 1.76)<0.0010.080
AAA rupture2.96 (1.79 to 4.90)<0.0011.66 (1.00 to 2.73)0.0460.152
TAA1.09 (1.01 to 1.17)0.0221.30 (1.20 to 1.41)<0.0010.002
Aortic dissection1.18 (0.95 to 1.48)0.1421.57 (1.27 to 1.94)<0.0010.074
Non-aortic aneurysm1.14 (1.06 to 1.23)0.0010.97 (0.92 to 1.03)0.3140.001
Mortality1.50 (1.47 to 1.54)<0.0011.47 (1.44 to 1.50)<0.0010.202

AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

TABLE 5

Age-stratified subgroup analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

Outcomes50 to 65 years
(n = 200,926 for each group)
>65 years
(n = 264,850 for each group)
P for
interaction
HR (95% CI)P-valueHR (95% CI)P-value
AAA1.30 (1.11 to 1.53)0.0011.49 (1.40 to 1.58)<0.0010.103
AAA ruptureNANA3.18 (2.08 to 4.84)<0.001NA
TAA1.06 (0.95 to 1.18)0.3291.21 (1.13 to 1.28)<0.0010.032
Aortic dissection1.29 (0.93 to 1.78)0.1311.39 (1.17 to 1.65)<0.0010.688
Non-aortic aneurysm1.10 (1.00 to 1.20)0.0511.04 (0.99 to 1.10)0.1130.303
Mortality1.43 (1.37 to 1.49)<0.0011.47 (1.45 to 1.50)<0.0010.228

AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; NA, not available; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

Source: Sensitivity and subgroup analyses Source: Sensitivity and subgroup analyses Source: Sensitivity and subgroup analyses Source: Sensitivity and subgroup analyses Source: Sensitivity and subgroup analyses

Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis

Results: Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis

In the multivariable Cox model fitted to the unmatched cohort, VDD remained associated with a higher risk of incident AAA after adjustment for demographic, cardiometabolic, vascular, and clinical risk factors (HR 1.52, p < 0.001; Table 6). The model reproduced several expected associations with AAA, including higher risks associated with male sex, older age, nicotine dependence, chronic obstructive pulmonary disease, ischemic heart disease, peripheral vascular disease, and hypertension. Diabetes mellitus was inversely associated with incident AAA (HR 0.78, p < 0.001), consistent with prior epidemiological observations showing a lower AAA risk among patients with diabetes.

TABLE 6

Multivariable Cox regression model for incident abdominal aortic aneurysm in the unmatched cohort.

VariablesHR (95% CI)P-value
VDD vs. control group1.52 (1.45 to 1.59)<0.001
Male3.33 (3.19 to 3.47)<0.001
Age at index1.06 (1.06 to 1.06)<0.001
Dyslipidemia1.10 (1.04 to 1.15)<0.001
Overweight and obesity1.00 (0.94 to 1.06)0.913
Diabetes mellitus0.78 (0.74 to 0.82)<0.001
Nicotine dependence3.12 (2.94 to 3.30)<0.001
Ischemic heart diseases1.54 (1.45 to 1.62)<0.001
COPD1.72 (1.61 to 1.83)<0.001
Chronic kidney disease (CKD)1.00 (0.93 to 1.07)0.916
Cerebrovascular diseases1.19 (1.11 to 1.27)<0.001
Heart failure0.97 (0.89 to 1.05)0.427
Peripheral vascular diseases1.41 (1.31 to 1.53)<0.001
Alcohol-related disorders0.82 (0.73 to 0.92)0.001
Malnutrition0.93 (0.76 to 1.13)0.456
Essential (primary) hypertension1.30 (1.23 to 1.37)<0.001
White1.35 (1.28 to 1.42)<0.001

Hazard ratios were estimated using a multivariable Cox proportional hazards model fitted in the unmatched cohort. CI, confidence interval; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; HR, hazard ratio; VDD, vitamin D deficiency.

As a supplemental descriptive analysis constrained by the TriNetX platform, Aalen: Johansen competing-risk cumulative incidence estimates were generated in the full unmatched cohort, with all-cause mortality treated as a competing event for incident AAA. This analysis showed a cumulative AAA incidence of 1.09% in the VDD cohort and 0.84% in the control cohort (Table 7). The corresponding cumulative mortality estimates were 13.70% and 11.91%, respectively. Because this analysis was performed in the unmatched cohort, it should be interpreted only as an unadjusted descriptive assessment of absolute cumulative incidence and not as a substitute for the primary propensity score: matched cause-specific Cox analysis.

TABLE 7

Supplemental descriptive competing-risk cumulative incidence of abdominal aortic aneurysm and all-cause mortality in the full unmatched cohort.

CohortOutcome% of cohortAalen: Johansen cumulative incidence at end of time window
VDDAAA0.64%1.09%
VDDMortality8.20%13.70%
Normal vitamin D statusAAA0.46%0.84%
Normal vitamin D statusMortality6.45%11.91%

Cumulative incidence was estimated using the Aalen: Johansen estimator, with all-cause mortality treated as a competing event for incident abdominal aortic aneurysm. AAA, abdominal aortic aneurysm; VDD, vitamin D deficiency.

In the exposure-gradient analysis, vitamin D insufficiency was associated with a higher risk of incident AAA than normal vitamin D status (HR 1.34, p < 0.001; Table 8), with a point estimate numerically lower than that observed for VDD in the primary analysis. Similar exposure-gradient patterns were observed for ruptured AAA, aortic dissection, and secondary hyperparathyroidism, whereas appendicitis remained near null.

TABLE 8

Exposure-gradient analysis of vitamin D insufficiency and 10-year risks of abdominal aortic aneurysm, severe aortic events, control outcomes, and mortality.

OutcomeVDI group (n = 615,798)Control group (n = 615,798)HR (95% CI)P-value
Events (%)Events (%)
AAA3,851 (0.63)2,788 (0.45)1.34 (1.28 to 1.41)<0.001
AAA rupture94 (0.02)58 (0.01)1.56 (1.12 to 2.16)0.007
TAA4,379 (0.71)3,480 (0.57)1.22 (1.17 to 1.28)<0.001
Aortic dissection469 (0.08)377 (0.06)1.20 (1.05 to 1.38)0.007
Non-aortic aneurysm4,938 (0.80)4,643 (0.75)1.04 (1.00 to 1.08)0.072
Mortality43,698 (7.10)33,249 (5.40)1.28 (1.26 to 1.30)<0.001
Hyperparathyroidism976 (0.16)879 (0.14)1.09 (0.99 to 1.19)0.076
Appendicitis3,052 (0.50)2,915 (0.47)1.02 (0.97 to 1.07)0.460

Data are presented as number of events (%). Vitamin D insufficiency was defined as serum 25(OH)D 20 to 30 ng/mL (n = 615,798), and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL (n = 615,798). AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDI, vitamin D insufficiency.

Source: Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis Source: Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis Source: Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis Source: Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis Source: Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis Source: Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis

Discussion

In this large multicenter propensity score: matched cohort study, clinically defined VDD was associated with a higher long-term risk of incident AAA than normal vitamin D status. This association was consistent across sensitivity analyses addressing survival, follow-up ascertainment, imaging detection, and laboratory balance and was further supported by multivariable adjustment in the unmatched cohort. Exploratory analyses also suggested associations with ruptured AAA and aortic dissection, whereas the association with non-aortic aneurysms was comparatively weaker, supporting possible aortic specificity. However, given the modest absolute risk difference and observational design, these findings should be interpreted as hypothesis-generating evidence of association rather than a causal relationship.

The observed association between VDD and incident AAA extends prior epidemiological evidence, which has been limited in scale and inconsistent in direction. The ARIC prospective cohort followed 12,770 participants for a median of 19.7 years and found no significant association between VDD and incident AAA (24), whereas an Australian community-based study of 4,233 older men (29) reported inverse associations between 25(OH)D concentrations and AAA diagnosis or aneurysm size. Recent cardiovascular literature also supports a broader link between vitamin D status and vascular inflammation, endothelial function, oxidative stress, hypertension, and atherosclerosis, although such evidence should be viewed as mechanistic background rather than direct proof of an AAA-specific effect (30).

Compared with these earlier studies, the present analysis used a clinically defined VDD threshold, included more than 460,000 matched patients per cohort, and applied a 1-year landmark design to reduce reverse causation bias. In addition, the propensity score model balanced a broad range of clinically relevant AAA risk factors, including smoking-related diagnoses, chronic obstructive pulmonary disease, peripheral vascular disease, chronic kidney disease, cardiovascular comorbidities, healthcare utilization, and the use of medications. These design features may partly explain why our study was able to detect a clinically defined VDD: AAA association that was not consistently observed in smaller cohorts using conventional multivariable adjustments.

Among the exploratory secondary outcomes, VDD was associated with ruptured AAA and aortic dissection. Because these analyses were not adjusted for multiplicity, they should be interpreted strictly as hypothesis-generating and not as confirmatory evidence of independent associations. The higher relative estimate for ruptured AAA is biologically plausible and compatible with experimental evidence (22), linking VDD to aneurysm enlargement and rupture-prone phenotypes. The lower HRs for ruptured AAA in the sensitivity analyses likely reflected rare-event instability and attenuation after restricting the analytic population; therefore, the rupture finding should be interpreted as a hypothesis-generating signal rather than a precise estimate of absolute rupture risk.

VDD was also associated with a high risk of thoracic aortic aneurysm, although the magnitude of the association (HR 1.13) was weaker than that observed for AAA. This attenuation is biologically plausible because the pathogenesis of thoracic aortic aneurysm involves distinct mechanisms, including genetic predisposition and cystic medial degeneration (31, 32), which may be less directly influenced by vitamin D: mediated inflammatory and matrix-remodeling pathways.

A notable feature of this study was the inclusion of non-aortic aneurysms as anatomic specificity comparators. In contrast to the stronger association observed for AAA, VDD was not significantly associated with non-aortic aneurysms. This pattern suggests that VDD may be more closely related to aortic-specific degenerative processes than to a generalized aneurysmal predisposition, although this interpretation remains to be confirmed. This interpretation is biologically plausible because AAA pathogenesis is characterized by localized aortic wall inflammation, vascular smooth muscle cell dysfunction, and extracellular matrix degradation.

The positive control outcome of secondary hyperparathyroidism supported the biological validity of the definition of VDD exposure. Conversely, the negative-control outcome of appendicitis showed a near-null effect estimate. Because control outcomes were exploratory and no multiplicity correction was applied, the nominal p-value for appendicitis was not interpreted as evidence of a meaningful association. Instead, the small effect size suggested that nonspecific residual bias, if present, was limited in magnitude and unlikely to explain the substantially larger primary AAA association (33).

The sensitivity analyses provided methodological support for the primary findings. The association between VDD and incident AAA remained present in models restricted to patients who survived throughout follow-up and those with documented healthcare encounters during follow-up, partially mitigating the concern that the finding was driven solely by differential mortality or incomplete observation. Because AAA is frequently asymptomatic and often detected incidentally during abdominal imaging performed for unrelated indications (34, 35), the imaging-restricted model was particularly relevant for addressing potential ascertainment biases.

The association persisted among patients who underwent abdominal computed tomography or ultrasonography during follow-up. The modestly higher HR in this model may reflect enrichment of a clinically higher-risk subgroup and residual differences in imaging indication, timing, or frequency. Because requiring at least one imaging examination did not fully equalize surveillance intensity, this analysis cannot fully eliminate selection or ascertainment bias.

Subgroup analyses suggested that the association between VDD and incident AAA was present in both males and females and in both age strata, without statistically significant interactions with the primary outcome. These findings suggest that the observed association was not confined to any single demographic subgroup. However, the subgroup analyses were exploratory, and the study was not designed to redefine sex: or age-specific screening strategies. The findings in patients aged 50 to 65 years are clinically interesting because conventional AAA screening strategies primarily focus on older adults, particularly men aged 65 years or older (1).

Nevertheless, the absolute risk of AAA in younger individuals is lower, and these findings should not be interpreted as evidence to expand AAA screening eligibility criteria. They suggest that VDD may identify a vascular risk phenotype that warrants further investigation.

The multivariable Cox model fitted to the unmatched cohort provided an additional internal validity check. VDD remained associated with incident AAA after adjusting for demographic, cardiometabolic, vascular, and clinical risk factors. The model also reproduced several expected AAA associations, including higher risks associated with male sex, older age, nicotine dependence, chronic obstructive pulmonary disease, ischemic heart disease, peripheral vascular disease, and hypertension (1, 10, 36). Diabetes mellitus was inversely associated with incident AAA, consistent with prior epidemiological studies and meta-analyses reporting a lower prevalence, incidence, or progression of AAA among patients with diabetes (37Not reported39).

Proposed mechanisms include diabetes-related extracellular matrix glycation and collagen cross-linking, reduced matrix metalloproteinase activity, altered inflammatory remodeling of the aortic wall, and potential effects of antidiabetic therapies, all of which may limit aneurysmal dilation despite the broader vascular risk associated with diabetes (37Not reported39). This internal consistency supports the validity of our analytical framework. However, because the multivariable model was performed in the unmatched cohort, it should be considered complementary to, rather than a replacement for, the primary propensity score: matched analysis.

The competing risk analysis further showed that the direction of the association persisted when death was considered a competing event; however, this analysis was descriptive and based on the full unmatched cohort. Similarly, the exposure-gradient analysis suggested a graded pattern across vitamin D insufficiency and deficiency, but it should be considered exploratory because the exposure categories were evaluated in separate matched comparisons rather than in a single continuous dose: response model.

Clinical implications should be interpreted cautiously. The present findings do not support VDD as a stand-alone indication for AAA screening, nor do they justify expanding current guideline-based AAA screening criteria. Instead, VDD may be considered a supplementary vascular risk marker in patients who already have established AAA risk factors. In general clinical practice, the presence of VDD may prompt clinicians to review traditional AAA risk factors, prior abdominal imaging, and eligibility for guideline-based screening. From a perioperative perspective, particularly before major noncardiac surgery, VDD in patients with conventional vascular risk factors may highlight the need for careful cardiovascular and vascular risk review and, when aneurysmal disease is known or suspected, multidisciplinary evaluation with vascular specialists.

However, vitamin D correction should not be assumed to reduce AAA risk or severe aortic events based on the present observational data.

Several limitations should be acknowledged. First, although both cohorts included only patients with documented 25(OH)D testing, the clinical indication for testing was unavailable and may have differed between groups. Therefore, indication-specific residual confounding cannot be excluded. Furthermore, important factors such as sunlight exposure, dietary intake, physical activity, frailty, socioeconomic status, over-the-counter supplement use, and detailed smoking history were not fully considered. Although race was adequately balanced after matching, the substantial pre-matching racial differences may reflect unmeasured social, geographic, or healthcare-access factors, and residual confounding related to these factors cannot be excluded.

Second, although the 3-year look-back was intended to identify a more persistent VDD phenotype, exposure classification remained based on a single qualifying 25(OH)D measurement. Harmonized data on testing season, assay methods, laboratory reference ranges, testing indications, and post-index vitamin D levels or supplementation were unavailable. Therefore, residual exposure misclassification cannot be excluded, and differential vitamin D repletion may have attenuated the observed association over time. Third, incident AAA, ruptured AAA, and aortic dissection were identified using electronic health record diagnosis codes without chart-level adjudication or algorithm validation against imaging-confirmed disease; therefore, outcome misclassification cannot be excluded and could have biased the effect estimates in either direction.

Although the imaging-restricted sensitivity analysis partially reduced concerns about differential ascertainment, selection by indication could not be excluded. Fourth, data on aneurysm diameter, growth rate, morphology, rupture presentation, operative repair, imaging findings, and cause-specific mortality were unavailable. In particular, the TriNetX platform does not provide structured diagnostic imaging details, such as baseline aneurysm diameter or serial aortic measurements. Therefore, we could not verify whether small, subclinical aneurysms were already present but unrecorded at the index date. Therefore, we could not evaluate aneurysm severity, longitudinal expansion, treatment pathways, or whether the deaths were directly attributable to aneurysm-related complications.

Fifth, secondary outcomes, subgroup analyses, control outcomes, sensitivity analyses, and exposure gradient analyses were exploratory and not adjusted for multiplicity. Therefore, these findings should be interpreted as supportive or hypothesis-generating, rather than as confirmatory evidence. Finally, the requirement for at least two healthcare encounters during the preceding 5 years may have selected more healthcare-engaged and comorbid patients while excluding healthier infrequent users. Accordingly, the findings may not be fully generalizable to community-dwelling populations or individuals with limited healthcare access.

Source: Discussion Source: Discussion Source: Discussion Source: Discussion Source: Discussion Source: Discussion Source: Discussion Source: Discussion Source: Discussion

Results

3.1

Patient selection, follow-up duration, and baseline characteristics

Before propensity score matching, 510,334 patients met the criteria for VDD, and 1,232,018 patients were classified as controls with normal vitamin D status. During the 365-day landmark period, 20,115 patients in the VDD cohort and 45,518 patients in the control cohort were excluded because they met the landmark exclusion criteria. After applying the exclusion criteria and 1:1 propensity score matching, 466,332 patients were included in each cohort. Following matching, all standardized mean differences were below 0.10, indicating a satisfactory covariate balance across demographic characteristics, comorbidities, healthcare utilization variables, and medication use (Table 1).

Propensity score density plots also demonstrated improved overlap between the cohorts after matching (Figure 1). The mean follow-up duration was 6.01 ± 3.00 years in the VDD cohort and 5.77 ± 2.91 years in the control group. The median follow-up duration was 5.98 years in the VDD cohort and 5.45 years in the control cohort, with corresponding interquartile ranges of 5.54 and 5.10 years.

TABLE 1

Baseline characteristics of patients with low vitamin D status and matched controls before and after propensity score matching.

VariablesBefore matchingAfter matching
VDD group (n = 510,334)Control group (n = 1,232,018)SMDVDD group (n = 466,332)Control group (n = 466,332)SMD
Patient characteristics
Age at index (years)59.8 ± 12.164.1 ± 11.60.36760.5 ± 12.261.0 ± 11.50.038
BMI ≥ 30 (kg/m2)189078 (37.1)343109 (27.8)0.197162372 (34.8)168763 (36.2)0.029
Female343412 (67.3)899176 (73.0)0.125318216 (68.2)316001 (67.8)0.010
White282776 (55.4)964601 (78.3)0.501280423 (60.1)275175 (59.0)0.023
Black or African American109770 (21.5)103066 (8.4)0.37580703 (17.3)80589 (17.3)0.001
Asian17496 (3.4)44991 (3.7)0.01217311 (3.7)17366 (3.7)0.001
Comorbidities and healthcare utilization
Essential (primary) hypertension221166 (43.3)520294 (42.2)0.022196059 (42.0)200208 (42.9)0.018
Dyslipidemia201600 (39.5)561700 (45.6)0.123185793 (39.8)191169 (41.0)0.023
Encounter for general adult medical examination155447 (30.5)432608 (35.1)0.099143713 (30.8)144114 (30.9)0.002
Neoplasms128474 (25.2)363885 (29.5)0.098119076 (25.5)118598 (25.4)0.002
Overweight and obesity110174 (21.6)180261 (14.6)0.18191690 (19.7)93797 (20.1)0.011
Diabetes mellitus101044 (19.8)188192 (15.3)0.11986732 (18.6)88387 (19.0)0.009
Disorders of thyroid gland76131 (14.9)253035 (20.5)0.14872271 (15.5)73143 (15.7)0.005
Nicotine dependence62334 (12.2)78009 (6.3)0.20448433 (10.4)48392 (10.4)0.000
Ischemic heart diseases51613 (10.1)122496 (9.9)0.00645979 (9.9)46245 (9.9)0.002
Diseases of liver32113 (6.3)64059 (5.2)0.04728160 (6.0)28084 (6.0)0.001
Other chronic obstructive pulmonary disease32831 (6.4)59221 (4.8)0.07127895 (6.0)27837 (6.0)0.001
Chronic kidney disease (CKD)30521 (6.0)79091 (6.4)0.01827440 (5.9)27963 (6.0)0.005
Cerebrovascular diseases29108 (5.7)69689 (5.7)0.00225676 (5.5)25997 (5.6)0.003
Iron deficiency anemia26330 (5.2)50802 (4.1)0.04922103 (4.7)22189 (4.8)0.001
Heart failure23527 (4.6)43375 (3.5)0.05519387 (4.2)19354 (4.2)0.000
Peripheral vascular diseases15420 (3.0)36457 (3.0)0.00413667 (2.9)13518 (2.9)0.002
Alcohol related disorders16621 (3.3)22445 (1.8)0.09112626 (2.7)12510 (2.7)0.002
COVID-1912934 (2.5)34610 (2.8)0.01712016 (2.6)11934 (2.6)0.001
Malnutrition4048 (0.8)8512 (0.7)0.0123508 (0.8)3524 (0.8)0.000
Medications
Glucocorticoids184146 (36.1)452424 (36.7)0.013165636 (35.5)166231 (35.6)0.003
Antilipemic agents141035 (27.6)370032 (30.0)0.053127784 (27.4)130640 (28.0)0.014
Beta blockers/related108933 (21.3)256833 (20.8)0.01296469 (20.7)97726 (21.0)0.007
ACE inhibitors97000 (19.0)190668 (15.5)0.09483025 (17.8)83403 (17.9)0.002
Calcium channel blockers88015 (17.2)183927 (14.9)0.06375139 (16.1)76008 (16.3)0.005
Angiotensin II inhibitor55008 (10.8)146624 (11.9)0.03550122 (10.7)50924 (10.9)0.006
Vitamin D supplementation47976 (9.4)192693 (15.6)0.18946351 (9.9)48821 (10.5)0.017
Insulins and analogues50551 (9.9)78051 (6.3)0.13141197 (8.8)41229 (8.8)0.000

Data are presented as number (%) or mean ± standard deviation. Standardized mean differences <0.10 were considered acceptable for covariate balance. ACE, angiotensin-converting enzyme; BMI, body mass index; CKD, chronic kidney disease; SMD, standardized mean difference; VDD, vitamin D deficiency.

FIGURE 1

Propensity score density distributions before and after matching. Propensity score density distributions are shown for patients with vitamin D deficiency and controls with normal vitamin D status before and after 1:1 propensity score matching. Improved overlap after matching indicates better covariate balance between the two cohorts. Vitamin D deficiency was defined as serum 25(OH)D < 20 ng/mL, and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL.

Figure image at the original source: open the original paper. This file is not included in this text import.
3.2

Primary, secondary, and control outcomes

During the 10-year observation window, VDD was associated with a higher risk of incident AAA than normal vitamin D status (3,086 [0.66%] vs. 2,087 [0.45%]; HR 1.40, 95% CI 1.33 to 1.48; p < 0.001; Figure 2 and Table 2). The absolute event rate difference was 0.21 percentage points. The E-value was 2.15 for the point estimate and 1.99 for the lower bound of the 95% CI.

FIGURE 2

Kaplan: Meier curves for abdominal aortic aneurysm: free probability. Kaplan: Meier curves show abdominal aortic aneurysm (AAA): free probability over the 10-year follow-up period in the propensity score: matched cohort. Follow-up began after the 1-year landmark period. Patients with vitamin D deficiency had a lower AAA: free probability than controls with normal vitamin D status. Shaded areas represent 95% confidence intervals. Vitamin D deficiency was defined as serum 25(OH)D < 20 ng/mL, and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL. AAA, abdominal aortic aneurysm; VDD, vitamin D deficiency.

Figure image at the original source: open the original paper. This file is not included in this text import.
TABLE 2

Association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

OutcomeVDD group (n = 466,332)Control group (n = 466,332)HR (95% CI)P-value
Events (%)Events (%)
Primary outcome
AAA3,086 (0.66)2,087 (0.45)1.40 (1.33 to 1.48)<0.001
Secondary outcomes
AAA rupture102 (0.02)31 (0.01)3.06 (2.05 to 4.58)<0.001
TAA2,990 (0.64)2,488 (0.53)1.13 (1.08 to 1.20)<0.001
Aortic dissection402 (0.09)284 (0.06)1.34 (1.15 to 1.56)<0.001
Non-aortic aneurysm3,825 (0.82)3,493 (0.75)1.05 (1.00 to 1.10)0.053
Mortality39,942 (8.57)25,782 (5.53)1.48 (1.46 to 1.50)<0.001
Positive control outcome
Secondary hyperparathyroidism1,147 (0.25)641 (0.14)1.72 (1.56 to 1.89)<0.001
Negative control outcome
Appendicitis2,339 (0.50)2,096 (0.45)1.06 (1.00 to 1.13)0.046

Data are presented as number of events (%). AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

Among the secondary outcomes, VDD was associated with ruptured AAA (HR 3.06, p < 0.001), aortic dissection (HR 1.34, p < 0.001), thoracic aortic aneurysm (HR 1.13, p < 0.001), and all-cause mortality (HR 1.48, p < 0.001). The association with non-aortic aneurysms was weaker and did not reach statistical significance (HR 1.05, p = 0.053). For control outcomes, VDD was associated with secondary hyperparathyroidism (HR 1.72, p < 0.001), whereas appendicitis showed only a small borderline association (HR 1.06, p = 0.046).

3.3

Sensitivity and subgroup analyses

The association between VDD and incident AAA remained consistent across all sensitivity models, with HRs ranging from 1.36 to 1.52 (Table 3). Similar patterns were observed for ruptured AAA and aortic dissection in the sensitivity analyses. In the imaging-restricted model, which included only patients who underwent at least one abdominal computed tomography or ultrasonography examination during follow-up, the association between VDD and incident AAA remained, suggesting that the primary finding was not entirely explained by differences in abdominal imaging ascertainment.

In the laboratory-adjusted matching model, VDD also remained associated with incident AAA after additional balancing of hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate. Before laboratory-adjusted matching, missingness differed across structured laboratory and measurement variables (Supplementary Table 4). The missing proportions in the VDD and control cohorts were 19.3% and 27.1% for estimated glomerular filtration rate (eGFR), 23.2% and 33.1% for hemoglobin, 34.2% and 39.6% for albumin, 39.2% and 35.4% for body mass index, and 58.2% and 66.8% for hemoglobin A1c, respectively.

These data show that laboratory and measurement completeness varied across variables, supporting the decision to evaluate these markers in a sensitivity analysis rather than in the primary matching model.

TABLE 3

Sensitivity analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm and related aortic outcomes.

OutcomesModel IModel IIModel IIIModel IV
HR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-value
AAA1.36 (1.27 to 1.44)<0.0011.42 (1.34 to 1.50)<0.0011.52 (1.42 to 1.63)<0.0011.40 (1.33 to 1.48)<0.001
AAA rupture2.24 (1.44 to 3.48)<0.0012.45 (1.70 to 3.54)<0.0012.45 (1.60 to 3.73)<0.0012.30 (1.61 to 3.29)<0.001
TAA1.09 (1.03 to 1.16)0.0021.13 (1.07 to 1.19)<0.0011.27 (1.18 to 1.36)<0.0011.13 (1.07 to 1.19)<0.001
Aortic dissection1.40 (1.17 to 1.68)<0.0011.37 (1.18 to 1.60)<0.0011.29 (1.06 to 1.56)0.0121.23 (1.06 to 1.43)0.006
Non-aortic aneurysm0.99 (0.94 to 1.04)0.7141.04 (0.99 to 1.09)0.0971.10 (1.04 to 1.17)0.0011.05 (1.01 to 1.10)0.030
MortalityNANA1.47 (1.44 to 1.49)<0.0011.42 (1.39 to 1.46)<0.0011.48 (1.46 to 1.51)<0.001

Model I was restricted to patients who survived throughout follow-up (n = 425,660 for each group). Model II was restricted to patients with at least one recorded healthcare encounter during follow-up (n = 445,290 for each group). Model III was restricted to patients with at least one recorded abdominal computed tomography or ultrasonography examination during follow-up (n = 120,142 for each group). Model IV additionally included hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate in propensity score matching (n = 461,521 for each group).

AAA, abdominal aortic aneurysm; CI, confidence interval; CT, computed tomography; eGFR, estimated glomerular filtration rate; HR, hazard ratio; NA, not available; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

In sex-stratified analyses, VDD was associated with incident AAA in both males (HR 1.45, p < 0.001) and females (HR 1.61, p < 0.001), without a significant interaction by sex (p for interaction = 0.080; Table 4). Age-stratified analyses showed similar associations in patients aged 50 to 65 years (HR 1.30, p = 0.001) and those older than 65 years (HR 1.49, p < 0.001), without significant interaction by age group (p for interaction = 0.103; Table 5).

TABLE 4

Sex-stratified subgroup analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

OutcomesMale (n = 147,947 for each group)Female (n = 317,309 for each group)P for interaction
HR (95% CI)P-valueHR (95% CI)P-value
AAA1.45 (1.34 to 1.55)<0.0011.61 (1.47 to 1.76)<0.0010.080
AAA rupture2.96 (1.79 to 4.90)<0.0011.66 (1.00 to 2.73)0.0460.152
TAA1.09 (1.01 to 1.17)0.0221.30 (1.20 to 1.41)<0.0010.002
Aortic dissection1.18 (0.95 to 1.48)0.1421.57 (1.27 to 1.94)<0.0010.074
Non-aortic aneurysm1.14 (1.06 to 1.23)0.0010.97 (0.92 to 1.03)0.3140.001
Mortality1.50 (1.47 to 1.54)<0.0011.47 (1.44 to 1.50)<0.0010.202

AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

TABLE 5

Age-stratified subgroup analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

Outcomes50 to 65 years
(n = 200,926 for each group)
>65 years
(n = 264,850 for each group)
P for
interaction
HR (95% CI)P-valueHR (95% CI)P-value
AAA1.30 (1.11 to 1.53)0.0011.49 (1.40 to 1.58)<0.0010.103
AAA ruptureNANA3.18 (2.08 to 4.84)<0.001NA
TAA1.06 (0.95 to 1.18)0.3291.21 (1.13 to 1.28)<0.0010.032
Aortic dissection1.29 (0.93 to 1.78)0.1311.39 (1.17 to 1.65)<0.0010.688
Non-aortic aneurysm1.10 (1.00 to 1.20)0.0511.04 (0.99 to 1.10)0.1130.303
Mortality1.43 (1.37 to 1.49)<0.0011.47 (1.45 to 1.50)<0.0010.228

AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; NA, not available; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

3.4

Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis

In the multivariable Cox model fitted to the unmatched cohort, VDD remained associated with a higher risk of incident AAA after adjustment for demographic, cardiometabolic, vascular, and clinical risk factors (HR 1.52, p < 0.001; Table 6). The model reproduced several expected associations with AAA, including higher risks associated with male sex, older age, nicotine dependence, chronic obstructive pulmonary disease, ischemic heart disease, peripheral vascular disease, and hypertension. Diabetes mellitus was inversely associated with incident AAA (HR 0.78, p < 0.001), consistent with prior epidemiological observations showing a lower AAA risk among patients with diabetes.

TABLE 6

Multivariable Cox regression model for incident abdominal aortic aneurysm in the unmatched cohort.

VariablesHR (95% CI)P-value
VDD vs. control group1.52 (1.45 to 1.59)<0.001
Male3.33 (3.19 to 3.47)<0.001
Age at index1.06 (1.06 to 1.06)<0.001
Dyslipidemia1.10 (1.04 to 1.15)<0.001
Overweight and obesity1.00 (0.94 to 1.06)0.913
Diabetes mellitus0.78 (0.74 to 0.82)<0.001
Nicotine dependence3.12 (2.94 to 3.30)<0.001
Ischemic heart diseases1.54 (1.45 to 1.62)<0.001
COPD1.72 (1.61 to 1.83)<0.001
Chronic kidney disease (CKD)1.00 (0.93 to 1.07)0.916
Cerebrovascular diseases1.19 (1.11 to 1.27)<0.001
Heart failure0.97 (0.89 to 1.05)0.427
Peripheral vascular diseases1.41 (1.31 to 1.53)<0.001
Alcohol-related disorders0.82 (0.73 to 0.92)0.001
Malnutrition0.93 (0.76 to 1.13)0.456
Essential (primary) hypertension1.30 (1.23 to 1.37)<0.001
White1.35 (1.28 to 1.42)<0.001

Hazard ratios were estimated using a multivariable Cox proportional hazards model fitted in the unmatched cohort. CI, confidence interval; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; HR, hazard ratio; VDD, vitamin D deficiency.

As a supplemental descriptive analysis constrained by the TriNetX platform, Aalen: Johansen competing-risk cumulative incidence estimates were generated in the full unmatched cohort, with all-cause mortality treated as a competing event for incident AAA. This analysis showed a cumulative AAA incidence of 1.09% in the VDD cohort and 0.84% in the control cohort (Table 7). The corresponding cumulative mortality estimates were 13.70% and 11.91%, respectively. Because this analysis was performed in the unmatched cohort, it should be interpreted only as an unadjusted descriptive assessment of absolute cumulative incidence and not as a substitute for the primary propensity score: matched cause-specific Cox analysis.

TABLE 7

Supplemental descriptive competing-risk cumulative incidence of abdominal aortic aneurysm and all-cause mortality in the full unmatched cohort.

CohortOutcome% of cohortAalen: Johansen cumulative incidence at end of time window
VDDAAA0.64%1.09%
VDDMortality8.20%13.70%
Normal vitamin D statusAAA0.46%0.84%
Normal vitamin D statusMortality6.45%11.91%

Cumulative incidence was estimated using the Aalen: Johansen estimator, with all-cause mortality treated as a competing event for incident abdominal aortic aneurysm. AAA, abdominal aortic aneurysm; VDD, vitamin D deficiency.

In the exposure-gradient analysis, vitamin D insufficiency was associated with a higher risk of incident AAA than normal vitamin D status (HR 1.34, p < 0.001; Table 8), with a point estimate numerically lower than that observed for VDD in the primary analysis. Similar exposure-gradient patterns were observed for ruptured AAA, aortic dissection, and secondary hyperparathyroidism, whereas appendicitis remained near null.

TABLE 8

Exposure-gradient analysis of vitamin D insufficiency and 10-year risks of abdominal aortic aneurysm, severe aortic events, control outcomes, and mortality.

OutcomeVDI group (n = 615,798)Control group (n = 615,798)HR (95% CI)P-value
Events (%)Events (%)
AAA3,851 (0.63)2,788 (0.45)1.34 (1.28 to 1.41)<0.001
AAA rupture94 (0.02)58 (0.01)1.56 (1.12 to 2.16)0.007
TAA4,379 (0.71)3,480 (0.57)1.22 (1.17 to 1.28)<0.001
Aortic dissection469 (0.08)377 (0.06)1.20 (1.05 to 1.38)0.007
Non-aortic aneurysm4,938 (0.80)4,643 (0.75)1.04 (1.00 to 1.08)0.072
Mortality43,698 (7.10)33,249 (5.40)1.28 (1.26 to 1.30)<0.001
Hyperparathyroidism976 (0.16)879 (0.14)1.09 (0.99 to 1.19)0.076
Appendicitis3,052 (0.50)2,915 (0.47)1.02 (0.97 to 1.07)0.460

Data are presented as number of events (%). Vitamin D insufficiency was defined as serum 25(OH)D 20 to 30 ng/mL (n = 615,798), and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL (n = 615,798). AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDI, vitamin D insufficiency.

Source: Results Source: Results Source: Results Source: Results

Conclusion

Clinically defined VDD was associated with a higher long-term risk of incident AAA in this large propensity score: matched cohort study. The association was supported by sensitivity models and multivariable adjustment, whereas secondary outcomes, subgroup analyses, and exposure gradient analyses should be interpreted as exploratory and hypothesis-generating. These findings should not be interpreted as supporting VDD-based AAA screening, but they suggest that VDD may serve as an adjunctive risk marker when evaluating patients with established vascular risk factors in general or perioperative clinical settings.

Given the modest absolute risk difference and the observational design, further prospective studies with systematic aortic imaging, repeated vitamin D measurements, detailed lifestyle data, and mechanistic endpoints are needed to determine whether VDD is causally involved in AAA development or severe aortic complications.

Source: Conclusion
Authors’ abstract

Background

Vitamin D deficiency (VDD) is prevalent among older adults and may influence vascular inflammation, extracellular matrix remodeling, and aortic wall integrity; however, its association with incident abdominal aortic aneurysm (AAA) remains uncertain.

Methods

This multicenter retrospective propensity score: matched cohort study used the TriNetX Global Collaborative Network to evaluate the association between vitamin D deficiency (VDD; serum 25-hydroxyvitamin D < 20 ng/mL) and the long-term risk of incident AAA in adults aged ≥50 years, compared with normal vitamin D status (≥30 ng/mL). A 1-year landmark design was applied, with follow-up extending up to 10 years after the index date. The primary outcome was incident AAA. The exploratory secondary outcomes included ruptured AAA, aortic dissection, thoracic aortic aneurysm, non-aortic aneurysm, and all-cause mortality.

Results

After propensity score matching, 466,332 patients were included in each cohort. VDD was associated with a higher risk of incident AAA than normal vitamin D status (0.66% vs. 0.45%; hazard ratio [HR] 1.40, 95% confidence interval [CI], 1.33 to 1.48; p < 0.001). Among the exploratory secondary outcomes, VDD showed higher nominal hazards of ruptured AAA (HR 3.06, 95% CI 2.05 to 4.58; p < 0.001), aortic dissection (HR 1.34, 95% CI 1.15 to 1.56; p < 0.001), thoracic aortic aneurysm (HR 1.13, 95% CI 1.08 to 1.20; p < 0.001), and all-cause mortality (HR 1.48, 95% CI 1.46 to 1.50; p < 0.001), whereas non-aortic aneurysms showed a weaker near-null association (HR 1.05, 95% CI 1.00 to 1.10; p = 0.053).

This association persisted across sensitivity analyses and multivariable adjustments.

Conclusion

Vitamin D deficiency was associated with an increased long-term risk of incident AAA, whereas the association with ruptured AAA was exploratory and should be interpreted cautiously because of the small number of events. Given the observational design and modest absolute risk difference, prospective studies are needed to determine whether this association is causal or clinically actionable.

Read the full paper
1

Introduction

Abdominal aortic aneurysm (AAA) is a clinically silent but potentially fatal vascular disease that may progress to rupture, emergency surgery, and death (1Not reported4). Its pathogenesis is characterized by chronic vascular inflammation, extracellular matrix degradation, vascular smooth muscle cell dysfunction, elastin fragmentation, and progressive loss of aortic wall integrity (5Not reported9). Although advanced age, male sex, smoking, hypertension, and atherosclerotic disease are well-established risk factors (10Not reported12), additional potentially modifiable risk markers related to aortic wall degeneration remain undefined.

Vitamin D deficiency (VDD) is common in older adults and patients with cardiometabolic comorbidities (13, 14). Beyond its classical role in calcium and bone metabolism, vitamin D signaling may influence vascular inflammation, oxidative stress, endothelial function, vascular smooth muscle cell phenotype, and extracellular matrix remodeling (15Not reported18). These pathways are directly relevant to aneurysm biology because AAA formation involves inflammatory cell infiltration, matrix metalloproteinase activation, medial degeneration, and weakening of the aortic wall (19Not reported21).

Experimental evidence (22) further suggests that VDD may promote larger and more rupture-prone AAA phenotypes, potentially through enhanced extracellular matrix remodeling and upregulation of osteopontin and matrix metalloproteinases, thereby providing a biologically plausible link between low vitamin D status and impaired aortic wall stability in patients with VDD.

However, human epidemiological evidence remains limited and inconsistent. A recent UK Biobank cohort study reported an inverse association between serum 25-hydroxyvitamin D concentrations and incident AAA (23), whereas the ARIC prospective cohort found little evidence that vitamin D-related biomarkers were associated with incident AAA (24). Moreover, previous studies have focused mainly on AAA occurrence (23, 24), leaving uncertainty regarding whether clinically defined VDD is associated with severe aortic outcomes, including AAA rupture and aortic dissection.

Therefore, we conducted a multicenter propensity score: matched cohort study using a large real-world electronic health record network to evaluate the association between VDD and the long-term risk of incident AAA. We also examined severe aortic events, including ruptured AAA and aortic dissection, as secondary outcomes. We hypothesized that patients with VDD would have a higher long-term risk of incident AAA and severe aortic events than those with normal vitamin D status.

2

Materials and methods

2.1

Study design and data source

This retrospective propensity score: matched cohort study was conducted using the TriNetX Global Collaborative Network, a federated research platform that aggregates longitudinal, de-identified electronic health records from multiple participating healthcare organizations. The TriNetX network has been widely used in peer-reviewed epidemiological and clinical outcomes research, supporting its utility for large-scale real-world evidence studies (25Not reported28). Because the dataset contained no directly identifiable patient information, the need for individual informed consent was waived. The study protocol was approved by the Institutional Review Board of Chi Mei Medical Center (Tainan, Taiwan) and was conducted in accordance with the Declaration of Helsinki and relevant institutional guidelines.

2.2

Study population and exposure classification

The study enrolled adults aged ≥50 years with at least one documented serum 25-hydroxyvitamin D [25(OH)D] measurement recorded between January 1, 2010, and December 31, 2023. To confirm ongoing engagement with the healthcare system and allow reliable ascertainment of baseline comorbidities, each individual was required to have a minimum of two healthcare encounters within the preceding 5-year window.

Participants were stratified according to their initial qualifying 25(OH)D values. Those with a concentration below 20 ng/mL were assigned to the VDD cohort, whereas individuals with a value of 30 ng/mL or higher formed the control cohort. The date of the qualifying measurement was used as the index date. To strengthen exposure specificity, individuals in the VDD cohort who had any recorded 25(OH)D value at or above 20 ng/mL within the preceding 3 years were removed, and control participants with any value below 30 ng/mL during the same look-back window were excluded.

This restriction was intended to identify consistently documented VDD and reduce misclassification from transiently low 25(OH)D values. This approach better aligned with the hypothesized chronic vascular effects of VDD, although potential selection related to frailty, chronic illness, or greater clinical monitoring cannot be fully excluded.

2.3

Exclusion criteria

Individuals with a pre-existing diagnosis of aortic aneurysm or aortic dissection before the index date were excluded to ensure an aortic disease: free baseline. Conditions known to markedly alter vitamin D metabolism, including end-stage renal disease, stage 4 or 5 chronic kidney disease, renal dialysis dependence, prior bariatric surgery, hepatic failure, Crohn’s disease, and celiac disease, also warranted exclusion. To reduce confounding from acute illness: driven vitamin D testing or transient fluctuations in vitamin D levels, individuals who experienced acute kidney failure, sepsis, severe sepsis, critical care utilization, or hospital admission within 30 days before the index date were excluded.

Finally, patients with hereditary, congenital, inflammatory, or infectious conditions associated with aortopathy, namely Marfan syndrome, bicuspid aortic valve, Ehlers: Danlos syndrome, Takayasu arteritis, systemic connective tissue disorders, syphilis, and human immunodeficiency virus infection, were excluded. The diagnostic codes for all exclusion criteria are detailed in Supplementary Table 1.

2.4

Landmark design

To mitigate reverse causation and immortal-time bias, a 1-year landmark period was imposed. Although the index date corresponded to the qualifying 25(OH)D measurement, follow-up for outcome events commenced only 365 days later and extended up to 10 years from the index date. Patients diagnosed with aortic aneurysm or aortic dissection or who died during this landmark interval were excluded from the analytic cohort.

2.5

Propensity score matching

Cohort comparability was optimized through 1:1 propensity score matching using a greedy nearest-neighbor algorithm with a caliper width of 0.1 pooled standard deviation. The propensity model incorporated prespecified baseline covariates encompassing demographic factors, established AAA risk factors, cardiometabolic and vascular comorbidities, renal and hepatic conditions, and relevant medication. As TriNetX operates as a federated analytics platform, investigators do not have access to raw patient-level data to perform multiple imputation. Consequently, structured laboratory and measurement variables were summarized based on their availability, while diagnosis-based covariates were treated as documented EHR indicators within the 5-year baseline period.

The absence of a diagnostic code was interpreted as the absence of a documented condition, although some degree of under-ascertainment cannot be excluded.

Laboratory values (hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate) were not included in the primary matching model because these parameters have variable completeness in electronic health record databases, and the TriNetX platform does not support imputation of missing values. More importantly, laboratory availability in EHR data is unlikely to be random and may reflect healthcare utilization, comorbidity burden, disease severity, and monitoring intensity. Requiring complete laboratory data for the primary matching model could therefore restrict the analysis to a more intensively monitored subgroup and introduce selection bias.

Balance was appraised using standardized mean differences, with a threshold of <0.10 deemed satisfactory. Propensity score density distributions were visually compared before and after matching to confirm adequate overlap between the groups. The diagnostic codes for all matching variables are summarized in Supplementary Table 2.

2.6

Outcomes

The primary endpoint was incident AAA, including both ruptured and non-ruptured AAA. The secondary endpoints included ruptured AAA, thoracic aortic aneurysm, aortic dissection, non-aortic aneurysm, and all-cause mortality. Non-aortic aneurysms included cerebral, peripheral arterial, and visceral arterial aneurysms, which were examined to assess the anatomical specificity of the observed association. A substantially weaker association with non-aortic aneurysms would support the interpretation that VDD is more closely related to aortic-specific pathobiology than to a generalized predisposition to aneurysms. Secondary hyperparathyroidism, a recognized consequence of VDD, was evaluated as a positive control outcome to support the biological validity of the exposure definition, whereas appendicitis was used as a negative-control outcome.

The diagnostic codes for all outcomes are provided in Supplementary Table 3.

2.7

Sensitivity, subgroup, and multivariable analyses

Four sensitivity models were used to evaluate the consistency of the primary findings. Model I was confined to patients who survived the entire observation period to assess whether the differential mortality could account for the observed association. Model II required at least one documented healthcare encounter during follow-up, thereby restricting the analysis to individuals with confirmed ongoing clinical observation. Model III was limited to individuals who underwent abdominal computed tomography or ultrasonography during follow-up, addressing potential detection bias, whereby AAA diagnoses may be more frequent in patients receiving more intensive evaluation.

Model IV additionally incorporated laboratory parameters (hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate) into the propensity score model to assess whether the primary findings persisted after balancing these laboratory markers.

Subgroup analyses stratified by sex and age (50 to 65 vs. >65 years) were performed using interaction terms to evaluate potential effect modification. A multivariable Cox proportional hazards model was additionally fitted, simultaneously adjusting for established AAA risk factors, to assess the consistency of the association between VDD and incident AAA and to verify that the model reproduced the expected direction and magnitude of known risk factors for internal validity.

2.8

Exposure-gradient analysis

To examine a potential exposure: gradient relationship, a supplementary analysis was performed comparing patients with vitamin D insufficiency (25(OH)D 20 to 30 ng/mL) with controls with normal vitamin D status (≥30 ng/mL). The propensity score matching variables and analytical approach were identical to those used in the primary analyses.

2.9

Statistical methods

Time-to-event data were modeled using Cox proportional hazards regression, yielding hazard ratios (HRs) with 95% confidence intervals (CIs). A cause-specific hazard framework was adopted for the primary analysis, as the aim was to quantify the association between VDD and AAA occurrence rather than estimating the absolute cumulative incidence. Because the TriNetX analytics platform does not currently permit Aalen: Johansen competing-risk estimation within propensity score: matched cohorts, the Aalen: Johansen estimator was applied only as a supplemental descriptive analysis in the full unmatched cohort, with all-cause mortality treated as a competing event for incident AAA.

These estimates were obtained separately for each cohort and were not interpreted as a confounding-adjusted between-group comparison. The proportional hazards assumption was assessed using Schoenfeld residual analysis. To gauge vulnerability to unmeasured confounding, E-values were computed for the primary outcome. The E-value represents the minimum strength of association that an unmeasured confounder would need to have with both the exposure and the outcome, above and beyond measured covariates, to fully explain the observed HR. Statistical significance for the primary outcome was defined as a two-sided α of 0.05.

Formal statistical inference was restricted to the primary outcome. Analyses of secondary endpoints, control outcomes, sensitivity models, exposure-gradient analyses, and subgroup comparisons were considered exploratory and hypothesis-generating. Therefore, p-values for these non-primary analyses were reported as nominal descriptive measures only, and no multiplicity correction was applied. These p-values should not be interpreted as confirmatory evidence of independent associations. Missing data were handled based on the available information within the database, with no imputation performed.

3

Results

3.1

Patient selection, follow-up duration, and baseline characteristics

Before propensity score matching, 510,334 patients met the criteria for VDD, and 1,232,018 patients were classified as controls with normal vitamin D status. During the 365-day landmark period, 20,115 patients in the VDD cohort and 45,518 patients in the control cohort were excluded because they met the landmark exclusion criteria. After applying the exclusion criteria and 1:1 propensity score matching, 466,332 patients were included in each cohort. Following matching, all standardized mean differences were below 0.10, indicating a satisfactory covariate balance across demographic characteristics, comorbidities, healthcare utilization variables, and medication use (Table 1).

Propensity score density plots also demonstrated improved overlap between the cohorts after matching (Figure 1). The mean follow-up duration was 6.01 ± 3.00 years in the VDD cohort and 5.77 ± 2.91 years in the control group. The median follow-up duration was 5.98 years in the VDD cohort and 5.45 years in the control cohort, with corresponding interquartile ranges of 5.54 and 5.10 years.

TABLE 1

Baseline characteristics of patients with low vitamin D status and matched controls before and after propensity score matching.

VariablesBefore matchingAfter matching
VDD group (n = 510,334)Control group (n = 1,232,018)SMDVDD group (n = 466,332)Control group (n = 466,332)SMD
Patient characteristics
Age at index (years)59.8 ± 12.164.1 ± 11.60.36760.5 ± 12.261.0 ± 11.50.038
BMI ≥ 30 (kg/m2)189078 (37.1)343109 (27.8)0.197162372 (34.8)168763 (36.2)0.029
Female343412 (67.3)899176 (73.0)0.125318216 (68.2)316001 (67.8)0.010
White282776 (55.4)964601 (78.3)0.501280423 (60.1)275175 (59.0)0.023
Black or African American109770 (21.5)103066 (8.4)0.37580703 (17.3)80589 (17.3)0.001
Asian17496 (3.4)44991 (3.7)0.01217311 (3.7)17366 (3.7)0.001
Comorbidities and healthcare utilization
Essential (primary) hypertension221166 (43.3)520294 (42.2)0.022196059 (42.0)200208 (42.9)0.018
Dyslipidemia201600 (39.5)561700 (45.6)0.123185793 (39.8)191169 (41.0)0.023
Encounter for general adult medical examination155447 (30.5)432608 (35.1)0.099143713 (30.8)144114 (30.9)0.002
Neoplasms128474 (25.2)363885 (29.5)0.098119076 (25.5)118598 (25.4)0.002
Overweight and obesity110174 (21.6)180261 (14.6)0.18191690 (19.7)93797 (20.1)0.011
Diabetes mellitus101044 (19.8)188192 (15.3)0.11986732 (18.6)88387 (19.0)0.009
Disorders of thyroid gland76131 (14.9)253035 (20.5)0.14872271 (15.5)73143 (15.7)0.005
Nicotine dependence62334 (12.2)78009 (6.3)0.20448433 (10.4)48392 (10.4)0.000
Ischemic heart diseases51613 (10.1)122496 (9.9)0.00645979 (9.9)46245 (9.9)0.002
Diseases of liver32113 (6.3)64059 (5.2)0.04728160 (6.0)28084 (6.0)0.001
Other chronic obstructive pulmonary disease32831 (6.4)59221 (4.8)0.07127895 (6.0)27837 (6.0)0.001
Chronic kidney disease (CKD)30521 (6.0)79091 (6.4)0.01827440 (5.9)27963 (6.0)0.005
Cerebrovascular diseases29108 (5.7)69689 (5.7)0.00225676 (5.5)25997 (5.6)0.003
Iron deficiency anemia26330 (5.2)50802 (4.1)0.04922103 (4.7)22189 (4.8)0.001
Heart failure23527 (4.6)43375 (3.5)0.05519387 (4.2)19354 (4.2)0.000
Peripheral vascular diseases15420 (3.0)36457 (3.0)0.00413667 (2.9)13518 (2.9)0.002
Alcohol related disorders16621 (3.3)22445 (1.8)0.09112626 (2.7)12510 (2.7)0.002
COVID-1912934 (2.5)34610 (2.8)0.01712016 (2.6)11934 (2.6)0.001
Malnutrition4048 (0.8)8512 (0.7)0.0123508 (0.8)3524 (0.8)0.000
Medications
Glucocorticoids184146 (36.1)452424 (36.7)0.013165636 (35.5)166231 (35.6)0.003
Antilipemic agents141035 (27.6)370032 (30.0)0.053127784 (27.4)130640 (28.0)0.014
Beta blockers/related108933 (21.3)256833 (20.8)0.01296469 (20.7)97726 (21.0)0.007
ACE inhibitors97000 (19.0)190668 (15.5)0.09483025 (17.8)83403 (17.9)0.002
Calcium channel blockers88015 (17.2)183927 (14.9)0.06375139 (16.1)76008 (16.3)0.005
Angiotensin II inhibitor55008 (10.8)146624 (11.9)0.03550122 (10.7)50924 (10.9)0.006
Vitamin D supplementation47976 (9.4)192693 (15.6)0.18946351 (9.9)48821 (10.5)0.017
Insulins and analogues50551 (9.9)78051 (6.3)0.13141197 (8.8)41229 (8.8)0.000

Data are presented as number (%) or mean ± standard deviation. Standardized mean differences <0.10 were considered acceptable for covariate balance. ACE, angiotensin-converting enzyme; BMI, body mass index; CKD, chronic kidney disease; SMD, standardized mean difference; VDD, vitamin D deficiency.

FIGURE 1

Propensity score density distributions before and after matching. Propensity score density distributions are shown for patients with vitamin D deficiency and controls with normal vitamin D status before and after 1:1 propensity score matching. Improved overlap after matching indicates better covariate balance between the two cohorts. Vitamin D deficiency was defined as serum 25(OH)D < 20 ng/mL, and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL.

Figure image at the original source: open the original paper. This file is not included in this text import.
3.2

Primary, secondary, and control outcomes

During the 10-year observation window, VDD was associated with a higher risk of incident AAA than normal vitamin D status (3,086 [0.66%] vs. 2,087 [0.45%]; HR 1.40, 95% CI 1.33 to 1.48; p < 0.001; Figure 2 and Table 2). The absolute event rate difference was 0.21 percentage points. The E-value was 2.15 for the point estimate and 1.99 for the lower bound of the 95% CI.

FIGURE 2

Kaplan: Meier curves for abdominal aortic aneurysm: free probability. Kaplan: Meier curves show abdominal aortic aneurysm (AAA): free probability over the 10-year follow-up period in the propensity score: matched cohort. Follow-up began after the 1-year landmark period. Patients with vitamin D deficiency had a lower AAA: free probability than controls with normal vitamin D status. Shaded areas represent 95% confidence intervals. Vitamin D deficiency was defined as serum 25(OH)D < 20 ng/mL, and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL. AAA, abdominal aortic aneurysm; VDD, vitamin D deficiency.

Figure image at the original source: open the original paper. This file is not included in this text import.
TABLE 2

Association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

OutcomeVDD group (n = 466,332)Control group (n = 466,332)HR (95% CI)P-value
Events (%)Events (%)
Primary outcome
AAA3,086 (0.66)2,087 (0.45)1.40 (1.33 to 1.48)<0.001
Secondary outcomes
AAA rupture102 (0.02)31 (0.01)3.06 (2.05 to 4.58)<0.001
TAA2,990 (0.64)2,488 (0.53)1.13 (1.08 to 1.20)<0.001
Aortic dissection402 (0.09)284 (0.06)1.34 (1.15 to 1.56)<0.001
Non-aortic aneurysm3,825 (0.82)3,493 (0.75)1.05 (1.00 to 1.10)0.053
Mortality39,942 (8.57)25,782 (5.53)1.48 (1.46 to 1.50)<0.001
Positive control outcome
Secondary hyperparathyroidism1,147 (0.25)641 (0.14)1.72 (1.56 to 1.89)<0.001
Negative control outcome
Appendicitis2,339 (0.50)2,096 (0.45)1.06 (1.00 to 1.13)0.046

Data are presented as number of events (%). AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

Among the secondary outcomes, VDD was associated with ruptured AAA (HR 3.06, p < 0.001), aortic dissection (HR 1.34, p < 0.001), thoracic aortic aneurysm (HR 1.13, p < 0.001), and all-cause mortality (HR 1.48, p < 0.001). The association with non-aortic aneurysms was weaker and did not reach statistical significance (HR 1.05, p = 0.053). For control outcomes, VDD was associated with secondary hyperparathyroidism (HR 1.72, p < 0.001), whereas appendicitis showed only a small borderline association (HR 1.06, p = 0.046).

3.3

Sensitivity and subgroup analyses

The association between VDD and incident AAA remained consistent across all sensitivity models, with HRs ranging from 1.36 to 1.52 (Table 3). Similar patterns were observed for ruptured AAA and aortic dissection in the sensitivity analyses. In the imaging-restricted model, which included only patients who underwent at least one abdominal computed tomography or ultrasonography examination during follow-up, the association between VDD and incident AAA remained, suggesting that the primary finding was not entirely explained by differences in abdominal imaging ascertainment.

In the laboratory-adjusted matching model, VDD also remained associated with incident AAA after additional balancing of hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate. Before laboratory-adjusted matching, missingness differed across structured laboratory and measurement variables (Supplementary Table 4). The missing proportions in the VDD and control cohorts were 19.3% and 27.1% for estimated glomerular filtration rate (eGFR), 23.2% and 33.1% for hemoglobin, 34.2% and 39.6% for albumin, 39.2% and 35.4% for body mass index, and 58.2% and 66.8% for hemoglobin A1c, respectively.

These data show that laboratory and measurement completeness varied across variables, supporting the decision to evaluate these markers in a sensitivity analysis rather than in the primary matching model.

TABLE 3

Sensitivity analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm and related aortic outcomes.

OutcomesModel IModel IIModel IIIModel IV
HR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-value
AAA1.36 (1.27 to 1.44)<0.0011.42 (1.34 to 1.50)<0.0011.52 (1.42 to 1.63)<0.0011.40 (1.33 to 1.48)<0.001
AAA rupture2.24 (1.44 to 3.48)<0.0012.45 (1.70 to 3.54)<0.0012.45 (1.60 to 3.73)<0.0012.30 (1.61 to 3.29)<0.001
TAA1.09 (1.03 to 1.16)0.0021.13 (1.07 to 1.19)<0.0011.27 (1.18 to 1.36)<0.0011.13 (1.07 to 1.19)<0.001
Aortic dissection1.40 (1.17 to 1.68)<0.0011.37 (1.18 to 1.60)<0.0011.29 (1.06 to 1.56)0.0121.23 (1.06 to 1.43)0.006
Non-aortic aneurysm0.99 (0.94 to 1.04)0.7141.04 (0.99 to 1.09)0.0971.10 (1.04 to 1.17)0.0011.05 (1.01 to 1.10)0.030
MortalityNANA1.47 (1.44 to 1.49)<0.0011.42 (1.39 to 1.46)<0.0011.48 (1.46 to 1.51)<0.001

Model I was restricted to patients who survived throughout follow-up (n = 425,660 for each group). Model II was restricted to patients with at least one recorded healthcare encounter during follow-up (n = 445,290 for each group). Model III was restricted to patients with at least one recorded abdominal computed tomography or ultrasonography examination during follow-up (n = 120,142 for each group). Model IV additionally included hemoglobin, hemoglobin A1c, albumin, and estimated glomerular filtration rate in propensity score matching (n = 461,521 for each group).

AAA, abdominal aortic aneurysm; CI, confidence interval; CT, computed tomography; eGFR, estimated glomerular filtration rate; HR, hazard ratio; NA, not available; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

In sex-stratified analyses, VDD was associated with incident AAA in both males (HR 1.45, p < 0.001) and females (HR 1.61, p < 0.001), without a significant interaction by sex (p for interaction = 0.080; Table 4). Age-stratified analyses showed similar associations in patients aged 50 to 65 years (HR 1.30, p = 0.001) and those older than 65 years (HR 1.49, p < 0.001), without significant interaction by age group (p for interaction = 0.103; Table 5).

TABLE 4

Sex-stratified subgroup analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

OutcomesMale (n = 147,947 for each group)Female (n = 317,309 for each group)P for interaction
HR (95% CI)P-valueHR (95% CI)P-value
AAA1.45 (1.34 to 1.55)<0.0011.61 (1.47 to 1.76)<0.0010.080
AAA rupture2.96 (1.79 to 4.90)<0.0011.66 (1.00 to 2.73)0.0460.152
TAA1.09 (1.01 to 1.17)0.0221.30 (1.20 to 1.41)<0.0010.002
Aortic dissection1.18 (0.95 to 1.48)0.1421.57 (1.27 to 1.94)<0.0010.074
Non-aortic aneurysm1.14 (1.06 to 1.23)0.0010.97 (0.92 to 1.03)0.3140.001
Mortality1.50 (1.47 to 1.54)<0.0011.47 (1.44 to 1.50)<0.0010.202

AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

TABLE 5

Age-stratified subgroup analyses for the association between vitamin D deficiency and 10-year risks of abdominal aortic aneurysm.

Outcomes50 to 65 years
(n = 200,926 for each group)
>65 years
(n = 264,850 for each group)
P for
interaction
HR (95% CI)P-valueHR (95% CI)P-value
AAA1.30 (1.11 to 1.53)0.0011.49 (1.40 to 1.58)<0.0010.103
AAA ruptureNANA3.18 (2.08 to 4.84)<0.001NA
TAA1.06 (0.95 to 1.18)0.3291.21 (1.13 to 1.28)<0.0010.032
Aortic dissection1.29 (0.93 to 1.78)0.1311.39 (1.17 to 1.65)<0.0010.688
Non-aortic aneurysm1.10 (1.00 to 1.20)0.0511.04 (0.99 to 1.10)0.1130.303
Mortality1.43 (1.37 to 1.49)<0.0011.47 (1.45 to 1.50)<0.0010.228

AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; NA, not available; TAA, thoracic aortic aneurysm; VDD, vitamin D deficiency.

3.4

Multivariable Cox regression, competing-risk analysis, and exposure-gradient analysis

In the multivariable Cox model fitted to the unmatched cohort, VDD remained associated with a higher risk of incident AAA after adjustment for demographic, cardiometabolic, vascular, and clinical risk factors (HR 1.52, p < 0.001; Table 6). The model reproduced several expected associations with AAA, including higher risks associated with male sex, older age, nicotine dependence, chronic obstructive pulmonary disease, ischemic heart disease, peripheral vascular disease, and hypertension. Diabetes mellitus was inversely associated with incident AAA (HR 0.78, p < 0.001), consistent with prior epidemiological observations showing a lower AAA risk among patients with diabetes.

TABLE 6

Multivariable Cox regression model for incident abdominal aortic aneurysm in the unmatched cohort.

VariablesHR (95% CI)P-value
VDD vs. control group1.52 (1.45 to 1.59)<0.001
Male3.33 (3.19 to 3.47)<0.001
Age at index1.06 (1.06 to 1.06)<0.001
Dyslipidemia1.10 (1.04 to 1.15)<0.001
Overweight and obesity1.00 (0.94 to 1.06)0.913
Diabetes mellitus0.78 (0.74 to 0.82)<0.001
Nicotine dependence3.12 (2.94 to 3.30)<0.001
Ischemic heart diseases1.54 (1.45 to 1.62)<0.001
COPD1.72 (1.61 to 1.83)<0.001
Chronic kidney disease (CKD)1.00 (0.93 to 1.07)0.916
Cerebrovascular diseases1.19 (1.11 to 1.27)<0.001
Heart failure0.97 (0.89 to 1.05)0.427
Peripheral vascular diseases1.41 (1.31 to 1.53)<0.001
Alcohol-related disorders0.82 (0.73 to 0.92)0.001
Malnutrition0.93 (0.76 to 1.13)0.456
Essential (primary) hypertension1.30 (1.23 to 1.37)<0.001
White1.35 (1.28 to 1.42)<0.001

Hazard ratios were estimated using a multivariable Cox proportional hazards model fitted in the unmatched cohort. CI, confidence interval; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; HR, hazard ratio; VDD, vitamin D deficiency.

As a supplemental descriptive analysis constrained by the TriNetX platform, Aalen: Johansen competing-risk cumulative incidence estimates were generated in the full unmatched cohort, with all-cause mortality treated as a competing event for incident AAA. This analysis showed a cumulative AAA incidence of 1.09% in the VDD cohort and 0.84% in the control cohort (Table 7). The corresponding cumulative mortality estimates were 13.70% and 11.91%, respectively. Because this analysis was performed in the unmatched cohort, it should be interpreted only as an unadjusted descriptive assessment of absolute cumulative incidence and not as a substitute for the primary propensity score: matched cause-specific Cox analysis.

TABLE 7

Supplemental descriptive competing-risk cumulative incidence of abdominal aortic aneurysm and all-cause mortality in the full unmatched cohort.

CohortOutcome% of cohortAalen: Johansen cumulative incidence at end of time window
VDDAAA0.64%1.09%
VDDMortality8.20%13.70%
Normal vitamin D statusAAA0.46%0.84%
Normal vitamin D statusMortality6.45%11.91%

Cumulative incidence was estimated using the Aalen: Johansen estimator, with all-cause mortality treated as a competing event for incident abdominal aortic aneurysm. AAA, abdominal aortic aneurysm; VDD, vitamin D deficiency.

In the exposure-gradient analysis, vitamin D insufficiency was associated with a higher risk of incident AAA than normal vitamin D status (HR 1.34, p < 0.001; Table 8), with a point estimate numerically lower than that observed for VDD in the primary analysis. Similar exposure-gradient patterns were observed for ruptured AAA, aortic dissection, and secondary hyperparathyroidism, whereas appendicitis remained near null.

TABLE 8

Exposure-gradient analysis of vitamin D insufficiency and 10-year risks of abdominal aortic aneurysm, severe aortic events, control outcomes, and mortality.

OutcomeVDI group (n = 615,798)Control group (n = 615,798)HR (95% CI)P-value
Events (%)Events (%)
AAA3,851 (0.63)2,788 (0.45)1.34 (1.28 to 1.41)<0.001
AAA rupture94 (0.02)58 (0.01)1.56 (1.12 to 2.16)0.007
TAA4,379 (0.71)3,480 (0.57)1.22 (1.17 to 1.28)<0.001
Aortic dissection469 (0.08)377 (0.06)1.20 (1.05 to 1.38)0.007
Non-aortic aneurysm4,938 (0.80)4,643 (0.75)1.04 (1.00 to 1.08)0.072
Mortality43,698 (7.10)33,249 (5.40)1.28 (1.26 to 1.30)<0.001
Hyperparathyroidism976 (0.16)879 (0.14)1.09 (0.99 to 1.19)0.076
Appendicitis3,052 (0.50)2,915 (0.47)1.02 (0.97 to 1.07)0.460

Data are presented as number of events (%). Vitamin D insufficiency was defined as serum 25(OH)D 20 to 30 ng/mL (n = 615,798), and normal vitamin D status was defined as serum 25(OH)D ≥ 30 ng/mL (n = 615,798). AAA, abdominal aortic aneurysm; CI, confidence interval; HR, hazard ratio; TAA, thoracic aortic aneurysm; VDI, vitamin D insufficiency.

4

Discussion

In this large multicenter propensity score: matched cohort study, clinically defined VDD was associated with a higher long-term risk of incident AAA than normal vitamin D status. This association was consistent across sensitivity analyses addressing survival, follow-up ascertainment, imaging detection, and laboratory balance and was further supported by multivariable adjustment in the unmatched cohort. Exploratory analyses also suggested associations with ruptured AAA and aortic dissection, whereas the association with non-aortic aneurysms was comparatively weaker, supporting possible aortic specificity. However, given the modest absolute risk difference and observational design, these findings should be interpreted as hypothesis-generating evidence of association rather than a causal relationship.

The observed association between VDD and incident AAA extends prior epidemiological evidence, which has been limited in scale and inconsistent in direction. The ARIC prospective cohort followed 12,770 participants for a median of 19.7 years and found no significant association between VDD and incident AAA (24), whereas an Australian community-based study of 4,233 older men (29) reported inverse associations between 25(OH)D concentrations and AAA diagnosis or aneurysm size. Recent cardiovascular literature also supports a broader link between vitamin D status and vascular inflammation, endothelial function, oxidative stress, hypertension, and atherosclerosis, although such evidence should be viewed as mechanistic background rather than direct proof of an AAA-specific effect (30).

Compared with these earlier studies, the present analysis used a clinically defined VDD threshold, included more than 460,000 matched patients per cohort, and applied a 1-year landmark design to reduce reverse causation bias. In addition, the propensity score model balanced a broad range of clinically relevant AAA risk factors, including smoking-related diagnoses, chronic obstructive pulmonary disease, peripheral vascular disease, chronic kidney disease, cardiovascular comorbidities, healthcare utilization, and the use of medications. These design features may partly explain why our study was able to detect a clinically defined VDD: AAA association that was not consistently observed in smaller cohorts using conventional multivariable adjustments.

Among the exploratory secondary outcomes, VDD was associated with ruptured AAA and aortic dissection. Because these analyses were not adjusted for multiplicity, they should be interpreted strictly as hypothesis-generating and not as confirmatory evidence of independent associations. The higher relative estimate for ruptured AAA is biologically plausible and compatible with experimental evidence (22), linking VDD to aneurysm enlargement and rupture-prone phenotypes. The lower HRs for ruptured AAA in the sensitivity analyses likely reflected rare-event instability and attenuation after restricting the analytic population; therefore, the rupture finding should be interpreted as a hypothesis-generating signal rather than a precise estimate of absolute rupture risk.

VDD was also associated with a high risk of thoracic aortic aneurysm, although the magnitude of the association (HR 1.13) was weaker than that observed for AAA. This attenuation is biologically plausible because the pathogenesis of thoracic aortic aneurysm involves distinct mechanisms, including genetic predisposition and cystic medial degeneration (31, 32), which may be less directly influenced by vitamin D: mediated inflammatory and matrix-remodeling pathways.

A notable feature of this study was the inclusion of non-aortic aneurysms as anatomic specificity comparators. In contrast to the stronger association observed for AAA, VDD was not significantly associated with non-aortic aneurysms. This pattern suggests that VDD may be more closely related to aortic-specific degenerative processes than to a generalized aneurysmal predisposition, although this interpretation remains to be confirmed. This interpretation is biologically plausible because AAA pathogenesis is characterized by localized aortic wall inflammation, vascular smooth muscle cell dysfunction, and extracellular matrix degradation.

The positive control outcome of secondary hyperparathyroidism supported the biological validity of the definition of VDD exposure. Conversely, the negative-control outcome of appendicitis showed a near-null effect estimate. Because control outcomes were exploratory and no multiplicity correction was applied, the nominal p-value for appendicitis was not interpreted as evidence of a meaningful association. Instead, the small effect size suggested that nonspecific residual bias, if present, was limited in magnitude and unlikely to explain the substantially larger primary AAA association (33).

The sensitivity analyses provided methodological support for the primary findings. The association between VDD and incident AAA remained present in models restricted to patients who survived throughout follow-up and those with documented healthcare encounters during follow-up, partially mitigating the concern that the finding was driven solely by differential mortality or incomplete observation. Because AAA is frequently asymptomatic and often detected incidentally during abdominal imaging performed for unrelated indications (34, 35), the imaging-restricted model was particularly relevant for addressing potential ascertainment biases.

The association persisted among patients who underwent abdominal computed tomography or ultrasonography during follow-up. The modestly higher HR in this model may reflect enrichment of a clinically higher-risk subgroup and residual differences in imaging indication, timing, or frequency. Because requiring at least one imaging examination did not fully equalize surveillance intensity, this analysis cannot fully eliminate selection or ascertainment bias.

Subgroup analyses suggested that the association between VDD and incident AAA was present in both males and females and in both age strata, without statistically significant interactions with the primary outcome. These findings suggest that the observed association was not confined to any single demographic subgroup. However, the subgroup analyses were exploratory, and the study was not designed to redefine sex: or age-specific screening strategies. The findings in patients aged 50 to 65 years are clinically interesting because conventional AAA screening strategies primarily focus on older adults, particularly men aged 65 years or older (1).

Nevertheless, the absolute risk of AAA in younger individuals is lower, and these findings should not be interpreted as evidence to expand AAA screening eligibility criteria. They suggest that VDD may identify a vascular risk phenotype that warrants further investigation.

The multivariable Cox model fitted to the unmatched cohort provided an additional internal validity check. VDD remained associated with incident AAA after adjusting for demographic, cardiometabolic, vascular, and clinical risk factors. The model also reproduced several expected AAA associations, including higher risks associated with male sex, older age, nicotine dependence, chronic obstructive pulmonary disease, ischemic heart disease, peripheral vascular disease, and hypertension (1, 10, 36). Diabetes mellitus was inversely associated with incident AAA, consistent with prior epidemiological studies and meta-analyses reporting a lower prevalence, incidence, or progression of AAA among patients with diabetes (37Not reported39).

Proposed mechanisms include diabetes-related extracellular matrix glycation and collagen cross-linking, reduced matrix metalloproteinase activity, altered inflammatory remodeling of the aortic wall, and potential effects of antidiabetic therapies, all of which may limit aneurysmal dilation despite the broader vascular risk associated with diabetes (37Not reported39). This internal consistency supports the validity of our analytical framework. However, because the multivariable model was performed in the unmatched cohort, it should be considered complementary to, rather than a replacement for, the primary propensity score: matched analysis.

The competing risk analysis further showed that the direction of the association persisted when death was considered a competing event; however, this analysis was descriptive and based on the full unmatched cohort. Similarly, the exposure-gradient analysis suggested a graded pattern across vitamin D insufficiency and deficiency, but it should be considered exploratory because the exposure categories were evaluated in separate matched comparisons rather than in a single continuous dose: response model.

Clinical implications should be interpreted cautiously. The present findings do not support VDD as a stand-alone indication for AAA screening, nor do they justify expanding current guideline-based AAA screening criteria. Instead, VDD may be considered a supplementary vascular risk marker in patients who already have established AAA risk factors. In general clinical practice, the presence of VDD may prompt clinicians to review traditional AAA risk factors, prior abdominal imaging, and eligibility for guideline-based screening. From a perioperative perspective, particularly before major noncardiac surgery, VDD in patients with conventional vascular risk factors may highlight the need for careful cardiovascular and vascular risk review and, when aneurysmal disease is known or suspected, multidisciplinary evaluation with vascular specialists.

However, vitamin D correction should not be assumed to reduce AAA risk or severe aortic events based on the present observational data.

Several limitations should be acknowledged. First, although both cohorts included only patients with documented 25(OH)D testing, the clinical indication for testing was unavailable and may have differed between groups. Therefore, indication-specific residual confounding cannot be excluded. Furthermore, important factors such as sunlight exposure, dietary intake, physical activity, frailty, socioeconomic status, over-the-counter supplement use, and detailed smoking history were not fully considered. Although race was adequately balanced after matching, the substantial pre-matching racial differences may reflect unmeasured social, geographic, or healthcare-access factors, and residual confounding related to these factors cannot be excluded.

Second, although the 3-year look-back was intended to identify a more persistent VDD phenotype, exposure classification remained based on a single qualifying 25(OH)D measurement. Harmonized data on testing season, assay methods, laboratory reference ranges, testing indications, and post-index vitamin D levels or supplementation were unavailable. Therefore, residual exposure misclassification cannot be excluded, and differential vitamin D repletion may have attenuated the observed association over time. Third, incident AAA, ruptured AAA, and aortic dissection were identified using electronic health record diagnosis codes without chart-level adjudication or algorithm validation against imaging-confirmed disease; therefore, outcome misclassification cannot be excluded and could have biased the effect estimates in either direction.

Although the imaging-restricted sensitivity analysis partially reduced concerns about differential ascertainment, selection by indication could not be excluded. Fourth, data on aneurysm diameter, growth rate, morphology, rupture presentation, operative repair, imaging findings, and cause-specific mortality were unavailable. In particular, the TriNetX platform does not provide structured diagnostic imaging details, such as baseline aneurysm diameter or serial aortic measurements. Therefore, we could not verify whether small, subclinical aneurysms were already present but unrecorded at the index date. Therefore, we could not evaluate aneurysm severity, longitudinal expansion, treatment pathways, or whether the deaths were directly attributable to aneurysm-related complications.

Fifth, secondary outcomes, subgroup analyses, control outcomes, sensitivity analyses, and exposure gradient analyses were exploratory and not adjusted for multiplicity. Therefore, these findings should be interpreted as supportive or hypothesis-generating, rather than as confirmatory evidence. Finally, the requirement for at least two healthcare encounters during the preceding 5 years may have selected more healthcare-engaged and comorbid patients while excluding healthier infrequent users. Accordingly, the findings may not be fully generalizable to community-dwelling populations or individuals with limited healthcare access.

5

Conclusion

Clinically defined VDD was associated with a higher long-term risk of incident AAA in this large propensity score: matched cohort study. The association was supported by sensitivity models and multivariable adjustment, whereas secondary outcomes, subgroup analyses, and exposure gradient analyses should be interpreted as exploratory and hypothesis-generating. These findings should not be interpreted as supporting VDD-based AAA screening, but they suggest that VDD may serve as an adjunctive risk marker when evaluating patients with established vascular risk factors in general or perioperative clinical settings.

Given the modest absolute risk difference and the observational design, further prospective studies with systematic aortic imaging, repeated vitamin D measurements, detailed lifestyle data, and mechanistic endpoints are needed to determine whether VDD is causally involved in AAA development or severe aortic complications.

Edited by: Giovanni Cangelosi, AST Fermo Marche Region Health System, Italy

Reviewed by: Jagannadha Rao Peela, Bioprist Institute of Medical and Health Sciences, Jamaica

Hamza Islam, Faisalabad Medical University, Pakistan

Data availability statement

The data supporting the findings of this study were obtained from the TriNetX Global Collaborative Network. Restrictions apply to the availability of these data, which were used under license for the current study and are not publicly accessible. Data access can be obtained through a direct agreement with TriNetX. Requests to access the datasets should be directed to https://live. trinetx.com.

Ethics statement

The studies involving humans were approved by Institutional Review Board of Chi Mei Medical Center. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent was waived by the Institutional Review Board of Chi Mei Medical Center because this retrospective study analyzed de-identified electronic health record data from the TriNetX Global Collaborative Network. The study involved no direct patient contact, no intervention, and no use of directly identifiable personal information.

Author contributions

K-CH: Software, Resources, Investigation, Writing: original draft, Writing: review & editing, Visualization, Data curation, Methodology, Validation, Project administration, Conceptualization, Supervision, Formal analysis. T-SY: Investigation, Software, Methodology, Writing: review & editing, Writing: original draft, Conceptualization, Project administration, Visualization, Validation. I-WC: Validation, Project administration, Formal analysis, Data curation, Supervision, Investigation, Writing: review & editing, Conceptualization, Writing: original draft, Software, Visualization, Methodology.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. During the preparation of this manuscript, the authors used ChatGPT to assist with language editing and manuscript refinement. The authors reviewed and edited the content as needed and took full responsibility for the final content of the manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnut.2026.1900016/full#supplementary-material

Supplement file at the original source: open the original paper. This file is not included in this text import.

References

1. Owens DK Davidson KW Krist AH Barry MJ Cabana M Caughey AB et al . Screening for abdominal aortic aneurysm: US preventive services task force recommendation statement. JAMA. (2019 ) 322 :2211Not reported8 . 10.1001/jama.2019.18928 31821437
2. Sprynger M Willems M Van Damme H Drieghe B Wautrecht JC Moonen M . Screening program of abdominal aortic aneurysm. Angiology. (2019 ) 70 :407Not reported13 . 10.1177/0003319718824940 30654619
3. Gunnarsson K Wanhainen A Björck M Djavani-Gidlund K Mani K . Nationwide study of ruptured abdominal aortic aneurysms during twenty years (1994-2013). Ann Surg. (2021 ) 274 :e160Not reported6 . 10.1097/SLA.0000000000003555 31425332
4. Malayala SV Raza A Vanaparthy R . Gender-based differences in abdominal aortic aneurysm rupture: a retrospective study. J Clin Med Res. (2020 ) 12 :794Not reported802 . 10.14740/jocmr4376 33447313 PMC7781278
5. Li Z Kong W . Cellular signaling in abdominal aortic aneurysm. Cell Signal. (2020 ) 70 :109575 . 10.1016/j.cellsig.2020.109575 32088371
6. Li Y Wang W Li L Khalil RA . MMPs and ADAMs/ADAMTS inhibition therapy of abdominal aortic aneurysm. Life Sci. (2020 ) 253 :117659 . 10.1016/j.lfs.2020.117659 32283055 PMC7385731
7. Stêpień KL Auguściak-Duma A Fus-Kujawa A Diak N Lesiak M Bogunia E et al . Abdominal aortic aneurysm: characteristics of extracellular matrix abnormalities caused by disorders of collagen types I and III and elastin. Int J Cardiol. (2026 ) 444 :133973 . 10.1016/j.ijcard.2025.133973 41115634
8. Cai D Chen SY . Elastin in the pathogenesis of abdominal aortic aneurysm. Cells. (2025 ) 14 :1597 . 10.3390/cells14201597 41148811 PMC12563134
9. Qian G Adeyanju O Olajuyin A Guo X . Abdominal aortic aneurysm formation with a focus on vascular smooth muscle cells. Life. (2022 ) 12 :191 . 10.3390/life12020191 35207478 PMC8880357
10. Kobeissi E Hibino M Pan H Aune D . Blood pressure, hypertension and the risk of abdominal aortic aneurysms: a systematic review and meta-analysis of cohort studies. Eur J Epidemiol. (2019 ) 34 :547Not reported55 . 10.1007/s10654-019-00510-9 30903463 PMC6497813
11. Zhang P Liu C Li J Gao H Fu Y Feng W et al . Risk factors for abdominal aortic aneurysm in general populations: a systematic review and meta-analysis. PLoS One. (2025 ) 20 :e0329500 . 10.1371/journal.pone.0329500 40982438 PMC12453230
12. Cho IY Han K Lee KN Koo HY Cho YH Lee JH et al . Risk factors for abdominal aortic aneurysm in patients with diabetes. J Vasc Surg. (2025 ) 81 :128eNot reported36e . 10.1016/j.jvs.2024.09.007 39303864
13. Wang TY Wang HW Jiang MY . Prevalence of vitamin D deficiency and associated risk of all-cause and cause-specific mortality among middle-aged and older adults in the United States. Front Nutr. (2023 ) 10 :1163737 . 10.3389/fnut.2023.1163737 37275650 PMC10232798
14. Adeniyi OV Masilela C George JA . Prevalence of vitamin D deficiency and its association with cardiometabolic risk factors among healthcare workers in the Eastern cape province, South Africa; cross-sectional study. Sci Rep. (2024 ) 14 :4756 . 10.1038/s41598-024-54977-y 38413628 PMC10899187
15. Zhang J McCullough PA Tecson KM . Vitamin D deficiency in association with endothelial dysfunction: implications for patients with COVID-19. Rev Cardiovasc Med. (2020 ) 21 :339Not reported44 . 10.31083/j.rcm.2020.03.131 33070539
16. Saponaro F Marcocci C Zucchi R . Vitamin D status and cardiovascular outcome. J Endocrinol Invest. (2019 ) 42 :1285Not reported90 . 10.1007/s40618-019-01057-y 31172459
17. Ao T Kikuta J Ishii M . The effects of vitamin d on immune system and inflammatory diseases. Biomolecules. (2021 ) 11 :1624 . 10.3390/biom11111624 34827621 PMC8615708
18. Wu Y Gong Y Ma Y Zhao Q Fu R Zhang X et al . Effects of vitamin D status on cutaneous wound healing through modulation of EMT and ECM. J Nutr Biochem. (2024 ) 134 :109733 . 10.1016/j.jnutbio.2024.109733 39127309
19. Quintana RA Taylor WR . Cellular mechanisms of aortic aneurysm formation. Circ Res. (2019 ) 124 :607Not reported18 . 10.1161/CIRCRESAHA.118.313187 30763207 PMC6383789
20. Golledge J Thanigaimani S Powell JT Tsao PS . Pathogenesis and management of abdominal aortic aneurysm. Eur Heart J. (2023 ) 44 :2682Not reported97 . 10.1093/eurheartj/ehad386 37387260 PMC10393073
21. Lu H Du W Ren L Hamblin MH Becker RC Chen YE et al . vascular smooth muscle cells in aortic aneurysm: from genetics to mechanisms. J Am Heart Assoc. (2021 ) 10 :e023601 . 10.1161/JAHA.121.023601 34796717 PMC9075263
22. Nsengiyumva V Krishna SM Moran CS Moxon JV Morton SK Clarke MW et al . Vitamin D deficiency promotes large rupture-prone abdominal aortic aneurysms and cholecalciferol supplementation limits progression of aneurysms in a mouse model. Clin Sci. (2020 ) 134 :2521Not reported34 . 10.1042/CS20200980 32936248 PMC7536319
23. Zhang Y Zhou C Ye Z Liu M He P Yang S et al . Serum 25-hydroxyvitamin D, genetic susceptibility, and abdominal aortic aneurysm risk. J Clin Endocrinol Metab. (2025 ) 110 :1989Not reported96 . 10.1210/clinem/dgae692 39364894
24. Lutsey PL Rooney MR Folsom AR Michos ED Alonso A Tang W . Markers of vitamin D metabolism and incidence of clinically diagnosed abdominal aortic aneurysm: the atherosclerosis risk in communities study. Vasc Med. (2018 ) 23 :253Not reported60 . 10.1177/1358863X17751258 29400142 PMC6190682
25. Chen IW Chang LC Lai YC Hung KC . Association between zinc deficiency and adverse outcomes in patients with dementia: a matched cohort study. Front Nutr. (2026 ) 13 :1801558 . 10.3389/fnut.2026.1801558 41939182 PMC13046538
26. Chang YJ Weng HL Hung KC Ho CN Lai YC Wu JY et al . Vitamin D deficiency and risk of heart failure in patients with obstructive sleep apnea: a cohort analysis. Front Nutr. (2026 ) 13 :1755607 . 10.3389/fnut.2026.1755607 41798844 PMC12960163
27. Chen IW Chang LC Chang YJ Lai YC Hung KC . Association between iron deficiency anemia and the risk of new-onset tuberculosis infection: a matched cohort analysis. Front Nutr. (2026 ) 13 :1727992 . 10.3389/fnut.2026.1727992 41641159 PMC12866892
28. Hung KC Yu TS Lai YC Yu CH Wu JY Tsai WW et al . Perioperative ketamine exposure and postoperative atrial fibrillation/flutter risk after video-assisted thoracoscopic surgery: a multi-institutional study. Drug Des Devel Ther. (2025 ) 19 :11173Not reported85 . 10.2147/DDDT.S569143 41425499 PMC12717812
29. Wong YY Flicker L Yeap BB McCaul KA Hankey GJ Norman PE . Is hypovitaminosis D associated with abdominal aortic aneurysm, and is there a dose-response relationship? Eur J Vasc Endovasc Surg. (2013 ) 45 :657Not reported64 . 10.1016/j.ejvs.2013.03.015 23602862
30. Islam H Hassaan SM Islam R Islam T Zaidi F Rehman HU et al . Vitamin D’s role in cardiovascular diseases. Discov Med. (2024 ) 36 :1973Not reported86 . 10.24976/Discov.Med.202436189.182 39463218
31. El-Hamamsy I Yacoub MH . Cellular and molecular mechanisms of thoracic aortic aneurysms. Nat Rev Cardiol. (2009 ) 6 :771Not reported86 . 10.1038/nrcardio.2009.191 19884902
32. Milewicz DM Prakash SK Ramirez F . Therapeutics targeting drivers of thoracic aortic aneurysms and acute aortic dissections: insights from predisposing genes and mouse models. Annu Rev Med. (2017 ) 68 :51Not reported67 . 10.1146/annurev-med-100415-022956 28099082 PMC5499376
33. Lipsitch M Tchetgen Tchetgen E Cohen T . Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology. (2010 ) 21 :383Not reported8 . 10.1097/EDE.0b013e3181d61eeb 20335814 PMC3053408
34. Castro-Ferreira R Vidoedo J Peixoto J Canedo A Teixeira J Leite-Moreira A et al . Incidental abdominal aortic aneurysms are largely undocumented and unmonitored. Ann Vasc Surg. (2021 ) 77 :182Not reported6 . 10.1016/j.avsg.2021.05.027 34411670
35. Mansoor SM Jørgensen JJ Hisdal J Rabben T . Thirty-nine percent of patients with a ruptured abdominal aortic aneurysm (AAA) have an incidentally detected AAA prior to rupture. Ann Vasc Surg. (2024 ) 108 :148Not reported56 . 10.1016/j.avsg.2024.04.017 38942371
36. Flessenkaemper IH Loddenkemper R Roll S Enke-Melzer K Wurps H Bauer TT . Screening of COPD patients for abdominal aortic aneurysm. Int J Chron Obstruct Pulmon Dis. (2015 ) 10 :1085Not reported91 . 10.2147/COPD.S81439 26089658 PMC4468935
37. Aune D Schlesinger S Norat T Riboli E . Diabetes mellitus and the risk of abdominal aortic aneurysm: a systematic review and meta-analysis of prospective studies. J Diabetes Complications. (2018 ) 32 :1169Not reported74 . 10.1016/j.jdiacomp.2018.09.009 30415876
38. Takagi H Umemoto T . Diabetes and abdominal aortic aneurysm growth. Angiology. (2016 ) 67 :513Not reported25 . 10.1177/0003319715602414 26311742
39. Ning X Ding N Ballew SH Hicks CW Coresh J Selvin E et al . Diabetes, its duration, and the long-term risk of abdominal aortic aneurysm: the atherosclerosis risk in communities (ARIC) study. Atherosclerosis. (2020 ) 313 :137Not reported43 . 10.1016/j.atherosclerosis.2020.09.031 33049655 PMC7655715
Publication and source details

Original title: Vitamin D deficiency and the long-term risk of new-onset abdominal aortic aneurysm and severe aortic events: a multicenter propensity score: matched cohort study

Kuo-Chuan Hung; Ting-Sian Yu; I-Wen Chen

Frontiers in Nutrition · 2026

* Correspondence: I-Wen Chen, MD, cheniwena60912@gmail.comCopyright © 2026 Hung, Yu and Chen.2026Hung, Yu and Chenhttps://creativecommons.org/licenses/by/4.0/This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.The author(s) declared that financial support was not received for this work and/or its publication.

Repository JATS snapshot; version as supplied

Source retrieved: 2026-11-08T11:47:38+00:00

DOI: 10.3389/fnut.2026.1900016 · PMC13581915