A man in his 40s came to my emergency department on a Sunday morning with an abscess in his groin. An abscess is a collection of infection under the skin, and it is painful and embarrassing and it is exactly the sort of thing you finally give in and come to an emergency department about on a weekend. We drained it. Because a groin abscess in a man his age makes me want to know his blood sugar, we checked a glucose. It came back at 440 mg/dL.
He went home with antibiotics, a prescription for metformin, and a diagnosis of type 2 diabetes he had no idea he had. He was 44. He had moved states a few years earlier and had not gotten around to finding a new doctor, so nothing about him had been measured by anybody in a long time.
A fasting glucose of 440 does not happen over a weekend. It takes years, and for most of those years a routine blood test would have looked normal. That is not a story about willpower. It is a story about measurement, and about what happens when nobody is looking.
This article is the companion to the second episode of Lab Results Explained. On paper it is about your A1c. Really it is about one idea: most numbers used to assess your blood sugar describe a single moment, and the A1c does not. It describes the last few months, and it does not care what you had for breakfast.
The physiology, and why none of the rest makes sense without it
When you eat, carbohydrates are broken down into glucose, which enters your bloodstream. Your pancreas sees that sugar, and specialized cells inside it called beta cells release insulin. Insulin acts like a key on the glucose gates of your muscle, fat, and liver, letting those tissues pull glucose out of the blood to use or store. Simply put, that is the whole system, and it can fail in two very different ways.
Type 1 diabetes
In type 1, the immune system destroys the beta cells, so the factory is gone and insulin has to come from outside the body as a medication. This is an autoimmune disease. It is not caused by diet, by eating too much sugar, or by anything the patient did. It usually appears in younger people, though it can happen at any age. The 2026 American Diabetes Association standards now recommend autoantibody screening for presymptomatic type 1 in people at elevated genetic risk.
Type 2 diabetes
Type 2 is a different failure. The factory works. The lock and the gate are faulty. Two things happen at once: insulin resistance, where muscle, fat, and liver stop responding properly to the same amount of insulin, and progressive beta cell failure, where the factory gets worse at producing. Both take years, which is why so many people are diagnosed at the same appointment they are diagnosed with a complication, exactly as happened in the case above.
Here is the part that matters most for everything below. When insulin resistance rises, your pancreas does not simply let your blood sugar climb. It compensates. Those beta cells crank output up several-fold to hold glucose in the normal range, and they do that without symptoms and without a single abnormal number on a standard panel, for a very long time. That is not controversial. It is how the machine works, and further down this page you will see it framed in a much less honest way.
A quick word on terminology. Insulin resistance, insulin sensitivity, and glucose tolerance all describe the same axis: how well insulin clears glucose from your blood. If you only ever hear one of those phrases, you may conclude that either glucose or insulin single-handedly causes diabetes. The cause is more nuanced than that. It is complex, heavily genetic, and shows up when a particular set of genes collides with an environment that makes excess energy easy to consume.
Why the fat matters
The best account of why the resistance starts comes from Roy Taylor and his group in Newcastle, England, and it is called the twin cycle hypothesis. When fat accumulates in the liver, the liver exports excess fat to the rest of the body, including the pancreas. Fat inside the pancreas interferes with how well beta cells work, which raises glucose, which drives more fat into the liver. It is not one organ failing. It is two organs pulling each other down in a cycle, and a cycle can in principle be run backwards.
Choose your parents wisely
More than 700 gene loci are associated with the heritability of type 2 diabetes, and family and twin studies report heritability estimates ranging from 20 to 80 percent. That is a wide range and it reflects real uncertainty, but the direction is not in doubt. So when somebody online tells you type 2 diabetes is purely a lifestyle disease, they either do not know the literature or there is something at the end of the story to buy.
The three tests we actually use
The first is a fasting plasma glucose. You do not eat for 8 hours, we take blood, we measure the sugar in it. This is the test that caught the patient above.
The second is an oral glucose tolerance test, which you may know from pregnant friends, because it is recommended in the second trimester. You fast, we draw blood, you drink a solution containing 75 grams of glucose, and we measure again 2 hours later. It is a stress test for your glucose disposal system, and it is the most sensitive of the three as well as the most inconvenient, which is why it is used far less often than it should be outside pregnancy. For scale, a small cookie dough blizzard contains roughly 78 grams of sugar, which is context for the corners of the internet describing this test as a poisonous dose.
The third is hemoglobin A1c, and it is the one this episode is named after.
What the A1c actually measures
Glucose in your blood sticks to the hemoglobin inside your red blood cells, a process called glycation. The more glucose there is, and the longer it is there, the more of your hemoglobin ends up carrying sugar. Red blood cells live about 120 days, so the percentage of your hemoglobin carrying glucose is an average of what your blood sugar has done over roughly the last 3 months. A fasting glucose is a photograph. The A1c is closer to a movie.
That buys you two things. You do not have to fast, which removes a real barrier. And the result is not thrown off by one meal or a short illness, which matters enormously in my world. Somebody arrives with pneumonia and a blood sugar of 240, and I cannot tell from that number alone whether they have diabetes or are simply sick, because illness and exercise both release stored glucose to supply fuel. An A1c tells me what the sugar has been doing outside the current event. The exception worth naming is diabetic ketoacidosis, where days of extreme hyperglycemia can push an A1c into the double digits.
The numbers
The American Diabetes Association and the US Preventive Services Task Force use the same diagnostic thresholds.
Test
Normal
Prediabetes
Diabetes
Fasting plasma glucose
Under 100 mg/dL
100 to 125 mg/dL
126 mg/dL or higher
Hemoglobin A1c
Under 5.7%
5.7% to 6.4%
6.5% or higher
2-hour glucose (OGTT)
Under 140 mg/dL
140 to 199 mg/dL
200 mg/dL or higher
Random glucose
Not diagnostic alone
Not diagnostic alone
200 mg/dL or higher with classic symptoms
One caveat matters. Unless the hyperglycemia is unequivocal, the diagnosis needs confirming with a repeat test. A single abnormal number is a reason to test again, not a reason to panic, which is the central message of the pillar episode in this series. The patient in the case did not need a second test. A fasting glucose of 440 alongside an abscess is unequivocal, and that is why he left with a prescription.
Prediabetes is the condition that ticks along silently in the background, and it is exactly what a regular primary care relationship is for. Somebody monitoring that band can intervene before the line is crossed. And sometimes, despite doing everything right, genetics get in the way and a person crosses it anyway.
Who should be tested
The two major bodies agree on the diagnostic numbers and differ slightly on who to screen. Neither is far from the other.
US Preventive Services Task Force
American Diabetes Association
Who to screen
Adults aged 35 to 70 who are overweight or obese
All adults from age 45, and at any age with overweight or obesity plus 1 risk factor
Strength
B recommendation
Standards of Care recommendation
Interval
Every 3 years if normal
Interval based on results and risk
Notes
Start age lowered from 40 to 35 in 2021. Screen earlier in higher-prevalence groups, and at a BMI of 23 or above in Asian American patients
Casts a deliberately wider net to reduce missed diagnoses
The Task Force also made a finding that surprises people. They concluded there is insufficient direct evidence that screening itself improves mortality or cardiovascular outcomes, and the 2 trials testing screening head to head found no mortality benefit over about 10 years, which may simply be too short a window. So why screen? Because of what sits on either side of the test. Intervening in prediabetes reduces progression to diabetes, and there is adequate evidence that treating newly diagnosed diabetes reduces all-cause mortality, diabetes-related death, and heart attacks over 10 to 20 years. You cannot treat what you have not found.
Where does the patient in the case sit? At 44 he was inside the Task Force window and just under the ADA universal age. Under either set of rules, if he had walked into a primary care office somebody would very likely have checked him. He did not fall through a gap in the guidelines. He fell through the gap between having a guideline and having a doctor.
When the A1c lies to you
The A1c is an indirect measurement. We are not measuring your glucose. We are measuring what your glucose has done to your red blood cells. So anything that changes your hemoglobin, or how long your red cells live, changes the answer without your blood sugar changing at all. The 2026 guideline lists the main offenders: anemia, glucose-6-phosphate dehydrogenase deficiency, recent transfusion, drugs that stimulate red cell production, kidney failure, pregnancy, thalassemia, and folate deficiency. In sickle cell disease and other homozygous hemoglobin variants, the A1c cannot be measured at all, because those patients do not have the hemoglobin the test is looking for.
The fix is not complicated. When something can interfere with the relationship between A1c and glucose, use the plasma glucose criteria instead. Do not force a test to be right. Use a different test.
One more, handled carefully. There are small differences in A1c between ancestry groups not explained by glucose, on the order of 0.3 percentage points. The relationship between A1c and actual complications still holds, so this is worth knowing about and is not a reason to distrust your result.
The fasting insulin argument, and why it is more persuasive than it is correct
The claim that fills my comment section usually arrives in some version of this: you never checked his fasting insulin, he had been insulin resistant for a decade, and you people only test once the pancreas has already failed.
Notice the shape first, because the shape is doing most of the work. The tactic is manufactured secret knowledge. You identify something a doctor did not do, you assert the reason is ignorance or bad faith, and the absence becomes the proof. The less commonly a test is ordered, the stronger the argument appears to get. Once you can see that structure, you will see it applied to dozens of tests.
Now the substance, starting where the argument is strongest. The biology is correct. Hyperinsulinemia genuinely does precede hyperglycemia, exactly as described above. So why is fasting insulin not on your standard panel? There are 3 reasons, and none of them is a conspiracy.
1. The measurement does not hold still
The American Diabetes Association convened a workgroup to look at this. They evaluated 12 commercial insulin assays from 9 manufacturers. Variation between assays ran from 12 percent to 66 percent, with a median of 24 percent, and giving every lab a common reference preparation to calibrate against did not fix it. In practice, the same tube of your blood, split and sent to 2 labs, can come back with meaningfully different insulin levels. That report is from 2007 and work has continued since, but it has never been solved to the standard we hold a diagnostic test to. Compare that with the A1c, which is standardized by a national program and traceable to the reference assay used in the trial that defined the outcomes we care about. When your A1c comes back at 6.5, it means the same thing in Portland as it does in Perth.
2. There is no line to draw
No endocrine society has ratified a diagnostic cutoff for fasting insulin. There is no number above which you are formally insulin resistant. The targets quoted confidently online are convention and expert opinion, not criteria backed by outcome data. That follows directly from the first problem. You cannot set a universal threshold for a measurement that does not mean the same thing in 2 different laboratories.
3. It does not change what you do
The Association for Diagnostics and Laboratory Medicine, in its guide on lab test utilization, states that clinical assessment of insulin resistance using fasting insulin and glucose, which is the HOMA-IR calculation, is not recommended and should be reserved for research purposes. These are the people whose entire job is deciding which laboratory tests are worth running.
The honest position
Fasting insulin is not meaningless. It is a real physiological quantity and a legitimate research tool. It is simply not a validated screening test, and it is not being withheld from you. Would it have caught the patient in this case earlier? No. Nothing would have, because he was not being tested for anything at all. What failed him was not the test menu.
If you want to know where you sit metabolically before your glucose becomes abnormal, there are standardized, inexpensive tests tied to real outcomes:
• Your waist circumference
• Your triglycerides and your HDL
• Your ALT, which is a window onto liver fat
• Your blood pressure
• Your fasting glucose and your A1c, including whether you sit in that 5.7 to 6.4 percent band
Every one of those points in the same direction, provided somebody is actually ordering them and acting on the answers.
The personal fat threshold, and the third door
People who receive a diagnosis like this one tend to see 2 doors. Either there is nothing they can do because it is genetic, or it is entirely their fault and they should feel terrible about it. Neither is right, and the third door is better than both.
Start with the genetics. Large-scale studies have identified more than 700 genetic locations linked to type 2 diabetes risk, roughly half discovered in the last few years as sample sizes and technology improved. The obvious next question is whether a genetic risk score built from those variants predicts who gets diabetes better than simply knowing a patient’s age and family history. Right now the answer is no. It may change. So the genetics are real, and they are also not a verdict.
The twin cycle, tested
In 2011, Taylor’s group took 11 people with type 2 diabetes and put them on a supervised 600-calorie-a-day diet to test the twin cycle hypothesis directly. If diabetes is driven by fat inside the liver and pancreas, removing that fat should reverse the problem. Within 7 days, fasting glucose had normalized. Over 8 weeks, liver fat fell from 12.8 percent to 2.9 percent and beta cell function climbed back toward that of people without diabetes. To be clear about scale, that was 11 people and it is not a treatment recommendation. But it did what a good experiment does. It tested a prediction, and the prediction held.
Then came DiRECT, which took the idea into ordinary primary care. It enrolled 306 people across 49 general practices in Scotland and England, all diagnosed within the previous 6 years, with a body mass index between 27 and 45, and none on insulin. Half the practices delivered a structured weight-management program and half delivered guideline-standard care. At 12 months, 46 percent of the intervention group were in remission, meaning an A1c below 6.5 percent with no diabetes medication at all, compared with 4 percent of controls. The dose-response is the part that should stay with you.
Weight change at 12 months
Proportion in remission
Gained weight
0%
Lost 0 to 5 kg
7%
Lost 5 to 10 kg
34%
Lost 10 to 15 kg
57%
Lost 15 kg or more
86%
At 2 years, 36 percent were still in remission, and durability tracked with sustained weight loss. So this is not permanent by default. It holds for as long as the thing that caused it stays gone, which is true of most of medicine.
The personal fat threshold
If type 2 diabetes were simply a matter of being above a certain weight, everybody with the disease would be heavy and everybody heavy would have the disease. Neither is true, and you have probably seen that for yourself. This is where Taylor’s concept of a personal fat threshold earns its place. Each of us has an individual limit on how much fat we can store before it spills into the liver and pancreas, and that limit varies widely between people. Some carry a great deal and stay metabolically fine. Others cross their own line at a weight that looks completely unremarkable from the outside.
His group tested it. They took people with type 2 diabetes whose body mass index sat between 21 and 27, the people nobody would point at in the street, and put them through stepwise weight loss. Seventy percent went into remission, across a wide range of individual thresholds. Taylor describes type 2 diabetes as a condition with a homogeneous cause in genetically diverse individuals, which is the best sentence I have read on the subject.
That is the answer to both dead ends. Your genes largely set where your threshold sits. Your weight relative to that threshold is the part that can move. It explains why the person next to you, eating the same food, gets a different result. I do not want to oversell it: DiRECT enrolled people within 6 years of diagnosis and excluded anyone on insulin, and the likelihood of remission drops with longer disease duration. Remission is also not a cure. It requires the weight loss to be maintained, which is exactly why support matters more than information.
The Tier 1 answer, which is boring and wins anyway
If you have prediabetes, the intervention with the best evidence is not a supplement and it is not a wearable. The Task Force pooled 23 trials involving 12,915 people and found that lifestyle interventions focused on diet and physical activity reduced progression to type 2 diabetes, with a pooled relative risk of 0.78. The same interventions also lowered weight, blood pressure, and triglycerides, and raised HDL.
Metformin also works for prevention, with a pooled relative risk of 0.73 across 3 trials, though it is not FDA-approved for that use. In the Diabetes Prevention Program, lifestyle interventions outperformed metformin and worked across all subgroups, while metformin performed best in specific groups: people under 60, a body mass index of 35 or above, a fasting glucose of 110 or higher, and women with a history of gestational diabetes.
So the Tier 1 foundations win again. Move your body every day. Lift something heavy at least once a week. Eat a pattern built mostly on plants and fiber. Sleep like it is your Olympic sport. Manage your stress. And know your numbers, by working with a primary care physician who knows you.
The patient in the case did not need more willpower or better information. An abscess got him a diagnosis that a 15-minute appointment could have delivered 5 years earlier, for a fraction of the cost and none of the fear. That is not a character flaw. Establishing with a new doctor in this country is genuinely hard: phone calls, insurance, waiting lists, and time off work. But it is the highest-yield thing on this page, and it is the one nobody puts in a headline, because there is nothing at the end of it to sell you.
How should this modify your practice?
If you are a patient
• If you are between 35 and 70, carrying extra weight, and cannot remember the last time anybody checked your blood sugar, ask for a fasting glucose or an A1c at your next appointment. That is one sentence in one visit.
• If the reason you cannot remember is that you do not currently have a doctor, then establishing with one is the actual task. Everything else on this page depends on it.
• If your A1c comes back in the 5.7 to 6.4 percent band, that is a window, not a sentence. It is the point at which lifestyle intervention has the best evidence behind it.
• If you have anemia, kidney disease, a hemoglobin variant, or you are pregnant, ask whether a glucose-based test would be more reliable for you than an A1c.
• You do not need to fast for an A1c. If fasting has been the barrier, it is no longer a reason to skip the test.
If you are a student, resident, or clinician
• An elevated glucose in an acutely unwell patient is not a diagnosis. An A1c drawn in the department separates chronic hyperglycemia from stress hyperglycemia and changes your discharge conversation.
• A groin or perineal abscess in a middle-aged adult is a reasonable prompt to check a glucose. In this case it was the entire diagnostic pathway.
• Unless the hyperglycemia is unequivocal, the diagnosis requires a confirmatory test. Say so out loud, so a single number does not become a lifelong label by accident.
• Know the A1c interference list well enough to recognize it at the bedside, and default to plasma glucose criteria when any of it applies.
• When a patient raises fasting insulin or HOMA-IR, agree with the biology first, then explain assay standardization and the absence of a ratified cutoff. Dismissing the question cedes the ground to whoever answered it first.
• For a patient leaving the ED with a new diagnosis, the highest-value referral is not a specialist. It is a primary care physician who will still be there in 3 months.
The one thing this week
Ask for a fasting glucose or an A1c if you are in the screening window and nobody has checked. If you do not have a doctor to ask, getting one is the one thing.
AUTHOR
Dr Adrian Cois is a board-certified Emergency Physician and Assistant Professor practicing in a high-acuity emergency department. He writes and hosts Overheard in the Emergency Room, an evidence-based health education series that uses deidentified emergency cases as anchors for preventive health. Before medicine he worked in exercise physiology and elite sports performance.
DISCLOSURE
Dr Cois has no financial relationship with any laboratory, device manufacturer, pharmaceutical company, testing service, or product named in this article. There are no affiliate links anywhere on DrCois.com. This article is for educational purposes only. It does not provide medical advice and does not establish a physician-patient relationship. If you have symptoms concerning for a medical condition, please seek care from a qualified clinician.
PATIENT PRIVACY
The case described has been deidentified. Identifying details have been altered or removed, and no protected health information appears in this article or in the corresponding episode.
RELATED READING
• S2E2 (17), How to Read Your Own Lab Results, the pillar episode for this series
• S2E5 (20), Your Cholesterol Panel, Explained
• S2E7 (22), Stroke: Know the Signs
• S2E9 (24), Statins, publishing next
• Season 1 Episode 6, Metabolic Health
• Season 1 Episode 10, Building Your Healthcare Team
All episodes and every source below are collected at DrCois.com.
REFERENCES
1. Taylor R. Aetiology of type 2 diabetes: an experimental medicine odyssey. Diabetologia. 2025;68(7):1375-1389. doi:10.1007/s00125-025-06428-0
2. Goyal S, Rani J, Bhat MA, Vanita V. Genetics of diabetes. World J Diabetes. 2023;14(6):656-679. doi:10.4239/wjd.v14.i6.656
3. ✻ American Diabetes Association Professional Practice Committee. 2. Diagnosis and classification of diabetes: Standards of Care in Diabetes 2026. Diabetes Care. 2026;49(Suppl 1):S27-S49. doi:10.2337/dc26-S002
4. US Preventive Services Task Force. Screening for prediabetes and type 2 diabetes: US Preventive Services Task Force recommendation statement. JAMA. 2021;326(8):736-743. doi:10.1001/jama.2021.12531
5. Lamsal R, Klyachman L, Adeyinka A, Kondamudi N, Pierre L. Is HbA1c an indicator of diabetic ketoacidosis severity in the pediatric population? Pediatr Emerg Care. 2023;39(4):216-218. doi:10.1097/PEC.0000000000002859
6. Marcovina S, Bowsher RR, Miller WG, et al. Standardization of insulin immunoassays: report of the American Diabetes Association Workgroup. Clin Chem. 2007;53(4):711-716. doi:10.1373/clinchem.2006.082214
7. Association for Diagnostics and Laboratory Medicine. Insulin. Optimal Testing: ADLM’s Guide to Lab Test Utilization. Updated April 23, 2024. Accessed September 7, 2026. https://myadlm.org/advocacy-and-outreach/optimal-testing-guide-to-lab-test-utilization/g-s/insulin
8. DeForest N, Majithia AR. Genetics of type 2 diabetes: implications from large-scale studies. Curr Diab Rep. 2022;22(5):227-235. doi:10.1007/s11892-022-01462-3
9. Lim EL, Hollingsworth KG, Aribisala BS, Chen MJ, Mathers JC, Taylor R. Reversal of type 2 diabetes: normalisation of beta cell function in association with decreased pancreas and liver triacylglycerol. Diabetologia. 2011;54(10):2506-2514. doi:10.1007/s00125-011-2204-7
10. Lean ME, Leslie WS, Barnes AC, et al. Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial. Lancet. 2018;391(10120):541-551. doi:10.1016/S0140-6736(17)33102-1
11. Lean MEJ, Leslie WS, Barnes AC, et al. Durability of a primary care-led weight-management intervention for remission of type 2 diabetes: 2-year results of the DiRECT open-label, cluster-randomised trial. Lancet Diabetes Endocrinol. 2019;7(5):344-355. doi:10.1016/S2213-8587(19)30068-3
