The study, published in Frontiers in Clinical Diabetes and Healthcare, was conducted at two hospitals in Thai Binh province by a team from Thai Binh University of Medicine and Pharmacy and Hai Phong University of Medicine and Pharmacy. Between January 2024 and September 2025, the researchers recruited 400 adults with type 2 diabetes who had been taking metformin for at least six months. They deliberately excluded anyone whose vitamin B12 status could be compromised by other factors: patients with gastrectomy, ileal resection, Crohn disease, or celiac disease; those taking proton pump inhibitors, H2 receptor antagonists, or colchicine for more than a year; people with pernicious anemia or malignancies; strict vegetarians; and anyone already taking vitamin B12 supplements. The exclusions were designed to isolate the relationship between metformin itself and the vitamin’s availability in the bloodstream.

The numbers tell a layered story. Of the 400 participants, 103, or 25.75 percent, fell below the study-defined threshold of 300 picograms per milliliter. But only 29 of them, 7.25 percent, had absolute deficiency, defined as concentrations under 200 picograms per milliliter. The remaining 74, or 18.50 percent, occupied the marginal band between 200 and 300 picograms per milliliter, a zone whose clinical meaning is far less certain. The authors are careful to note that this serum-based definition includes marginal values and does not, by itself, establish functional deficiency, which would normally require additional biomarkers such as methylmalonic acid or homocysteine that were not measured in this study.

When the researchers examined which clinical characteristics tracked with low vitamin B12, two factors stood out in their exploratory adjusted analysis. Patients who had been taking metformin for more than five years had nearly twice the odds of falling below the threshold compared with those on shorter courses, with an adjusted odds ratio of 1.84 and a 95 percent confidence interval spanning 1.07 to 3.18. Anemia, defined by conventional World Health Organization hemoglobin cutoffs of under 130 grams per liter in men and under 120 grams per liter in women, carried even greater weight, with an adjusted odds ratio of 3.43 and a confidence interval of 1.62 to 7.23. Male sex and self-reported alcohol consumption showed higher proportions of low vitamin B12 in unadjusted comparisons, but neither retained statistical significance in the reported multivariable model.

The duration signal is consistent with what randomized and longitudinal evidence has long suggested. In a placebo-controlled trial published in BMJ in 2010, de Jager and colleagues observed a greater occurrence of biochemical vitamin B12 deficiency among insulin-treated patients with type 2 diabetes during long-term metformin treatment. The Diabetes Prevention Program Outcomes Study, a longitudinal analysis published in 2016, associated longer cumulative metformin exposure with low vitamin B12 status, though that population was originally recruited for diabetes prevention rather than treatment of established disease. The proposed mechanism centers on the gut: metformin appears to impair the calcium-dependent absorption of the vitamin B12-intrinsic factor complex in the ileum. In a small physiological study published in Diabetes Care in 2000, Bauman and colleagues demonstrated that calcium supplementation could reverse metformin-associated malabsorption, providing mechanistic support for the calcium-dependent hypothesis, though the authors of the new study emphasize that this does not constitute evidence for routine calcium supplementation in clinical practice.

The Vietnamese findings sit within a wider international landscape of prevalence estimates that vary with thresholds, supplementation, and sampling. Kim and colleagues reported vitamin B12 deficiency in 22.2 percent of 1,111 metformin-treated patients in Korea using the same threshold of under 300 picograms per milliliter. A multicenter Chinese study by Gao and colleagues evaluated definite and borderline categories separately and found associations with higher daily metformin doses but not with their selected duration categories. Within Vietnam itself, previous studies have reported prevalences of 18.6 percent in 156 patients by Huynh and colleagues and 22.1 percent among 145 patients at a southern tertiary hospital by Phan and colleagues, with the latter identifying adjusted associations with treatment duration, dose, and HbA1c. A study of elderly patients at Friendship Hospital by Vu and colleagues found that nearly 80 percent of participants reported vitamin B12 supplementation, making that population fundamentally incomparable with the supplement-free sample in the new study.

One of the most clinically consequential aspects of the study concerns the relationship between biochemical deficiency and symptoms that patients actually experience. Vitamin B12 deficiency can cause anemia and macrocytosis, an enlargement of red blood cells defined here as a mean corpuscular volume above 100 femtoliters. But macrocytosis was not statistically significant in either the univariable analysis or the reported multivariable model, and its adjusted confidence interval, spanning 0.79 to 79.37, is so wide that the point estimate cannot be reliably interpreted. More troubling is the neurological dimension: vitamin B12 deficiency can produce neuropathic symptoms that overlap with diabetic peripheral neuropathy, making it easy to attribute one condition’s damage to the other. Observational evidence on this overlap is inconsistent. Ahmed and colleagues did not identify a significant association between biochemical deficiency and peripheral neuropathy in their cross-sectional study, whereas Didangelos and colleagues reported improvement in several neurological measures after vitamin B12 supplementation in a randomized trial of selected metformin-treated patients with neuropathy. The authors caution that a low serum result and clinically manifest deficiency should not be treated as interchangeable outcomes.

The alcohol finding adds another layer of biological complexity. In unadjusted comparisons, participants reporting alcohol consumption had a low vitamin B12 proportion of 37.3 percent versus 22.7 percent among those not reporting consumption, with a crude odds ratio of 2.03. But the association attenuated in the adjusted model, and the binary self-reported measure did not characterize quantity, duration, or drinking pattern. Evidence from other populations suggests the relationship between alcohol and measured vitamin B12 status is more complex than a uniform decrease in serum concentration. Gibson and colleagues found that a controlled alcohol intervention reduced serum vitamin B12 in healthy male volunteers, while Fragasso and colleagues described alcohol-dependent patients with megaloblastic anemia who responded to vitamin B12 treatment despite normal or borderline serum concentrations. The authors cannot determine whether alcohol explains the higher unadjusted proportion in male participants or whether sex modifies an alcohol-related association, and no mediation or interaction analyses were presented.

The study’s limitations are substantial and the authors are unusually transparent about them. The cross-sectional design precludes temporal or causal conclusions, and hospital-based convenience sampling limits generalizability. All participants received metformin, so excess risk relative to nonuse could not be estimated. Baseline vitamin B12 values, dietary intake, methylmalonic acid, and homocysteine were unavailable. The reported adjusted model omitted diabetes duration and daily metformin dose, leaving potential residual confounding by treatment history. The macrocytosis estimate was highly imprecise, and no sensitivity analyses were performed to establish the stability of the adjusted estimates under alternative model specifications. The authors explicitly state that the adjusted findings remain exploratory and do not establish causality or model robustness, and that significance for one exposure and nonsignificance for another are not a test of the difference between their effects.

What practical guidance emerges from this evidence? The American Diabetes Association and the UK Medicines and Healthcare products Regulatory Agency already advise consideration of periodic vitamin B12 assessment during long-term metformin treatment, particularly in patients with risk factors or suggestive hematological or neurological findings. The new study’s five-year duration category can serve as a pragmatic prompt to prioritize review of vitamin B12 status, but the authors stress it is not a validated testing threshold and should not be a reason to defer testing earlier when deficiency is suspected. Patients with unexplained anemia or new neuropathic symptoms should be assessed regardless of treatment duration. Importantly, a low vitamin B12 result does not itself require discontinuation of metformin when the drug remains tolerated and otherwise appropriate. The study cannot specify an optimal testing interval or demonstrate a benefit of routine supplementation, and longitudinal research with detailed exposure measurement, functional biomarkers, and documented analytical plans is needed to clarify how metformin’s benefits and its biochemical side effects can be balanced over decades of treatment.

Subject of Research: Vitamin B12 deficiency prevalence and clinical factors among Vietnamese patients with type 2 diabetes receiving metformin therapy

Article Title: Prevalence and associated factors of vitamin B12 deficiency among Vietnamese patients with type 2 diabetes receiving metformin therapy: a cross-sectional study

Article References: Nguyen, T. K. D., Ke, T. L. A., Pham, V. L., Tran, X. T., & Vu, T. B. (2026). Prevalence and associated factors of vitamin B12 deficiency among Vietnamese patients with type 2 diabetes receiving metformin therapy: a cross-sectional study. Frontiers in Clinical Diabetes and Healthcare, 7, Article 1924740. https://doi.org/10.3389/fcdhc.2026.1924740

Image Credits: AI Generated

DOI: 10.3389/fcdhc.2026.1924740

Keywords: metformin, vitamin B12, type 2 diabetes, anemia, Vietnam, cross-sectional study, diabetic neuropathy, Prevalence, associated, factors, vitamin, deficiency