Diabetes Diagnosis and Classification Pathway
Confirming the diagnosis, ranking the type, and sending the test that settles · v1Stage 1. Is This Diabetes, and Can the HbA1c Be Believed
Stage 2. The Phenotype
Everything below is answerable at the bedside. This stage alone produces the ranked differential; the tests in stage 3 refine it.Stage 3. The Tests the Phenotype Has Made Worth Sending
Evidence & Clinical Pearls
1. The Classification That Is Official, and the Numbers That Are Not
Patients and residents both arrive having read about type 3, type 4 and type 5 diabetes. Some of those terms describe real entities and some do not, and the difference matters because one of them is a hypothesis about Alzheimer's disease. The American Diabetes Association recognises four headings: type 1, type 2, specific types due to other causes, and gestational diabetes. WHO's 2019 revision adds a fifth grouping, hybrid forms, which is where slowly evolving immune-mediated diabetes of adults and ketosis-prone type 2 diabetes sit.
| Term | What it is used to mean | Standing |
|---|---|---|
| Type 1 | Autoimmune beta cell destruction, 1A with antibodies and 1B idiopathic | Official, ADA and WHO |
| Type 2 | Insulin resistance with progressive secretory failure | Official, ADA and WHO |
| Type 1.5 | Almost always LADA. A minority of writers use it for double diabetes instead | Colloquial. The entity is real; ADA files it inside type 1, WHO under hybrid forms as slowly evolving immune-mediated diabetes of adults |
| Type 2.5 | Nothing fixed | No definition in the peer-reviewed literature. The entity people usually mean is double diabetes, named by Teupe and Bergis in 1991: type 1 diabetes with superimposed insulin resistance. Use that name |
| Type 3 | Two unrelated things | As a name for Alzheimer's disease and brain insulin resistance it is a research hypothesis, proposed by Steen, de la Monte and colleagues in 2005, and is not a form of diabetes mellitus. Used loosely for pancreatogenic diabetes it is imprecise; say type 3c |
| Type 3c | Diabetes of the exocrine pancreas | The entity is official under "diseases of the exocrine pancreas". The ADA's preferred name is pancreatic diabetes; 3c is the popular label |
| Type 4 | Age-related insulin resistance in lean individuals | Not a recognised human category. The term comes from a 2015 mouse study from the Salk Institute on fat-resident regulatory T cells. Human work is ongoing |
| Type 5 | Malnutrition-related diabetes in lean people with a history of undernutrition | Newly named. Classified at an expert meeting at CMC Vellore in January 2025, endorsed by the International Diabetes Federation at the 2025 Bangkok congress. Formal diagnostic criteria have not yet been published; a working group is writing them |
2. The Four Questions That Do Most of the Typing
Before any antibody or C-peptide, four answers narrow the field further than most people expect. How old were they at diagnosis. How thin are they. Did they need insulin, and how soon. What is the family pattern. A lean 22 year old with a parent and a grandparent affected, no ketosis and control on a tablet is a different problem from a lean 22 year old with ketoacidosis and no family history, and the two need opposite investigations.
The difficulty is the middle of the range. A 35 year old with a body mass index of 24 who was labelled type 2 four years ago and is failing tablets is the commonest classification error in Indian practice, and the answer is a GAD65 and a C-peptide, not another agent.
3. Latent Autoimmune Diabetes in Adults, and the Drug It Inverts
The Immunology of Diabetes Society definition is three items: age at diagnosis above 30, at least one islet autoantibody, and no insulin requirement for at least six months after diagnosis. The 2020 international consensus adds that none of the three is categorical.
To decide whom to test, the Fourlanos clinical screening tool is five yes-or-no items: onset before 50, acute osmotic symptoms, body mass index below 25, personal autoimmune disease, family autoimmune disease. Two or more means send the antibody, at a sensitivity of 90 per cent and a specificity of 71 per cent. Note what those numbers mean: with a positive predictive value of about 21 per cent this is a rule-out tool. A score of 0 or 1 has a negative predictive value of 99 per cent and lets you stop; a score of 2 does not make the diagnosis, it buys the test. Note also that its age item is 50 while the diagnostic criterion is 30. They are different numbers doing different jobs.
A classification error here inverts the correct drug. The consensus panel does not recommend sulfonylureas in LADA, because accelerated loss of beta cell function on them cannot be ruled out; the evidence is one small Japanese trial in which C-peptide fell faster on a sulfonylurea than on insulin. In HNF1A and HNF4A MODY a low-dose sulfonylurea is first-line and often replaces insulin. Two lean, young, antibody-negative-looking patients, and the same tablet is the wrong answer for one and the right answer for the other.
4. Type 2 Diabetes in India Is Not the Textbook Picture
ICMR-INDIAB, reported in 2023 from 113,043 people across mainland India, found 11.4 per cent with diabetes and 15.3 per cent with prediabetes, roughly 101 million and 136 million people. Two features of the Indian phenotype matter for classification rather than for management.
It arrives about a decade early. Prevalence peaks at 52 to 56 years in South Asians against 64 to 68 in white Europeans, and INDIAB documented a sharp rise between 25 and 34. A diagnosis at 30 is therefore much weaker evidence against type 2 here than the Western literature implies. And body mass index misleads. A nationwide Indian registry of 37,927 people with type 2 diabetes found only 16 per cent had a body mass index below 25, yet around three quarters met a body-fat-percentage definition of obesity: 77 per cent of men and 71 per cent of women. That is the thin-fat phenotype, and it is why this tool reads the body mass index against Asian cut-offs and weighs acanthosis and a metabolic profile alongside it rather than trusting the number alone. Estimates of how much Indian type 2 diabetes is lean vary widely with the setting, from 16 per cent in that urban registry to 52 per cent in a rural tertiary series, and there is no reliable national figure.
5. MODY, and Why This Tool Links Out Instead of Calculating
Maturity-onset diabetes of the young is fewer than 5 per cent of all diabetes and is dominated by three genes. The Exeter best-practice criteria for sending a test are: diagnosis before 25 in at least one family member, diabetes in two or three successive generations in a parent-to-child pattern, absent islet autoantibodies, and preserved C-peptide beyond the honeymoon period, in someone who is not obese and not ketosis-prone.
| Gene | The clue | Treatment |
|---|---|---|
| GCK | Mild fasting hyperglycaemia present from birth and stable for years, HbA1c rarely above 7.5 per cent, small increment on a tolerance test, one parent with a mildly raised fasting glucose. Often found incidentally or in pregnancy | None. Treatment does not improve it and may harm. Microvascular complications are minimal. The exception is pregnancy where the fetus has not inherited the variant |
| HNF1A | Progressive, young onset, multigenerational, glycosuria at a blood glucose below 10 mmol/L because the renal threshold is low, antibodies absent, striking sulfonylurea sensitivity | Low-dose sulfonylurea first-line. Patients frequently come off insulin. Watch for hypoglycaemia at small doses |
| HNF4A | As HNF1A but without the low renal threshold, plus macrosomia in about 56 per cent and transient neonatal hypoglycaemia in about 15 per cent | Sulfonylurea sensitive, managed as HNF1A |
The Exeter MODY probability calculator is the right tool and this module deliberately does not clone it. Its published paper gives the beta coefficients but not the intercept, so the equation cannot be reconstructed to produce a calibrated probability, and its own instructions record that the model is not validated in non-White populations. Both of those are reasons to use it directly and read its output with that caveat in mind, rather than to reproduce it here with a number invented to fill the gap. The result panel links to it.
Indian MODY may not be the same disease genetically. A CMC Vellore study using targeted sequencing in 56 clinically diagnosed patients found a pathogenic variant in only 19 per cent, against about 30 per cent or more in Western laboratories, and among those found the commonest gene was NEUROD1 rather than GCK or HNF1A, with HNF4A next. That was eleven mutation-positive patients in total, so it is not a frequency estimate and should not be quoted as one, but it is a real signal that Western gene panels and Western prediction models may underperform here.
6. The Pancreas: Fibrocalculous Pancreatic Diabetes and the Rest of Type 3c
Fibrocalculous pancreatic diabetes is a tropical disease and southern India has the highest known prevalence. The criteria are occurrence in a tropical country, diabetes by standard criteria, evidence of chronic pancreatitis, which means either pancreatic calculi on a plain film or at least three of abnormal pancreatic morphology on imaging, chronic abdominal pain since childhood, steatorrhoea and an abnormal pancreatic function test, and the absence of other causes such as alcohol. The picture is a lean adolescent or young adult with years of recurrent abdominal pain, then brittle insulin-requiring diabetes that is characteristically ketosis-resistant despite marked hyperglycaemia. Diabetes usually appears one to two decades after the pain begins.
It is uncommon and becoming less common: a Chennai population survey put it at 0.36 per cent of self-reported diabetes and 0.019 per cent of the general population. Two things make it worth recognising anyway. Pancreatic enzyme replacement changes how the patient lives. And the risk of pancreatic carcinoma is the highest of any form of pancreatitis, quoted at around a hundredfold, which makes new pain, jaundice or weight loss in an established case an urgent problem rather than a routine one.
The wider category, diabetes of the exocrine pancreas, is dominated by chronic pancreatitis, which is about 80 per cent of it, with pancreatic cancer, pancreatectomy, cystic fibrosis and haemochromatosis making up most of the rest. The Ewald and Bretzel criteria are three: exocrine insufficiency on faecal elastase or a direct function test, an abnormal pancreas on imaging, and absent type 1 autoantibodies. What distinguishes it from type 2 at the bedside is that glucagon and pancreatic polypeptide are lost alongside insulin, so control is brittle and hypoglycaemia is frequent, and that fat-soluble vitamin deficiency needs looking for.
7. Type 5 Diabetes, and the History Behind It
WHO created a category of malnutrition-related diabetes in 1985 with two subtypes, fibrocalculous pancreatic diabetes and protein-deficient pancreatic diabetes, then deleted the whole category in 1999 over a disagreement about whether undernutrition alone could cause diabetes. Type 5 diabetes is the return of the second limb, not the first: fibrocalculous pancreatic diabetes remained in the classification under diseases of the exocrine pancreas throughout, and is an explicit exclusion when considering type 5.
The phenotype in the consensus statement is a lean young person, body mass index below 18.5, with a history of early-life undernutrition, marked impairment of insulin secretion but normal hepatic and peripheral insulin sensitivity, resistance to ketosis, and negative islet autoantibodies. Formal diagnostic criteria have not been published. A working group co-chaired from Albert Einstein College of Medicine and CMC Vellore was announced in April 2025 to write them. Until they appear, this tool raises type 5 as a consideration and says plainly that the criteria are not settled, rather than inventing a threshold.
8. Ketosis-Prone Diabetes: the Patient Who Presents in Ketoacidosis and Does Not Have Type 1
Ketoacidosis at presentation does not prove type 1. Ketosis-prone diabetes, first described in Brooklyn and still sometimes called Flatbush diabetes, presents with ketoacidosis and then behaves like type 2, with many patients coming off insulin entirely within months. The Aβ classification sorts them on two axes: A is autoantibody status, positive or negative, and β is beta cell reserve on C-peptide, with 1 ng/mL as the divide. The two β-positive groups can usually be withdrawn from insulin with monitoring; the two β-negative groups stay insulin-dependent.
This is not a Western curiosity. A CMC Vellore series of 51 newly diagnosed patients admitted with ketoacidosis found 11, about a fifth, were antibody-negative with preserved beta cell function. That denominator is admissions with ketoacidosis, not all diabetes, so it is not a prevalence figure. The practical point stands: in a patient who presented in ketoacidosis, an antibody and a C-peptide at three months decide whether insulin is for life or for a season.
9. Drugs, Transplantation, and the Glucose Profile Steroids Produce
Glucocorticoids are the commonest cause, with reported rates from 2 per cent in general practice to 32 per cent after transplantation or in rheumatoid arthritis. Second-generation antipsychotics roughly double to treble the background rate. Checkpoint inhibitors are the newest and the most abrupt, and can precipitate genuine autoimmune type 1 diabetes with ketoacidosis. Also on the list, and easy to forget on an Indian ward: isoniazid, thiazides at higher dose, protease inhibitors, pentamidine, interferon and diazoxide.
The steroid glucose profile is the practical pearl. With a single morning dose of an intermediate-acting steroid, hyperglycaemia appears in the afternoon and evening while the fasting glucose stays normal. A fasting sample therefore underestimates it, and so does a tolerance test, which is performed fasting. HbA1c is useful only after about two months of therapy. The recommended screen is a post-lunch or pre-dinner glucose, and any value above 200 mg/dL at any time of day is the most useful single criterion.
Post-transplant diabetes mellitus replaced the older term new-onset diabetes after transplantation, because effective pre-transplant screening is impractical in most centres and the old name assumed it had been done. It is diagnosed once the patient is stable on maintenance immunosuppression with stable graft function and no acute infection, not in the first weeks; the 2024 international consensus screens with a tolerance test at 10 to 13 weeks, and notes that HbA1c lacks sensitivity here. In the first six weeks, afternoon capillary glucose is the more sensitive monitor. Tacrolimus carries more risk than ciclosporin; steroid avoidance reduces the risk with a risk ratio of 0.70.
10. Hyperglycaemia in Pregnancy: Two Tests That Disagree
The Indian national programme uses the DIPSI test: 75 g of glucose irrespective of the last meal, one venous sample at two hours, and 140 mg/dL or above is gestational diabetes. It is done at the first antenatal contact and repeated at 24 to 28 weeks with at least four weeks between. It exists because bringing a fasting woman back for a three-sample test is often not feasible, and it is national guidance.
It is also less sensitive than the alternative, because it captures no fasting value. One analysis found it missed about 16 per cent of women identified by the IADPSG criteria, and a Sri Lankan study reported a sensitivity of 40.6 per cent. The IADPSG and WHO 2013 criteria use a fasting 75 g test, and one abnormal value of three diagnoses gestational diabetes: fasting 92, one hour 180, two hours 153 mg/dL.
The distinction that changes management is gestational diabetes versus diabetes in pregnancy. A fasting glucose of 126 or above, a two-hour value of 200 or above, or a random of 200 or above with symptoms is overt diabetes, which usually means pre-existing undiagnosed diabetes: it needs immediate treatment intensity and fetal anomaly surveillance, and it will not resolve after delivery. Reclassify everyone with a 75 g tolerance test at six weeks post partum, and read it against the ordinary non-pregnant criteria rather than the simplified table in the national guideline, which calls a fasting glucose of 115 mg/dL normal.
11. Reading the Antibodies and the C-peptide
If you can afford one antibody, send GAD65. A negative antibody does not exclude type 1: about 5 to 10 per cent of type 1 diabetes is antibody-negative and is classified as type 1B. A positive antibody in an adult who did not need insulin for six months is LADA.
C-peptide is only informative after about three years, because residual secretion persists in genuine type 1 for longer than most people expect. The ADA and EASD consensus reads a random sample taken within five hours of eating, with the glucose measured on the same sample: below 200 pmol/L is severe insulin deficiency and consistent with type 1; 200 to 600 pmol/L is compatible with type 1 or MODY; 600 pmol/L or above strongly suggests type 2. A low value is only interpretable if the paired glucose was 72 mg/dL or above, otherwise the low C-peptide may simply be an appropriate response and the test should be repeated. The urinary C-peptide creatinine ratio is an alternative at five years or more, taken two hours after the largest meal, with below 0.2 nmol/mmol indicating absolute deficiency at a sensitivity of 100 per cent and a specificity above 95 per cent.
A note on units, because it has caused published confusion: the urinary ratio is nmol/mmol, not nmol/mol. And to convert C-peptide, 1 ng/mL is about 331 pmol/L.
12. Which Threshold, and Why the Fasting Glucose Answer Depends on It
Diabetes is agreed everywhere: fasting 126 mg/dL or above, two-hour 200 or above, HbA1c 6.5 per cent or above, or a random of 200 or above with classical symptoms, and in the absence of unequivocal hyperglycaemia two abnormal results are needed, either from the same sample or from two samples.
Prediabetes is not agreed. The ADA calls a fasting glucose of 100 to 125 mg/dL impaired fasting glucose. WHO, and following it the RSSDI 2022 recommendations, use 110 to 125. A patient with a fasting glucose of 105 is prediabetic by the ADA and normal by RSSDI. ICMR's own 2018 guideline prints both ranges side by side without adjudicating between them. This tool reports the fasting glucose against both and says which gives which answer, because there is no honest way to pick one for a reader who may be following either.
13. What This Tool Deliberately Does Not Do
It does not compute a probability. The ranking it produces is a structured clinical weighting written for this tool, and it says so on every run: no validated model exists for typing diabetes from clinical features alone, and inventing a percentage would be inventing precision. It does not reimplement the Exeter MODY calculator, for the reasons in pearl 5. It does not make a molecular diagnosis or predict a gene. It does not manage the diabetes it classifies, does not calculate an insulin dose, and does not stage complications. And it does not screen populations: every weighting in it assumes a patient who already has hyperglycaemia and a question about what kind.
Algorithm References & Evidence Base
- American Diabetes Association Professional Practice Committee. Diagnosis and classification of diabetes: standards of care in diabetes 2026. Diabetes Care. 2026;49(suppl 1):S27-S49. The publisher's full text could not be read during the building of this module; the thresholds used here were taken from a transcription of the 2026 criteria table and are the long-standing values.
- World Health Organization. Classification of diabetes mellitus. Geneva: WHO; 2019. Source of the hybrid forms grouping, read here through secondary summaries because the WHO text returned an access error.
- Buzzetti R, Tuomi T, Mauricio D, et al. Management of latent autoimmune diabetes in adults: a consensus statement from an international expert panel. Diabetes. 2020;69(10):2037-2047.
- Fourlanos S, Perry C, Stein MS, Stankovich J, Harrison LC, Colman PG. A clinical screening tool identifies autoimmune diabetes in adults. Diabetes Care. 2006;29(4):970-975. The five items, the cut-off of 2 and the figures of 90 and 71 per cent were confirmed from two independent secondary sources; the primary paper could not be read directly.
- Ellard S, Bellanne-Chantelot C, Hattersley AT. Best practice guidelines for the molecular genetic diagnosis of maturity-onset diabetes of the young. Diabetologia. 2008;51(4):546-553.
- Shields BM, McDonald TJ, Ellard S, Campbell MJ, Hyde C, Hattersley AT. The development and validation of a clinical prediction model to determine the probability of MODY in patients with young-onset diabetes. Diabetologia. 2012;55(5):1265-1272. Publishes the beta coefficients but not the model intercept, which is why this module links to the Exeter calculator rather than reproducing it.
- Chapla A, Mruthyunjaya MD, Asha HS, et al. Maturity onset diabetes of the young in India: a distinctive mutation pattern identified through targeted next-generation sequencing. Clin Endocrinol (Oxf). 2015;82(4):533-542. Eleven mutation-positive patients; a signal about the Indian gene distribution, not a frequency estimate.
- Holt RIG, DeVries JH, Hess-Fischl A, et al. The management of type 1 diabetes in adults: a consensus report by the American Diabetes Association and the European Association for the Study of Diabetes. Diabetologia. 2021;64(12):2609-2652. Source of the C-peptide bands and the paired-glucose rule.
- Hope SV, Wienand-Barnett S, Shepherd M, et al. Practical classification guidelines for diabetes in patients treated with insulin: a cross-sectional study of the accuracy of diabetes diagnosis. Br J Gen Pract. 2016;66(646):e315-e322. Source of the urinary C-peptide creatinine ratio threshold.
- Balasubramanyam A, Garza G, Rodriguez L, et al. Accuracy and predictive value of classification schemes for ketosis-prone diabetes. Diabetes Care. 2006;29(12):2575-2579. The A-beta scheme; the primary paper could not be read directly and the scheme is quoted here from secondary reviews.
- Das Gupta R, Ramachandran R, Gangadhara P, et al. Clinical characteristics, beta-cell dysfunction and treatment outcomes in patients with A-negative beta-positive ketosis-prone diabetes. J Diabetes Complications. 2017;31(9):1401-1407. Christian Medical College, Vellore.
- Unnikrishnan AG, Bhake RC, Kumaran S, Kalra S. Fibrocalculous pancreatic diabetes. In: Endotext. South Dartmouth: MDText.com; updated 13 September 2024. Source of the diagnostic criteria used here.
- Mohan V, Farooq S, Deepa M. Prevalence of fibrocalculous pancreatic diabetes in Chennai in south India. JOP. 2008;9(4):489-492.
- Ewald N, Bretzel RG. Diabetes mellitus secondary to pancreatic diseases (type 3c): are we neglecting an important disease? Eur J Intern Med. 2013;24(3):203-206. Criteria read through a secondary reproduction; the primary returned an access error.
- Wadivkar P, Jebasingh F, Thomas N, et al. Classifying a distinct form of diabetes in lean individuals with a history of undernutrition: an international consensus statement. Lancet Glob Health. 2025;13(10):e1771-e1776. The Vellore Declaration. Abstract read; the full text returned an access error.
- International Diabetes Federation. New IDF type 5 diabetes working group. Brussels: IDF; 15 April 2025.
- Sharif A, Chakkera H, de Vries APJ, et al. International consensus on post-transplantation diabetes mellitus. Nephrol Dial Transplant. 2024;39(3):531-549.
- Suh S, Park MK. Glucocorticoid-induced diabetes mellitus: an important but overlooked problem. Endocrinol Metab (Seoul). 2017;32(2):180-189. Source of the afternoon and evening glucose profile and of the screening test to use.
- Scheen AJ, Feve B. When therapeutic drugs lead to diabetes. Diabetologia. 2022;65(5):751-762.
- Anjana RM, Unnikrishnan R, Deepa M, et al. Metabolic non-communicable disease health report of India: the ICMR-INDIAB national cross-sectional study. Lancet Diabetes Endocrinol. 2023;11(7):474-489. Figures taken from a secondary report; the primary returned an access error.
- Research Society for the Study of Diabetes in India. RSSDI clinical practice recommendations for the management of type 2 diabetes mellitus 2022. Source of the Indian impaired fasting glucose range of 110 to 125 mg/dL.
- Ministry of Health and Family Welfare, Government of India. Diagnosis and management of gestational diabetes mellitus: national guidelines. New Delhi: National Health Mission; 2018. Source of the DIPSI protocol and the six-week post-partum reclassification.
- World Health Organization. Diagnostic criteria and classification of hyperglycaemia first detected in pregnancy. Geneva: WHO; 2013. Source of the IADPSG thresholds and of the distinction between gestational diabetes and diabetes in pregnancy.
- Steen E, Terry BM, Rivera EJ, et al. Impaired insulin and insulin-like growth factor expression and signaling mechanisms in Alzheimer's disease: is this type 3 diabetes? J Alzheimers Dis. 2005;7(1):63-80. Cited to identify the origin of a term that is not a form of diabetes mellitus.
- Bapat SP, Myoung Suh J, Fang S, et al. Depletion of fat-resident Treg cells prevents age-associated insulin resistance. Nature. 2015;528(7580):137-141. The mouse study behind the term type 4 diabetes.
How to Cite This Tool
AMA Style:
Umakanth S. Diabetes Diagnosis and Classification Pathway. MEDiscuss. Published 2026. Accessed .
Vancouver Style:
Umakanth S. Diabetes Diagnosis and Classification Pathway [Internet]. MEDiscuss.org; 2026 [cited ]. Available from:
Save this calculation
Whatever you type here stays on this device and is not sent anywhere. The server receives only a scrambled code made from it, so nobody with access to the server can tell which patient a saved calculation belongs to. Enter the same nickname the next time to see that patient's earlier values. What is stored