Metformin Side Effects: How to Reduce Them, and the B12 Problem

How common are metformin’s GI side effects?

Gastrointestinal side effects affect roughly a quarter to nearly half of people who start metformin. A Cochrane review of 41 trials in 4,552 women with PCOS found gastrointestinal side effects in 22–40% of women on metformin versus 10% on placebo — an odds ratio of 4.00 (95% CI 2.63–6.09). Nausea, cramping, bloating and diarrhoea are the specific complaints, and they cluster overwhelmingly in the first weeks of treatment rather than appearing evenly across months of use. This is also the single biggest reason people stop taking the drug before it has had time to do anything measurable, which makes “how do I get through the first month” a more practically useful question than the side-effect statistic on its own.

What actually reduces metformin side effects?

Slow dose titration is the standard, evidence-backed first move, not a folk remedy. A 2024 practical review on optimising metformin therapy states plainly that GI side effects “usually resolve if the dose is carefully titrated, or by switching to the extended-release formulation”, and notes that these effects can appear even after years of otherwise stable therapy, not only at the start. Taking each dose with food is the other standard, low-cost step recommended alongside titration in the same review. Neither claim rests on a single flashy trial — it is the accumulated clinical experience behind current prescribing guidance, and it is also why a prescriber’s response to early GI symptoms is usually “let’s slow down the increase,” not “let’s stop.”

Probiotics have an actual trial base behind them for this specific problem, which is more than most home remedies circulating for metformin nausea can claim. A 2024 meta-analysis pooled 26 RCTs of different metformin add-on therapies — sulfonylureas, glitazones, DPP-IV inhibitors, and probiotics, analysed as separate subgroups rather than one pooled figure. Within that, a subset of 5 trials and 1,325 people testing probiotics specifically found lower risk of diarrhoea (RR 0.37, 95% CI 0.27–0.52), bloating (RR 0.26, 95% CI 0.12–0.60), and constipation (RR 0.56, 95% CI 0.42–0.73) in people taking metformin plus probiotics versus metformin alone — with no significant effect on nausea or vomiting specifically. That is a general type 2 diabetes evidence base rather than a PCOS-specific one, but the mechanism — gut microbiome disruption from metformin’s action in the intestine — does not depend on why someone is taking the drug, so it is reasonable to expect the same benefit either way.

Does extended-release metformin actually cause fewer side effects?

Here the evidence pulls in two different directions depending on how you ask the question, and both directions are worth knowing. Across randomised head-to-head trials at equal doses, a 2021 meta-analysis of 9 RCTs in 2,609 adults with type 2 diabetes found extended-release metformin only reduced dyspepsia (RR 0.58, 95% CI 0.34–0.98), with no significant difference from immediate-release on the other GI symptoms measured. Averaged across an unselected group of patients, switching formulation is a small, narrow benefit — not the dramatic fix its reputation suggests.

But a retrospective cohort told a different story for the specific patients who switch because of GI trouble. Among 205 patients who moved from immediate-release to extended-release after struggling with side effects, GI adverse events fell from 26.34% to 11.71%, and diarrhoea specifically fell from 18.05% to 8.29% after the switch. That is a large, clinically meaningful improvement — in exactly the population most likely to actually make the switch in real life. The honest read is that extended-release is not a guaranteed improvement for the average person starting metformin, but for someone who has already tried immediate-release and struggled with it, the real-world data on switching looks considerably better than the head-to-head trial average alone would predict. Both formulations are metformin; this is a delivery-mechanism decision, not a different drug, and it is worth raising directly if GI symptoms are the reason you are considering stopping.

Metformin and vitamin B12: how big is the risk?

Long-term metformin use measurably lowers vitamin B12, and the effect grows the longer you take it. In the Diabetes Prevention Program Outcomes Study, participants taking metformin 850 mg twice daily had low or borderline-low B12 in 19.1% of cases at five years, versus 9.5% on placebo, and each additional year of use raised the odds of deficiency by 13%. That trial population was adults at risk of type 2 diabetes rather than PCOS specifically, but the mechanism — reduced B12 absorption in the terminal ileum — is a property of the drug itself, not of the condition it is treating, so the finding transfers directly to anyone on long-term metformin regardless of why they started it.

B12 deficiency matters here for a reason beyond fatigue: its symptoms — tiredness, brain fog, tingling in the hands or feet, mood changes — overlap heavily with symptoms already common in PCOS, which makes it easy to attribute a treatable deficiency to “just PCOS” and miss it entirely.

What should B12 monitoring actually look like?

Regular B12 screening for anyone on long-term metformin is the recommendation, not an occasional afterthought. The same 2024 practical review that covers titration and formulation switching states plainly that “vitamin B-12 should be screened regularly in long-time metformin users because metformin may induce clinical vitamin B-12 deficiency”. Neither that review nor the DPPOS trial specifies an exact interval that applies to everyone — what a reasonable cadence looks like depends on your baseline level, your diet, and how long you have already been on the drug — which is precisely why this is a question for your prescriber rather than a fixed rule this article can hand you.

Is metformin’s lactic-acidosis risk something to actually worry about?

Lactic acidosis is the side effect most associated with metformin in the public imagination, and it is also the rarest one by a wide margin. Pooled data across 347 trials and cohort studies found no cases of lactic acidosis in 70,490 patient-years of metformin use, and a JAMA systematic review put the overall incidence at roughly 3 to 10 cases per 100,000 person-years — statistically indistinguishable from the background rate in the general population. The same review found metformin remains reasonable to use in mild-to-moderate kidney impairment with dose adjustment and monitoring, which is the opposite of the blanket “avoid if kidneys are imperfect” rule some patients are given informally. The precautions prescribers apply — checking kidney function, pausing around severe dehydration or contrast imaging — exist to keep this risk close to zero, not because the baseline risk is high.

Rarity is not the same as unrecognisable, and this is the page someone experiencing an actual symptom is most likely to land on — so the statistics above are not the whole answer to “should I worry about this.” Lactic acidosis has a distinct pattern: deep or unusually rapid breathing, severe muscle pain or cramping, severe nausea and vomiting, cold or clammy skin, and an extreme, out-of-proportion weakness. That is different from the routine GI upset covered earlier on this page — ordinary nausea and diarrhoea build up gradually over the first weeks of treatment and do not involve breathing changes, while this does not build gradually and does not pass with food or time. If you notice that combination, especially the breathing change, that is same-day medical attention, not something to mention at your next scheduled appointment.

Alcohol is one of the few genuinely modifiable risk factors here, and it is worth naming directly rather than leaving it implied. Heavy alcohol use impairs the liver’s ability to clear lactate — the same pathway metformin already slows — which is why heavy or binge drinking is among the situations where metformin is typically paused or avoided. That is not a claim that an occasional drink is dangerous for someone with normal kidney function; it is a reason to flag heavy drinking with whoever prescribes the drug, the same way kidney function and contrast scans get flagged.

Signs metformin is not working for PCOS

Metformin is targeting specific, measurable things — insulin resistance, a handful of metabolic markers, and ovulation frequency — and “not working” means those specific markers have not moved, not that you feel different. The 2023 international guideline’s evidence base shows metabolic markers like HOMA-IR typically shift within the first 8 to 12 weeks in trials that found an effect at all, while ovulation-related outcomes generally need 3 to 6 months of continuous use to assess fairly. If a repeat HOMA-IR, fasting insulin, or cycle count at those checkpoints looks unchanged from your starting point, that absence of movement is itself useful information — not proof the drug has failed forever, but a legitimate prompt to review the plan with whoever prescribed it.

For ovulation specifically, the guideline is direct about where metformin sits in the order of options: letrozole, not metformin, is named as the first-line drug for ovulation induction in PCOS. If cycle regularity or ovulation was the goal and months of metformin have not produced it, that is not a personal failure of the drug or of you — it is consistent with what the trial evidence already predicts, and it is the guideline’s own basis for moving to a different first-line option rather than raising the metformin dose indefinitely.

Is metformin safe in pregnancy or while breastfeeding?

This is worth addressing directly rather than leaving it unstated, since restored ovulation is often the point of taking metformin in the first place. Metformin crosses the placenta, but a review of oral antidiabetic agents in pregnancy found no developmental toxicity in infants when metformin was used before and throughout pregnancy in women with PCOS. For breastfeeding, metformin transfers into breast milk at a relative infant dose of roughly 0.3% of the mother’s weight-adjusted dose, well under the 10% level generally used as a threshold of concern. Neither of those findings makes the decision for you — whether to continue, pause, or combine metformin with an ovulation agent around conception, pregnancy, or breastfeeding depends on why you are taking it and what else is going on, which is exactly why it belongs in the conversation with your prescriber, ideally before a positive test rather than after one.

What happens when you stop metformin?

Metformin manages markers; it does not cure the underlying condition, and the clearest evidence for that is what happens when people stop taking it. A PCOS-specific observational study followed 44 women who had previously responded well to metformin after they discontinued it. At 6 months, women who had used it long-term gained weight, had fewer menstrual bleeds in the following 6 months, and showed a borderline rise in androstenedione — but in a genuine surprise against the simple “everything reverts” narrative, HOMA-IR and glucose homeostasis stayed stable in both the long-term and short-term groups over that same 6-month window. In this small study, weight and cycle regularity drifted back faster than the insulin- resistance marker most people assume would be the first thing to relapse.

A separate line of evidence, from metformin’s use in diabetes prevention rather than PCOS, points the same general direction from a different angle. In the Diabetes Prevention Program, participants who stopped metformin for a short washout period saw fasting glucose rise by about 4.5 mg/dL after roughly 11 days off the drug, and the medication’s overall effect on preventing diabetes fell from 31% to 25% once the pharmacological, on-drug component was removed. Roughly a quarter of metformin’s benefit in that trial existed only while the drug was actively in the body — it was not a durable change left behind after stopping. That trial was in adults with prediabetes, not PCOS, but it demonstrates the same underlying principle the PCOS withdrawal study suggests: some of what metformin does is a standing effect of the drug being present, not a lasting correction.

None of this means stopping is wrong, or that everyone who stops will relapse identically — the PCOS study above was small, observational, and followed a group selected for having responded well in the first place, which is not the same as every PCOS patient on the drug. What it does mean is that “should I come off metformin” is a question with a real, individual answer that depends on why you started it and what you are hoping happens next — which makes it a conversation with your prescriber, not a protocol this article, or anyone else online, can hand you.

You may see PCOS referred to as polyendocrine metabolic ovarian syndrome (PMOS) after a 2026 global consensus of more than 50 organisations renamed it. Nothing about the evidence above changes with the name — this article uses PCOS because that is still what people search.

Common questions

If you are weighing a supplement instead of, or alongside, metformin, supplements to avoid with PCOS covers the interactions worth checking first, and cinnamon for PCOS is the closest supplement comparison on insulin resistance — with a far smaller evidence base than metformin’s.

Curious whether metformin fits your specific PCOS pattern in the first place? Take the quiz — and if the side effects above are new to you, what metformin actually does covers the trial evidence behind why a prescriber suggested it.