Outcome ink, not A1c theater
UKPDS 34 still sits under every 1000 mg split because death and diabetes-related events moved, not because a percentage point of A1c looks pretty on a poster.
Plenty of glucose-lowering drugs can bend a lab. Fewer have a randomised, decade-scale file in newly diagnosed type 2 diabetes that also moved all-cause mortality. That is why this study page starts with UKPDS 34 (Lancet, 12 September 1998) and not with a mechanism cartoon.
The Glucophage 1000 mg tablet is a common US twice-daily strength taken with meals. It is not the exact tablet the 1977-1997 UK centres dispensed. The outcome argument still attaches to the molecule in overweight new type 2, which is why guidelines kept metformin in the first chair long after newer classes arrived for high-risk hearts.
Newer SGLT2 and GLP-1 outcome trials in established atherosclerotic or kidney disease are real and large. They do not delete the UKPDS overweight-metformin arm. They add columns for people whose risk looks different than a 53-year-old newly diagnosed cohort from the 1980s.
EGFR stops, iodinated-contrast holds, and B12 surveillance belong on the formulary Glucophage page. Here we stay with what the outcome trials actually counted.
Prevention ink after UKPDS
The US Diabetes Prevention Program later showed metformin reduced progression from high-risk glycemia to diabetes by about 31 percent, with a stronger mark in younger and more overweight participants. That is a prevention column, not a reprint of UKPDS 34 mortality.
Do not merge those files in a consult. A person with established type 2 and a 1000 mg split is in the outcome chair. A person with prediabetes considering metformin is in the prevention chair. Different endpoints. Different consent.
SGLT2 and GLP-1 outcome trials in people with established cardiovascular or kidney disease now often lead the add-on conversation. Metformin usually stays unless eGFR or intolerance removes it. That is layering, not a funeral for UKPDS.
B12 decline in long-term users is a monitoring footnote with enough cohort support to check levels when anemia or neuropathy appears. Neuropathy work-up still needs its own file - do not assume a low B12 is the only reason feet burn, and do not assume gabapentin will reprint UKPDS.
Gestational diabetes continuation after delivery is a separate, smaller file than UKPDS 34. Do not cite 1998 mortality numbers to keep a postpartum 1000 mg split without a clinician who owns that indication.
Why 1000 mg is a meal tablet
Gastrointestinal intolerance is the usual reason a 1000 mg split fails in the first month. Taking it with food, starting lower, or using an extended-release form is how clinics keep people on the molecule that has the outcome file.
Lactic acidosis is rare and almost always a sick-patient story - hypoxia, severe kidney collapse, liver failure, binge alcohol - not a healthy outpatient who swallowed a 1000 mg tablet with dinner. Hold during dehydrating illness. That sick-day hold is safety, not a contradiction of UKPDS.
EGFR below 30 mL/min/1.73 m² is a stop on modern labels. The 30 to 45 band is a caution and a dose-review zone. UKPDS did not enrol dialysis. Do not cite 1998 to override a current kidney row.
Iodinated contrast and hypoxic hospitalisations are temporary hold columns. Restart when perfusion and eGFR recover. Losing the 1000 mg split for a week in hospital is not 'stopping metformin forever' unless the kidney file changed.
Immediate-release 1000 mg and extended-release 1000 mg are not interchangeable on a whim. Gut comfort often improves with the extended-release form; missed-meal timing still matters. Do not double a forgotten breakfast tablet at supper and call it the UKPDS dose.
The sulphonylurea add-on wrinkle
A supplementary randomisation put 537 people already on maximum sulphonylurea with a still-high fasting glucose onto continued sulphonylurea alone (269) or added metformin (268). Early add-on was associated with a higher risk of diabetes-related death in that slice (96 percent increase, wide CI 2 to 275, p=0.039).
The same paper's combined look and an epidemiologic pass across 4416 people did not show a metformin-plus-sulphonylurea death signal. The wrinkle stayed in the file because it was randomised, not because it became the whole story.
Modern combination practice is not 1990s chlorpropamide. Still, the honest teaching is that UKPDS 34 is not a blank check for every add-on sequence. Read the arm you are quoting.
If a headline says 'metformin plus older sulfonylureas kills,' ask whether they cited the supplementary slice, the combined analysis, or a later observational dump. Those are different stamps.
Holman kept the ledger open a decade later
Holman, Paul, Bethel, Matthews, and Neil reported 10-year post-trial monitoring in the New England Journal of Medicine in 2008. Between-group A1c differences faded after the first year of the watch. The metformin risk reductions did not.
In the metformin group versus overweight conventional therapy, reductions persisted for any diabetes-related endpoint (21 percent, p=0.01), myocardial infarction (33 percent, p=0.005), and death from any cause (27 percent, p=0.002). Diabetes-related death stayed down about 30 percent (p=0.01).
That is the legacy line: early intensive metformin policy left a mark after the lab gaps closed and after people drifted onto mixed therapies. It is the opposite of 'only the current A1c matters.'
Microvascular rows were not the metformin arm's loud victory in the same way. Do not invent an eye-and-kidney guarantee the post-trial file did not print. Retinal and renal protection in UKPDS more famously followed the broader intensive-glucose and blood-pressure policies.
The overweight metformin arm
Of 4075 people recruited across 15 UKPDS centres, 1704 were overweight - more than 120 percent of ideal body weight - with new type 2 diabetes, mean age 53, fasting glucose 6.1 to 15.0 mmol/L after three months of diet without hyperglycaemic symptoms.
Seven hundred fifty-three entered the randomised comparison: conventional policy, mostly diet (411 people), versus intensive metformin aimed at fasting glucose below 6 mmol/L (342 people). Median follow-up 10.7 years. A secondary look compared those 342 with 951 overweight people allocated intensive chlorpropamide, glibenclamide, or insulin.
Median HbA1c landed at 7.4 percent on metformin and 8.0 percent on conventional policy. That half-point is not the story people should tattoo on a fridge. The story is the event aggregates the protocol named in advance: any diabetes-related clinical endpoint, diabetes-related death, and death from any cause.
This was not a two-year A1c beauty contest. It was a long policy trial in people who had just met the diagnosis. That design is why the ink still gets cited when a 1000 mg split is written on a first prescription.
Thirty-two, forty-two, thirty-six
Versus conventional diet policy, metformin cut any diabetes-related endpoint by 32 percent (95 percent CI 13 to 47, p=0.002), diabetes-related death by 42 percent (9 to 63, p=0.017), and all-cause mortality by 36 percent (9 to 55, p=0.011).
Among people already allocated to intensive glucose control, metformin looked better than the sulphonylurea or insulin intensive policies for any diabetes-related endpoint (p=0.0034), all-cause mortality (p=0.021), and stroke (p=0.032). Less weight gain and fewer hypoglycaemic attacks than insulin or sulphonylureas sat in the same paper.
Those three percentages are why a Boston ledger still stamps metformin early in overweight type 2 when the gut and the eGFR allow it. They are not a promise that your personal 1000 mg breakfast tablet will reprint 1998.
Myocardial infarction reductions in the original metformin arm were part of the clinical picture and became even clearer in later follow-up. Do not flatten the file to 'it lowers sugar a bit.'
Weight-neutral or modest-loss behavior in that arm mattered next to insulin and sulphonylurea gain. A 1000 mg split that a patient can stay on without another 4 kg is part of why the intensive metformin policy looked different on the mortality line.
First-line because events moved
UKPDS 34 gave overweight new type 2 a rare oral-drug mortality file: 32 percent fewer diabetes-related endpoints, 42 percent fewer diabetes-related deaths, 36 percent lower all-cause mortality versus diet policy. Holman 2008 kept MI and death reductions visible after A1c gaps closed.
The Glucophage 1000 mg meal tablet is how many US charts still cash that ink. Gut, eGFR, and B12 decide whether the person can stay on it. Newer heart-and-kidney classes add columns. They do not shred the overweight-metformin arm.
Do not quote the sulphonylurea add-on wrinkle as if it cancelled the main randomisation. Do not quote DPP prevention as if it were UKPDS death data.
Formulary holds remain on the Glucophage page. This study file only keeps the outcome numbers from dissolving into 'it is cheap and it lowers A1c.'
Talk with your own clinician or pharmacist before you change a tablet or a dose. Open the ledger disclaimer.
Updated: