Metformin vs insulin: how much each lowers A1C, and how they differ

The best single estimate comes from a 2012 meta-analysis of trials lasting at least 12 weeks ^[1]:

• Metformin alone vs placebo: A1C 1.12 points lower.
• Added to another tablet: 0.95 points lower than the other tablet alone.
• Added to insulin: 0.60 points lower than insulin alone.

Higher doses lowered A1C more than lower doses, without a significant increase in side effects in those trials ^[1]. The figures are averages, and the trials behind them differed a great deal: the meta-analysis reported high heterogeneity, with an I² of about 80% for the monotherapy estimate ^[1].

Over the long term, the UK Prospective Diabetes Study followed overweight adults with newly diagnosed type 2 diabetes for a median of 10.7 years. Median A1C was 7.4% on metformin and 8.0% on conventional, mainly diet-based treatment ^[4]. A network meta-analysis of 453 trials later found that insulin regimens and some GLP-1 receptor agonists added to metformin produced the largest A1C reductions of all the options ^[5].

Metformin does not add insulin to the body. Its label describes three actions: it lowers the amount of glucose the liver releases, reduces how much glucose the gut absorbs, and makes muscle and other tissue respond better to the insulin already there ^[6]. Because it does not push insulin levels up, it rarely drives glucose too low on its own ^[2].

Insulin is the hormone itself. Injected insulin moves glucose into cells whatever the person's own insulin output, which is why it works when the pancreas makes too little and why too large a dose for the food eaten produces hypoglycemia ^[2]. That difference in mechanism explains most of the differences in the table below: metformin has a natural limit to how far it lowers A1C, and insulin does not, but insulin needs dose adjustment and glucose monitoring.

The table compares the drugs as classes. Insulin comes in several forms, from long-acting basal insulin to mealtime insulin, and the hypoglycemia risk differs between regimens. The hypoglycemic drugs guide lists which classes cause low blood sugar.

On the two risks people ask about most, low blood sugar and weight, metformin comes out ahead in the trials. In UKPDS, intensive treatment caused more hypoglycemia than conventional treatment; major episodes occurred in 1.8% of people a year on insulin, against 1.0% on chlorpropamide, 1.4% on glibenclamide and 0.7% with conventional care ^[3]. People assigned insulin gained 4.0 kg on average ^[3]. The metformin analysis found less weight gain and fewer hypoglycemic attacks than with insulin or sulfonylureas ^[4].

Metformin also showed outcome benefits in that population. Compared with conventional care, overweight adults allocated metformin had a 32% lower risk of any diabetes-related endpoint and a 36% lower risk of death from any cause ^[4].

Metformin has its own risks. Its label carries a boxed warning for lactic acidosis, which is rare and linked mainly to kidney impairment and other conditions that reduce oxygen supply; gut side effects are common; and long-term use can lower vitamin B12 ^[6]. The ADA likewise advises periodic B12 checks in long-term users ^[2].

"Safer" also depends on the person. In type 1 diabetes, insulin is the treatment that keeps people alive, and metformin is not a substitute ^[2]. In type 2 diabetes with very high glucose or weight loss from uncontrolled diabetes, the ADA treats insulin as the appropriate choice ^[2].

The ADA Standards of Care 2026 describe how clinicians choose ^[2]:

• Metformin is a common first medicine when no heart, kidney or weight-related condition points to another class, because it is effective, does not cause hypoglycemia on its own and is inexpensive.
• For people with heart disease, heart failure or chronic kidney disease, an SGLT2 inhibitor or GLP-1 receptor agonist with proven benefit is recommended independent of A1C, such as Jardiance for heart failure or kidney disease.
• When an injectable medicine is needed, a GLP-1 receptor agonist is generally preferred to insulin.
• Insulin is considered when there is evidence of catabolism (weight loss), symptoms of high blood sugar, A1C above 10% or glucose of 300 mg/dL or more.

The two are not exclusive, and many people with type 2 diabetes use both. In the meta-analysed trials, adding metformin to insulin lowered A1C by a further 0.60 points compared with insulin alone ^[1].

Metformin is not the only drug studied as an add-on to insulin. Pramlintide, the one approved amylin analogue, is labelled only for people who use mealtime insulin, with a boxed warning for severe hypoglycemia ^[7]. In REIMAGINE 3, adults on basal insulin who received CagriSema lowered A1C by 2.33 points against 0.66 on placebo, without severe hypoglycemia ^[8]. CagriSema is not approved; Novo Nordisk filed it with the FDA for weight management, not diabetes, in December 2025 ^[9]. Its amylin component is covered on the cagrilintide page.