Perhaps some elderly falls are due to Diabetes

Type 2 Diabetes and Falls Incidents among the Elderly: An Umbrella

Review of their Associations with Vitamin D Levels and Other Salient Peer-Reviewed Research Observations - 2026

Many things are flaky in the publication: it is not an umbrella review; it says responsibility is shared by all authors, but only one author was listed; many errors in references; only two weeks between submission and acceptance (little time for peer review and editing), etc.

PDF

However, some of the references look promising

Rout et al. 2026, Nutrients: a multi-ancestry Mendelian randomization study finding that T2D causally lowers 25(OH)D (ref 102)
Chen & Feng 2026, Front Endocrinol: lower 25(OH)D predicts diabetic peripheral neuropathy, with a nonlinear link to severity (ref 24)
Aghamohammadzadeh et al. 2026, J Nutr Metab: a triple-blind RCT of high-dose D3 plus flaxseed oil in elderly T2D (ref 67)
Abera et al. 2026, PLoS One: a meta-analysis of vitamin D deficiency in African T2D patients (ref 93)
Ma et al. 2024, Sci Rep: vitamin D level as a predictor of dysmobility syndrome in T2D (ref 49)

Dysmobility Syndrome = have at least 3 of 6 mobility problems

Claude - AI Sept 2026

Dysmobility syndrome is a composite diagnosis that Binkley and colleagues proposed in 2013 as a broader alternative to "sarcopenia." Instead of looking at muscle alone, it scores bone, muscle, fat and physical function together. A person has it if they meet 3 or more of these 6 criteria:

Component	Criterion
Osteoporosis	BMD T-score ≤ −2.5
Low lean mass	Appendicular lean mass/height² below sex-specific cutoffs
Falls	At least one fall in the past year
Slow gait	Walking speed < 1.0 m/s
Weak grip	< 20 kg (women); < 30 kg (men)
High body fat	> 40% (women); > 30% (men)

Why it's studied in type 2 diabetes: T2D tends to push on almost every component at once:

Fat and muscle: Sarcopenic obesity combines high fat with low lean mass.
Gait and falls: Neuropathy slows walking and raises fall risk.
Strength: Glycation and poor muscle quality reduce grip strength.

One nuance matters a lot for T2D. People with T2D break bones more easily, yet they often have normal or even higher BMD, because the problem is bone quality rather than density. So the osteoporosis criterion misses many of them, and dysmobility scoring may underestimate their true fragility.

Why it matters clinically: In general older populations, meeting the criteria is linked to more falls and fractures and to higher mortality. In older T2D patients, Chinese hospital-based studies report it as common. The garbled "50.4%" figure in the Marks review was most likely the dysmobility syndrome rate in the Zhang 2025 sample, not new-onset diabetes.

The vitamin D angle: Low 25(OH)D is associated with several components on its own: weaker muscle, more falls, lower bone density and higher fat mass. Ma et al. 2024 (Sci Rep) found that lower vitamin D predicted dysmobility syndrome in T2D patients. That is an observational association, though, and it can run both ways: the Rout 2026 Mendelian randomization study suggests T2D itself lowers 25(OH)D. As far as I know, no RCT has used dysmobility syndrome as an outcome for vitamin D supplementation. That gap would be worth a "What this does NOT show" note if you build a page around the Ma study.

Interaction between T2D, Falls, and Vitamin D - research of studies

Claude AI - Deep Research Sept 2026

People with type 2 diabetes (T2D) fall more often and break bones more often than people without diabetes, and they often have low vitamin D. But no randomized trial has yet shown that vitamin D supplementation prevents falls in people with T2D, and the general older-adult trials show no fall benefit in vitamin D–replete people and harm from large intermittent (bolus) doses.

TL;DR

T2D → falls is solid: meta-analysis and cohort data put fall risk roughly 19–76% higher in older adults with T2D (Freire 2024 HR 1.63; Rasmussen 2021 HR 1.19 for hospital-treated falls; Lin 2024 HR 1.757), driven mainly by neuropathy, hypoglycemia, insulin use, polypharmacy, frailty and poor vision, with fracture risk raised despite normal or high bone density.
T2D ↔ low vitamin D is mostly one-way: the newest multi-ancestry Mendelian randomization (Rout 2026) suggests T2D genetic liability lowers 25(OH)D (β −9.1 nmol/L per SD of T2D polygenic risk) while genetically higher vitamin D does not lower T2D risk; yet pooled RCTs in prediabetes (Pittas 2023 IPD) show a modest 12% reduction in progression to diabetes, so the two lines are not fully reconciled.
Vitamin D → fewer falls is unproven in T2D: in general older adults, daily 800–1,000 IU may modestly reduce falls mainly in people who are deficient (<50 nmol/L), VITAL (2,000 IU/d) and the USPSTF review show no benefit in replete community-dwellers, and bolus regimens (500,000 IU/yr; 60,000 IU/month) and ≥2,000 IU/d in STURDY signalled more falls; in T2D/prediabetes the only fall-outcome trial evidence is a post-hoc analysis of an active vitamin D analogue (eldecalcitol, DPVD: HR 0.78).

Practical bottom line

For readers with T2D: Fall prevention in diabetes is primarily about avoiding hypoglycemia, reviewing sedating and blood-pressure drugs, foot and eye care, treating neuropathy, and strength-and-balance exercise. Correcting genuine vitamin D deficiency with modest daily doses is reasonable and low-risk, but it should not be expected to prevent falls by itself.

For clinicians: Check 25(OH)D in older adults with T2D who have neuropathy, falls, sarcopenia, or osteoporosis; replace deficiency with daily dosing (typically 800–2,000 IU/day), avoid large intermittent boluses in fall-prone patients, and pair repletion with exercise and medication de-intensification. Do not use bone density alone to judge fracture risk in T2D.

For researchers: The decisive trial — vitamin D3 in deficient older adults with T2D, daily dosing, falls as primary endpoint, with neuropathy and sarcopenia stratification — has not been done.

Key Findings

T2D roughly doubles some fall-related risks, but most of the excess is explained by complications and treatment. The Freire 2024 BMC Geriatrics meta-analysis (12 studies, 8 pooled) found a 63% higher fall risk in older adults with T2D (HR 1.63; 95% CI 1.30–2.05), with insulin use the most important associated factor.\
A JAMDA meta-analysis (2026;27(3):106012, published online January 2026; 32 studies, 23,666 adults ≥60 with diabetes) found an aggregated fall prevalence of 29.5% (95% CI 21.2–38.6), with hypoglycemia (OR 2.22; 1.63–3.01), diabetic peripheral neuropathy (OR 2.15; 1.35–3.42) and impaired cognition (OR 1.77) as key risk factors.
Tight HbA1c control on insulin or sulfonylureas was not, by itself, linked to more serious falls. In Shabnam 2025 (Diabetes Care; 21,365 UK adults ≥70), sustained HbA1c <7% on insulin or sulfonylureas gave HR 1.04 (0.96–1.11) for fall hospitalization and 1.07 (0.97–1.17) for fractures;[2] severe hypoglycemia itself remained associated with falls.[3] The message is "avoid hypoglycemia," not "avoid glucose control."
T2D appears to lower vitamin D more than low vitamin D causes T2D. Rout, Blackett and Sanghera (Nutrients 2026;18(12):1944; UK Biobank n=471,861 plus 3,486 Asian Indians) reported that a vitamin D polygenic score did not affect T2D risk, whereas each SD of T2D polygenic risk reduced 25(OH)D by 9.1 nmol/L.[4] The same group's 2021 bidirectional MR had found the opposite (no effect of T2D liability on 25(OH)D),[5] so this direction should be considered probable rather than settled.
Supplementing vitamin D in prediabetes modestly slows progression to diabetes. The Pittas 2023 IPD meta-analysis of D2d, Tromsø and DPVD (4,190 adults) found HR 0.88 (0.77–0.99), a 3.3% absolute risk reduction over 3 years, and 30% higher likelihood of regression to normal glucose;[6][7] the striking 76% lower risk at intratrial 25(OH)D ≥125 nmol/L is a within-arm observational comparison, not a randomized result.[8] VITAL found no effect on incident T2D in a general population (HR 0.91; 0.76–1.09).[9]
Low 25(OH)D tracks with falls, neuropathy and dysmobility in T2D — but only in observational data. Cross-sectional Chinese studies link low 25(OH)D to dysmobility syndrome (Ma 2024: OR 2.99 for 25(OH)D <36.2 nmol/L; Zhang 2025: ≤43.7 nmol/L an independent factor)[10][11] and to distal symmetric polyneuropathy (Chen & Feng 2026: severity accelerates below ~26.1 ng/mL).[12][13]
General-population trials: dose and schedule matter more than "vitamin D yes/no." Daily 800–1,000 IU in deficient older adults is the only regimen with a consistent signal;[14] large boluses increased falls (Sanders 2010; Bischoff-Ferrari 2016), and STURDY found higher fall-event rates at 2,000–4,000 IU/day than at 1,000 IU/day.[15][16] Replete populations (VITAL) show no benefit.[17]
No RCT of vitamin D in T2D has used falls as a pre-specified primary endpoint. The closest evidence is the DPVD ancillary trial in prediabetes, where the active analogue eldecalcitol 0.75 µg/day reduced incident sarcopenia (HR 0.51; 0.31–0.83) and, post hoc, falls (HR 0.78; 0.62–0.97)[18] — results that cannot be transferred to ordinary vitamin D3.[19]
There is no real evidence that falls cause T2D. The idea that injury-related immobility, inactivity and muscle loss could precipitate or worsen T2D is biologically plausible but, in this review, no prospective study testing falls or fractures as an exposure for incident T2D was identified.

Details

1. T2D → falls, injuries and fractures

Magnitude of risk. Three independent data sources agree on direction, with different magnitudes reflecting how falls were counted:

Freire 2024 (BMC Geriatr 24:201, doi 10.1186/s12877-024-04668-0): systematic review and meta-analysis; HR 1.63 (1.30–2.05)[20] for falls in older adults with T2D vs without; insulin use was the strongest associated factor. The authors note a paucity of studies and heterogeneous results.[1]
Rasmussen 2021 (Curr Drug Saf 16(1):52–61, doi 10.2174/1574886315666200908110058): Danish National Patient Register, 407,009 people with T2D (and 12,975 with T1D) each matched 1:1 to controls, 1996–2017; outcome = first fall requiring hospital treatment. Cumulative fall incidence was 11.9% in T2D; adjusted HR 1.19 (1.16–1.22)[21] for T2D and 1.33 (1.25–1.43) for T1D. Fall-related fracture incidence rate ratios in T2D: hip 1.02 (1.01–1.04), radius 1.39 (1.18–1.61), humerus 1.24 (1.12–1.37), skull/face 1.15 (1.07–1.24).[21] One age-subgroup HR in the abstract is internally inconsistent with its CI and is likely a typographical error.
Lin 2024 (BMC Geriatr 24:717, doi 10.1186/s12877-024-05314-5): Guangzhou prospective cohort of community-dwelling adults ≥65 (6,169 analysed; 1,970 with T2D), median follow-up 2.42 years; falls occurred in 588 with T2D vs 767 without (about 29.8% vs 18.3%); adjusted HR 1.757 (1.577–1.957).[22] Daily exercise was protective (HR 0.793; 0.686–0.918).[22]
JAMDA 2026 meta-analysis (27(3):106012; 32 studies, 23,666 adults ≥60): aggregated fall prevalence 29.5% (95% CI 21.2–38.6; recurrent falls 16.6%) in older adults with diabetes; hypoglycemia OR 2.22, peripheral neuropathy OR 2.15, impaired cognition OR 1.77.

Interpretation: registry data capturing only hospital-treated falls (Rasmussen) show smaller relative risks than self-reported fall cohorts (Lin), which is expected; the consistent 20–75% excess is real but heavily mediated by complications and frailty.

Mechanisms (strength of evidence varies):

Mechanism	How it raises falls/injury	Evidence quality
Peripheral neuropathy / proprioception loss	Impaired foot sensation, ankle strategy, postural sway	Meta-analytic association (OR 2.15)
Hypoglycemia (insulin, sulfonylureas)	Dizziness, confusion, syncope	Meta-analytic association (OR 2.22); Shabnam 2025: severe hypoglycemia linked, tight HbA1c alone not
Insulin use	Marker of longer, more complicated disease plus hypoglycemia	Freire 2024: strongest factor
Polypharmacy, orthostatic hypotension	Antihypertensives, sedatives, diuretics	Cohort/cross-sectional data
Retinopathy / vision loss	Missed hazards, depth perception	Plausible; cited by Freire 2024
Sarcopenia / sarcopenic obesity	Weakness, slow gait, poor recovery from trips	Cross-sectional; part of dysmobility
Vestibular dysfunction	Diabetes-associated vestibular impairment	Limited observational data
Foot ulcers, offloading boots, amputation	Altered gait and base of support	Clinical observation; limited quantitative data
Fear of falling / kinesiophobia	Activity restriction → deconditioning	Limited data in T2D specifically
Cognitive impairment	Poorer hazard judgment, medication errors	Meta-analytic association (OR 1.77)

The T2D bone-quality paradox. People with T2D tend to have normal or higher areal bone mineral density (BMD) yet higher risk of several fractures. Proposed explanations include accumulation of advanced glycation end-products in collagen, low bone turnover, cortical porosity, and the higher fall rate itself. The practical consequence is that DXA T-scores underestimate fracture risk in T2D, so falls and bone fragility have to be assessed together rather than relying on BMD.

Dysmobility syndrome. Binkley and colleagues proposed in 2013 a "dysmobility syndrome" analogous to the metabolic syndrome, diagnosed when ≥3 of six components are present: osteoporosis, low lean (muscle) mass, low grip strength, slow gait speed, a fall in the past year, and high body fat.[12] The concept is attractive for T2D because it captures sarcopenic obesity and fall history together, but it is a research construct rather than a validated clinical diagnosis.

2. T2D ↔ low vitamin D

T2D → lower 25(OH)D (Mendelian randomization).

Rout, Blackett, Sanghera 2026 (Nutrients 18(12):1944, doi 10.3390/nu18121944): genome-wide univariate and polygenic-score bidirectional MR in UK Biobank (471,861, multi-ethnic) and the Asian Indian Diabetic Heart Study/Sikh Diabetes Study (3,486). A polygenic score of vitamin D-raising alleles did not affect T2D or cardiovascular risk. Each SD increase in T2D polygenic risk predicted 25(OH)D lower by 9.1 nmol/L (reported 95% CI −8.9 to −9.3; p = 9.5 × 10⁻³¹), consistent across ancestries. Of note, the instrument for T2D explained a very small share of covariance (0.049%), and the printed CI is unusually narrow.[4][23]
Earlier conflicting result from the same group: a 2021 bidirectional MR in South Asians and Europeans (Nutr J, doi 10.1186/s12937-021-00725-1) concluded that genetically instrumented T2D risk did not change 25(OH)D.[5] The 2026 study's much larger sample likely explains the difference, but readers should know the finding is not yet replicated by independent groups.
Plausible mechanisms (not proven): sequestration of vitamin D in adipose tissue, reduced outdoor activity, kidney disease and other diabetes complications.[4] Because obesity is genetically correlated with T2D, horizontal pleiotropy through adiposity is a real concern for the T2D → vitamin D estimate.

Low vitamin D → T2D (reverse direction).

D2d (US): 2,423 adults with prediabetes, vitamin D3 4,000 IU/day vs placebo, median ~2.5 years, mean baseline 25(OH)D 28 ng/mL; the reduction in diabetes incidence was not statistically significant.[24]
Pittas 2023 IPD meta-analysis (Ann Intern Med 176:355–363, doi 10.7326/M22-3018): D2d + Tromsø (n=511, weekly cholecalciferol) + Japanese DPVD (eldecalcitol); 4,190 participants; mean baseline 25(OH)D 63 nmol/L. Vitamin D reduced new-onset diabetes by 12% (HR 0.88; 0.77–0.99), 3-year absolute risk reduction 3.3%, and increased regression to normal glucose regulation by 30%.[7] Among cholecalciferol-treated participants, intratrial 25(OH)D ≥125 nmol/L vs 50–74 nmol/L was associated with HR 0.24 (0.16–0.36) — a post-randomization comparison open to confounding by adherence and adiposity.[8]
VITAL (general population): vitamin D3 2,000 IU/day did not reduce incident T2D (HR 0.91; 0.76–1.09; median 5.3 years).[9]
Synthesis: the MR evidence argues against a large causal effect of lifelong vitamin D status on T2D, while trials suggest a modest pharmacological benefit in high-risk prediabetes. These can coexist (for example, effects confined to deficient or high-risk groups, or MR instruments that capture only small differences in 25(OH)D).

3. Low vitamin D → falls

Observational evidence.

Sim 2021 (J Am Geriatr Soc 69(11):3114–3123, doi 10.1111/jgs.17442): community-dwelling older Australian women followed 14.5 years; the highest 25(OH)D category had lower hazard of falls-related hospitalization than the lowest (HR 0.76; 0.61–0.95), with a graded spline relationship; the authors suggest ≥75 nmol/L may be beneficial.[25] Prospective, but residual confounding by health, activity and outdoor time cannot be excluded.
Not all cohorts agree: the Japanese Murakami cohort found low plasma 25(OH)D was not associated with recurrent falls in middle-aged and older adults.[26]

Randomized trials and meta-analyses (general older adults).

Bischoff-Ferrari 2009 meta-analysis (BMJ): supplemental vitamin D 700–1,000 IU/day reduced falls by 19% (7 RCTs, n=1,921; RR 0.81, 95% CI 0.71–0.92); lower doses did not. This older analysis predates the large null trials.
Sanders 2010 (JAMA 303(18):1815–22): 2,256 community-dwelling women ≥70, a single annual oral dose of 500,000 IU cholecalciferol for 3–5 years: falls increased (RR 1.15; 1.02–1.30) and fractures increased (RR 1.26; 1.00–1.59), with the fall incidence RR 1.31 in the first 3 months after each dose versus 1.13 during the following 9 months (P = .02); in total there were 2,892 vs 2,512 falls (83.4 vs 72.7 per 100 person-years).
Bischoff-Ferrari 2016 (JAMA Intern Med 176(2):175–83): 200 adults ≥70 with a prior fall, 12 months: 24,000 IU/month vs 60,000 IU/month vs 24,000 IU + 300 µg calcifediol/month. The higher-dose arms more often reached ≥30 ng/mL but had no functional benefit and more falls (66.9% and 66.1% vs 47.9% with 24,000 IU).[16][27]
VITAL falls ancillary (LeBoff 2020, J Clin Endocrinol Metab 105(9):2929–38, doi 10.1210/clinem/dgaa311): 25,871 adults (men ≥50, women ≥55; mean age 67.1), vitamin D3 2,000 IU/day vs placebo for median 5.3 years, baseline 25(OH)D 77 nmol/L. No reduction in ≥2 falls (OR 0.97; 0.90–1.05), injurious falls/doctor visits (OR 1.03; 0.94–1.13) or falls requiring hospital visits (OR 1.04; 0.90–1.19). The population was not selected for deficiency.[17][28][29]
DO-HEALTH (Bischoff-Ferrari, JAMA 2020): 2,157 European adults ≥70, of whom 40.7% had vitamin D deficiency (25(OH)D <20 ng/mL) and 41.9% had fallen in the year before enrollment, received vitamin D3 2,000 IU/day, omega-3 and/or a home exercise programme for 3 years; vitamin D did not significantly reduce falls.
STURDY (Appel 2021, Ann Intern Med 174:145–156, doi 10.7326/M20-3812): 688 adults ≥70 with elevated fall risk and 25(OH)D 10–29 ng/mL (25–72.5 nmol/L); daily vitamin D3 200 (control), 1,000, 2,000 or 4,000 IU. In the dose-finding stage, the 2,000 and 4,000 IU/day arms had higher fall/death event rates than 1,000 IU/day (HR 1.86; 1.16–2.97 and 1.68; 1.05–2.69); 1,000 IU/day then failed to prevent falls vs 200 IU/day (HR 0.94; 0.76–1.15; P = 0.54), with the higher-dose groups showing more serious falls (HR 1.87; 1.03–3.41) and falls with hospitalization (HR 2.48; 1.13–5.46) than 200 IU/day, and the trial stopped for futility. A secondary analysis suggested ≥1,000 IU/day may increase first falls with fracture (flagged by the authors as possible type 1 error), and achieved 25(OH)D was not associated with falls.
Bolland, Grey & Avenell 2018 (Lancet Diabetes Endocrinol): meta-analysis and trial sequential analysis concluding that vitamin D does not prevent fractures or falls or meaningfully affect BMD, with no difference between higher and lower doses.[30]
Ling 2021 (Clin Nutr 40:5531–7): updated meta-analysis concluding vitamin D supplementation reduces fall risk specifically in vitamin D–deficient older adults.[31]
Kong 2022 (Endocrinol Metab): only daily 800–1,000 IU was associated with fewer falls (about 19% lower risk); intermittent dosing showed no benefit (RR 1.01; 0.94–1.09).[32][33]
2024 network meta-analysis (BMC Geriatr, doi 10.1186/s12877-024-05009-x): 35 RCTs, 58,937 older adults; 800–1,000 IU/day reduced falls (RR 0.85; 0.74–0.95), daily dosing RR 0.78, and benefit in those with baseline 25(OH)D ≤50 nmol/L (RR 0.69) but not >50 nmol/L; >1,000 IU/day increased falls relative to 800–1,000 IU/day.[14][34]
2025 community-dwelling meta-analysis (J Clin Med, doi 10.3390/jcm14176117): no effect in adults ≥65 living in the community (≤1,000 IU/day OR 0.96; 0.90–1.02; >1,000 IU/day OR 1.02; 0.96–1.09).[35]
USPSTF (draft statement, 17 December 2024): with moderate certainty, vitamin D (± calcium) has no net benefit for primary prevention of falls or fractures in community-dwelling postmenopausal women and men ≥60 (D recommendation); pooled RR for ≥1 fall 0.99 (0.97–1.01; 8 RCTs; 36,744 participants followed 9 months to 5.3 years), or 5 fewer fallers per 1,000 supplemented, and rate-of-falls IRR 0.98 (0.94–1.03; 4 RCTs; 28,519 participants). It does not apply to people with known deficiency, osteoporosis, or those in institutions.[36][37]

Muscle function trials.

Bislev 2022 IPD meta-analysis (J Clin Endocrinol Metab 107(5):1317–27, doi 10.1210/clinem/dgac004): 4 RCTs, n=260, daily vitamin D3 2,800–7,000 IU for 3–9 months; no effect on isometric strength, Timed Up and Go, chair rising, body composition or balance, including in those <50 nmol/L (n=176); a possible strength benefit in severely obese individuals.[38] A companion 2021 meta-analysis found a tendency toward worse SPPB scores with vitamin D (MD −0.18; −0.37 to 0.01).[39]
Ozsoy-Unubol 2021 (Int J Clin Pract 75(12):e14851): single-blind RCT in postmenopausal women 50–70 with severe deficiency (<25 nmol/L); vitamin D 50,000 IU/week for 8 weeks (n=21), core and balance exercise (n=18), or both (n=20). Vitamin D improved balance and quality of life; exercise was essential for fall-risk reduction; neither was superior to the other and the combination was not clearly additive.[40][41] Small, short, surrogate outcomes.

What the pattern means: benefit, where seen, is confined to deficient people taking moderate daily doses; replete people gain nothing; and large intermittent doses or ≥2,000 IU/day in frail fallers carry a signal of harm. This is the most important practical lesson for T2D, where intermittent high-dose prescribing is common.

4. Vitamin D, diabetic neuropathy and muscle in T2D

Chen & Feng 2026 (Front Endocrinol, doi 10.3389/fendo.2026.1890018): patients with distal symmetric polyneuropathy (DSPN) had lower total 25(OH)D (23.95 ± 5.14 vs 32.31 ± 5.97 ng/mL), driven by 25(OH)D3 (25(OH)D2 did not differ).[42] DSPN risk fell linearly with rising 25(OH)D, while neuropathy severity (MNSI-PE) was nonlinear, with a turning point at about 26.1 ng/mL below which severity accelerated.[13] Observational and single-centre; the threshold is data-derived and needs external validation.
Sun 2024 (Front Endocrinol, doi 10.3389/fendo.2024.1354511): among T2D inpatients undergoing nerve conduction, 1,620 (44.6%) had subclinical neuropathy; only the DSPN subtype had lower 25(OH)D (36.18 vs 41.03 nmol/L), with a negative linear association.[43]
Yao 2026 (J Int Med Res) on vitamin D and diabetic peripheral neuropathic pain: could not be verified in detail for this review; treat any specific figures attributed to it with caution until checked against the primary source.
Neuropathy RCTs: a multicentre Chinese placebo-controlled RCT of high-dose intramuscular vitamin D2 for T2D with DSPN and vitamin D insufficiency has a published protocol (Front Endocrinol 2023, doi 10.3389/fendo.2023.1202917; ChiCTR2200062266);[44] results were not located. No adequately powered RCT has shown vitamin D improves neuropathy-related balance or falls.
Muscle RCTs in T2D (small, mixed):
- Cavalcante 2015: 38 postmenopausal women with T2D, 25(OH)D <30 ng/mL and weak grip; 6,600 IU/week (~942 IU/day) for 3 months; 25(OH)D rose only to 22.98 ng/mL (nonsignificant); handgrip improved within the vitamin D group with no effect on glycemia.[45]
- Chinese 4-arm trial (Front Physiol 2022, doi 10.3389/fphys.2022.1003572): 61 T2D patients (mean age 50), 1,000 IU/day ± endurance exercise for 12 weeks; vitamin D added no benefit on strength or physical performance beyond exercise.[46]
- Korean 2026 trial in elderly women with T2D and deficiency: 12 weeks of resistance training plus vitamin D; strength gains were mainly exercise-driven, with a significant exercise × vitamin D interaction only for functional reach (a balance measure).[47]

Bottom line for this link: low vitamin D is a consistent marker of neuropathy in T2D, but whether it contributes causally — or simply reflects longer, worse-controlled, more obese, less mobile disease — is unknown.

5. Vitamin D and dysmobility syndrome in T2D

Ma 2024 (Sci Rep 14:19792, doi 10.1038/s41598-024-70400-y): cross-sectional, 330 hospitalized T2D patients (mean age 67.0) in Qinhuangdao, China. Dysmobility syndrome prevalence 25.5%; vitamin D deficiency 67.9%. 25(OH)D was lower in DMS (41.74 ± 14.60 vs 47.19 ± 13.01 nmol/L). Adjusted ORs: 25(OH)D <36.2 nmol/L (lowest quartile) 2.990 (1.284–6.964); age 1.160 per year; HbA1c 1.262 per %.[12] The population had poorly controlled diabetes; the authors explicitly state whether supplementation reduces DMS needs confirmation.[10][48]
Zhang 2025 (Diabetes Metab Syndr Obes 18:2815–2823, doi 10.2147/DMSO.S518190): cross-sectional, 262 T2D patients >60 years. 25(OH)D fell across groups with increasing DMS components (47.15 → 43.30 → 39.66 nmol/L; p=0.004); 25(OH)D ≤43.7 nmol/L (printed as "mmol/L," an evident unit typo), age ≥70 and prior falls were independent risk factors.[11][49] Note that falls are both a DMS component and a "risk factor" here — a circularity that inflates associations.
Interpretation: two small cross-sectional single-country studies, both in hospital/clinic populations, both adjusting for limited confounders (not physical activity, sun exposure or BMI in all models). They show co-occurrence, not that vitamin D prevents dysmobility.

6. Vitamin D RCTs in T2D/prediabetes with falls, fracture or function outcomes

Trial	Population	Intervention	Function/fall outcome	Result
DPVD ancillary (Kawahara 2024, Lancet Healthy Longev)	1,094 Japanese adults with prediabetes, no sarcopenia; baseline 25(OH)D 20.7 ng/mL	Eldecalcitol 0.75 µg/day (active analogue) vs placebo, median 2.9 y	Sarcopenia (primary); falls (post hoc); grip	Sarcopenia 4.6% vs 8.8% (HR 0.51; 0.31–0.83); falls 24.6% vs 32.8% (HR 0.78; 0.62–0.97); grip +1.85% vs +0.45%[18]
D2d	2,423 US adults with prediabetes	D3 4,000 IU/day, ~2.5 y	Falls/physical function not a reported main outcome	No published falls result located
VITAL diabetes subgroup	Adults with diabetes within VITAL	D3 2,000 IU/day, 5.3 y	Falls by diabetes status	No diabetes-specific falls result located
Aghamohammadzadeh 2026 (J Nutr Metab)	104 Iranian T2D patients ≥65 (102 completed)[50]	D3 50,000 IU every 2 weeks + flaxseed oil 1,500 mg/day, 8 weeks	None measured (SF-36 only)	hs-CRP −0.98 µg/mL;[51] achieved 25(OH)D not measured[51]
Cavalcante 2015	38 postmenopausal women with T2D	D3 6,600 IU/week, 3 months	Handgrip	Within-group grip gain; achieved 22.98 ng/mL
Front Physiol 2022	61 Chinese T2D adults	D3 1,000 IU/day ± exercise, 12 weeks	Strength, performance	No added benefit of vitamin D

Explicit answer: no randomized trial of ordinary vitamin D (D3 or D2) in people with T2D has used falls or fractures as a pre-specified primary outcome. The DPVD falls finding comes from an active vitamin D analogue (which did not change 25(OH)D and lowered 1,25(OH)₂D by 21.4%), in prediabetes rather than T2D, was defined after study completion, did not record fall severity or fractures, and was not adjusted for multiple testing.[18][52]

On Aghamohammadzadeh 2026: this triple-blind trial gave 50,000 IU D3 fortnightly (about 200,000 IU over 8 weeks) together with flaxseed oil, so vitamin D's effect cannot be isolated; it enrolled people already at 30–125 nmol/L, did not measure post-intervention 25(OH)D, measured no falls, strength or gait outcomes, and had striking baseline imbalances (for example, COPD 73.1% vs 2.0% as printed, and SF-36 physical function medians 87.5 vs 45.0).[51] It is irrelevant to falls and should be cited only for its inflammatory endpoint.

7. Do falls or injuries precipitate T2D?

No prospective study treating falls, fall injuries or fractures as an exposure for incident T2D, or for worsening glycemic control, was identified in this review. The hypothesis is plausible — immobility and inactivity reduce muscle glucose uptake, and loss of muscle mass is linked to insulin resistance — and short-term bed-rest physiology studies support the general principle that inactivity worsens insulin sensitivity. But going from that to "a fall causes diabetes" is speculation. Any observational association would be heavily confounded, because frailty, obesity and undiagnosed dysglycemia precede both. The claim should be presented on VitaminDWiki as a hypothesis, not a finding.

8. Other relevant considerations

Abera 2026 (PLoS One) meta-analysis of vitamin D deficiency prevalence and genetic associations in African T2D: could not be verified in detail for this review; no figures are reported here.
Vitamin D receptor (VDR) in muscle: VDR is expressed in skeletal muscle and mediates effects on muscle cell proliferation and function in experimental studies; this is mechanistic support only and does not translate into the null muscle-function trials summarized above.
GLP-1 receptor agonists and muscle: rapid weight loss with GLP-1 drugs includes lean-mass loss; in older adults with T2D this could theoretically worsen sarcopenia and fall risk, which argues for pairing these drugs with resistance training and adequate protein. Fall-outcome data for this specific concern are limited.
Vitamin D + exercise: across Ozsoy-Unubol 2021, the Korean 2026 trial and the Chinese 2022 trial, exercise drives most functional benefit; vitamin D adds little or nothing measurable except possibly in severe deficiency.[40][53]
Magnesium: magnesium is a cofactor for the hydroxylation steps that activate vitamin D, and hypomagnesemia is common in T2D (urinary losses, metformin and proton-pump inhibitor use). This is a reasonable rationale for checking magnesium in deficient T2D patients who respond poorly to vitamin D, but no trial has tested magnesium plus vitamin D for falls in T2D.
Low-credibility source: the 2026 narrative "umbrella review" by Ray Marks in the Journal of Orthopaedic Science and Research (Athenaeum Scientific Publishers, doi 10.46889/JOSR.2026.7302) was excluded as low-credibility and is not relied on here.

Summary evidence table

Study	Design	N / population	Vitamin D form, dose, schedule	Baseline → achieved 25(OH)D	Outcome	Result
Freire 2024	Meta-analysis	12 studies (8 pooled); older adults	—	—	Falls, T2D vs no T2D	HR 1.63 (1.30–2.05)
Rasmussen 2021	Registry cohort	407,009 T2D + matched controls; Denmark	—	—	Hospital-treated falls	HR 1.19 (1.16–1.22)
Lin 2024	Prospective cohort	6,169 adults ≥65; Guangzhou	—	—	Falls	HR 1.757 (1.577–1.957)
Shabnam 2025	Cohort	21,365 T2D ≥70; UK	—	—	Falls, fractures with HbA1c <7% on insulin/SU	HR 1.04 (0.96–1.11); 1.07 (0.97–1.17)
Rout 2026	Mendelian randomization	471,861 UK Biobank + 3,486 Asian Indians	—	—	25(OH)D per SD T2D PRS	β −9.1 nmol/L; vitamin D PGS no effect on T2D[4]
Pittas 2023	IPD meta-analysis of 3 RCTs	4,190 with prediabetes	D3 4,000 IU/d; D3 20,000 IU/wk; eldecalcitol	Mean 63 nmol/L → not pooled	New-onset T2D	HR 0.88 (0.77–0.99)
VITAL T2D	RCT ancillary	General US population	D3 2,000 IU/d, 5.3 y	—	Incident T2D	HR 0.91 (0.76–1.09)[9]
Sim 2021	Prospective cohort	Older Australian women, 14.5 y	—	Categories of plasma 25(OH)D	Falls hospitalization	HR 0.76 (0.61–0.95) highest vs lowest
Sanders 2010	RCT	2,256 women ≥70	D3 500,000 IU once yearly	Not reported here	Falls; fractures	RR 1.15 (1.02–1.30); RR 1.26 (1.00–1.59) — harm
Bischoff-Ferrari 2016	RCT	200 adults ≥70 with prior fall	D3 24,000 vs 60,000 IU/month vs 24,000 IU + calcifediol	Higher arms reached ≥30 ng/mL	Falls incidence	66.9% and 66.1% vs 47.9% — harm at higher doses
VITAL falls	RCT	25,871 adults, mean 67 y	D3 2,000 IU/d, 5.3 y	77 nmol/L → Not reported here	≥2 falls	OR 0.97 (0.90–1.05) — null
DO-HEALTH	RCT (2×2×2)	2,157 adults ≥70, Europe	D3 2,000 IU/d, 3 y	Not reported here	Falls	No significant reduction
STURDY	Adaptive RCT	688 adults ≥70, 25(OH)D 10–29 ng/mL	D3 200/1,000/2,000/4,000 IU/d	Achieved levels not linked to falls	Time to first fall	No benefit; 2,000 and 4,000 IU/d HR 1.86 and 1.68 vs 1,000
BMC Geriatr 2024 NMA	Network meta-analysis	35 RCTs, 58,937	Various	Benefit only if ≤50 nmol/L	Falls	800–1,000 IU/d RR 0.85 (0.74–0.95)
USPSTF 2024 (draft)	Systematic review	Community-dwelling ≥60	Various ± calcium	—	≥1 fall	RR 0.99 (0.97–1.01); D recommendation
Bislev 2022	IPD meta-analysis	4 RCTs, n=260	D3 2,800–7,000 IU/d, 3–9 months	Subgroup <50 nmol/L	Muscle health	Null
Ozsoy-Unubol 2021	RCT	59 postmenopausal women, <25 nmol/L	D 50,000 IU/week, 8 weeks ± exercise	Not reported here	Balance, fall risk	Balance/QoL improved; no superiority vs exercise
Chen & Feng 2026	Observational	T2D patients	—	23.95 vs 32.31 ng/mL (DSPN vs none)	DSPN severity	Turning point ~26.1 ng/mL
Ma 2024	Cross-sectional	330 T2D, mean 67 y	—	41.74 vs 47.19 nmol/L	Dysmobility syndrome	OR 2.99 (1.28–6.96) for <36.2 nmol/L[10]
Zhang 2025	Cross-sectional	262 T2D >60 y	—	47.15 → 39.66 nmol/L across groups	Dysmobility syndrome	≤43.7 nmol/L independent risk factor[11]
DPVD ancillary 2024	RCT	1,094 with prediabetes, Japan	Eldecalcitol 0.75 µg/d, 2.9 y	20.7 ng/mL → unchanged	Sarcopenia; falls (post hoc)	HR 0.51 (0.31–0.83); HR 0.78 (0.62–0.97)[18]
Aghamohammadzadeh 2026	Triple-blind RCT	104 T2D ≥65, Iran	D3 50,000 IU every 2 weeks + flaxseed oil, 8 weeks	30–125 nmol/L entry → not measured	hs-CRP; no falls/function	hs-CRP −0.98 µg/mL

Recommendations

For VitaminDWiki editorial framing:

State clearly that the T2D–falls link is established, the T2D → low vitamin D direction is supported by new MR evidence, and the vitamin D → fewer falls link is unproven in T2D and largely null in replete older adults.
Always pair any fall claim with dose and schedule; flag bolus-dose harm (Sanders 2010; Bischoff-Ferrari 2016) and STURDY's ≥2,000 IU/day signal in frail fallers prominently.
Present DPVD as an analogue/prediabetes finding, not as evidence for over-the-counter vitamin D3.
Present dysmobility and neuropathy associations as markers that justify testing, not as proof that supplementation will help.

For clinical practice (older adults with T2D):

Prioritize proven fall-prevention levers: hypoglycemia avoidance and de-intensification where appropriate, medication review (sedatives, antihypertensives causing orthostasis), foot care and appropriate footwear/offloading, vision checks, and progressive strength-and-balance training.
Measure 25(OH)D in those with falls, neuropathy, sarcopenia, osteoporosis, CKD, or obesity; treat deficiency (especially <50 nmol/L) with daily vitamin D3 rather than large intermittent boluses.
Avoid annual or large monthly boluses in fall-prone patients; there is no evidence that targeting very high 25(OH)D reduces falls.
Assess fracture risk with clinical risk factors and fall history, not BMD alone, given the T2D bone paradox.
In patients losing weight on GLP-1 drugs, add resistance exercise and adequate protein to protect muscle.

What this does NOT show / limitations

Association is not causation. All vitamin D–neuropathy, vitamin D–dysmobility and vitamin D–falls data in T2D are cross-sectional or cohort studies; low 25(OH)D can be a consequence of obesity, inactivity, indoor living, CKD and frailty, all of which independently cause falls (reverse causation and confounding).
MR has its own limits. The Rout 2026 T2D instrument explains a tiny fraction of variance, may act through adiposity (pleiotropy), and conflicts with the same group's 2021 null result. MR estimates lifelong genetic differences, not the effect of supplementing an older adult.
Pittas 2023 is about diabetes incidence, not falls, and its strongest numbers (≥125 nmol/L) are non-randomized.
General-population fall trials may not generalize to T2D. VITAL and DO-HEALTH enrolled mostly replete, healthy people; STURDY enrolled frail fallers with low-normal levels; none reported T2D-specific falls results that were located for this review.
DPVD used eldecalcitol, a potent active analogue unavailable in many countries, that did not raise 25(OH)D; its fall result is post hoc and unadjusted for multiplicity.[18]
Surrogate outcomes dominate the T2D trials (grip, balance scores, inflammatory markers), and trials are small and short.
Some included journals are lower tier (for example, Curr Drug Saf for Rasmussen 2021) and several abstracts contain internal numeric inconsistencies, flagged above.
Unverified items: Yao 2026 and Abera 2026 could not be checked against primary sources in this review; no figures from them are reported.

Research gaps

A falls-endpoint RCT of vitamin D3 in older adults with T2D and 25(OH)D <50 nmol/L, using daily dosing (for example 800–2,000 IU/day), powered for falls and injurious falls, stratified by neuropathy, insulin/sulfonylurea use and sarcopenia.
Factorial trials of vitamin D × resistance/balance exercise in T2D with falls, not just strength, as the outcome.
Adequately powered neuropathy RCTs (the Chinese intramuscular D2 trial results are awaited) measuring balance, gait and falls rather than nerve conduction alone.
Independent replication of T2D → low 25(OH)D MR, using multivariable MR adjusting for BMI.
T2D-specific secondary analyses of existing mega-trials (VITAL diabetes subgroup falls, D2d physical function and fractures), which could be done with existing data.
Prospective studies of falls/fractures as exposures for incident T2D or glycemic deterioration, to test the speculative falls → T2D pathway.
Magnesium–vitamin D interaction trials in T2D with musculoskeletal endpoints.
Fall and muscle outcomes with GLP-1 receptor agonists in older adults with T2D, including whether vitamin D repletion modifies lean-mass loss.

Caveats

This review used a limited search budget; some requested studies (Yao 2026, Abera 2026, Binkley 2013 original, DO-HEALTH and D2d exact secondary figures) were described from secondary sources or not verified, and are labelled accordingly. The USPSTF statement cited is the December 2024 draft; readers should check whether the final statement differs. Numeric data were transcribed from abstracts or full texts where accessible; where abstracts contained internal inconsistencies, these are noted rather than corrected.

Sources

Related in VitaminDWiki

Hypothesis by VitaminDWiki

Infrequent Vitamin D dosing (2+ months) may be the problem
Diabetics feel good for weeks after dosing, but fail to notice the reduction in good feeling
Falls increase as the Vitamin D levels fade due to drug haze, dysmobility, slower muscle reaction time