There has been long-standing uncertainty about whether treatment with metformin causes vitamin B12 deficiency and/or anaemia, and therefore whether screening for these complications should occur in the very high proportion of our patients prescribed metformin as first-line therapy for type 2 diabetes. The DPPOS (Diabetes Prevention Program Outcomes Study) demonstrated that metformin use in people with impaired glucose tolerance was associated with anaemia at 5 years, whether or not vitamin B12 deficiency was present (Aroda et al, 2016). A recent meta-analysis confirmed that people with diabetes taking metformin have a significantly greater risk of vitamin B12 deficiency than those not taking it, and have significantly lower B12 levels that vary with the dose and duration of metformin use, suggesting that this is related to metformin (Yang et al, 2019). The authors recommended that further studies should explore the associations between metformin and anaemia, and that, in the meantime, vitamin B12 levels should be measured annually, and treated if deficient, in those prescribed metformin.
The present study used post hoc data from more than 8000 newly or recently diagnosed people from the ADOPT (A Diabetes Outcome Progression Trial) and UKPDS (UK Prospective Diabetes Study) randomised controlled trials to ascertain whether there was an association between metformin and anaemia, and if so, to explore the time frame for anaemia development. Anaemia was defined as the first haemoglobin level after diabetes diagnosis less than 11 g/dL. Women with levels <12 g/dL and men with levels <13 g/dL at diagnosis were deemed to have pre-existing anaemia and were excluded from the analysis. The identified falls in haemoglobin level were small, averaging 0.5 g/dL, and had occurred by the first measurement at 3 years in UKPDS. In ADOPT, they also occurred early, with no additional change after 2 years. The authors state that the early onset of anaemia associated with metformin was not likely to be related to vitamin B12 deficiency, as stores should last 2–5 years and there was no increase in mean corpuscular volume during the ADOPT study. This early reduction in haemoglobin was also seen in those treated with a thiazoledinedione, as described in the Summaries of Product Characteristics for this class of drugs. The authors also examined real-world, routinely collected data from more than 10 000 participants in the GoDARTS (Genetics of Diabetes Audit and Research in Tayside Scotland) study to identify whether cumulative exposure to metformin is associated with increasing anaemia. This analysis demonstrated that, for each 1 g per day of metformin, there was a 2% higher risk of moderate anaemia per year. Moderate anaemia occurred in 41.8% of individuals in GoDARTS, compared with only 3.4% and 2.2% in ADOPT and UKPDS, respectively. This was in line with the higher mean age at diagnosis in GoDARTS and the longer duration of monitoring (up to 20 years). The key limitation of this study was that vitamin B12 levels were not available because, although full blood counts were included in pre-planned monitoring, B12 levels were not routinely measured during these studies as the potential association between metformin and B12 metabolism was unknown at that time. The authors conclude that they would not advocate avoidance or discontinuation of metformin due to anaemia risk, because the benefits of metformin are proven, the effects on anaemia are modest and the mechanisms are unknown. However, prescribers should anticipate a reduction in haemoglobin levels in the first few years after metformin initiation. Click here to read the publication.