What is metformin, really?
Metformin is the pill most people with type 2 diabetes meet first. In the United States it shows up as Glucophage, as the extended-release Glucophage XR, Glumetza and Fortamet, or as a liquid called Riomet. In the UK the box usually just says metformin hydrochloride. The molecule inside every one of them is the same.
It belongs to a class called the biguanides, and it is worth sitting with the fact that it is the only member still standing. Its two siblings, phenformin and buformin, were withdrawn in the 1970s because they caused lactic acidosis often enough to kill people. Metformin survived that cull because it behaves quite differently in the body — a distinction we will return to, because the reputation of the dead siblings still follows the survivor around (Bailey, Diabetologia, 2017).
The family history runs further back. The chemistry traces to Galega officinalis — goat's rue, French lilac — used in medieval Europe for exactly the thirst-and-endless-urination pattern we now call diabetes. Metformin reached the clinic in the late 1950s. By the standards of modern pharmacology it is an antique, and it is still first in line.
Here is the sentence to hold onto, because almost every myth about this drug dissolves against it: metformin is not a hormone, and it is not a stimulant. It contains no insulin. It does not lean on your pancreas and ask for more.
How it works — the simple version
Your liver is a sugar factory that never fully closes. Between meals, and especially overnight, it builds glucose from scratch out of amino acids, lactate and glycerol, so your brain has fuel while you sleep. The process is called gluconeogenesis, and in a healthy body it is beautifully regulated: demand goes down, production goes down. In type 2 diabetes that regulation slips, and the liver keeps producing through the night as though you had been fasting for days. This is why so many people with diabetes wake up to a high reading before they have eaten a single thing — the number did not come from dinner, it came from the factory.
Metformin's main job is to talk that factory down.
The way it does it is genuinely elegant. Metformin concentrates inside liver cells and mildly throttles complex I, the first turnstile in the mitochondrial chain that turns food into usable energy. Not blocking it — throttling it. The cell ends up making slightly less ATP, its energy currency, and holding slightly more AMP, the spent version. The ratio between those two is the cell's fuel gauge, and the needle has just moved.
That movement flips a switch called AMPK. Think of AMPK as low-battery mode. When your phone drops into low-power mode it does not shut off; it suspends the expensive background processes and keeps the essentials running. Cells behave the same way. And building glucose from raw materials is one of the most expensive background processes a liver runs. Low-battery mode cancels it.
That is the centrepiece, and the majority of metformin's blood-sugar effect traces back to it. But the modern picture has grown a few more rooms (Foretz et al., Nature Reviews Endocrinology, 2023):
- Muscle and fat listen better. Metformin improves insulin sensitivity in peripheral tissue, so the insulin you already make moves more glucose out of the blood and into cells that can use it.
- The gut slows the delivery. Metformin slows glucose absorption from the intestine, flattening the spike after a meal rather than removing it.
- The microbiome joins in. Much of an oral dose never really leaves the intestinal wall, and the gut turns out to be a far more active site than anyone assumed. Metformin shifts the composition of gut bacteria, and transplant experiments suggest those shifts contribute to the glucose-lowering effect rather than merely accompanying it (Wu et al., Nature Medicine, 2017). Young science, but no longer a footnote.
One consequence deserves its own line. Metformin adjusts a thermostat; it does not pull a lever. It brings down glucose that is inappropriately high without driving normal glucose lower — which is why the drug's most famous absence, hypoglycaemia, is an absence by design.
What else it does to your body, beyond lowering blood sugar
The gut takes the first hit. Nausea, cramping, bloating, a loose and urgent relationship with the bathroom. This is the single most common reason people abandon metformin, and it is usually front-loaded into the first weeks. It is also mechanistically unsurprising: a drug that spends much of its working life in the intestinal wall, changing how glucose is absorbed and which bacteria thrive, is going to be noticed by the intestine. The effect is dose-related and, for most people, temporary. Extended-release formulations were designed for exactly this problem and are a standard move when the immediate-release version is intolerable (ADA, Standards of Care in Diabetes, 2024).
Vitamin B12 quietly drifts down. Metformin interferes with the calcium-dependent uptake of B12 in the last stretch of the small intestine, and over years that shows up in blood work. In the Diabetes Prevention Program Outcomes Study, people on long-term metformin had measurably more B12 deficiency than those on placebo, and the gap widened with duration of use (Aroda et al., JCEM, 2016). ADA recommends periodic B12 measurement in people on long-term metformin, with a lower threshold in anyone who has anaemia or peripheral neuropathy (ADA, 2024). That last clause matters, because B12-deficiency neuropathy and diabetic neuropathy feel identical from the inside — same numbness, same tingling, same feet — and only one of them is fixed by replacing a vitamin. Proton pump inhibitors nudge B12 absorption in the same direction, so anyone taking both has two reasons to check.
Lactic acidosis: the famous one. Here is the actual chain. Metformin shifts some metabolism toward lactate production in the gut, and at the same time reduces the liver's capacity to recycle lactate back into glucose — because recycling lactate is gluconeogenesis, the very process the drug is suppressing. In a body with working kidneys, this is a non-event: the drug is not metabolised at all, it is cleared unchanged by the kidneys, and lactate stays balanced.
The picture changes when clearance fails. In advanced kidney disease the drug accumulates. In decompensated heart failure, shock, sepsis, severe liver disease or heavy alcohol use, the body's ability to clear lactate drops on its own. Stack those and the arithmetic stops working. That is why metformin is contraindicated below a certain level of kidney function and paused during acute illness — not because it poisons anything, but because it depends entirely on the exit door staying open.
How rare is rare? A Cochrane review pooling hundreds of comparative trials found no cases attributable to metformin and no difference in lactate levels compared with other diabetes treatments (Salpeter et al., Cochrane Database of Systematic Reviews, 2010). Real-world estimates land at a handful of cases per 100,000 patient-years — rare enough that it should not shape a healthy person's decision, serious enough that it shapes the rules around imaging, surgery and sick days.
Hypoglycaemia, and why it mostly does not happen. Sulfonylureas tell the pancreas to release insulin whether or not there is glucose to handle. Metformin says nothing to the pancreas, and used on its own carries a very low risk of hypoglycaemia (ADA, 2024). The risk arrives when it is paired with insulin or a secretagogue — and if a beta-blocker is also in the mix, the early warning symptoms of a low can be muffled. A combination worth discussing rather than discovering.
Weight, and the long game. Metformin is weight-neutral to modestly weight-lowering, which sets it apart from much of the diabetes cabinet. And UKPDS 34, the trial that made it first-line, found that in overweight patients metformin reduced diabetes-related endpoints and all-cause mortality compared with conventional management (UKPDS Group, Lancet, 1998). Nearly three decades on, that result is still doing work.
What people usually take with it, and why
Metformin is a foundation, not a whole building. Type 2 diabetes is progressive, and most people eventually need a second agent. What changed in the last decade is how that second agent gets picked. The ADA/EASD consensus report and the current Standards of Care are explicit: the choice is no longer driven by HbA1c alone, but by what else is going on in the body — established cardiovascular disease, heart failure, chronic kidney disease, weight (Davies et al., Diabetologia, 2022; ADA, 2024). Metformin generally stays as the glucose-lowering backbone; the partner is chosen for the organ that needs protecting.
- SGLT-2 inhibitors (empagliflozin, dapagliflozin) make the kidneys spill glucose into the urine. Their real headline is what they do beyond glucose: benefit in heart failure and slowing of kidney disease progression, which is why guidelines recommend them for people with those conditions independently of how close HbA1c already is to target (ADA, 2024). Fixed-dose combinations with metformin exist — Synjardy, Xigduo XR.
- GLP-1 receptor agonists (semaglutide, liraglutide, dulaglutide) imitate a gut hormone: insulin release only when glucose is high, slower stomach emptying, less appetite. Substantial weight effect, plus cardiovascular benefit in established atherosclerotic disease.
- DPP-4 inhibitors, the gliptins (sitagliptin, saxagliptin, linagliptin) protect your own GLP-1 from being broken down. Gentle, weight-neutral, low hypoglycaemia risk, but without the cardiovascular or kidney bonus above. Combination tablets include Janumet and Kombiglyze XR.
- Sulfonylureas (gliclazide, glimepiride, glipizide) are cheap and effective, but they are secretagogues — real hypoglycaemia risk and some weight gain. Guidelines moved them down the list without removing them; cost still matters enormously in much of the world.
Combination tablets are a convenience trade: fewer pills to remember, less freedom to adjust one component without touching the other. Which side of that trade suits a given person is a conversation, not a rule.
Two interaction themes deserve a place in your head. Anything that suddenly hurts kidney function turns a safe drug into an accumulating one: dehydration from vomiting or diarrhoea, iodinated contrast for CT scans, and NSAIDs, especially alongside diuretics or ACE inhibitors in an older person. And heavy alcohol, which impairs lactate clearance and is the classic co-factor in the case reports.
Red flags — when to call a doctor
Most of metformin's day-to-day complaints are annoying, not dangerous. These are the exceptions.
The lactic acidosis picture is an emergency, not a wait-and-see. Unusual muscle aching or pain. Profound weakness out of proportion to anything you did. Trouble breathing, or breathing gone unusually deep and fast. Stomach pain with nausea and vomiting. Feeling cold, especially in hands and feet. A slow or irregular heartbeat, or lightheadedness. Any combination of these in someone taking metformin needs urgent medical assessment. It is rare. It also does not improve on its own.
Sick days. Vomiting, diarrhoea, fever, anything that has you drinking far less than usual — these are the situations where metformin is commonly paused, because dehydration and reduced kidney perfusion are exactly what the drug is not built for. Every prescriber should have given you a sick-day plan. If nobody has, ask for one before you need it.
Before general anaesthesia, and before contrast imaging. A hold is sometimes required and sometimes not. The American College of Radiology's guidance is deliberately graded rather than blanket: with normal kidney function and no acute kidney injury, stopping metformin is generally not required, while in reduced kidney function, acute kidney injury or arterial contrast injection, a temporary hold with function rechecked before restarting is advised (ACR Manual on Contrast Media, 2023). Your radiology department and your prescriber decide this together, from your actual numbers.
Signs pointing at B12. New numbness or tingling in feet or hands, unsteadiness, a sore or unusually smooth tongue, fatigue that does not track with sleep, memory or mood changes. These warrant a blood test rather than a guess — assuming it is "just the diabetes" is how a treatable deficiency gets to run for years.
Gut symptoms that arrive late. The settling-in period is early and improves. New, persistent digestive trouble after months or years of stable treatment is a different signal and deserves a look rather than a shrug.
What people get wrong
"Metformin damages your kidneys." The most durable myth in the class, and it has the causality backwards. Metformin does not injure kidney tissue; it is cleared by the kidneys, unchanged, so failing kidneys let it accumulate. The restriction protects you from the drug piling up, not your kidneys from the drug. A JAMA systematic review pushed the field toward exactly this reframing, and the eligibility threshold was subsequently loosened rather than tightened — from a crude creatinine cut-off to a kidney-function-based one that lets many people with moderate impairment stay on treatment (Inzucchi et al., JAMA, 2014; FDA labelling revision, 2016).
"Metformin causes hypoglycaemia, like all diabetes drugs." It is not a secretagogue. It does not tell the pancreas to release insulin. Used alone, it carries a very low risk of hypoglycaemia — this is one of the reasons it earned and kept its first-line position (ADA, 2024). Where lows do occur, look at the other drug on the list.
"You have to stop it 48 hours before any CT scan." The old blanket rule, and not current for years. The ACR's position is stratified by kidney function and type of contrast administration, and for many people with normal function no interruption is needed at all (ACR, 2023).
"It's really a weight-loss drug." The weight effect is real but modest, and it is a side-benefit rather than the point. Metformin is licensed and studied as a treatment for type 2 diabetes, with prevention of progression from prediabetes as a secondary role. Treating it as a shortcut for weight management without a metabolic indication is using a specific tool for a job it was never sized for.
"It's hard on the liver." The worry got imported from the general anxiety around lactic acidosis, but the evidence leans the other way. Metformin is not a hepatotoxin, and in fatty liver disease — NAFLD, now more often called MASLD — its metabolic effects are studied as potentially favourable rather than harmful. Advanced liver failure is a reason to avoid metformin, but that is about lactate clearance, not about the drug attacking liver cells.
"The newer drugs have replaced it." The GLP-1 agonists and SGLT-2 inhibitors are genuinely important and genuinely better at things metformin cannot do. They have not displaced it. The 2022 ADA/EASD consensus and the 2024 Standards of Care both keep metformin as first-line pharmacotherapy for most people with type 2 diabetes, precisely because it combines decades of outcome data, negligible hypoglycaemia risk, weight neutrality and a price that works everywhere rather than only in wealthy health systems (Davies et al., 2022; ADA, 2024).
And the quietest mistake of all: stopping it because you feel fine. Type 2 diabetes does not hurt while it does damage. Feeling well on metformin is not evidence you no longer need it — more often, it is evidence that it is working.