Vitamin B12 deficiency may occur with metformin treatment, but few studies have assessed risk with long-term use.

To assess the risk of B12 deficiency with metformin use in the Diabetes Prevention Program (DPP)/DPP Outcomes Study (DPPOS).

Secondary analysis from the DPP/DPPOS. Participants were assigned to the placebo group (PLA) (n = 1082) or the metformin group (MET) (n = 1073) for 3.2 years; subjects in the metformin group received open-label metformin for an additional 9 years.

Twenty-seven study centers in the United States.

DPP eligibility criteria were: elevated fasting glucose, impaired glucose tolerance, and overweight/obesity. The analytic population comprised participants with available stored samples. B12 levels were assessed at 5 years (n = 857, n = 858) and 13 years (n = 756, n = 764) in PLA and MET, respectively.

Metformin 850 mg twice daily vs placebo (DPP), and open-label metformin in the metformin group (DPPOS).

B12 deficiency, anemia, and peripheral neuropathy.

Low B12 (≤ 203 pg/mL) occurred more often in MET than PLA at 5 years (4.3 vs 2.3%; P = .02) but not at 13 years (7.4 vs 5.4%; P = .12). Combined low and borderline-low B12 (≤ 298 pg/mL) was more common in MET at 5 years (19.1 vs 9.5%; P < .01) and 13 years (20.3 vs 15.6%; P = .02). Years of metformin use were associated with increased risk of B12 deficiency (odds ratio, B12 deficiency/year metformin use, 1.13; 95% confidence interval, 1.06–1.20). Anemia prevalence was higher in MET, but did not differ by B12 status. Neuropathy prevalence was higher in MET with low B12 levels.

Long-term use of metformin in DPPOS was associated with biochemical B12 deficiency and anemia. Routine testing of vitamin B12 levels in metformin-treated patients should be considered.