Abstract
Background
Older adults with atrial fibrillation (AF)–associated acute ischemic stroke are prone to disability, depressive symptoms, and falls; the prognostic value of vitamin D status is uncertain.
Methods
We conducted a single-center prospective cohort study of patients aged ≥65 years with ECG-confirmed AF and imaging-confirmed acute ischemic stroke at a tertiary hospital in China (January 2022–August 2024). Serum 25-hydroxyvitamin D [25(OH)D] within 48 hours was modeled per 10 ng/mL decrement and as <20 vs ≥20 ng/mL. Outcomes were 3-month modified Rankin Scale (mRS) shift, Barthel Index (BI), and PHQ-9 ≥10, plus 12-month time to first fall (death as a competing event) and total falls rate. Models adjusted for prespecified baseline confounders.
Results
Among 802 participants, median age was 76.2 years (IQR, 71.6–80.6); 60.7% had 25(OH)D <20 ng/mL. Three-month mRS/BI, 3-month PHQ-9, and falls follow-up data were available for 776 (96.8%), 762 (95.0%), and 742 (92.5%) participants, respectively. Each 10 ng/mL lower 25(OH)D was associated with worse 3-month mRS shift (common OR 1.17, 95% CI 1.05–1.31) and lower BI (β −3.9, 95% CI −6.2 to −1.6). Depressive symptoms occurred in 16.8% and were more common with 25(OH)D <20 ng/mL (OR 1.54, 95% CI 1.07–2.22). Among 742 with falls follow-up, low 25(OH)D predicted first fall (HR 1.29, 95% CI 1.07–1.55) and a higher recurrent falls rate (IRR 1.33, 95% CI 1.10–1.61).
Conclusion
Lower baseline 25(OH)D was consistently associated with poorer functional recovery, greater depressive symptom burden, and higher fall risk and recurrence, supporting 25(OH)D as a pragmatic prognostic marker; multicenter validation and targeted trials are warranted.
Keywords: 25(OH)D, atrial fibrillation, acute ischemic stroke, post-stroke depression, falls
Introduction
Atrial fibrillation (AF)–associated ischemic stroke in older adults poses a major clinical challenge because of its substantial disability burden, rehabilitation and secondary prevention needs, and the co-occurrence of geriatric syndromes that can hinder long-term recovery.1,2 AF prevalence rises sharply with age and is a leading risk factor for ischemic stroke; compared with strokes of other etiologies, AF-associated strokes are associated with higher mortality and greater disability.1,2 In this population, frailty, falls, depressive symptoms, and cognitive impairment are common and often coexist, further complicating post-stroke management and rehabilitation.3,4 Frailty, in particular, is highly prevalent among older adults with AF and is consistently linked to worse functional recovery and greater dependence.5 Although clinical guidelines recommend anticoagulation for stroke prevention, its use is frequently lower in older patients with geriatric syndromes because clinicians and patients may be concerned about bleeding risk and treatment complexity.4,6 This descriptive observation should not be interpreted as support for withholding anticoagulation or other evidence-based secondary prevention; anticoagulation when indicated, lipid management when appropriate, and vascular risk-factor control remain central to care after AF-associated ischemic stroke.1,2,7 Accordingly, optimizing AF management in older adults requires a comprehensive, patient-centered approach that integrates geriatric assessment and addresses these syndromes alongside standard cardiovascular and stroke care to improve outcomes and quality of life.7
Vitamin D status, most commonly assessed by circulating 25-hydroxyvitamin D [25(OH)D], is increasingly recognized as a pragmatic biomarker in older adults with multimorbidity.8,9 The high prevalence of low 25(OH)D in older populations—estimated at 25% to 65%—underscores its clinical relevance,8 particularly when concentrations fall below 20 ng/mL, a threshold linked to higher risks of depressive symptoms, cognitive decline, and falls.9–12 Biological plausibility is supported by vitamin D’s neurosteroid-like properties and the widespread expression of vitamin D receptors in the brain.10 Vitamin D deficiency has also been associated with poorer physical performance and increased fall risk, and supplementation trials have evaluated musculoskeletal and fall outcomes with mixed findings.11,12 However, prospective evidence remains limited in older adults with non-valvular atrial fibrillation (NVAF)–associated stroke, and few studies have concurrently evaluated depressive symptoms, falls, and functional recovery within this high-risk cohort.13 Clarifying the prognostic relevance of 25(OH)D in this setting may inform integrated geriatric–stroke care strategies and help reduce the downstream burden of age-related morbidity and healthcare utilization.14
The primary aim was to assess the association between baseline serum 25(OH)D and 3-month (post-discharge) global disability using an ordinal shift analysis of the modified Rankin Scale (mRS). Secondary aims were to evaluate associations with 3-month (post-discharge) activities of daily living (Barthel Index), clinically significant post-stroke depressive symptoms at 3 months (post-discharge; PHQ-9 ≥10), and fall outcomes over 12 months. We also prespecified an exploratory assessment of nonlinearity and an exploratory mediation hypothesis whereby early depressive symptom severity (PHQ-9 at 1 month) may partially mediate the association between baseline 25(OH)D and subsequent falls. Clinically, the study was intended to clarify whether 25(OH)D can help identify older stroke survivors who may benefit from closer nutritional, rehabilitation, mood, and fall-risk follow-up while continuing standard evidence-based stroke and AF care.
Methods
Study Design and Setting
This non-interventional, single-center prospective cohort study was conducted at Tianjin Huanhu Hospital. Consecutive eligible patients were enrolled between January 2022 and August 2024. Follow-up assessments were scheduled at approximately 1, 3, and 12 months after discharge. Participants were followed from discharge until 12 months, death, or loss to follow-up, whichever occurred first. Baseline serum 25(OH)D, measured within 48 hours of admission, was the prespecified exposure. The study protocol conformed to the Declaration of Helsinki and was approved by the Ethics Committee of Tianjin Huanhu Hospital. Written informed consent was obtained from all participants before any study procedures.
Participants and Clinical Definitions
Patients were eligible if they were aged ≥65 years, had imaging-confirmed acute ischemic stroke, and had ECG-documented AF.
AF Definition
AF was defined by rhythm documentation on (1) a 12-lead ECG demonstrating AF or (2) a rhythm strip/single-lead ECG showing AF for ≥30 seconds, characterized by irregularly irregular RR intervals and the absence of distinct, consistent P waves. AF could be documented either before admission or during the index hospitalization, including via in-hospital telemetry or ambulatory monitoring.
Stroke Definition
Acute ischemic stroke was diagnosed by the treating stroke neurologist as an episode of acute focal neurological dysfunction attributable to focal cerebral infarction, supported by brain imaging (non-contrast CT and/or MRI) supporting an ischemic stroke diagnosis and excluding intracranial hemorrhage. Stroke severity was quantified using the National Institutes of Health Stroke Scale (NIHSS), assessed on admission or within 24 hours. Pre-stroke functional status was recorded using the modified Rankin Scale (mRS), reflecting function immediately prior to the index event.
Valve Exclusions
Patients with moderate-to-severe rheumatic mitral stenosis or a mechanical prosthetic heart valve were excluded.
Vitamin D Exposure Assessment
Baseline 25(OH)D was measured from venous blood drawn within 48 hours of admission and analyzed in the hospital central laboratory using a standardized automated chemiluminescent immunoassay as total 25(OH)D (25[OH]D2 + 25[OH]D3).
The primary exposure specification treated 25(OH)D as a continuous variable per 10 ng/mL lower concentration. Potential nonlinearity was evaluated using restricted cubic splines. For clinical interpretability, a secondary exposure specification categorized vitamin D status as low/inadequate (<20 ng/mL) versus higher (≥20 ng/mL).
Outcomes and Follow-Up Procedures
Functional recovery at 3 months was assessed using the modified Rankin Scale (mRS) and the Barthel Index (BI; 0–100). The 3-month mRS and BI assessments were performed by trained study neurologists or research nurses using standardized case-report forms; baseline 25(OH)D results were not provided to outcome assessors. The primary functional endpoint was the full ordinal distribution of mRS scores, analyzed using shift methods.
Depressive symptoms were assessed using the Patient Health Questionnaire-9 (PHQ-9), a previously validated measure of depressive symptom severity,15 at 1 and 3 months. PHQ-9 questionnaires were self-completed when possible or administered by trained study staff using standardized instructions; baseline 25(OH)D results were not provided to staff conducting these assessments. Clinically significant depressive symptoms were defined a priori as PHQ-9 ≥10 for the primary binary endpoint, and PHQ-9 ≥9 was evaluated as a prespecified sensitivity definition.
Falls were ascertained after discharge through structured follow-up interviews with the patient and/or caregiver. Falls were recorded prospectively at scheduled 1-, 3-, and 12-month contacts; at each contact, respondents were asked about events since discharge or the previous contact. Formal fall diaries were not used. When participants could not respond reliably, caregivers served as proxy respondents. A fall was defined as an unexpected event in which the participant came to rest inadvertently on the ground, floor, or a lower level. The date (or best-estimated date) of the first fall was recorded for time-to-event analyses. Fall etiology was not adjudicated beyond the standard definition (eg, syncope-related falls were not separately classified). Fall burden was modeled as a recurrent-event count outcome (total number of falls), and incidence rate ratios (IRRs) were estimated using negative binomial regression. Deaths were identified through hospital records and follow-up contacts and were treated as a competing event in time-to-first-fall analyses. The structured falls interview guide was developed for this study (Supplementary Figure S1).
Covariates
Covariates were prespecified a priori as potential confounders of the association between vitamin D status and post-stroke outcomes. The primary models adjusted for the following baseline covariates: age, sex, season of blood draw, admission National Institutes of Health Stroke Scale (NIHSS) score, pre-stroke modified Rankin Scale (mRS), serum albumin, C-reactive protein (CRP), estimated glomerular filtration rate (eGFR), and vascular/cardiac comorbidities (hypertension, diabetes mellitus, heart failure, coronary artery disease, and prior stroke or transient ischemic attack).
Care-pathway variables, including anticoagulation at discharge (prescription of an oral anticoagulant at discharge) and rehabilitation intensity (higher vs lower intensity based on prespecified criteria), may be downstream of the index admission and potentially on the causal pathway. Therefore, these variables were evaluated in sensitivity analyses rather than included in the primary adjustment set.
Variables not available in a standardized format, including prior falls, baseline gait/balance impairment, cognitive status, sarcopenia or muscle strength, visual impairment, body mass index, pre-admission or post-discharge vitamin D supplementation, medication burden, and home-environment hazards, were not included in the primary adjustment set.
Statistical Analysis
Baseline characteristics were summarized as mean ± SD for approximately normally distributed variables, median [IQR] for skewed variables, and n (%) for categorical variables. Between-group comparisons by vitamin D category (<20 vs ≥20 ng/mL) were performed using Welch’s t test, the Wilcoxon rank-sum test, or the χ2-test, as appropriate. P values are reported, and false discovery rate (FDR) q values were calculated for the prespecified set of baseline comparisons using the Benjamini–Hochberg procedure.
Primary Disability Endpoint
Proportional-odds ordinal logistic regression was used to estimate a common odds ratio (OR) for a worse 3-month modified Rankin Scale (mRS) score (shift analysis). The proportional-odds assumption was assessed using a global score/Brant-type test; if the assumption had been materially violated, a dichotomized mRS sensitivity analysis was planned.
Barthel Index at 3 months: The Barthel Index (BI) was analyzed using multivariable linear regression with robust standard errors.
Depressive Symptoms
Clinically significant depressive symptoms at 3 months (PHQ-9 ≥10) were modeled using multivariable logistic regression.
First Fall
Time to first fall was analyzed using cause-specific Cox regression from discharge to the first fall, with censoring at loss to follow-up or end of follow-up; death was treated as a competing event. Absolute fall risk over time by vitamin D category was displayed using competing-risk cumulative incidence curves. Fine–Gray subdistribution hazard models were used as a competing-risk sensitivity analysis.
Recurrent Falls Burden
Recurrent falls were analyzed using negative binomial regression with a log(person-time) offset to estimate incidence rate ratios (IRRs), allowing participants to contribute follow-up time until death or censoring. A prespecified sensitivity analysis used a zero-inflated negative binomial model when excess zeros were present.
Across outcomes, 25(OH)D was modeled primarily as a continuous predictor (per 10 ng/mL lower) and secondarily as a dichotomous exposure (<20 vs ≥20 ng/mL). Adjusted effect estimates are reported with 95% confidence intervals (CIs) and Wald-test p values. To limit false-positive findings across the prespecified outcome set and the two exposure parameterizations, FDR-adjusted q values were computed across 10 hypothesis tests (five outcomes × two exposure specifications). The mRS shift endpoint was prespecified as the primary inference; FDR results are presented as a complementary approach to address multiplicity across the broader outcome set.
Missing Data, Sensitivity Analyses, and Mediation Analysis
Analyses were performed using participants with available outcome data for each endpoint; outcomes were not imputed. Missing covariates were handled using multiple imputation by chained equations under a missing-at-random assumption. The imputation model included the exposure, outcome indicators, and all covariates used in the analytic models, and estimates were pooled using Rubin’s rules (m = 10).
Prespecified sensitivity analyses included complete-case analysis; exclusion of participants with marked inflammation (CRP >50 mg/L); additional adjustment for care-pathway variables (discharge anticoagulation and rehabilitation intensity); Fine–Gray subdistribution hazard modeling for time to first fall; and alternative count-model specifications for falls burden when excess zeros were present.
The exploratory mediation analysis examined whether depressive symptom severity at 1 month partially mediated the association between baseline 25(OH)D and subsequent falls using a counterfactual framework with bootstrap confidence intervals. A 1-month landmark approach was applied, excluding participants who experienced a fall or died before the 1-month PHQ-9 assessment to preserve temporal ordering.
Results
Participant Flow, Vitamin D Distribution, and Baseline Characteristics
Among 802 participants with baseline 25(OH)D measured within 48 hours of admission, 3-month (post-discharge) functional outcomes (mRS and BI) were available for 776 (96.8%), 3-month PHQ-9 for 762 (95.0%), and falls follow-up data for 742 (92.5%). The falls follow-up cohort included 61 deaths (8.2%), and 60 participants were lost to follow-up for falls outcomes (Figure 1). The overall cohort had a median age of 76.2 years (IQR, 71.6–80.6) and included 379 women (47.3%) (Table 1). Median baseline 25(OH)D was 18.1 ng/mL (IQR, 12.6–24.7), and 487 participants (60.7%) had low/inadequate vitamin D status (<20 ng/mL).
Figure 1.
Participant flow, outcome availability, and falls frequency. The cohort included 802 older adults (≥65 years) with ECG-documented AF and imaging-confirmed acute ischemic stroke with baseline serum 25(OH)D measured within 48 hours. Three-month functional outcomes were available for 776 participants (mRS and Barthel Index), and three-month PHQ-9 data were available for 762 participants. Falls follow-up data were available for 742 participants (including 61 deaths); 60 participants were lost to follow-up for falls outcomes. Among participants with falls follow-up data, 327 (44.1%) reported ≥1 fall and 191 (25.7%) reported recurrent falls (≥2).
Table 1.
Baseline Characteristics by Vitamin D Status
| Characteristic | Overall (N=802) | Low 25(OH)D <20 (n=487) | Higher 25(OH)D ≥20 (n=315) | p Value | FDR q |
|---|---|---|---|---|---|
| Age (years), median [IQR] | 76.2 [71.6–80.6] | 77.3 [72.5–81.5] | 74.7 [70.2–78.8] | <0.001 | <0.001 |
| Female, n (%) | 379 (47.3) | 250 (51.3) | 129 (41.0) | 0.005 | 0.011 |
| Serum 25(OH)D (ng/mL), median [IQR] | 18.1 [12.6–24.7] | 13.7 [10.8–17.1] | 27.6 [23.3–33.4] | – | – |
| Season of blood draw, n (%) | <0.001 | 0.001 | |||
| Winter | 214 (26.7) | 150 (30.8) | 64 (20.3) | ||
| Spring | 187 (23.3) | 113 (23.2) | 74 (23.5) | ||
| Summer | 201 (25.1) | 100 (20.5) | 101 (32.1) | ||
| Autumn | 200 (24.9) | 124 (25.5) | 76 (24.1) | ||
| NIHSS on admission, median [IQR] | 8 [4–12] | 8 [5–12] | 6 [4–10] | <0.001 | <0.001 |
| Pre-stroke mRS, n (%) | 0.050 | 0.084 | |||
| 0–1 | 606 (75.6) | 354 (72.7) | 252 (80.0) | ||
| 2 | 108 (13.5) | 71 (14.6) | 37 (11.7) | ||
| 3–5 | 88 (11.0) | 62 (12.7) | 26 (8.3) | ||
| Albumin (g/L), mean ± SD | 39.0 ± 3.8 | 38.5 ± 3.9 | 39.8 ± 3.5 | <0.001 | <0.001 |
| C-reactive protein (mg/L), median [IQR] | 7.2 [4.1–12.4] | 8.1 [4.7–13.6] | 6.2 [3.5–10.5] | <0.001 | <0.001 |
| eGFR (mL/min/1.73m2), mean ± SD | 63.7 ± 18.2 | 61.3 ± 17.4 | 67.5 ± 18.9 | <0.001 | <0.001 |
| Hypertension, n (%) | 603 (75.2) | 374 (76.8) | 229 (72.7) | 0.219 | 0.274 |
| Diabetes, n (%) | 227 (28.3) | 144 (29.6) | 83 (26.3) | 0.364 | 0.390 |
| Heart failure, n (%) | 208 (25.9) | 136 (27.9) | 72 (22.9) | 0.129 | 0.176 |
| Coronary artery disease, n (%) | 254 (31.7) | 155 (31.8) | 99 (31.4) | 0.967 | 0.967 |
| Prior stroke/TIA, n (%) | 175 (21.8) | 113 (23.2) | 62 (19.7) | 0.275 | 0.317 |
| Anticoagulation at discharge, n (%) | 489 (61.0) | 280 (57.5) | 209 (66.3) | 0.015 | 0.028 |
| High rehabilitation intensity, n (%) | 371 (46.3) | 212 (43.5) | 159 (50.5) | 0.064 | 0.096 |
Notes: Values are median [IQR], mean ± SD, or n (%), as appropriate. P values compare vitamin D categories; FDR q values were calculated using the Benjamini-Hochberg procedure across prespecified baseline comparisons.
Abbreviations: 25(OH)D, 25-hydroxyvitamin D; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; FDR, false discovery rate; IQR, interquartile range; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; SD, standard deviation; TIA, transient ischemic attack.
Compared with participants with 25(OH)D ≥20 ng/mL, those with 25(OH)D <20 ng/mL were older (median, 77.3 vs 74.7 years), more often female (51.3% vs 41.0%), and had more severe strokes (NIHSS median, 8 vs 6) (Table 1). The low vitamin D group also had lower serum albumin, higher CRP, and lower eGFR. Oral anticoagulation at discharge was less frequent among participants with low vitamin D status (57.5% vs 66.3%) (Table 1).
Functional Recovery at 3 Months
At 3 months, 411 of 776 participants (53.0%) had an unfavorable functional outcome, descriptively defined as mRS >2, with a higher proportion among those with low vitamin D status than among those with higher vitamin D status (57.3% [270/471] vs 46.2% [141/305]) (Table 2).
Table 2.
Outcomes by Vitamin D Status and Adjusted Associations
| Outcome (Timepoint) | N Analyzed | Low <20 | High ≥20 | Adjusted Effect (Continuous, per 10 ng/mL Lower) | 95% CI | p | FDR q | Adjusted Effect (Categorical) | 95% CI | p | FDR q |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 3-month disability (mRS shift; mRS>2 descriptive) | 776 | 270/471 (57.3) | 141/305 (46.2) | Common OR 1.17 | 1.05–1.31 | 0.005 | 0.008 | Common OR 1.31 | 1.09–1.57 | 0.004 | 0.006 |
| Barthel Index at 3 months (0–100) | 776 | 75.0 ± 22.0 | 81.1 ± 21.0 | β −3.9 | −6.2 to −1.6 | <0.001 | 0.002 | β −6.1 | −9.4 to −2.8 | <0.001 | 0.002 |
| Clinically significant depressive symptoms (PHQ-9 ≥10) at 3 months | 762 | 88/463 (19.0) | 40/299 (13.4) | OR 1.30 | 1.06–1.58 | 0.010 | 0.011 | OR 1.54 | 1.07–2.22 | 0.020 | 0.020 |
| First fall within 12 months (cause-specific HR; death competing event) | 742 | 213/451 (47.2) | 114/291 (39.2) | HR 1.17 | 1.07–1.28 | <0.001 | 0.002 | HR 1.29 | 1.07–1.55 | 0.007 | 0.009 |
| Total falls rate within 12 months (total events; negative binomial IRR) | 742 | 100.1/100 PY (433 falls/432.5 PY) | 75.3/100 PY (210/279.0 PY) | IRR 1.19 | 1.08–1.31 | <0.001 | 0.002 | IRR 1.33 | 1.10–1.61 | 0.003 | 0.006 |
Notes: Continuous effects are reported per 10 ng/mL lower 25(OH)D; categorical effects compare <20 vs ≥20 ng/mL. Adjusted models included age, sex, season of blood draw, NIHSS, pre-stroke mRS, albumin, CRP, eGFR, hypertension, diabetes, heart failure, coronary artery disease, and prior stroke/TIA.
Abbreviations: BI, Barthel Index; CI, confidence interval; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; FDR, false discovery rate; HR, hazard ratio; IRR, incidence rate ratio; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; PHQ-9, Patient Health Questionnaire-9; PY, person-years; TIA, transient ischemic attack.
In the prespecified multivariable mRS shift analysis across the full ordinal scale, lower baseline 25(OH)D was associated with greater disability (adjusted common OR per 10 ng/mL lower 25(OH)D, 1.17; 95% CI, 1.05–1.31; p=0.005; FDR q=0.008). The proportional-odds assumption was not rejected for the primary ordinal model (global score/Brant-type test p=0.41). Low vitamin D status (<20 vs ≥20 ng/mL) was similarly associated with a worse mRS shift (adjusted common OR, 1.31; 95% CI, 1.09–1.57; p=0.004; q=0.006) (Table 2 and Figure 2).
Figure 2.
Adjusted associations of low vitamin D status with key outcomes. Forest plot of adjusted effect estimates comparing low 25(OH)D (<20 ng/mL) versus higher 25(OH)D (≥20 ng/mL). Effect estimates are displayed on a logarithmic scale with the null at 1.0. Models adjust for baseline covariates: age, sex, season of blood draw, NIHSS, pre-stroke mRS, albumin, CRP, eGFR, and comorbidities (hypertension, diabetes, heart failure, coronary artery disease, and prior stroke/TIA). Care-pathway variables (discharge anticoagulation and rehabilitation intensity) were evaluated in sensitivity analyses.
The mean BI at 3 months was lower in the low vitamin D group (75.0±22.0 vs 81.1±21.0). In adjusted analyses, each 10 ng/mL lower baseline 25(OH)D was associated with a 3.9-point lower BI (β, −3.9; 95% CI, −6.2 to −1.6), and low vitamin D status was associated with a 6.1-point lower BI (β, −6.1; 95% CI, −9.4 to −2.8) (Table 2). Because a published minimal clinically important difference for the Barthel Index in stroke patients is 1.85 points on the 0–20 version (approximately 9.25 points when converted to a 0–100 scale), the adjusted 6.1-point categorical difference should be interpreted as statistically significant but modest and not necessarily reaching a conservative clinical-importance threshold.16 Restricted cubic spline analyses suggested a steeper increase in the odds of worse disability at lower 25(OH)D concentrations, with relative flattening around and above the 20 ng/mL reference level (Figure 3).
Figure 3.
Restricted cubic spline association between baseline 25(OH)D and 3-month disability (mRS shift). Adjusted common odds ratio for worse 3-month mRS (shift analysis) across baseline 25(OH)D concentrations, modeled using restricted cubic splines with 25(OH)D = 20 ng/mL as the reference. The curve is adjusted for baseline covariates as specified in the primary mRS model.
Depressive Symptoms at 3 Months
Clinically significant depressive symptoms (PHQ-9 ≥10) were present in 128 of 762 participants (16.8%) at 3 months and were more common among those with low vitamin D status than among those with higher vitamin D status (19.0% [88/463] vs 13.4% [40/299]) (Table 2). In adjusted models, lower baseline 25(OH)D was associated with higher odds of clinically significant depressive symptoms (adjusted OR per 10 ng/mL lower 25(OH)D, 1.30; 95% CI, 1.06–1.58; p=0.010; FDR q=0.011). The categorical comparison showed a concordant association (adjusted OR for <20 vs ≥20 ng/mL, 1.54; 95% CI, 1.07–2.22; p=0.020; q=0.020) (Table 2 and Figure 2).
Falls and Mortality Over 12 Months
Among the 742 participants with falls follow-up data, 61 deaths occurred (8.2%). Overall, 327 participants (44.1%) reported at least one fall, and 191 (25.7%) were recurrent fallers (≥2 falls). The proportion experiencing at least one fall was higher among participants with low vitamin D status than among those with higher vitamin D status (47.2% [213/451] vs 39.2% [114/291]) (Table 2). Consistent with these descriptive findings, unadjusted competing-risk cumulative incidence curves showed a higher absolute risk of first fall over time in the low vitamin D group (Figure 4).
Figure 4.
Cumulative incidence of first fall by baseline vitamin D status. Unadjusted competing-risk cumulative incidence functions for first fall within 12 months after discharge by baseline vitamin D status (<20 vs ≥20 ng/mL). Death is treated as a competing event.
In cause-specific Cox models, lower baseline 25(OH)D was associated with a higher hazard of first fall (adjusted HR per 10 ng/mL lower 25(OH)D, 1.17; 95% CI, 1.07–1.28; p<0.001; FDR q=0.002). Low vitamin D status was likewise associated with a higher hazard of first fall (adjusted HR for <20 vs ≥20 ng/mL, 1.29; 95% CI, 1.07–1.55; p=0.007; q=0.009) (Table 2 and Figure 2).
For fall burden allowing recurrent events, the crude fall rate was higher in the low vitamin D group than in the higher vitamin D group (100.1 per 100 person-years [433 falls/432.5 PY] vs 75.3 per 100 person-years [210 falls/279.0 PY]). In adjusted negative binomial models, lower baseline 25(OH)D was associated with a higher fall rate (IRR per 10 ng/mL lower, 1.19; 95% CI, 1.08–1.31; p<0.001; q=0.002), and low vitamin D status was associated with a higher total-fall rate (IRR, 1.33; 95% CI, 1.10–1.61; p=0.003; q=0.006) (Table 2).
Sensitivity Analyses
Across prespecified sensitivity analyses, the direction and magnitude of associations were generally consistent with the primary models (Table 3). For disability, the association between low vitamin D status and a worse mRS shift was similar in complete-case analyses, after excluding participants with marked inflammation (CRP >50 mg/L), and after additional adjustment for care-pathway variables (anticoagulation at discharge and rehabilitation intensity).
Table 3.
Sensitivity Analyses for Key Outcomes (Low vs Higher Vitamin D)
| Outcome | Analysis | Effect (Low vs High) | 95% CI | P Value |
|---|---|---|---|---|
| 3-month disability (mRS shift) | Primary adjusted model | 1.31 | 1.09–1.57 | 0.004 |
| 3-month disability (mRS shift) | Complete-case analysis | 1.29 | 1.06–1.57 | 0.011 |
| 3-month disability (mRS shift) | Exclude CRP >50 mg/L | 1.27 | 1.04–1.55 | 0.019 |
| 3-month disability (mRS shift) | Care-adjusted (add anticoagulation + rehab) | 1.28 | 1.06–1.55 | 0.011 |
| 3-month depressive symptoms (PHQ-9 ≥10) | Primary adjusted model | 1.54 | 1.07–2.22 | 0.020 |
| 3-month depressive symptoms (PHQ-9 ≥10) | Complete-case analysis | 1.49 | 1.02–2.18 | 0.040 |
| 3-month depressive symptoms (PHQ-9 ≥10) | Exclude CRP >50 mg/L | 1.46 | 1.00–2.14 | 0.051 |
| 3-month depressive symptoms (PHQ-9 ≥9) | Sensitivity endpoint (lower threshold) | 1.41 | 1.05–1.90 | 0.023 |
| 12-month first fall (cause-specific) | Primary adjusted model | 1.29 | 1.07–1.55 | 0.007 |
| 12-month first fall (cause-specific) | Complete-case analysis | 1.27 | 1.05–1.54 | 0.014 |
| 12-month first fall (cause-specific) | Exclude falls/death before 1 month | 1.24 | 1.02–1.50 | 0.029 |
| 12-month first fall (competing risk) | Fine–Gray subdistribution model | 1.26 | 1.04–1.52 | 0.019 |
| 12-month total falls rate | Primary negative binomial model | 1.33 | 1.10–1.61 | 0.003 |
| 12-month total falls rate | Complete-case analysis | 1.31 | 1.08–1.59 | 0.007 |
| 12-month total falls rate | Zero-inflated NB (sensitivity) | 1.30 | 1.07–1.58 | 0.009 |
Notes: Effects are adjusted estimates comparing low versus higher vitamin D status (<20 vs ≥20 ng/mL). Complete-case analyses excluded participants with missing covariate data.
Abbreviations: CI, confidence interval; CRP, C-reactive protein; mRS, modified Rankin Scale; NB, negative binomial; PHQ-9, Patient Health Questionnaire-9.
For depressive symptoms, results were consistent in complete-case analyses and when applying the more inclusive sensitivity definition of PHQ-9 ≥9. Excluding participants with CRP >50 mg/L attenuated the association, yielding borderline statistical significance (Table 3).
For first fall, findings were consistent in complete-case models, in analyses excluding participants who experienced a fall or died before 1 month, and in competing-risk sensitivity analyses using a Fine–Gray subdistribution hazard model (sHR, 1.26; 95% CI, 1.04–1.52; p=0.019). Associations with fall burden remained robust across alternative count-model specifications (Table 3).
Exploratory Mediation Analysis
In an exploratory counterfactual mediation analysis within the 1-month landmark cohort, depressive symptom severity at 1 month mediated a modest proportion of the association between lower baseline 25(OH)D and subsequent first fall during months 1–12 (Table 4). The total effect was HR 1.17 per 10 ng/mL lower 25(OH)D (95% CI, 1.07–1.28). The natural direct effect was HR 1.15 (95% CI, 1.05–1.26), and the natural indirect effect through 1-month PHQ-9 was HR 1.02 (95% CI, 1.01–1.04), corresponding to an estimated proportion mediated of 14.0% (Table 4).
Table 4.
Exploratory Mediation Analysis (1-Month PHQ-9 Mediating Vitamin D to Subsequent Falls)
| Effect | Measure | Estimate | 95% CI | p Value |
|---|---|---|---|---|
| Total effect on subsequent falls (1–12 months) | HR per 10 ng/mL lower 25(OH)D | 1.17 | 1.07–1.28 | <0.001 |
| Natural direct effect (not via 1-month PHQ-9) | HR per 10 ng/mL lower 25(OH)D | 1.15 | 1.05–1.26 | 0.003 |
| Natural indirect effect (mediated via 1-month PHQ-9) | HR per 10 ng/mL lower 25(OH)D | 1.02 | 1.01–1.04 | 0.008 |
| Proportion mediated via 1-month PHQ-9 | % | 14.0 | 6.0–24.0 | 0.003 |
Notes: Mediation analyses used the 1-month landmark cohort and excluded participants who experienced a fall or died before the 1-month PHQ-9 assessment.
Abbreviations: CI, confidence interval; HR, hazard ratio; PHQ-9, Patient Health Questionnaire-9.
Discussion
In this single-center prospective cohort of 802 older adults with AF–associated acute ischemic stroke, low baseline vitamin D status was common and was consistently associated with worse outcomes spanning functional recovery, depressive symptom burden, and falls. Lower baseline 25(OH)D was associated with greater 3-month disability across the full ordinal mRS distribution and with poorer activities of daily living at 3 months. Clinically significant depressive symptoms at 3 months were more prevalent among participants with low vitamin D status. Over 12 months, falls were frequent, and low vitamin D status was associated with a higher risk of first fall and a greater burden of total falls. These associations were broadly robust across prespecified sensitivity analyses, supporting the interpretation of 25(OH)D as a pragmatic risk marker for multiple post-stroke outcomes in this high-risk population.
Our findings on functional recovery are consistent with prior observational studies in acute ischemic stroke showing that admission 25(OH)D is independently associated with short-term disability and mortality.17 The present study extends this literature by focusing specifically on older adults with AF-associated acute ischemic stroke—a subgroup characterized by high baseline vulnerability and complex rehabilitation and secondary prevention needs—and by jointly evaluating disability, mood, and falls within a single cohort. However, recent Mendelian randomization analyses have not supported a causal relationship between genetically predicted 25(OH)D and post-stroke functional outcomes,18 and interventional evidence remains mixed.19,20 Taken together, the overall evidence supports interpreting admission 25(OH)D primarily as a tool for risk stratification in AF-associated acute ischemic stroke, rather than as evidence of a treatment effect from vitamin D supplementation. The spline finding further suggests that risk stratification may be most efficient when screening focuses on patients with clearly low 25(OH)D (<20 ng/mL), rather than assuming a uniform incremental risk across higher concentrations.
The observed burden of post-stroke depressive symptoms is consistent with contemporary guidance and epidemiologic data indicating that post-stroke depression is common, clinically consequential, and warrants systematic screening and longitudinal follow-up after stroke.21,22 Prior cohort studies have linked early low vitamin D status to subsequent post-stroke depression,23 and a meta-analysis has likewise supported an association between low circulating vitamin D and increased risk of post-stroke depression.24 Mechanistically, vitamin D may plausibly influence neuroinflammation and neurotransmission pathways relevant to mood regulation; however, a balanced interpretation is particularly important in older adults with AF-associated acute ischemic stroke, because 25(OH)D may also serve as a marker of shared upstream factors, including frailty, reduced outdoor activity, comorbidity burden, kidney dysfunction, and poorer nutritional and inflammatory status. In our cohort, lower 25(OH)D clustered with lower serum albumin and higher CRP, supporting this “shared pathway” explanation. The mediation analysis was consistent with this caution: 1-month PHQ-9 explained only 14.0% of the association between lower 25(OH)D and later first fall, suggesting that mood-independent pathways such as neuromuscular impairment, frailty, mobility limitation, and environmental exposure may account for a larger share of fall risk.
Falls were common after AF-associated acute ischemic stroke in this older cohort, closely mirroring prospective community-based stroke data such as the FREESE cohort25 and aligning with meta-analytic evidence that post-stroke falls are multifactorial, with strong contributions from balance and mobility impairment, self-care disability, cognitive impairment, depression, and medication effects.26 The observed association between low 25(OH)D and higher fall risk is biologically plausible, potentially reflecting impaired neuromuscular function that may exacerbate stroke-related gait and balance deficits. However, supplementation data caution against inferring a direct preventive benefit from observational associations, as high-dose bolus vitamin D regimens have been associated with increased falls in older adults.27 Accordingly, these data cannot be used to infer that vitamin D supplementation will prevent falls or improve post-stroke mobility in this population.
Accordingly, the clinical value of 25(OH)D in older adults with AF-associated acute ischemic stroke is best framed within integrated geriatric–stroke care rather than as a stand-alone intervention target. Key components directly supported by the present findings include: (1) early nutritional assessment and dysphagia-aware medical nutrition therapy;28 (2) rehabilitation consultation with fall-risk evaluation and mobility support;29,30 (3) systematic screening and management of post-stroke depressive symptoms.21,22 Vitamin D supplementation decisions should be individualized and should align with guideline indications.31 Standard evidence-based secondary prevention for AF-associated stroke should proceed according to guideline indications.1,2,7 This focused pathway positions vitamin D status as a practical marker to help prioritize supportive, multidisciplinary interventions, while maintaining appropriate caution regarding causality and underscoring the need for external validation and targeted trials in AF-associated acute ischemic stroke.
Several limitations warrant cautious interpretation. First, although we adjusted for major prognostic factors in acute ischemic stroke and atrial fibrillation including age, sex, seasonality, stroke severity (NIHSS), pre-stroke disability (pre-stroke mRS), inflammatory/nutritional markers (CRP and albumin), renal function (eGFR), and comorbidities, residual confounding remains possible because key determinants of both vitamin D status and post-stroke outcomes were not measured or not available in a standardized format. These include prior falls and osteoporosis/fracture history, baseline gait/balance performance, muscle strength and sarcopenia, cognition/dementia and visual impairment, BMI and lifestyle factors (smoking and alcohol use), sunlight exposure/outdoor activity, vitamin D supplementation (pre-admission or post-discharge), socioeconomic conditions and home-environment hazards, and medication profiles that materially influence fall and mood risk (eg, sedatives/hypnotics, antidepressants, antipsychotics, and blood-pressure–lowering agents). The absence of these standardized variables means that residual confounding, particularly for falls and depressive symptoms, cannot be excluded. Second, the single-center design may limit generalizability to other populations and care systems, and the acute-phase response may affect measured 25(OH)D despite adjustment for CRP and albumin. Third, falls ascertainment may be subject to reporting and recall bias, and we did not serially measure 25(OH)D, precluding evaluation of time-varying exposure and potential changes during rehabilitation. Collectively, these limitations support framing 25(OH)D as a pragmatic risk marker that may capture a broader vulnerability phenotype (frailty, reduced mobility, and limited outdoor activity) rather than implying a treatment effect. Future work should therefore prioritize multicenter validation in older adults with nonvalvular atrial fibrillation–associated acute ischemic stroke, incorporate standardized assessments of premorbid falls risk, physical performance, cognition and vision, medication exposures, and supplementation behaviors, and evaluate whether targeted correction of clearly low vitamin D within integrated geriatric-stroke pathways (nutrition support, depression care, and multifactorial fall prevention) improves patient-centered outcomes in pragmatic trials.
Conclusion
Lower baseline 25(OH)D was consistently associated with poorer functional recovery, a greater burden of depressive symptoms, and higher risk of first fall and total falls burden after AF-associated acute ischemic stroke in older adults. Because low 25(OH)D is closely linked to frailty, limited outdoor activity, mobility impairment, and comorbidity burden, part of the observed associations, particularly for falls and depressive symptoms may reflect underlying vulnerability rather than a direct biological effect of vitamin D. They should not be interpreted as evidence that vitamin D correction alone improves post-stroke recovery or prevents falls. These findings support vitamin D status as a pragmatic risk marker within integrated geriatric–stroke care. Multicenter validation and targeted trials are needed to determine whether correction of clearly low vitamin D within structured care pathways can improve patient-centered outcomes.
Funding Statement
There is no funding to report.
Data Sharing Statement
Data sets generated during the current study are available from the corresponding author on reasonable request.
Ethics Approval and Consent to Participate
The protocol conformed to the Declaration of Helsinki and was approved by the ethics committee of Tianjin Huanhu Hospital (LL2021051206). All participants provided written informed consent prior to any study procedures.
Disclosure
The authors have no conflicts of interest to declare for this work.
References
- 1.Joglar JA, Chung MK, Armbruster AL. et al. 2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation: a Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2024;149(1):e1–14. doi: 10.1161/CIR.0000000000001193 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Kleindorfer DO, Towfighi A, Chaturvedi S, et al. 2021 Guideline for the Prevention of Stroke in Patients With Stroke and Transient Ischemic Attack: a Guideline From the American Heart Association/American Stroke Association. Stroke. 2021;52(7):e364–e467. doi: 10.1161/STR.0000000000000375 [DOI] [PubMed] [Google Scholar]
- 3.Pierre-Louis IC, Saczynski JS, Lopez-Pintado S, et al. Characteristics associated with poor atrial fibrillation-related quality of life in adults with atrial fibrillation. J Cardiovasc Med. 2023;24(7):422–429. doi: 10.2459/jcm.0000000000001479 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Shah SJ, Fang MC, Jeon SY, Gregorich SE, Covinsky KE. Geriatric syndromes and atrial fibrillation: prevalence and association with anticoagulant use in a national cohort of older Americans. J Am Geriatr Soc. 2021;69(2):349–356. doi: 10.1111/jgs.16822 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.O’Caoimh R, Morrison L, Hanley M, McManus C, Donlon K, Galvin P. Impact of frailty on healthcare outcomes after cardioembolic ischaemic stroke due to atrial fibrillation. Int J Environ Res Public Health. 2024;21(3):270. doi: 10.3390/ijerph21030270 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wagner G, Smeikal M, Gisinger C, et al. Epidemiology, risk profile, management, and outcome in geriatric patients with atrial fibrillation in two long-term care hospitals. Sci Rep. 2022;12(1):Article18725. doi: 10.1038/s41598-022-22013-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.ESC Scientific Document Group; Hindricks G, Potpara T, Dagres N, et al. 2020 ESC Guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS). Eur Heart J. 2021;42(5):373–498. doi: 10.1093/eurheartj/ehaa612 [DOI] [PubMed] [Google Scholar]
- 8.Meshkin A, Badiee F, Salari N, et al. The Global Prevalence of Vitamin D Deficiency in the Elderly: a Meta-analysis. Indian J Orthop. 2024;58(3):223–230. doi: 10.1007/s43465-023-01089-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Williams J, Sink KM, Tooze JA, et al. Low 25-hydroxyvitamin D concentrations predict incident depression in well-functioning older adults: the Health, Aging, and Body Composition Study. J Gerontol a Biol Sci Med Sci. 2015;70(6):757–763. doi: 10.1093/gerona/glu184 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Eyles DW. Vitamin D: brain and Behavior. JBMR Plus. 2020;5(1):e10419. doi: 10.1002/jbm4.10419 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Appel LJ, Michos ED, Mitchell CM, et al. The Effects of Four Doses of Vitamin D Supplements on Falls in Older Adults: a Response-Adaptive, Randomized Clinical Trial. Ann Intern Med. 2021;174(2):145–156. doi: 10.7326/M20-3812 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bischoff-Ferrari HA, Dawson-Hughes B, Orav EJ, et al. Monthly High-Dose Vitamin D Treatment for the Prevention of Functional Decline: a Randomized Clinical Trial. JAMA Intern Med. 2016;176(2):175–183. doi: 10.1001/jamainternmed.2015.7148 [DOI] [PubMed] [Google Scholar]
- 13.Lelli D, Pérez Bazán LM, Calle Egusquiza A, et al. 25(OH) vitamin D and functional outcomes in older adults admitted to rehabilitation units: the SAFARI study. Osteoporosis Int. 2019;30(4):887–895. doi: 10.1007/s00198-019-04845-7 [DOI] [PubMed] [Google Scholar]
- 14.Hannemann A, Wallaschofski H, Nauck M, et al. Vitamin D and health care costs: results from two independent population-based cohort studies. Clin Nutr. 2018;37(6 Pt A):2149–2155. doi: 10.1016/j.clnu.2017.10.014 [DOI] [PubMed] [Google Scholar]
- 15.Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med. 2001;16(9):606–613. doi: 10.1046/j.1525-1497.2001.016009606.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hsieh YW, Wang CH, Wu SC, Chen PC, Sheu CF, Hsieh CL. Establishing the minimal clinically important difference of the Barthel Index in stroke patients. Neurorehabil Neural Repair. 2007;21(3):233–238. doi: 10.1177/1545968306294729 [DOI] [PubMed] [Google Scholar]
- 17.Tu W-J, Zhao S-J, Xu D-J, Chen H. Serum 25-hydroxyvitamin D predicts the short-term outcomes of Chinese patients with acute ischaemic stroke. Clin Sci. 2014;126(5):339–346. doi: 10.1042/cs20130284 [DOI] [PubMed] [Google Scholar]
- 18.Wu Y, Jin T, Pang Q, Cheng Y. Genetic association between serum 25-hydroxyvitamin D levels and functional outcome after ischemic stroke. BMC Neurol. 2024;24(1):Article467. doi: 10.1186/s12883-024-03972-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Momosaki R, Abo M, Urashima M. Vitamin D supplementation and post-stroke rehabilitation: a randomized, double-blind, placebo-controlled trial. Nutrients. 2019;11(6):Article1295. doi: 10.3390/nu11061295 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Fleet JL, McIntyre A, Janzen S, et al. A systematic review examining the effect of vitamin D supplementation on functional outcomes post-stroke. Clin rehabilitat. 2023;37(11):1451–1466. doi: 10.1177/02692155231174599 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.American Heart Association Stroke Council, Council on Cardiovascular and Stroke Nursing, and Council on Quality of Care and Outcomes Research; Towfighi A, Ovbiagele B, El Husseini N, et al. Poststroke depression: a scientific statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2017;48(2):e30–e43. doi: 10.1161/str.0000000000000113 [DOI] [PubMed] [Google Scholar]
- 22.Liu L, Xu M, Marshall IJ, Wolfe CD, Wang Y, O’Connell MD. Prevalence and natural history of depression after stroke: a systematic review and meta-analysis of observational studies. PLoS Med. 2023;20(3):e1004200. doi: 10.1371/journal.pmed.1004200 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Han B, Lyu Y, Sun H, Wei Y, He J. Low serum levels of vitamin D are associated with post-stroke depression. Eur J Neurol. 2015;22(9):1269–1274. doi: 10.1111/ene.12607 [DOI] [PubMed] [Google Scholar]
- 24.Hung K-C, Wu J-Y, Illias AM, et al. Association of a low vitamin D status with risk of post-stroke depression: a meta-analysis and systematic review. Frontiers in Nutrition. 2023;10:Article1142035. doi: 10.3389/fnut.2023.1142035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Walsh ME, Galvin R, Williams DJ, et al. Falls-related events in the first year after stroke in Ireland: results of the multi-centre prospective FREESE cohort study. Eur Stroke J. 2018;3(3):246–253. doi: 10.1177/2396987318764961 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Xu T, Clemson L, O’Loughlin K, Lannin NA, Dean C, Koh G. Risk factors for falls in community stroke survivors: a systematic review and meta-analysis. Arch Phys Med Rehabil. 2018;99(3):563–573.e5. doi: 10.1016/j.apmr.2017.06.032 [DOI] [PubMed] [Google Scholar]
- 27.Sanders KM, Stuart AL, Williamson EJ, et al. Annual high-dose oral vitamin D and falls and fractures in older women: a randomized controlled trial. JAMA. 2010;303(18):1815–1822. doi: 10.1001/jama.2010.594 [DOI] [PubMed] [Google Scholar]
- 28.Burgos R, Bretón I, Cereda E, et al. ESPEN guideline clinical nutrition in neurology. Clin Nutr. 2018;37(1):354–396. doi: 10.1016/j.clnu.2017.09.003 [DOI] [PubMed] [Google Scholar]
- 29.American Heart Association Stroke Council, Council on Cardiovascular and Stroke Nursing, Council on Clinical Cardiology, and Council on Quality of Care and Outcomes Research; Winstein CJ, Stein J, Arena R, et al. Guidelines for adult stroke rehabilitation and recovery: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2016;47(6):e98–e169. doi: 10.1161/str.0000000000000098 [DOI] [PubMed] [Google Scholar]
- 30.Montero-Odasso M, van der Velde N, Martin FC, et al. World guidelines for falls prevention and management for older adults: a global initiative. Age Ageing. 2022;51(9):afac205. doi: 10.1093/ageing/afac205 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Demay MB, Pittas AG, Bikle DD, et al. Vitamin D for the prevention of disease: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2024;109(8):1907–1947. doi: 10.1210/clinem/dgae290 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data sets generated during the current study are available from the corresponding author on reasonable request.