Are skin tags a true sign of diabetes? No — a skin tag (acrochordon) is not a diagnostic sign of diabetes. The association is real but modest and nonspecific: pooled studies put skin tags at roughly 1.5 to 2.5 times higher odds in people with diabetes, and a large US survey found about 27 percent prevalence in diabetics versus 19 percent in nondiabetics. Most people with skin tags do not have diabetes, and most people with diabetes never develop them. A skin tag is a soft clue that raises pretest probability, never a verdict.
The Mechanics: Why Insulin Leaves a Mark on Skin
Skin tags are small fibroepithelial polyps — a core of loose collagen and blood vessels wrapped in epidermis. They cluster in skin folds: the neck, armpits, groin, eyelids, and under the breasts. Friction in these zones is the classic mechanical trigger, which is why the same lesions appear in lean people who simply have a folding pattern.
The deeper mechanism links to insulin. Insulin and insulin-like growth factor 1 (IGF-1) signal through receptors on keratinocytes and fibroblasts. When insulin is chronically high — the hallmark of insulin resistance — those receptors get sustained stimulation. The result is localized epidermal thickening and fibroblast proliferation, the two tissue events that build a tag.
In plain terms: the tag is a visible footprint of a growth signal that is normally quiet.
Two details sharpen this. First, the mechanism is plausible and supported by receptor biology, but it is not proven as causation in humans. It remains a strong hypothesis, not an established fact. Second, tags are not caused by high glucose directly. The growth factor is insulin, not sugar, which is why a person can have normal fasting glucose and still form tags during years of compensated insulin resistance.
This distinction matters for interpretation. A skin tag is an insulin signal, not a glucose signal — and those can diverge.
Quantified Context: How Strong Is the Link?
The association is consistent across studies but far from a diagnostic threshold. The largest US data come from NHANES, the national health survey, which found skin tag prevalence of about 27 percent in adults with diabetes compared with 19 percent in those without — an absolute difference of roughly 8 percentage points. Pooled case-control and cross-sectional data place the odds ratio for diabetes in people with skin tags at roughly 1.5 to 2.5, depending on how diabetes and obesity are adjusted.
| Measure | With skin tags | Without skin tags | Interpretation |
|---|---|---|---|
| Diabetes prevalence (NHANES-type data) | 27% | 19% | +8 points absolute |
| Odds ratio for diabetes (pooled) | 1.5–2.5 | Reference | Modest, nonspecific |
| Skin tag prevalence in adults overall | 25–50% | — | Common, often benign |
| Positive predictive value, low-risk adult | ~25–35% | — | Most positives are not diabetic |
| Positive predictive value, high-risk adult | ~50–65% | — | Still not diagnostic |
Note the positive predictive value row. Even at higher risk, a tag predicts diabetes in fewer than two of three cases. That is the honest ceiling of the sign.
The comparison most articles omit: skin tags versus acanthosis nigricans, the dark velvety neck patches. Acanthosis nigricans has a stronger and more specific association with insulin resistance and is visible to the naked eye. A neck tag plus acanthosis is a far louder signal than tags alone. If you are weighing which skin finding deserves attention, the dark patch outranks the tag.
Where Tags Appear: Body Site Changes the Signal
Location is not cosmetic — it changes the pretest probability. Tags in friction zones without other markers are mostly mechanical. Tags in unusual sites or in high numbers are more likely to reflect a metabolic driver.
- Neck and axillae: highest yield, most common, and most confounded by friction and obesity.
- Eyelids: common and frequently mechanical, but a cluster of multiple eyelid tags in a younger adult is a recognized soft marker of insulin resistance.
- Groin and perianal: friction-dominated, lower metabolic signal.
- Under the breasts: friction plus obesity overlap, weak independent signal.
- Unusual sites (back, torso, extremities): rare for tags; when present, warrant a closer metabolic look.
The count matters too. A single tag is noise. Ten or more tags, especially with acanthosis nigricans or a rising waist circumference, shifts the odds meaningfully. This is a graded clue, not a binary one.
Variations and Edge Cases: Who the Rule Breaks For
The tag-diabetes link behaves differently across groups, and ignoring this produces bad advice.
Pregnancy is the clearest edge case. Tags often appear or multiply during the second and third trimesters under hormonal and growth-factor shifts, then regress after delivery. A new tag in pregnancy says almost nothing about long-term diabetes risk.
Children and adolescents are a second exception. Tags are uncommon before adulthood, and when they appear in a child, especially with obesity, they carry a stronger relative signal than in adults — but the absolute numbers are small, so the practical yield is limited.
Lean adults are the counterintuitive case. A 2017 analysis of national survey data reported that skin tags were associated with diabetes even in normal-weight adults, though the effect was weaker. This undercuts the common claim that tags in lean people are "just friction." They are often friction, but the insulin axis does not require obesity.
Skin color and genetics matter as well. The association holds across groups, but acanthosis nigricans prevalence varies, which changes how much a tag adds to the picture.
The reconciliation sentence: prevalence runs 25 to 50 percent in adults overall, 27 versus 19 percent in diabetics versus nondiabetics, and the odds ratio sits at 1.5 to 2.5 — consistent figures once you separate prevalence from risk ratio.
Misconceptions: What the Tag Does Not Mean
Three beliefs do most of the damage.
First, the myth that a tag "means" diabetes. It does not. It shifts probability by a few percentage points. Most tag-bearers have normal glucose.
Second, the myth that removing the tag changes risk. Removing a tag is cosmetic and does not alter insulin resistance. The tag is a marker, not a cause.
Third, the myth that tags are harmless and ignore-worthy. They are benign growths, but a rapidly enlarging, bleeding, or irregularly pigmented lesion is not a routine tag and needs evaluation. The Dermoscopy literature treats changing pigmented lesions as a different category entirely.
A fourth, subtler error: assuming a normal fasting glucose clears the question. Insulin resistance can be present for years before fasting glucose rises. If you are concerned, the more sensitive tests are HbA1c and, when indicated, an oral glucose tolerance test.
What to Actually Do: Concrete Actions
If you have skin tags and want to act on the signal without overreacting, the useful steps are measurable.
- Get an HbA1c and a fasting glucose. These two cover most of the diagnostic ground; add an OGTT if results are borderline.
- Measure waist circumference. A waist above 40 inches in men or 35 inches in women strengthens the metabolic case.
- Check for acanthosis nigricans at the neck and axillae; its presence doubles the value of the tag as a clue.
- Ask about a family history of type 2 diabetes; a first-degree relative raises baseline risk substantially.
- See a clinician promptly for any tag that bleeds, grows rapidly, or changes color.
- Do not remove tags expecting a metabolic benefit; removal is cosmetic only.
In plain terms: use the tag as a reason to check a number, not as a reason to panic.
Limitations of the Evidence
The evidence base is observational. Most studies are cross-sectional, which cannot establish that insulin resistance causes tags — only that the two travel together. Confounding by obesity and age is substantial and only partly adjusted for.
Screening performance is weak. No major guideline body endorses skin tags as a diabetes screening trigger. The 2024 American Diabetes Association Standards of Care recommend screening based on age, BMI, and risk factors, not dermatologic findings. A tag is a nudge, not an entry criterion.
Mechanism claims deserve the same caution. The insulin-IGF-1 receptor story is biologically sound and supported by tissue studies, but human causal data are thin. Treat it as the best current explanation, not settled science.
Finally, prevalence figures vary widely by population and by how tags are counted at examination. The 25 to 50 percent adult range reflects that measurement problem, and the 27 versus 19 percent diabetes contrast is the most defensible single comparison available from national data.
Timeline: From Insulin Resistance to Visible Tag
The tag does not appear overnight. The metabolic drift that drives it unfolds over years, which is why timing matters for interpretation.
| Stage | Typical timing | What is happening | What you would see |
|---|---|---|---|
| Compensated insulin resistance | Years 0–5 | Insulin rises to hold glucose normal | Often no symptoms; waist may creep up |
| Early dysglycemia | Years 3–8 | Fasting glucose and HbA1c begin to drift | Acanthosis nigricans may appear |
| Tag formation | Years 5–10 | Sustained growth-factor signaling in folds | First tags on neck and axillae |
| Tag accumulation | Years 8–15 | Multiple tags, larger size | Clusters in friction zones |
| Overt type 2 diabetes | Variable, often years 10+ | Diagnostic glucose thresholds crossed | Tags may be long established |
| Post-treatment | Months after control | Insulin falls with treatment | Tags usually persist; some regress |
The table's central lesson: tags often appear before the diabetes diagnosis, which is why they can be a useful early nudge — and also why they are never proof. The lag works both ways. A tag today may reflect insulin resistance that began a decade ago, and a diabetes diagnosis today may arrive long after the tags did.
Are skin tags a true sign of diabetes? The accurate answer is that they are a soft, graded, mechanism-backed clue with modest predictive power. They raise the odds by roughly 1.5 to 2.5 times and appear in about 27 percent of diabetics versus 19 percent of nondiabetics. They are worth a blood test, not a diagnosis.