- Type 2 diabetes is diagnosed with a blood test. An HbA1c of 48 mmol/mol (6.5%) or more, a fasting glucose of 7.0 mmol/L or more, or a 2-hour glucose-tolerance result of 11.1 mmol/L or more all count. Unless symptoms are obvious, two abnormal results are needed (ADA Standards of Care 2026, section 2; WHO 2011).
- England's NICE guideline was rewritten in February 2026. Most adults are now offered modified-release metformin plus an SGLT-2 inhibitor from the start. People with heart or artery disease are also offered semaglutide (NICE NG28, Initial medicines).
- Remission is possible for some people. In the Diabetes UK-funded DiRECT trial, 46% of people on a total diet replacement programme were in remission at 12 months, against 4% on usual care. The figure was 36% at 2 years, and 13% of the extension group at 5 years (Lean 2018; Lean 2019; Lean 2024).
- Weight loss alone did not prevent heart attacks or strokes in the large NIH-funded Look AHEAD lifestyle trial. Its hazard ratio was 0.95 (95% CI 0.83 to 1.09) after a median of 9.6 years (Look AHEAD 2013).
- Seek urgent help for heavy thirst, passing lots of urine, vomiting, confusion, drowsiness or fast breathing. These can be signs of HHS or DKA, which can kill if not treated quickly (NHS).
- Money trail: The trials behind SGLT-2 and GLP-1 medicines were paid for by their makers (Boehringer Ingelheim and Eli Lilly, Novo Nordisk, Janssen) (EMPA-REG; LEADER; CREDENCE). The American Diabetes Association disclosed $24.8 million from drug and device makers in 2023 (ADA disclosure). Several NICE committee members declared industry payments; one was partly excluded (NICE register of interests).
- Evidence grade: Strong for metformin plus an SGLT-2 inhibitor or GLP-1 agonist to cut heart and kidney events in higher-risk people. Strong for remission through large, sustained weight loss early in the disease. Moderate for structured exercise lowering HbA1c. Insufficient for any supplement replacing medicines.
Independent evidence review · Condition
Type 2 diabetes is common and easy to diagnose. Many people have no symptoms for years, which is why testing matters. For people diagnosed recently and living with excess weight, sustained weight loss of 10–15 kg can put the condition into remission. Medicines now aim to protect the heart and kidneys, not just to lower blood glucose. The cleanest evidence comes from publicly funded trials (DiRECT, Look AHEAD, GRADE) and from NICE. The strongest evidence for the newer drug classes comes from trials paid for by their makers. Independent meta-analyses broadly confirm those results. Our confidence: high on diagnosis and on the benefits of metformin, SGLT-2 inhibitors and GLP-1 agonists in higher-risk people. Moderate on how long remission lasts. Low on supplements.
Table of contents
- Evidence summary
- What type 2 diabetes is
- Symptoms, diagnosis and red flags
- Causes and risk factors
- Treatments that work
- Medicines: the 2026 NICE pathway and drug classes
- Monitoring and targets
- Supplements and lifestyle claims
- Risks of common treatments
- Who to see
- Follow the money
- Related research
- Frequently asked questions
- Sources and funding notes
Evidence summary
| Claim | Evidence | Source | Funding / conflict | Strength |
|---|---|---|---|---|
| HbA1c of 48 mmol/mol (6.5%) or more diagnoses diabetes | WHO and ADA both adopt 6.5%. ADA requires two abnormal results unless hyperglycaemia is unequivocal. HbA1c is unreliable in pregnancy, haemoglobin variants, anaemia and other conditions that change red-cell turnover. | WHO 2011; ADA 2026 | WHO: member-state funded. ADA guideline paid from ADA general revenue; ADA itself takes industry money (see below). | Strong |
| Intensive weight management can put early type 2 diabetes into remission | 306 people, 49 practices. Remission at 12 months: 46% vs 4% (OR 19.7, 95% CI 7.8 to 49.8). At 2 years: 36% vs 3%. Remission was 86% in people who lost 15 kg or more. | DiRECT 2018; 2019 | Funded by Diabetes UK. Some authors employed by, or held shares in, Counterweight, the programme provider. | Strong |
| Remission lasts | At 5 years, 13% of the extension group were in remission. Of those in remission at 2 years, 26% were still in remission at 5 years. | DiRECT 5-year | Diabetes UK. Authors declared links to Counterweight, Cambridge Weight Plan and drug makers, outside this work. | Moderate |
| Lifestyle weight loss prevents heart attacks and strokes | 5,145 adults, median 9.6 years. No difference in cardiovascular events (HR 0.95, 95% CI 0.83 to 1.09). Stopped early for futility. | Look AHEAD 2013 | Funded by the US National Institutes of Health and others | Moderate (against the claim) |
| Structured exercise lowers HbA1c | 47 trials, 8,538 people. Structured exercise: −0.67 percentage points (95% CI −0.84 to −0.49). Advice alone without diet: no change. | Umpierre 2011 | Academic (Brazil); funding statement not verified | Moderate |
| SGLT-2 inhibitors and GLP-1 agonists reduce deaths, heart attacks and kidney failure | 764 trials, 421,346 people. Both classes lowered all-cause death, cardiovascular death, non-fatal heart attack and kidney failure (high certainty). For SGLT-2 inhibitors, absolute benefit ranged from 3 to 40 fewer deaths per 1,000 over 5 years, depending on baseline risk. | Palmer, BMJ 2021 | No funding. Some authors declared payments from drug makers. | Strong |
| Empagliflozin, liraglutide and canagliflozin cut events in high-risk people | EMPA-REG: primary outcome HR 0.86 (0.74 to 0.99). LEADER: HR 0.87 (0.78 to 0.97). CREDENCE (kidney disease): HR 0.70 (0.59 to 0.82). | EMPA-REG; LEADER; CREDENCE | Funded by the makers: Boehringer Ingelheim and Eli Lilly; Novo Nordisk (with NIH); Janssen | Strong (maker-funded, independently confirmed) |
| Add-on medicines differ in how well they hold HbA1c | 5,047 people on metformin, 5 years. Glargine and liraglutide held targets better than glimepiride or sitagliptin. Severe hypos: 2.2% glimepiride vs 0.7% sitagliptin. | GRADE 2022 | NIDDK (US government) "and others" | Strong |
| Metformin lowers vitamin B12 | MHRA reclassified low B12 as a common side effect, more likely with higher doses and longer use | MHRA 2022 | UK medicines regulator | Risk |
| GLP-1 agonists and tirzepatide can cause severe pancreatitis | Strengthened warnings in January 2026, including rare necrotising and fatal cases | MHRA 2026 | UK medicines regulator | Risk (rare) |
| Supplements (cinnamon, chromium, berberine and others) can replace treatment | Small, mixed trials. The US government's complementary-health agency says not to replace medical treatment with unproven products. | NCCIH 2023 | US government (NIH) | Insufficient |
Independent evidence and credibility scorecard
Independence tiers: Tier 1 = no identifiable financial stake in the outcome; Tier 2 = indirect or partial ties (for example, some authors with industry links, or an otherwise independent funder with commercial partners); Tier 3 = an interested party with a clear stake, but not the seller; Tier 4 = funded or run by the company that sells the product. Credibility: A = high, with incentives that reward accuracy; B = solid, but watch the bias; C = interested; D = self-interested.
| Source | Who funds it | Country | Independence tier (1–4) | Credibility (A–D) | Why it would tell the truth / residual bias |
|---|---|---|---|---|---|
| NICE NG28 (2026 update) | UK Department of Health and Social Care (framework agreement) | UK | Tier 1 | A | Statutory body that must justify NHS spending in public. Residual bias: a cost lens (it backs whichever SGLT-2 inhibitor is cheapest). Some committee members declared industry payments. |
| WHO HbA1c consultation (2011) | WHO (member-state and donor funding) | Switzerland (intergovernmental) | Tier 1 | B | Global standard-setter with no product to sell. Residual bias: we did not review the experts' declarations of interest. |
| ADA Standards of Care 2026 | ADA general revenue; ADA received $24.8m from drug and device makers in 2023 | USA | Tier 2 | B | Has rules: a majority of the committee and the chair must be conflict-free. Residual bias: the organisation depends partly on the companies whose products it grades. |
| Shigeta 2023, audit of ADA guideline authors | No external funding statement found; Medical Governance Research Institute | Japan | Tier 1 | B | Uses public payment databases and has nothing to sell. Residual bias: a watchdog's incentive to find conflicts; covers 2021 only. |
| DiRECT trial (2018–2024) | Diabetes UK (charity, mostly public donations, with pharma partners) | UK | Tier 2 | B | Large, randomised and published in full at 1, 2 and 5 years. Residual bias: the funder campaigns on remission, and some authors worked for the programme provider. |
| Look AHEAD (2013) | US National Institutes of Health and others | USA | Tier 1 | A | Publicly funded, and reported a null result that went against its own hypothesis |
| GRADE trial (2022) | NIDDK (US government) "and others" (not itemised in the abstract) | USA | Tier 2 | B | Head-to-head comparison of four drug classes, with no single maker in charge. Residual bias: the other funders were not checked. |
| Palmer et al., BMJ network meta-analysis (2021) | No funding; several authors declared payments from drug makers | New Zealand / Australia / Italy | Tier 2 | B | Pre-registered and reports absolute benefits and harms. Residual bias: author ties, and it pools mostly maker-funded trials. |
| Maker-funded outcome trials (EMPA-REG, LEADER, CREDENCE) | Boehringer Ingelheim and Eli Lilly; Novo Nordisk (with NIH); Janssen | Germany / USA / Denmark / USA–Belgium | Tier 4 | C | Pre-registered, regulator-required designs with independent data committees. Residual bias: the sponsor controls the design, the comparator and the framing. |
| MHRA Drug Safety Updates | UK government regulator | UK | Tier 1 | A | Legally accountable for drug safety. Residual bias: it acts on spontaneous reports, which undercount harms. We did not verify its fee income from industry. |
| RAND insulin price comparison (2024) | Sponsored by HHS Office of the Assistant Secretary for Planning and Evaluation | USA | Tier 1 | B | Non-profit whose reputation rests on its methods. Residual bias: commissioned by a government pushing drug-price reform. |
| IDF Diabetes Atlas, 11th edition (2025) | Sanofi (2022–2024), Novo Nordisk (2022–2023), Pfizer-MSD Alliance grant (2022) | Belgium | Tier 3 | C | Global prevalence modelling by expert committee. Residual bias: sponsors benefit from larger estimates of disease burden. |
Country clustering: 5 of the 12 sources are US-based and 3 are UK-based. On treatment the article leans on UK (NICE, MHRA, DiRECT) and US (ADA, NIH trials) perspectives. Where they agree – on diagnosis thresholds, on metformin, and on SGLT-2 inhibitors and GLP-1 agonists for higher-risk people – confidence is higher. They differ in places. ADA sets a general HbA1c goal under 7%. NICE uses 48 or 53 mmol/mol depending on hypo risk, and recommends SGLT-2 inhibitors for almost everyone partly on cost grounds after dapagliflozin became generic.
What type 2 diabetes is
In type 2 diabetes the body cannot make enough insulin, or the insulin it makes does not work properly (insulin resistance). Glucose then builds up in the blood (NHS). Type 1 diabetes is different: it is an autoimmune disease that destroys insulin-making cells. Type 2 is the far more common form. It develops more slowly and, unlike type 1, it may be preventable and can sometimes go into remission (NHS).
The International Diabetes Federation projects that 853 million adults (1 in 8) will have diabetes by 2050. It estimates that diabetes-related health spending now exceeds USD 1 trillion, about 12% of global health budgets (IDF Diabetes Atlas FAQ). Read these figures knowing that the Atlas is sponsored by insulin and diabetes-drug makers (see Follow the money).
Complications
High glucose over years damages blood vessels and nerves. The NHS lists heart disease and stroke, foot ulcers and infections that can lead to amputation, diabetic retinopathy and sight loss, kidney disease and gum disease (NHS complications). There are also two acute emergencies. HHS develops over days to weeks, often with infection and dehydration. DKA is a build-up of ketones when insulin is too low (NHS). Long-term NICE checks cover the eyes, feet, kidneys and heart. Since 2022 NICE has also asked clinicians to consider gum disease (NICE update history).
Symptoms, diagnosis and red flags
Symptoms
The most common symptoms are feeling very tired, peeing more than usual, constant thirst and losing weight without trying. Others are blurred vision, slow-healing cuts, and itching or repeated thrush. Symptoms come on gradually, and many people have none (NHS symptoms). This is why testing people at risk matters more than waiting for symptoms.
How it is diagnosed
| Test | Diabetes | Prediabetes / raised risk (ADA ranges) |
|---|---|---|
| HbA1c (average glucose over about 2–3 months) | 48 mmol/mol (6.5%) or more | 39–47 mmol/mol (5.7–6.4%) |
| Fasting plasma glucose | 7.0 mmol/L (126 mg/dL) or more | 5.6–6.9 mmol/L (100–125 mg/dL) |
| 2-hour glucose in a 75 g glucose tolerance test | 11.1 mmol/L (200 mg/dL) or more | 7.8–11.0 mmol/L (140–199 mg/dL) |
| Random glucose with classic symptoms or a crisis | 11.1 mmol/L (200 mg/dL) or more | – |
Source: ADA Standards of Care 2026, section 2. In the UK, "non-diabetic hyperglycaemia" is usually defined as HbA1c 42–47 mmol/mol, a narrower band than the ADA's; we could not open the NICE CKS page that sets this out from outside the UK. Without unmistakable high glucose, the ADA requires two abnormal results, either from two tests or one test repeated (ADA). WHO accepted HbA1c for diagnosis in 2011, with 6.5% as the cut-off, provided the assays are quality-assured (WHO).
When HbA1c misleads: the ADA advises glucose-based tests instead in pregnancy, haemoglobin variants such as sickle-cell trait, G6PD deficiency, HIV, anaemia, dialysis, recent blood loss or transfusion, and haemolysis (ADA). For people already diagnosed, NICE says to use glucose profiles or fructosamine if HbA1c is invalid, and to investigate unexplained mismatches (NICE 1.5.3–1.5.4).
Red flags
- Possible DKA: thirst, frequent urination, stomach pain, nausea or vomiting, deep breathing, fruity breath, drowsiness or confusion. The NHS says to call 999 or go to A&E if you have these symptoms and have not been diagnosed with diabetes, or if blood ketones are over 3 mmol/L (NHS DKA).
- Possible HHS: passing a lot of urine, extreme thirst, nausea, confusion, sleepiness or shortness of breath (NHS).
- Severe hypo: glucose usually below 4 mmol/L, with shaking, sweating and confusion. Call 999 if the person is unconscious or does not recover (NHS).
- Foot changes: a new ulcer, a hot red swollen foot, or a wound that is not healing needs same-day assessment (NHS).
Causes and risk factors
The NHS says you may be at higher risk if you are white and over 40, or over 25 and of South Asian, Chinese, Black African or Black Caribbean background. Other risks are a parent or sibling with diabetes, being overweight and not very active, and having a condition such as high blood pressure (NHS). In England, adults aged 40–74 without an existing condition can ask for a free NHS Health Check (NHS).
Weight is the main factor that can be changed. In DiRECT, remission was achieved by none of the 76 people who gained weight, 7% of those who lost up to 5 kg, and 86% of those who lost 15 kg or more (Lean 2018). That gradient supports the view that excess fat drives much of the disease in this group. Prevention is covered in our type 2 diabetes prevention and management guide.
Treatments that work
| Approach | Verdict | Best evidence |
|---|---|---|
| Total diet replacement (about 800–850 kcal/day for 12–20 weeks) with support to keep weight off | Works | DiRECT: 46% remission at 1 year, 36% at 2 years (Lean 2019) |
| Metformin plus an SGLT-2 inhibitor as first treatment | Works | NICE 2026 analysis; BMJ network meta-analysis (NICE; Palmer 2021) |
| Adding a GLP-1 agonist (semaglutide) for people with heart or artery disease | Works | LEADER (maker-funded) and independent pooling (LEADER; Palmer) |
| Structured exercise (aerobic, resistance or both) | Works | HbA1c −0.67 points; greater with more than 150 min/week (Umpierre 2011) |
| Low-carbohydrate diets | Mixed | Remission (HbA1c <6.5%) 57% vs 31% at 6 months; the effect faded by 12 months and when remission was defined as off medication (Goldenberg 2021) |
| Metabolic (bariatric) surgery for obesity with type 2 diabetes | Works | 10-year remission 50% after biliopancreatic diversion and 25% after gastric bypass vs 5.5% with medical therapy; small trial, n=60 (Mingrone 2021) |
| Intensive lifestyle programme to prevent heart attacks and strokes | Mixed | Better weight and HbA1c, but no fewer cardiovascular events (Look AHEAD) |
| Supplements as a replacement for medicines | Insufficient | NCCIH |
Remission: what DiRECT showed
DiRECT was carried out in 49 GP practices in Scotland and Tyneside. It recruited people aged 20–65 with a BMI of 27–45, diagnosed within the previous 6 years and not using insulin. In the intervention group, diabetes and blood-pressure medicines were stopped. Participants then had 825–853 kcal/day of formula diet for 3–5 months, a staged return to food, and long-term support. Remission was defined as an HbA1c below 48 mmol/mol at least 2 months after stopping all diabetes medicines (Lean 2018).
Remission tends to slip as weight returns. At 5 years, 13% of the 85 people in the extension group were in remission. Of those in remission at 2 years, 26% were still in remission at 5 years. Even so, the original intervention group had about half the rate of serious adverse events of the control group (4.8 vs 10.2 per 100 patient-years) (Lean 2024).
The NHS in England now runs this approach as the Type 2 Diabetes Path to Remission Programme, and NICE links to it directly (NICE 1.3.4). Real-world results are lower than in the trial. Of 1,740 people who started before 2022, 55% completed. Among those with two HbA1c results, 27% were in remission at 1 year; the figure was 32% among completers (Valabhji 2024). That evaluation was led by NHS England staff, with academic and Diabetes UK co-authors, and reports no funding.
Diet and activity for everyone
NICE's dietary advice for type 2 diabetes is the same healthy-eating advice given to the general population. That means high-fibre, low-glycaemic-index carbohydrates, low-fat dairy and oily fish, and limits on saturated and trans fats. NICE also advises against foods marketed specifically for people with diabetes (NICE 1.3.3, 1.3.8). Low-carbohydrate diets improved remission at 6 months in a BMJ meta-analysis. The effect was smaller when remission required being off medication, and the data at 12 months were sparse (Goldenberg 2021).
Medicines: the 2026 NICE pathway and drug classes
NICE changed its pathway in February 2026. The old model was "metformin first, then add-ons to hit an HbA1c number". The new model starts with medicines chosen as much for heart and kidney protection as for glucose (NICE update information). The earlier 2022 pathway kept metformin as the usual first drug and reserved SGLT-2 inhibitors for people with heart failure, established cardiovascular disease or high cardiovascular risk. The 2026 update replaced that.
| Group | NICE 2026 first-line offer |
|---|---|
| No relevant other conditions | Modified-release metformin and an SGLT-2 inhibitor; SGLT-2 inhibitor alone if metformin is unsuitable (1.13.1–1.13.2) |
| Heart failure | Modified-release metformin and an SGLT-2 inhibitor (1.14.1) |
| Atherosclerotic cardiovascular disease | Modified-release metformin, an SGLT-2 inhibitor and subcutaneous semaglutide up to 1 mg weekly (1.15.1) |
| Early-onset type 2 diabetes | Modified-release metformin and an SGLT-2 inhibitor; consider adding a GLP-1 agonist or tirzepatide (1.16.1) |
| Chronic kidney disease | eGFR 30 or more: metformin and an SGLT-2 inhibitor. eGFR 20–29: dapagliflozin or empagliflozin plus a DPP-4 inhibitor. Below 20: consider a DPP-4 inhibitor (1.18.1–1.18.3) |
| Frailty | Modified-release metformin; an SGLT-2 inhibitor only if frailty does not raise the risk of dehydration or low blood pressure (1.19.1) |
Source: NICE NG28, Initial medicines. If more treatment is needed, NICE now suggests adding a DPP-4 inhibitor first. If that is unsuitable, the options are a sulfonylurea, pioglitazone or insulin (NICE 1.25.1).
NICE's committee gave its reasoning in absolute terms. Over 3 years, 123 in 1,000 people not taking an SGLT-2 inhibitor had a major cardiovascular event, against 107–115 in 1,000 taking one. Heart-failure admissions fell from 54 to 26–40 in 1,000. In return, 10–100 in 1,000 would get genital infections, and 1 in 1,000 or fewer would have rare events such as DKA or Fournier's gangrene (NICE).
The drug classes in brief
- Metformin (biguanide; cheap and generic). In UKPDS 34, overweight people allocated metformin had 32% fewer diabetes-related endpoints (95% CI 13 to 47) and 36% lower all-cause mortality than those on conventional, mainly diet-only, treatment (UKPDS 34, 1998). NICE now prefers the modified-release form for most people (NICE).
- SGLT-2 inhibitors (dapagliflozin, empagliflozin, canagliflozin, ertugliflozin) make the kidneys pass glucose into the urine. They protect the heart and kidneys. NICE supports dapagliflozin while it is the cheapest option, now that generics are available (NICE).
- GLP-1 receptor agonists (semaglutide, liraglutide, dulaglutide) and the dual GLP-1/GIP agonist tirzepatide are mostly weekly or daily injections. They lower glucose and weight. Liraglutide and semaglutide have cardiovascular outcome data. NICE notes that it recommends tirzepatide for its glucose-lowering benefits only (NICE 1.16.1).
- DPP-4 inhibitors (sitagliptin, linagliptin, alogliptin and others) are tablets that are weight-neutral and rarely cause hypos. They were the least effective option for holding HbA1c in GRADE (GRADE).
- Sulfonylureas (gliclazide, glimepiride) are cheap and effective, but can cause hypos and weight gain (NICE).
- Pioglitazone lowers HbA1c but causes fluid retention. Avoid it in heart failure. It also raises the risk of fractures and weight gain (NICE).
- Insulin is effective for very high or rising glucose, with a higher risk of hypos and weight gain. NICE made pragmatic changes in 2026 because some insulin products were withdrawn or in short supply (NICE).
Heart risk is a large part of type 2 diabetes care. Statins and blood-pressure treatment are covered in our diabetes and heart disease review.
Monitoring and targets
- How often to test HbA1c: every 3–6 months until stable, then every 6 months (NICE 1.5.1).
- NICE targets: 48 mmol/mol (6.5%) if managed by diet or by medicines that do not cause hypos. 53 mmol/mol (7.0%) on a medicine that can cause hypos. If HbA1c rises to 58 mmol/mol (7.5%) or more, treatment should be stepped up and the target is 53 (NICE 1.5.7–1.5.8). Targets can be relaxed for older or frailer people, people at risk from falls, and people who drive for work (NICE 1.5.9).
- ADA targets: HbA1c below 7% (53 mmol/mol) for many non-pregnant adults. Lower (below 6.5%) if it can be reached safely; up to 8% where harms outweigh benefits. On a continuous glucose monitor, the goal is more than 70% of time between 3.9 and 10 mmol/L, with less than 4% of time below 3.9 (ADA 2026, section 6).
- Home finger-prick testing: NICE does not routinely offer it unless the person uses insulin, has hypos, takes hypo-causing tablets and drives, or is pregnant (NICE 1.6.2). Flash glucose monitoring is offered to people on multiple daily insulin injections who have recurrent or severe hypos, among other criteria (NICE 1.7.1).
- Blood pressure: managed as for other adults under NICE's hypertension guideline (NICE). Our home blood pressure monitoring guide covers technique.
Supplements and lifestyle claims
| Claim | What independent sources say | Grade |
|---|---|---|
| Cinnamon lowers blood sugar | Reviews found lower fasting glucose; study quality varied (NCCIH) | Weak |
| Berberine works "like metformin" | May lower glucose as an add-on; interacts with medicines (NCCIH). See our berberine review. | Weak |
| Chromium improves HbA1c | One review suggests a small benefit; high doses may harm the kidneys and liver (NCCIH) | Weak |
| Magnesium improves insulin sensitivity | Evidence judged insufficient for guidelines (NCCIH) | Insufficient |
| Omega-3 improves glucose control | Conflicting; helps triglycerides more than glucose (NCCIH) | Insufficient |
| "Diabetic" foods help | NICE advises against them (NICE 1.3.8) | Not supported |
| You can "reverse" diabetes with weight loss | Remission is real but not guaranteed, and it often ends if weight returns (DiRECT 5-year) | Strong (for remission, not cure) |
NCCIH's central advice is not to replace medical treatment for diabetes with an unproven product (NCCIH). Some supplements can add to the glucose-lowering effect of medicines and raise the risk of hypos. Our full grading is in diabetes supplements: benefits, limits and safety.
Risks of common treatments
- Metformin: stomach upset is common. Low vitamin B12 is now classed as a common side effect. The MHRA advises testing B12 if symptoms such as new neuropathy or anaemia appear, and considering regular checks in people at risk (MHRA 2022).
- SGLT-2 inhibitors: genital fungal infections (10–100 in 1,000 people), dehydration and low blood pressure. Rare Fournier's gangrene (NICE). EMPA-REG reported more genital infections and no rise in other adverse events (EMPA-REG).
- GLP-1 agonists: nausea and other gut effects, which were the most common reason for stopping liraglutide in LEADER (LEADER). The MHRA says these medicines must not be used in pregnancy, when trying to conceive or when breastfeeding. Tirzepatide may reduce how well the oral contraceptive pill works in people living with overweight (MHRA 2025).
- Pioglitazone: fluid retention, possible worsening of heart failure, fractures and weight gain (NICE).
- Very-low-calorie diets: in DiRECT, 4% of the intervention group had serious adverse events in year 1 (9 events in 7 people), against 1% of controls. Only two events, both gallbladder or tummy pain in one person, were judged possibly related (Lean 2018). Medicines are stopped under supervision on these programmes, never on your own.
- Bariatric surgery: serious adverse events were more frequent after biliopancreatic diversion than with medical therapy (OR 2.7) (Mingrone 2021).
Who to see
- GP or practice nurse: diagnosis, annual reviews, most prescribing and referral to NHS remission or education programmes.
- Diabetes specialist team: for insulin starts, early-onset type 2, pregnancy, kidney disease, recurrent hypos, or uncertainty about whether diabetes is type 1 or type 2.
- Dietitian: NICE says dietary advice should come from someone with specific expertise in nutrition (NICE 1.3.1).
- Podiatrist, eye-screening service, kidney and heart specialists: for complications.
- Pharmacist: medicine reviews, interactions, and checks on supplements.
- A&E or 999: for any of the red flags above.
Follow the money: who funds the research, the guidelines and the treatments
Who profits either way? Drug and device makers profit when more people are diagnosed and treated with newer, patented medicines, monitors and pumps. Formula-diet companies and weight-management providers profit from remission programmes. Health systems and payers profit, in a different sense, when cheaper generics such as metformin, gliclazide and now dapagliflozin are used. Patient charities need both public donations and corporate partners. Nobody in this picture is free of interests, so we looked for agreement between sources with opposing interests.
Who funded the trials
The most-cited trials of SGLT-2 inhibitors and GLP-1 agonists were funded by their makers. EMPA-REG OUTCOME was funded by Boehringer Ingelheim and Eli Lilly (Zinman 2015). LEADER was funded by Novo Nordisk with the US National Institutes of Health (Marso 2016). CREDENCE was funded by Janssen (Perkovic 2019). Their findings were later pooled in an unfunded BMJ network meta-analysis, which rated the benefits as high certainty. Some of its authors declared payments from drug makers (Palmer 2021; full text). NICE's own evidence review, run by a publicly funded body, reached the same direction of effect (NICE). That agreement between maker-funded trials and public reviewers is the strongest signal we have.
The cleanest trials on lifestyle and drug comparisons were publicly funded. Look AHEAD was funded by the NIH and others (Look AHEAD), and GRADE by the NIDDK and others (GRADE). DiRECT was funded by Diabetes UK (Lean 2018). In its 2- and 5-year papers, one author reported employment and shares in Counterweight, the weight-management company that delivered the programme. Others reported fees or support from Cambridge Weight Plan, a formula-diet maker, and from several drug companies, all stated as outside the submitted work (Lean 2024).
Guideline panels and their conflicts
NICE (England). NICE is an executive body sponsored by the Department of Health and Social Care (framework agreement). Its 2026 medicines committee published a 32-page register of interests, last updated 11 February 2026 (register). It records the following:
- A consultant diabetologist member had given industry-funded education sessions (sponsors included AstraZeneca, Novartis, Boehringer Ingelheim and Novo Nordisk). The member agreed to stop taking industry funding while on the committee. Because the member was a site investigator in Novo Nordisk's FLOW trial of semaglutide in kidney disease, they were excluded from drafting the recommendations for people with diabetes and kidney disease.
- A cardiologist member had given talks on empagliflozin for Boehringer Ingelheim.
- A topic adviser's university had received speaker fees from AstraZeneca and Novo Nordisk on SGLT-2 inhibitors, plus investigator-initiated research funding from BMS, Novo Nordisk and AstraZeneca.
Most of these were classed as "declare and participate". This shows NICE's process working as designed: interests are declared and some members are excluded. It also shows that the clinicians with most experience of these drugs often have industry links.
American Diabetes Association (USA). The ADA says its Standards of Care are paid for from general revenue with no industry support. It requires a majority of the committee, and the chair, to be free of relevant conflicts, and members with a relevant interest cannot vote on that recommendation (ADA Introduction and Methodology 2026). However, the ADA's "general revenue" includes industry money. Its own unaudited disclosure lists $24,844,486 from drug and device makers in 2023. The largest amounts came from Eli Lilly ($7.27m), Abbott ($6.40m), Novo Nordisk ($2.74m), Genentech ($2.0m), Xeris ($1.85m) and Sanofi ($1.12m) (ADA 2023 disclosure). The figure for 2022 was $19,288,599 (ADA 2022 disclosure). An independent audit of the 2021 Standards found that 8 of 25 authors (32%) received industry payments in the year before publication and 12 (48%) within three years. The median was $33,262 per author in 2020. Only 29.3% of recommendations rested on the highest level of evidence (Shigeta 2023).
Diabetes UK, which funded DiRECT (the trial behind the NHS remission programme), reported income of £38.4m in 2024. Of that, £27.93m came from donations and legacies, and about £1.7m from government contracts and grants (Charity Commission). It also has corporate partners. Novo Nordisk has funded its Clinical Champions programme since 2014 and was a Silver Sponsor of its 2025 professional conference. Diabetes UK states that Novo Nordisk had no say over content (Diabetes UK). Sanofi is also listed as a long-standing partner (Diabetes UK).
International Diabetes Federation. The global prevalence figures quoted almost everywhere come from the IDF Diabetes Atlas. Its 11th edition was supported by Sanofi (2022–2024), Novo Nordisk (2022–2023) and a Pfizer-MSD Alliance grant (IDF).
The treatment industry and drug prices
The three big insulin makers are Novo Nordisk (Denmark), Eli Lilly (USA) and Sanofi (France). The Novo Nordisk Foundation holds its stake in the Novo Group companies through its wholly owned subsidiary Novo Holdings A/S (Novo Nordisk Foundation). A RAND study sponsored by the US Department of Health and Human Services found that US manufacturer gross prices for insulin averaged 9.71 times those in 33 other OECD countries. After estimated rebates, US net prices were still 2.33 times higher (Mulcahy & Schwam, RAND 2024).
From 1 January 2026, Medicare's first negotiated prices cut four diabetes medicines sharply for a 30-day supply. Jardiance (empagliflozin) went from a list price of $573 to $197. Farxiga (dapagliflozin) went from $556 to $178.50. Januvia (sitagliptin) went from $527 to $113. NovoLog/Fiasp (insulin aspart) went from $495 to $119 (CMS fact sheet). In England, NICE explicitly backed dapagliflozin while generic versions make it the cheapest SGLT-2 inhibitor (NICE).
Research funding at a glance
12 sources graded. The counts below are the independence tiers assigned in the scorecard above. They count sources, not funding amounts or evidence quality.
Public funding alone does not prove independence, and commercial funding alone does not prove a claim false. Named funders, countries and relationships are in the scorecard and source notes.
Related research
For wider, evidence-graded context, see these Pure City Research guides:
- Type 2 diabetes: prevention and management evidence
- Diabetes supplements: benefits, limits and safety
- Diabetes and heart disease
- Berberine: glucose evidence and safety
- Chronic kidney disease and the heart
- MASLD and MASH (fatty liver)
- Statins and other cholesterol-lowering medicines
- Blood pressure medicines
- Home blood pressure monitoring
- Peripheral arterial disease
- Obstructive sleep apnoea
- Psyllium husk: cholesterol and blood sugar
Frequently asked questions
What HbA1c level means type 2 diabetes?
An HbA1c of 48 mmol/mol (6.5%) or more, usually confirmed by a second test unless symptoms are obvious. Results of 39–47 mmol/mol (5.7–6.4%) fall in the ADA's prediabetes range (ADA 2026; WHO).
Can type 2 diabetes be reversed?
It can go into remission, meaning HbA1c below 48 mmol/mol off all diabetes medicines. In DiRECT, 46% reached remission at 1 year with a total diet replacement programme. Remission was much more likely with more weight lost, and many people relapsed by 5 years (Lean 2018; Lean 2024). Doctors call this remission, not cure, because glucose can rise again.
Is metformin still the first medicine?
In England, since February 2026, most adults are offered modified-release metformin together with an SGLT-2 inhibitor. If metformin is not suitable, an SGLT-2 inhibitor alone is offered (NICE).
Will I be offered Ozempic (semaglutide) for type 2 diabetes?
Under NICE's 2026 pathway, semaglutide up to 1 mg weekly is offered first-line to people who also have atherosclerotic cardiovascular disease. A GLP-1 agonist or tirzepatide may be considered for early-onset type 2 diabetes (NICE). Weight-management prescribing is covered by separate guidance.
What HbA1c target should I aim for?
NICE suggests 48 mmol/mol on diet or on medicines that do not cause hypos, and 53 mmol/mol on medicines that can. Targets can be relaxed for frailty, fall risk or limited life expectancy (NICE). The ADA uses below 7% (53 mmol/mol) for many adults (ADA).
Do I need to check my blood sugar at home?
Usually not, unless you take insulin, have hypos, take tablets that can cause hypos and drive or operate machinery, or are pregnant (NICE 1.6.2).
Does metformin cause vitamin B12 deficiency?
It commonly lowers B12, especially at higher doses and with longer use. The MHRA advises testing if symptoms such as numbness or anaemia appear (MHRA).
Can supplements like cinnamon or berberine replace diabetes medicines?
No. The evidence is small and mixed, and the US NCCIH warns against replacing medical treatment with unproven products (NCCIH).
Sources and funding notes
What we could not verify: We could not open the full text of the WHO 2011 HbA1c report or its expert declarations, only the WHO landing page. We could not open NICE Clinical Knowledge Summaries from outside the UK, so the UK "non-diabetic hyperglycaemia" range is not cited. We could not view the full funding statements for UKPDS 34, the 2011 exercise meta-analysis, the full "and others" funders list for Look AHEAD and GRADE, or the full DiRECT 2018 declaration (the 2- and 5-year declarations are cited). We did not verify the MHRA's current share of income from industry fees, or WHO's donor breakdown for the 2011 consultation. The IDF's 2024 headline prevalence figure could not be confirmed from a page we opened, so we cite only its 2050 projection and spending estimate. Pure City Research does not yet have dedicated pages for individual diabetes medicines (metformin, SGLT-2 inhibitors, GLP-1 agonists), so drug classes are summarised here rather than linked. The ADA industry-revenue figures are the association's own unaudited drafts.
- NICE NG28: Type 2 diabetes in adults: management (last updated 18 February 2026), including chapters on initial medicines, further medication, blood glucose management, dietary advice and update information — guideline; UK government (DHSC).
- NICE Type 2 Diabetes (medicines update) committee register of interests (updated 11 February 2026) — conflicts disclosure; NICE.
- DHSC and NICE framework agreement — NICE governance and sponsorship; UK government.
- WHO. Use of glycated haemoglobin (HbA1c) in diagnosis of diabetes mellitus (13 January 2011) — diagnostic standard; WHO.
- ADA. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes—2026 (PubMed) — diagnostic criteria; ADA general revenue.
- ADA. Introduction and Methodology: Standards of Care in Diabetes—2026 — funding and conflict rules; ADA.
- ADA. 6. Glycemic Goals, Hypoglycemia, and Hyperglycemic Crises—2026 — targets; ADA.
- ADA. Revenue received from pharmaceutical/device makers, 2023 (unaudited draft) and 2022 — self-disclosure; ADA.
- Shigeta H et al. Financial conflicts of interest and quality of evidence underlying the ADA Standards of Medical Care in Diabetes, 2021. Cureus 2023 (full text) — independent audit; Japan; no external funding statement found.
- Lean MEJ et al. DiRECT 12-month results. Lancet 2018 — RCT; funded by Diabetes UK.
- Lean MEJ et al. DiRECT 2-year results. Lancet Diabetes Endocrinol 2019 — RCT; Diabetes UK.
- Lean MEJ et al. DiRECT 5-year extension. Lancet Diabetes Endocrinol 2024 — extension study; Diabetes UK; author ties to Counterweight and Cambridge Weight Plan declared.
- Valabhji J et al. Early findings from the NHS Type 2 Diabetes Path to Remission Programme. Lancet Diabetes Endocrinol 2024 — service evaluation; no funding; NHS England authors.
- Look AHEAD Research Group. Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes. NEJM 2013 — RCT; NIH and others.
- GRADE Research Group (Nathan DM et al.). Glycemia reduction in type 2 diabetes – glycemic outcomes. NEJM 2022 — RCT; NIDDK and others.
- UKPDS Group. Intensive blood-glucose control with metformin (UKPDS 34). Lancet 1998 — RCT; funding not verified.
- Umpierre D et al. Physical activity advice or structured exercise and HbA1c. JAMA 2011 — meta-analysis; Brazil; funding not verified.
- Goldenberg JZ et al. Low and very low carbohydrate diets for type 2 diabetes remission. BMJ 2021 — meta-analysis; Texas A&M University support.
- Mingrone G et al. Metabolic surgery vs medical therapy, 10-year follow-up. Lancet 2021 — RCT; Fondazione Policlinico Gemelli (Italy).
- Palmer SC et al. SGLT-2 inhibitors and GLP-1 receptor agonists: network meta-analysis. BMJ 2021 (full text) — meta-analysis; no funding; some authors with industry ties.
- Zinman B et al. EMPA-REG OUTCOME. NEJM 2015 — RCT; funded by Boehringer Ingelheim and Eli Lilly.
- Marso SP et al. LEADER. NEJM 2016 — RCT; funded by Novo Nordisk and NIH.
- Perkovic V et al. CREDENCE. NEJM 2019 — RCT; funded by Janssen.
- MHRA. Metformin and reduced vitamin B12 levels (20 June 2022) — regulator.
- MHRA. SGLT2 inhibitors: monitor ketones during treatment interruption (18 March 2020) — regulator.
- MHRA. GLP-1 and GLP-1/GIP receptor agonists: strengthened warnings on acute pancreatitis (29 January 2026) — regulator.
- MHRA. Women on "skinny jabs" must use effective contraception (5 June 2025) — regulator.
- NHS: Symptoms of type 2 diabetes and how it's diagnosed, What is type 2 diabetes?, Complications, DKA and hypoglycaemia — patient information; UK government (NHS England).
- NCCIH. Diabetes and Dietary Supplements: What You Need To Know (November 2023) — US government (NIH).
- Mulcahy AW, Schwam D. Comparing insulin prices in the United States to other countries. RAND Health Quarterly 2024 — sponsored by HHS ASPE.
- CMS. Medicare Drug Price Negotiation Program: negotiated prices for initial price applicability year 2026 — US government.
- IDF Diabetes Atlas FAQs (11th edition) — prevalence modelling; sponsored by Sanofi, Novo Nordisk, Pfizer-MSD Alliance.
- Charity Commission: The British Diabetic Association (Diabetes UK), financial year 2024 — regulator record.
- Diabetes UK: Our partnership with Novo Nordisk and Sanofi — charity self-disclosure.
- Novo Nordisk Foundation: about the Foundation — ownership structure; Denmark.
Educational research reviewed 7 October 2026. This article is not medical advice; talk to a qualified clinician or pharmacist before starting, stopping or combining treatments.
Have a question — or want us to cover something?
Ask about anything on this page, or request the next deep dive: an ingredient, a supplement, or a health concern. We use published research, evidence syntheses, and regulatory guidance, with clear source links.
One daily research roundup
Get the topics, key findings and links from our new articles in one email. At most one digest a day, only when there is something new.