When to Switch From Metformin to Insulin

Most people with type 2 diabetes do not face a single dramatic moment where metformin stops and insulin begins. Instead, the transition usually happens when blood sugar levels remain stubbornly high despite oral medications, with guidelines pointing to an HbA1c above roughly 10% or fasting glucose at or above 300 mg/dL as strong signals that insulin is needed. The underlying reason is that the pancreas gradually loses its ability to produce enough insulin on its own, and metformin can only do so much once that decline reaches a certain point. What surprises many people is that “switching” is something of a misnomer: in most cases, insulin gets added on top of metformin rather than replacing it entirely.

Why Metformin Eventually Stops Being Enough

Metformin works mainly by reducing how much sugar the liver releases into the bloodstream and by helping the body’s cells respond better to insulin. It does not make the pancreas produce more insulin. That distinction matters because type 2 diabetes involves a steady decline in the insulin-producing beta cells of the pancreas. The process typically starts with a loss of the quick burst of insulin that normally appears right after eating, then progresses to a reduced ability to produce insulin in response to any stimulus, and can eventually reach a stage of near-complete beta-cell failure that makes injected insulin the only option.

How quickly this happens varies enormously. Some people take metformin for decades without ever needing insulin. Others, especially those diagnosed at a younger age or with a strong family history, may see their beta cells decline within just a few years. The trajectory depends on genetics, how long blood sugars have been elevated, body weight, and other metabolic factors. So the honest answer to “when” is: when your body’s insulin production can no longer keep up, regardless of what oral medications you’re taking. The clinical markers that signal that threshold are more concrete.

The Clinical Triggers for Starting Insulin

Current guidelines converge on a few clear situations where insulin should be started, sometimes urgently. The most widely cited threshold is an HbA1c persistently above 10% or blood glucose levels at or above 300 mg/dL, especially when these numbers are present despite the person already taking two or three oral medications.² At that level, the pancreas is producing so little insulin on its own that no oral drug can close the gap.

There are also emergency scenarios. Diabetic ketoacidosis, where the body starts breaking down fat for fuel and produces dangerous levels of ketones, always requires insulin treatment, typically given intravenously in the hospital until the crisis resolves.³ A related condition called hyperosmolar hyperglycemic state, which involves extremely high blood sugar with severe dehydration, similarly demands insulin alongside aggressive fluid replacement.⁴ These situations can happen even to people who previously had well-controlled diabetes, triggered by illness, surgery, or certain medications.

Steroid medications are a common culprit. When someone with type 2 diabetes takes glucocorticoids for conditions like asthma, autoimmune disease, or joint inflammation, blood sugar often spikes dramatically. This steroid-induced hyperglycemia frequently needs temporary insulin to manage, even if the person was doing well on metformin alone beforehand.⁵ In many of these cases, insulin can be tapered off once the steroids are stopped, so the “switch” turns out to be temporary.

Adding Insulin Usually Means Keeping Metformin

One of the biggest misconceptions is that starting insulin means throwing away your metformin prescription. In practice, the standard approach is to keep metformin going and add a single daily injection of long-acting (basal) insulin on top of it. A randomized trial that compared continuing metformin versus switching to a placebo when patients started insulin found clear advantages to keeping metformin in the mix. Over 12 months, the patients who stayed on metformin gained less weight (about 6 kg versus nearly 8 kg), had a greater drop in HbA1c, and needed about 25 fewer units of insulin per day compared to those on placebo.⁶ Treatment satisfaction also improved more in the metformin group.

The rationale is straightforward. Metformin tackles insulin resistance at the tissue level while the injected insulin handles the shortfall in production. The two work on different parts of the problem. Stopping metformin when you add insulin is like removing one tool from your toolbox just because you picked up a new one. Unless there is a specific reason to discontinue metformin, your doctor will likely tell you to keep taking it.

When Metformin Itself Has to Go

There are situations where metformin genuinely needs to be stopped, and kidney function is the main one. Metformin is cleared from the body by the kidneys, so as kidney function declines, the drug can accumulate and raise the risk of a rare but serious side effect called lactic acidosis. Guidelines from multiple countries have settled on a consistent set of thresholds based on estimated glomerular filtration rate (eGFR), which measures how well the kidneys filter:

eGFR above 45: Metformin can be used safely at normal doses.
eGFR 30 to 44: If metformin is already being taken, the dose should be reduced (generally to half). New prescriptions should not be started in this range.⁷
eGFR below 30: Metformin should be stopped altogether.⁸

The U.K.’s National Institute for Health and Care Excellence uses similar cutoffs, recommending review when eGFR drops below 45 and discontinuation below 30.⁹ When kidney disease reaches these levels, insulin often becomes the backbone of glucose management because most other oral medications also carry restrictions in advanced kidney disease. This is one of the few scenarios where you truly are switching from metformin to insulin rather than adding one to the other.

Severe liver disease and certain acute illnesses that reduce oxygen delivery to tissues, such as sepsis or heart failure requiring hospitalization, are other circumstances where metformin is typically paused or discontinued. During these episodes, insulin is the safest way to control blood sugar until the acute problem resolves.

Newer Medications That Can Delay the Switch

The landscape of type 2 diabetes drugs has expanded considerably, and several classes of medication now sit between metformin and insulin in the treatment ladder. GLP-1 receptor agonists (drugs like semaglutide and liraglutide) and SGLT2 inhibitors (like empagliflozin and dapagliflozin) have changed the calculus for many patients. A meta-analysis of 13 randomized trials comparing these non-insulin medications to insulin found that the newer drugs achieved modestly better HbA1c reduction on average, led to about 3 kg more weight loss, and lowered systolic blood pressure. Insulin use was also tied to more than double the rate of hypoglycemic events compared to non-insulin alternatives.¹⁰

That does not mean insulin is always the wrong choice. In a head-to-head study comparing a GLP-1 receptor agonist to basal insulin, the GLP-1 drug did not significantly lower HbA1c over 16 weeks, while basal insulin did achieve a meaningful reduction. The GLP-1 agonist, however, produced significant weight loss that insulin did not.¹¹ The practical takeaway is that for people whose HbA1c is very high and who need rapid glucose lowering, insulin often works faster and more reliably. For those with moderately elevated HbA1c who also need to lose weight or have cardiovascular or kidney concerns, the newer drugs may be a better first escalation step after metformin.

For patients who eventually do need insulin, fixed-ratio combination products that pair basal insulin with a GLP-1 receptor agonist in a single daily injection have shown promise. Clinical trials have found that these combinations improve blood sugar control with less weight gain and fewer hypoglycemic episodes compared to basal insulin alone.¹² Real-world data from routine clinical practice have confirmed these findings, showing consistent improvements in glucose control along with modest weight loss or stability and low hypoglycemia risk across different patient groups.¹³ These combinations offer a way to get the glucose-lowering power of insulin while offsetting its typical downsides.

What Starting Insulin Actually Looks Like

If you’ve been picturing multiple daily injections with complicated dosing schedules, the reality is less daunting than you might expect. The standard starting point for most people with type 2 diabetes is a single injection of basal insulin, usually given at bedtime or in the morning. Basal insulin provides a slow, steady release over 24 hours (or longer, with newer formulations) and is designed to keep fasting blood sugar in check overnight and between meals. You typically keep taking metformin and any other oral medications alongside it.

The dose starts low and gets adjusted upward every few days based on fasting blood sugar readings. A large observational study of managed-care patients found that those who started basal insulin after already trying one oral medication saw a bigger HbA1c drop (about 1.3 percentage points) compared to those who waited until they were on three or more oral drugs (about 0.9 percentage points), though patients who waited longer were also less likely to stop taking insulin.¹⁴ The finding hints at a tension in clinical practice: starting insulin earlier tends to produce bigger improvements, but patients who have exhausted more oral options first may be more committed to sticking with it.

Continuous glucose monitors have started to play a role in insulin titration. A recent phase 3 trial tested a once-weekly long-acting insulin (icodec) titrated using continuous glucose monitoring data rather than traditional fingerstick readings. Over 26 weeks, time spent in the target blood sugar range rose from about 54% to 76%, while time spent too high dropped by roughly half. No severe hypoglycemic episodes occurred during the trial.¹⁵ A once-weekly injection guided by a sensor on your arm is a very different experience from the multiple-daily-injection image that scares many patients away from insulin.

Weight Gain and Low Blood Sugar on Insulin

These are the two side effects people worry about most, and both are real but not inevitable. On weight gain, a retrospective study found that about one in ten patients gained 5 kg or more after starting insulin. The strongest predictor was an unusual pre-insulin pattern: people who had been losing weight while their HbA1c was rising in the two years before starting insulin were the most vulnerable, with roughly one in five of them gaining 5 kg or more.¹⁶ That pattern likely reflects beta cells failing to the point where calories were being lost through high blood sugar (glucose literally spilling into the urine), and insulin corrects that leak, which restores weight. It is not so much that insulin causes fat gain as that it stops a pathological calorie drain.

Hypoglycemia is a more nuanced story than many patients expect. A common assumption is that insulin carries the highest low-blood-sugar risk of any diabetes drug, but the comparison depends on what you’re comparing it to. When researchers looked at people already on metformin who either added insulin or added a sulfonylurea (an older class of pill that stimulates the pancreas to release more insulin), the insulin group actually had a somewhat higher risk of hypoglycemia than the sulfonylurea group, with about 31 versus 25 low-blood-sugar events per 1,000 person-years.¹⁷ However, compared to the newer drug classes like GLP-1 agonists and SGLT2 inhibitors, insulin carries a distinctly higher hypoglycemia risk, as noted earlier. The risk also depends heavily on the type of insulin regimen: a single daily basal injection carries far less hypoglycemia risk than a complex regimen with multiple mealtime doses.

The Fear Factor

Clinical thresholds and pharmacology aside, one of the biggest real-world barriers to starting insulin is psychological. Research has documented that nearly a third of people with type 2 diabetes are initially reluctant to begin insulin when their doctor first recommends it.¹⁸ This phenomenon, sometimes called psychological insulin resistance, stems from a cluster of fears: that needing insulin means you’ve “failed” at managing your disease, that injections will be painful, that insulin will make you gain weight or cause dangerous lows, or that it signals the beginning of the end.

The irony is that quality of life typically improves after insulin is started. Research on people who were insulin-naive and then transitioned to insulin found that overall quality of life scores went up after the switch, not down.¹⁹ This makes sense when you think about it: chronically high blood sugar causes fatigue, blurry vision, frequent urination, and a general feeling of malaise. Bringing those numbers down, even if it means an injection, often makes people feel markedly better. The delay in starting insulin, driven by fear, often prolongs a period of preventable poor health and accelerates the complications that people fear insulin will cause. This is one area where the evidence should genuinely reassure: the anticipation tends to be worse than the reality.

When the Diagnosis Itself Is Wrong

Some people labeled with type 2 diabetes actually have a slow-moving form of autoimmune diabetes called latent autoimmune diabetes of adults, or LADA. These individuals initially look like typical type 2 patients: they’re often diagnosed in middle age, and their blood sugar responds to metformin and lifestyle changes at first. But unlike true type 2 diabetes, LADA involves the immune system actively destroying beta cells, similar to type 1 diabetes but at a slower pace. Patients with LADA typically require insulin therapy within 6 months to 6 years of diagnosis, as their beta cells are being knocked out by an autoimmune process that metformin and similar drugs cannot stop.²⁰

LADA should be suspected in anyone whose diabetes is progressing unusually fast despite good adherence to treatment, who is lean or only mildly overweight, or who has a personal or family history of autoimmune conditions. Antibody testing can confirm it. A related blood test, C-peptide, measures how much insulin the pancreas is still producing. Research has shown that very low C-peptide levels correlate with worse metabolic control, more severe hypoglycemia, and a three-fold higher odds of developing a diabetic complication.²¹ If your blood sugars are rising quickly and oral medications are losing their effect faster than expected, asking about LADA and getting a C-peptide level checked is worthwhile. The treatment is straightforward: these patients need insulin, and delaying it while trying additional oral drugs is counterproductive because no pill can replace beta cells that are being destroyed by the immune system.

Practical Signals You Can Watch For

Outside of lab numbers and formal guidelines, certain day-to-day patterns often signal that metformin alone (or metformin plus other oral drugs) is no longer enough. Persistent morning fasting blood sugars above 130 mg/dL despite taking medications consistently suggest that overnight insulin production is flagging. Unexplained weight loss alongside rising blood sugars is another red flag. Frequent thirst, increased urination, and fatigue that creep back after a period of good control point the same way. None of these symptoms on their own mean you need insulin tomorrow, but together they paint a picture of declining beta-cell function that warrants a conversation with your doctor sooner rather than later.

If your HbA1c has crept above 8% or 9% on two or more oral medications and has been there for six months, the evidence favors acting rather than waiting. Each month spent with high blood sugar contributes to the cumulative damage in blood vessels, nerves, and kidneys that drives long-term complications. The fear of insulin is understandable, but the complications of prolonged high blood sugar are far more consequential than a daily injection. The modern insulin toolkit, with once-daily and even once-weekly options, combination products that limit weight gain, and sensors that make dose adjustments safer, has made the transition considerably less burdensome than it was even a decade ago.