GLP-1 Hair Loss: What the 2026 Research Actually Found

TL;DR: Two studies in July 2026 shifted the GLP-1 hair-loss picture. A BMJ target trial emulation found a 37% higher rate of alopecia than SGLT-2 inhibitors and 68% higher than DPP-4 inhibitors — and the association was confined to non-scarring alopecia, the kind where the follicle survives. A separate preprint found the risk held regardless of how much weight people lost, which complicates the tidy "it's just the weight loss" explanation this article used to lead with. None of that changes the practical picture much. What it changes is how honestly we can describe it.

You are standing in the shower and there is more hair in the drain than usual. Or you notice extra strands on your pillow. Or your ponytail feels thinner than it did a few months ago. And the timing lines up: you started a GLP-1 medication recently and the weight has been coming off.

Hair loss on GLP-1 medication is one of those side effects that catches people off guard. It was not the one your prescriber warned you about. It is not as discussed as nausea or constipation. But it is real, it can be distressing, and understanding what is happening makes it far less frightening.

What changed in July 2026

Three pieces of evidence landed within days of each other, and together they are the first time the question has been asked properly. (We read the three studies together in a separate piece; this one stays on hair.)

The BMJ study — the comparison that had been missing. On 22 July 2026, The BMJ published Risk of hair loss associated with glucagon-like peptide-1 receptor agonists in adults with type 2 diabetes: target trial emulation (Tang, Zhang, Cotsarelis, Chen et al.; BMJ 2026;394:e100077). Researchers used Penn Medicine electronic health records from January 2019 to September 2024 and compared 12,004 GLP-1 users against 15,221 people on SGLT-2 inhibitors, and a further 11,964 against 11,233 people on DPP-4 inhibitors. After adjusting for age, sex, ethnicity, pre-existing conditions, other medications and BMI:

Comparison	Alopecia rate (per 1,000 person-years)	Hazard ratio (95% CI)
GLP-1 vs SGLT-2 inhibitors	6.91 vs 5.04	1.37 (1.08 to 1.73)
GLP-1 vs DPP-4 inhibitors	6.53 vs 3.89	1.68 (1.28 to 2.20)

Why the comparator matters: SGLT-2 inhibitors also cause weight loss. If shedding were purely a consequence of losing weight, the two groups should have looked alike. They did not.

But look at the confidence intervals, not just the headline. The SGLT-2 comparison — the one that produced the widely-reported "37%" — has a lower bound of 1.08, which almost touches "no difference at all". The DPP-4 comparison (lower bound 1.28) is the sturdier of the two. The 37% figure is real, and it is also the weaker half of this study. Most coverage led with it anyway.

The signal was specific to non-scarring alopecia: hazard ratios of 1.53 (1.18 to 1.97) versus SGLT-2 inhibitors and 1.72 (1.28 to 2.31) versus DPP-4 inhibitors — higher than the all-alopecia figures, which is what "specific to" means here. Hold onto that word non-scarring. It is the most important word in this article, it is also the most widely misread, and it gets its own section below.

Read the denominators before the percentages. A 37% increase sounds enormous. The underlying rate moved from about 5 cases per 1,000 person-years to about 7. Dr Marie Spreckley of Cambridge, commenting independently via the Science Media Centre, put the absolute scale better than any percentage does: "around two to three additional clinically recorded cases of hair loss per 1,000 person-years."

Two caveats the authors state themselves. They had no clinical detail, so they could not assess "severity, extent, duration and reversibility of alopecia after stopping treatment", and they "cannot rule out the possibility that other unmeasured factors may have influenced their results." And the population is adults with type 2 diabetes — if you are taking a GLP-1 for weight alone, this cohort is adjacent to you, not identical.

The nference preprint — the finding that is harder to explain away. Reported by Reuters on 22 July, researchers at nference matched 11,046 tirzepatide users to 11,046 semaglutide users. New-onset alopecia occurred in 4.24% on tirzepatide versus 3.33% on semaglutide; among women, 5.44% versus 3.63%. The detail that matters: the difference held regardless of how much weight people lost or how much drug they received.

Handle this one carefully. It is a preprint, not peer-reviewed, and every figure above reaches us through Reuters' reporting — we could not locate the preprint itself on any preprint server, so there is no title, no author list and no stated follow-up window to check. Venky Soundararajan of nference framed the takeaway as supporting "more individualized counseling and surveillance rather than a generic message that weight loss itself inevitably causes hair loss." Treat it as a signal, not a finding.

The FAERS review — ten years of reports. In the Journal of Cosmetic Dermatology (2026;25(6):e70967), Gupta, Teasell, Bamimore, Mirmirani and Talukder ran a disproportionality analysis of FDA adverse-event reports from 2016 to 2025 (doi:10.1111/jocd.70967). Across seven GLP-1 medications they found 1,941 reports across 1,929 cases, and only semaglutide and tirzepatide produced statistically significant signals.

The signal is also not stable year to year, which is worth knowing before anyone tells you it is mounting. For semaglutide the reporting odds ratio was 1.43 (1.15–1.77) in 2022 and 1.55 (1.35–1.77) in 2024 — but 1.11 (0.96–1.28) in 2025, which is not significant.

One honest caveat, and it is a big one: a disproportionality analysis works on voluntarily-submitted reports. It cannot establish causation and cannot produce an incidence rate — it can only say that a pairing appears more often than the background of the database. Media attention alone inflates reporting. As the same lead author writes elsewhere, these signals "reflect disproportional reporting rather than true incidence" and "do not establish causality." It is a smoke detector, not a diagnosis.

The largest cohort of all, and the one that complicates the story most. Also in 2026, Vidal and colleagues published a TriNetX analysis of 547,993 matched adults in JAAD International (2026;25:133–135). At 12 months, risk was elevated for telogen effluvium (adjusted OR 1.76), androgenetic alopecia (1.64) and other non-scarring hair loss (1.40) — all P < .001. Keep that middle number in view. It is the reason the next section exists.

So is it the medication, or the weight loss?

An earlier version of this article answered that confidently: the weight loss, almost certainly. That answer is no longer clean enough to print.

The honest position now sits between two findings that pull in opposite directions.

Pulling toward the medication: the BMJ comparator group also lost weight and still shed less hair. The nference difference persisted regardless of weight lost. Semaglutide and tirzepatide are the only two GLP-1s producing FAERS signals.

Pulling toward the weight loss: the largest synthesis to date still lands the other way. A systematic review of 24 studies by Gupta and colleagues in Science Progress (April 2026) concluded that hair loss on these medications is "likely multifactorial, with acute metabolic perturbations, particularly rapid weight loss, caloric restriction and hormonal shifts, the most likely contributors." They add the sentence that should govern how anyone writes about this: "The direct role of GLP-1 RAs in hair loss remains difficult to disentangle from downstream metabolic effects."

A second systematic review, in the International Journal of Dermatology, found the pharmacovigilance association weak and inconsistent across drugs — reporting odds ratios of 1.24–2.46 for semaglutide, but 0.83–1.73 for tirzepatide and 0.61–1.53 for liraglutide. Those last two ranges cross 1.0, meaning no clear association at all.

Where that leaves you. Something about being on these medications is associated with more shedding than an equivalent amount of weight loss on a different drug. Whether that is the molecule, the speed of the loss, the drop in total food intake, or a combination, nobody can currently say. Anyone telling you they know is ahead of the evidence.

What is actually happening to your hair

The pattern most people experience is telogen effluvium — a well-documented response to physiological stress of many kinds, including rapid weight loss.

Your hair grows in cycles. At any moment about 85% of scalp follicles are in the anagen (growing) phase and about 15% are in telogen (resting). A significant physiological change — major caloric reduction, surgery, illness, childbirth, rapid weight loss — can push an unusually large share of follicles into the resting phase at once.

Then comes the delay that makes this so confusing. Shedding typically starts around two to four months after the trigger — DermNet gives two to four, StatPearls gives roughly three with a range of one to six, and the British Association of Dermatologists gives around three. The sources do not fully agree, and anyone quoting a single confident number is smoothing that over. What they do agree on is the shape: the hair leaving now was set in motion by something that happened months ago. People routinely conclude their shedding is unrelated to a medication they started in spring, or blame something that happened last week. Both are usually wrong.

The BAD's own list of triggers includes, in as many words, "marked weight loss and extreme dieting".

For scale, from the regulatory record rather than press coverage: hair loss appears in the Wegovy prescribing information at 3% on semaglutide 2.4 mg versus 1% on placebo (pooled across three trials, 2,116 patients versus 1,261). In SURMOUNT-1 for tirzepatide, alopecia occurred in about 5% of participants across all three dose arms — 5.24% at 5 mg, 5.03% at 10 mg, 5.87% at 15 mg, against 1.09% on placebo. Note that this is not a dose-response: the 10 mg arm was lower than the 5 mg arm.

Non-scarring: the most important word — and the most misread

Dermatology splits hair loss into two categories, and the difference between them is not a matter of degree.

In scarring (cicatricial) alopecia, the follicle is destroyed and replaced with fibrous tissue. StatPearls puts it plainly: "The hair follicles are irreversibly destroyed in scarring alopecia, leading to permanent hair loss."

In non-scarring alopecia, the follicle is still there: "The hair follicles are preserved in nonscarring alopecia; therefore, hair loss is potentially reversible, and hair regrowth is possible."

The BMJ signal fell entirely in the second category. That is genuinely good news, and it is worth sitting with: the machinery that grows hair is intact. Telogen effluvium is non-scarring by definition — the scalp shows no scarring even during active shedding.

Now the part that almost every article about this study leaves out.

"Non-scarring" does not mean "temporary". It is a statement about the follicle, not about the outcome. Androgenetic alopecia — ordinary pattern hair loss — is also non-scarring, and it is progressive. It does not resolve on its own. And the BMJ study's non-scarring category includes it.

This is not a hypothetical objection. It is exactly what the largest cohort found: in 547,993 matched adults, androgenetic alopecia rose alongside telogen effluvium (adjusted OR 1.64 versus 1.76). Dermatology has a word for the likely relationship — telogen effluvium "may unmask a genetic tendency to genetic balding" (DermNet), and the British Association of Dermatologists notes that in some cases telogen effluvium "can lead to an early diagnosis of androgenetic alopecia". The shedding does not necessarily create the pattern loss. It can reveal one that was already on its way.

So the honest reading of "non-scarring" is narrower than the reassuring one:

• Your follicles are preserved, and regrowth is possible. True.
• Most cases of telogen effluvium are self-limited. True.
• Therefore your hair will come back. Not supported — that inference skips the possibility that some of what is thinning is pattern loss becoming visible for the first time.

That is why this article does not promise you regrowth. Not out of caution for its own sake, but because the category the good news lives in is broader than the good news.

Why month two to four is not a coincidence

Line up two timelines that are usually discussed separately.

The shedding timeline. Trigger in the first weeks of treatment — the steepest part of the loss curve, the sharpest drop in food intake. Shedding starts about three months later.

The decision timeline. Nausea has faded. The dramatic early loss has slowed. The novelty is gone and the cost is not. Around 15% of people are still on a GLP-1 at two years — the second year is where most of the attrition happens, but the doubts start much earlier.

These two clocks overlap almost exactly. The hair starts coming out in the same window where people are quietly asking themselves whether this is worth it. That is not a coincidence of biology; it is a coincidence of calendars. And it means hair shedding does something out of proportion to its medical seriousness: it arrives as evidence, right when someone is looking for evidence.

Hair is tied to identity in a way that a number on a scale is not. When it starts leaving during a period you were already ambivalent about, it does not read as a side effect. It reads as a verdict.

So here is the useful thing to know. The shedding is delayed feedback about something that already happened — the fastest stretch of a process that has since slowed down. It is not a readout of how things are going now. That distinction is worth holding onto, because a decision made in month three based on month-one biology is a decision made on stale information.

What to do with that: any question about your medication belongs with your prescriber, not with a forum or an article. What this piece can offer is the timeline, so the conversation starts from the calendar instead of from panic.

What is actually in your control

Honest framing first: nothing here is established to prevent telogen effluvium, and this article is not going to pretend otherwise. What follows is the set of things that are supportive, low-risk, and yours to decide.

Protein is the one with a real mechanism. Hair is largely keratin, a protein. Under constrained intake the body triages amino acids toward organs that keep you alive; hair is not on that list. And intake really does fall — a 2026 analysis in the Journal of Translational Medicine found that among GLP-1 users, fewer than 10% met recommended protein intakes, with mean daily protein of 33.4 g. Our protein calculator gives you a target based on your own weight and activity; the same target does double duty for muscle, which is the other tissue that competes for the same amino acids.

Rate of loss, described accurately. Telogen effluvium tracks the speed of change more than the total. That is a fact about physiology, not an instruction about your prescription — nothing in this article is a reason to change how you take your medication. If the pace of your weight loss concerns you, that is a specific and reasonable thing to raise with your prescriber, who is the only person who should be adjusting anything.

Be gentle with the hair you have. Avoid tight styles that pull at the roots, reduce heat styling, use a wide-tooth comb on wet hair. To be clear about what this does: it does not prevent telogen effluvium. It avoids adding mechanical breakage on top of it.

Ask about labs rather than guessing. Iron deficiency is one of the most common nutritional causes of hair loss independent of any medication, and rapid weight change can worsen iron status. A panel including ferritin, vitamin D, zinc and thyroid function turns guesswork into information. Whether to supplement anything, and at what dose, is a decision for your prescriber — identifying a specific deficiency is a different thing from taking something in case.

Stress is an independent trigger. Psychological stress causes telogen effluvium on its own, and can stack with the weight-loss trigger. The emotional weight of watching your hair thin is itself real, and worth tending to.

Recovery: what the evidence supports, and what it does not

The reassurance the research actually supports is narrower — and more specific — than "it grows back".

The systematic review in Science Progress frames the counselling point this way: patients "can be reassured that there is no strong evidence of permanent follicular damage associated with GLP-1 RA use, but reported cases most commonly resemble TE, which is typically self-limited and reversible." Improvements are "generally observed within 3–6 months following stabilization of weight."

Acute telogen effluvium is defined as shedding lasting under six months. Once the trigger resolves, DermNet describes hair fall "gradually tapering back to normal over 6–9 months in most cases", and the American Academy of Dermatology gives the same window.

Now the parts an earlier version of this page left out, which are the parts you deserve:

• Shedding can "continue to be intermittently or continuously greater than normal for long periods of time, sometimes for years" (DermNet).
• "Recovery may be incomplete in some cases" (DermNet).
• Where a stressor persists, the shedding can persist with it (AAD).
• Three different clocks get collapsed into one. Shedding stopping, regrowth starting, and regrowth you can actually see are separate events. As one clinical review puts it: shedding takes 3–6 months to cease, regrowth can be noted 3–6 months after the trigger is removed, "but cosmetically significant regrowth can take 12-18 months".
• And per the section above: some of what is thinning may be pattern loss becoming visible, which follows its own course.

New growth often shows up first as short, fine hairs at the hairline and part. Those are a sign the follicles are cycling again.

We are not going to tell you your hair will come back on a schedule. What the evidence supports is this: the follicles are preserved, most cases are self-limited, and there is no strong evidence of permanent follicular damage from these medications. Note the shape of that last one — it is an absence of evidence of harm, not a promise of recovery. That is a reasonable thing to hold onto. It is not a guarantee, and anyone offering you one is selling something.

What this article cannot tell you

• Whether your shedding is caused by your medication. Nobody can tell you that from a study. Telogen effluvium has many triggers and they stack.
• Whether it will happen to you. These are population rates. Roughly 5% in the tirzepatide trials means roughly 95% did not report it.
• Whether to change anything about your medication. Not our lane, not this article's job. That conversation is yours and your prescriber's, and the July studies are not a reason to make a decision without them.
• Whether any supplement will help. The evidence for supplementing a person who is not deficient is weak. Testing first is the defensible order.
• Whether stopping would reverse it. The BMJ authors say directly that they could not assess reversibility after stopping treatment. Anyone extrapolating past that is guessing.
• Which kind of hair loss you have. Telogen effluvium and early pattern loss can look similar at the start and behave very differently over a year. That is a question for an examination, not an article.

When to see a dermatologist

Telogen effluvium from rapid weight loss is the most likely explanation, but it is not the only one, and some alternatives are time-sensitive. Seek a dermatology opinion if the hair loss:

• persists beyond six to nine months
• is patchy rather than diffuse — telogen effluvium thins all over, it does not make bald spots
• shows patterned loss or rapid miniaturisation (thinning concentrated at the crown or temples)
• comes with scalp redness, scaling, pain or a shiny surface — possible signs of a scarring process, which is the category where timing genuinely matters
• is severe enough to be affecting your daily life

Thyroid disorders, alopecia areata and androgenetic alopecia all deserve to be ruled in or out rather than assumed away.

The part that is not really about hair

Hair is tied to identity for a lot of people. Losing it, even temporarily, can feel like losing a piece of yourself — particularly during a period when your body is already changing faster than your sense of who you are can keep up.

If you are struggling with that, it is worth saying plainly: that reaction is proportionate. It does not make you vain and it does not cancel out what you are gaining. It means you are a person having a normal response to a visible change you did not choose.

The follicles are preserved. The shedding is, for most people, self-limited. And the things that support your hair through this — enough protein, enough sleep, attention to your labs, some patience with a slow process — are the same things that carry the rest of this journey. None of them are hair strategies. They are just the work, showing up in one more place.

Key takeaways

1. A BMJ target trial emulation (22 July 2026) found GLP-1 users had a 37% higher rate of alopecia than SGLT-2 users (6.91 vs 5.04 per 1,000 person-years) and 68% higher than DPP-4 users — but the absolute rates remain low.
2. The association was confined to non-scarring alopecia, where the follicle is preserved and regrowth is possible. But non-scarring is not the same as temporary — pattern hair loss is also non-scarring, and it rose too in the largest cohort (adjusted OR 1.64 across 547,993 adults).
3. An nference preprint found new-onset alopecia in 4.24% of tirzepatide users vs 3.33% of semaglutide users (5.44% vs 3.63% in women) — and the difference held regardless of weight lost.
4. The mechanism is unresolved. The largest systematic review still favours rapid weight loss and caloric restriction as the likeliest contributors.
5. Shedding typically starts about three months after the trigger — it is delayed feedback about an earlier phase, not a readout of the present.
6. Most cases are self-limited, with fullness returning over roughly 6–9 months after the trigger resolves — but shedding can persist longer and recovery is not always complete.
7. Nothing is established to prevent it. Adequate protein, gentle handling and lab testing are supportive and low-risk. Supplements and any medication question go to your prescriber.

Frequently asked questions

Do GLP-1 medications cause hair loss?

They are associated with more of it. The July 2026 BMJ study found a 37% higher rate of alopecia than SGLT-2 inhibitors and 68% higher than DPP-4 inhibitors, in adults with type 2 diabetes. Whether the medication causes it directly or acts through rapid weight loss and reduced intake is still unresolved — the largest systematic review to date leans toward the latter.

When does hair loss start on a GLP-1?

Typically about three months after the trigger, with a range of one to six months. Because the trigger is usually the earliest and fastest phase of weight loss, shedding often begins around months two to four of treatment — which is why it so often feels disconnected from anything recent.

Will my hair grow back?

The honest answer is that the evidence supports possibility, not promise. The alopecia signal in the BMJ study was confined to the non-scarring type, where follicles are preserved and regrowth is possible, and there is no strong evidence of permanent follicular damage from these medications. Most cases of telogen effluvium are self-limited, with fullness returning over roughly 6–9 months after the trigger resolves. Three caveats keep that from being a promise: shedding can persist longer, recovery is not always complete, and "non-scarring" also covers pattern hair loss — which is progressive and which rapid shedding can bring into view for the first time. If your thinning is patterned rather than diffuse, that is worth a dermatologist's opinion rather than a waiting game.

What does the research say about hair loss on tirzepatide versus semaglutide?

A 2026 preprint matching 11,046 pairs found new-onset alopecia in 4.24% of tirzepatide users versus 3.33% of semaglutide users, and 5.44% versus 3.63% among women. It is a preprint, not peer-reviewed, and a second systematic review found the pharmacovigilance association weak and inconsistent. This is a signal worth knowing about and not a basis for choosing a medication — that is a conversation for your prescriber.

Should I take biotin for hair loss on a GLP-1?

That is a question for your prescriber, and the evidence is weaker than the marketing. A review in Skin Appendage Disorders found biotin has no proven efficacy for hair growth in people without an underlying deficiency. It also interferes with common lab assays — including thyroid and troponin tests — which matters if you are having bloodwork done. Testing to identify an actual deficiency is the more defensible order.

Sources

1. Tang H, Zhang B, Lu Y, Zhang D, Liu R, Lu Y, Cotsarelis G, Asch DA, Chen Y. Risk of hair loss associated with glucagon-like peptide-1 receptor agonists in adults with type 2 diabetes: target trial emulation. The BMJ 2026;394:e100077, 22 July 2026. PMID 42486607 · BMJ Group summary · Science Media Centre expert reaction
2. nference matched-cohort preprint on new-onset alopecia with tirzepatide vs semaglutide, reported by Reuters (Nancy Lapid), 22 July 2026. The preprint itself could not be located on any preprint server; these figures are cited as reported. Reuters via CTV News · via KFGO
3. Gupta AK, Teasell E, Bamimore MA, Mirmirani P, Talukder M. Alopecia Risk With GLP-1 Receptor Agonists: A Disproportionality Analysis Using the FDA Adverse Event Reporting System (FAERS). J Cosmet Dermatol 2026;25(6):e70967. doi:10.1111/jocd.70967 · Dermatology Times summary
4. Vidal SI et al. Increased risk of hair loss with GLP-1 receptor agonists: a real-world multicenter TriNetX cohort study. JAAD Int 2026;25:133–135 (n=547,993). PMC12997224
5. Gupta AK, Teasell EM, Economopoulos V, Mirmirani P. GLP-1 therapies and hair loss: a systematic review of current evidence and implications for counseling. Science Progress, April 2026. doi:10.1177/00368504261444578
6. Branyiczky MK et al. Effects of GLP-1 Receptor Agonists on Hair Loss and Regrowth: A Systematic Review. Int J Dermatol, 2026. doi:10.1111/ijd.70133
7. Wegovy prescribing information, Table 3 (hair loss 3% vs 1% placebo). FDA label
8. SURMOUNT-1 posted results, alopecia by dose arm. NCT04184622
9. Syed HA, Zito PM. Alopecia. StatPearls (scarring vs non-scarring; follicle preservation). NBK538178 · Telogen Effluvium, NBK430848 · DermNet: telogen effluvium
10. Korus S, Cembrowska-Lech D, Kłoda K, Stachowska E. J Transl Med 2026;24:532 (protein intake adequacy on GLP-1 therapy). doi:10.1186/s12967-026-07702-4
11. Patel DP, Swink SM, Castelo-Soccio L. A Review of the Use of Biotin for Hair Loss. Skin Appendage Disord 2017. PMC5582478 · FDA biotin assay-interference safety communication
12. British Association of Dermatologists, Telogen Effluvium patient information leaflet, updated October 2025 (trigger list; regrowth expectations). PDF
13. Malkud S. Telogen Effluvium: A Review. J Clin Diagn Res 2015 (the three separate recovery clocks; cosmetically significant regrowth 12–18 months).

Matt Cole covers the chemistry and the studies. This article draws on the July 2026 BMJ target trial emulation, the nference preprint reported by Reuters, the FAERS disproportionality analysis in the Journal of Cosmetic Dermatology, two 2026 systematic reviews, FDA labelling, and StatPearls/DermNet/AAD dermatology references. It is not medical advice.

Find your protein target. The protein calculator gives you a daily number based on your weight and activity — the same one that supports muscle.

Earlier in the journey? What to expect in your first month covers the side effects that show up before this one, and nausea remedies that actually work covers the one most people meet first.

Track what the scale doesn't show. Gila logs the signals that decide whether people stay — hair shedding, muscle changes, food noise — alongside your doses. Join the free pilot or subscribe to the newsletter.