Want to apply this research to your care?
CED Clinic translates emerging research into individualized clinical care. Dr. Caplan has treated 30,000+ patients.
Book a consultation →- Which GLP-1 receptor agonists carry a statistically significant association with nonscarring hair loss, and which do not
- How large the magnitude of risk elevation is for semaglutide versus tirzepatide, with specific relative risks and confidence intervals
- Why the pattern of findings points toward weight loss-induced telogen effluvium rather than direct follicular toxicity
- What this data cannot yet tell us, and where the evidence genuinely stops
TL;DR: In a propensity score matched cohort of more than 400,000 patients, initiating semaglutide or tirzepatide was associated with a 43% and 68% higher relative risk of nonscarring hair loss, respectively, compared to initiating metformin; dulaglutide and liraglutide showed no statistically significant difference.
Table of Contents
Abstract
Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are medications approved for obesity and type 2 diabetes mellitus (T2DM). Previous investigations have suggested an association between GLP-1RAs and nonscarring hair loss; however, evaluations have been limited by sample size or inability to adjust for confounding factors. This comparative cohort analysis used the TriNetX research network to further characterize this association. Patients aged 18 years or older who were newly prescribed a GLP-1RA (dulaglutide, liraglutide, semaglutide, or tirzepatide) following first diagnosis of overweight/obesity or T2DM were identified using validated RxNorm and ICD-10-CM codes. Each GLP-1RA cohort was matched 1:1 to a comparator group initiating metformin using a greedy nearest neighbor propensity score algorithm. Relative risks for androgenetic alopecia, telogen effluvium, or other nonscarring hair loss in the year following drug initiation were calculated. Compared to those starting metformin, patients initiating semaglutide (RR 1.43; 95% CI 1.30–1.56; p<0.001) or tirzepatide (RR 1.68; 95% CI 1.44–1.97; p<0.001) experienced significantly increased risk of nonscarring hair loss. No significant difference was observed for dulaglutide or liraglutide. The authors propose that the association may be more consistent with weight loss-induced telogen effluvium than with direct follicular toxicity, given that semaglutide and tirzepatide are the most efficacious GLP-1RAs for weight loss.
Lanehart MH, Zinn Z, Beatty CJ. Association between GLP-1 receptor agonists and nonscarring hair loss: a population-based, propensity score matched cohort analysis. Archives of Dermatological Research. 2026;318:164.
DOI: 10.1007/s00403-026-04603-w
Published online: 19 May 2026
Study at a Glance
| Design | Population-based, propensity score matched cohort study using TriNetX |
| Population | Adults aged 18+ newly prescribed a GLP-1RA following first diagnosis of overweight/obesity (E66) or T2DM (E11); matched comparator group initiated metformin |
| Total N (semaglutide cohort) | 170,965 semaglutide + 170,965 matched controls |
| Total N (tirzepatide cohort) | 62,230 tirzepatide + 62,230 matched controls |
| Primary Endpoint | New diagnosis of nonscarring hair loss (androgenetic alopecia, telogen effluvium, or other; ICD-10-CM L64–L65.9) within 1 year of drug initiation |
| Key Finding | Semaglutide: RR 1.43 (95% CI 1.30–1.56); Tirzepatide: RR 1.68 (95% CI 1.44–1.97); both p<0.001 versus metformin |
| Funding | None declared |
| COI | No competing interests declared |
Study Snapshot
| Agent | N (drug) | N (control) | Events (drug) | Events (control) | RR (95% CI) | Significant? |
|---|---|---|---|---|---|---|
| Dulaglutide | 49,705 | 50,214 | 194 | 209 | 0.94 (0.77–1.14) | No |
| Liraglutide | 35,465 | 25,734 | 212 | 190 | 1.12 (0.93–1.37) | No |
| Semaglutide | 163,839 | 166,261 | 1,053 | 750 | 1.43 (1.30–1.56) | Yes, p<0.001 |
| Tirzepatide | 59,305 | 60,213 | 418 | 252 | 1.68 (1.44–1.97) | Yes, p<0.001 |
Study Facts Table
| Full Study Facts | |
| Authors | Matthew H. Lanehart, Zachary Zinn, Colleen J. Beatty |
| Journal | Archives of Dermatological Research |
| Year / Volume | 2026; 318:164 |
| Design | Population-based propensity score matched cohort analysis; database: TriNetX |
| Total Patients Identified (pre-match) | Dulaglutide 1,551,291; liraglutide 475,968; semaglutide 1,991,302; tirzepatide 2,601,723; metformin comparator 6,809 (post-match) |
| Intervention | Initiation of dulaglutide, liraglutide, semaglutide, or tirzepatide following first diagnosis of overweight/obesity or T2DM |
| Comparator | Metformin initiation, propensity score matched 1:1 |
| Matching Variables | Demographics, overweight/obesity status, T2DM status, cigarette smoking, essential hypertension, ischemic heart disease, polycystic ovarian disease, pregnancy, systemic lupus erythematosus, thyroid disease |
| Primary Endpoint | New nonscarring hair loss diagnosis (ICD-10-CM L64–L65.9, excluding alopecia areata L63–L63.9) within 1 year of drug initiation |
| Key Results | Semaglutide RR 1.43 (95% CI 1.30–1.56, p<0.001); Tirzepatide RR 1.68 (95% CI 1.44–1.97, p<0.001); Dulaglutide RR 0.94 (0.77–1.14, NS); Liraglutide RR 1.12 (0.93–1.37, NS) |
| Adverse Events | Not reported as a separate safety category; hair loss itself is the outcome of interest |
| Funding | None declared |
| Conflict of Interest | None declared |
Investigators at West Virginia University queried the TriNetX federated research network to identify adult patients newly prescribed one of four GLP-1RAs (dulaglutide, liraglutide, semaglutide, or tirzepatide) following their first coded diagnosis of overweight/obesity or type 2 diabetes. Patients with any prior diagnosis of nonscarring hair loss, any lifetime diagnosis of alopecia areata, or any prior prescription of the comparator drug classes were excluded. Each GLP-1RA cohort was then matched 1:1 to a metformin-initiation control group using a greedy nearest-neighbor propensity score algorithm, balancing across demographics, obesity status, T2DM status, and nine hair-loss-relevant comorbidities including thyroid disease, polycystic ovarian disease, systemic lupus erythematosus, and pregnancy.
The primary outcome was a new ICD-10-CM code for nonscarring hair loss (androgenetic alopecia L64–L64.9, telogen effluvium L65.0, or other L65.1–L65.9) in the 12 months following drug initiation. Relative risks and 95% confidence intervals were calculated with chi-squared tests; sex- and age-stratified sensitivity analyses were also performed.
Age- and sex-stratified sensitivity analyses demonstrated results consistent with the primary analysis for both semaglutide and tirzepatide. Dulaglutide and liraglutide showed no statistically significant signals in any subgroup stratum.
This study was designed to quantify hair loss as the outcome of interest, not to enumerate broader adverse events. No separate safety data were collected. Events counted represent new ICD-10-CM-coded diagnoses encompassing androgenetic alopecia, telogen effluvium, and other nonscarring alopecia subcategories. The paper does not report severity grading, duration of hair loss, or rate of resolution.
The greedy nearest-neighbor propensity score matching strategy with 1:1 matching is an appropriate and commonly applied method for reducing selection bias in large administrative database studies. Matching on nine clinically relevant comorbidities known to influence hair cycle addresses many major sources of confounding. Standardized differences near or below 0.10 across matched variables indicate adequate balance post-matching. The one-year observation window is clinically motivated, given the known latency of telogen effluvium following physiologic stressors.
However, patient-level weight loss could not be quantified, making it impossible to formally test whether magnitude of weight loss mediates the association. The study cannot establish causality, and ICD-10-CM code-based outcome ascertainment carries inherent misclassification risk in both directions.
Clinicians prescribing semaglutide or tirzepatide to patients with obesity or type 2 diabetes should incorporate the possibility of nonscarring hair loss into pre-treatment counseling. The relative risk elevations are statistically robust and consistent across age and sex subgroups, but the absolute event rates remain modest. The absence of a significant signal for dulaglutide and liraglutide — the two GLP-1RAs with comparatively lower weight-loss efficacy — supports a biologically coherent hypothesis that hair loss reflects weight loss-induced telogen effluvium rather than direct follicular toxicity. Patients should be informed that hair shedding, if it occurs, typically manifests several months after a significant caloric deficit, and that the signal is an association, not confirmed causation.
Semaglutide and tirzepatide are each associated with a statistically significant elevation in risk of nonscarring hair loss; the pattern fits telogen effluvium from weight loss more than direct follicular toxicity, but causality has not been established.
This study found that semaglutide and tirzepatide users were diagnosed with nonscarring hair loss more frequently than matched metformin users — a 43% and 68% higher relative risk, respectively. Dulaglutide and liraglutide showed no statistically significant difference, a pattern that tracks closely with each agent’s known weight-loss efficacy.
- The signal is real and robust: More than 400,000 propensity-matched patients; consistent across age and sex subgroups.
- Mechanism remains inferential: The study never measured actual weight loss, so the telogen effluvium hypothesis is biologically plausible but untested within this dataset.
- This is association, not causation: Observational cohort design cannot establish a direct mechanistic link.
- Clinical implication is counseling, not avoidance: The absolute event rates are modest; the goal is informed expectation-setting, not discouraging effective therapy.
If you are taking semaglutide or tirzepatide, this study does not mean you will lose your hair. It means that among hundreds of thousands of patients, hair loss diagnoses were more common in people starting those medications than in similar people starting metformin.
Importantly, the study points toward a phenomenon called telogen effluvium, a temporary shedding condition that often follows major physiologic stressors such as rapid weight loss, surgery, severe illness, or nutritional changes. If that explanation proves correct, the medication may not be damaging the follicle directly. Instead, the body’s response to significant weight reduction may be influencing the hair cycle.
The practical takeaway is awareness rather than alarm. If diffuse shedding develops several months after major weight loss, discuss it with your healthcare team. The study supports the possibility of a real association, but it does not prove that the medication itself is causing permanent hair damage.
This paper supports including hair shedding among counseling topics when initiating semaglutide or tirzepatide, particularly for patients with preexisting alopecia concerns or strong cosmetic priorities.
The absence of a statistically significant signal with dulaglutide and liraglutide is clinically interesting because those agents generally produce less weight loss than semaglutide and tirzepatide. The pattern is therefore consistent with weight-loss-mediated telogen effluvium rather than a uniform class-wide follicular toxicity.
Clinicians should continue to evaluate iron status, ferritin, thyroid function, nutritional adequacy, androgenetic alopecia, autoimmune disease, and other contributors rather than attributing all shedding to medication exposure.
The most important limitation is that the study never measured actual weight loss. The proposed mechanism centers on weight-loss-induced telogen effluvium, yet the central variable needed to test that hypothesis is absent.
A skeptic would also ask whether patients receiving semaglutide or tirzepatide have greater healthcare utilization than metformin users, creating more opportunities for hair loss diagnoses to be documented and coded. Even with excellent propensity matching, observational datasets cannot eliminate every source of residual confounding. The signal appears real, but the mechanism remains uncertain.
The study combines androgenetic alopecia, telogen effluvium, and other nonscarring alopecias into a single endpoint. Those conditions have different biological mechanisms and different clinical implications. Outcome identification relied on ICD-10 coding rather than direct dermatologic assessment. Coding errors, incomplete documentation, and diagnostic variability may affect outcome classification.
The analysis is strong for signal detection but weaker for determining causality, biological mechanism, or precise diagnosis. Duration, severity, and reversibility of hair loss are entirely absent from the dataset.
Earlier retrospective cohorts, pharmacovigilance analyses, and adverse-event reporting studies had already suggested possible links between semaglutide, tirzepatide, and hair loss. Burke et al. (2025, JAAD) conducted a comparable retrospective cohort. Godfrey et al. (2025, JEADV) used FDA FAERS disproportionality analysis and identified a signal for semaglutide and tirzepatide specifically. Kim et al. (2025, Diabetes, Obesity and Metabolism) performed a global pharmacovigilance analysis of GLP-1RAs approved for obesity.
This paper’s primary contribution is methodological. Large propensity-matched cohorts provide stronger adjustment for confounding than spontaneous reporting systems. Rather than overturning prior understanding, this study strengthens an existing signal and refines the conversation around possible mechanisms.
Patients beginning semaglutide or tirzepatide should understand that hair shedding may occur during substantial weight loss regardless of how that weight loss is achieved. Baseline hair concerns should be documented before treatment begins.
If shedding develops, nutritional adequacy, protein intake, ferritin, iron status, and thyroid function deserve review before treatment decisions are made. The practical lesson is preparation, expectation management, and thoughtful monitoring — not reflexive discontinuation.
Future studies should directly measure the amount and speed of weight loss to determine whether hair-loss risk scales proportionally with metabolic change. Researchers should also separate telogen effluvium from androgenetic alopecia and other nonscarring alopecias rather than combining them into one endpoint. Prospective studies with dermatologist-confirmed diagnoses would substantially improve confidence in causal interpretation.
Longer follow-up is also needed to assess whether shedding reverses after the initial weight-loss phase, stabilizes with weight maintenance, or represents a more persistent change.
Misreading: “GLP-1 drugs cause hair loss.” The study demonstrates association, not causation. No mechanism has been confirmed within this dataset.
Misreading: “Semaglutide damages hair follicles.” The study provides no evidence of direct follicular toxicity. The leading hypothesis is metabolic, not pharmacotoxic.
Misreading: “All GLP-1 drugs increase hair-loss risk.” Dulaglutide and liraglutide did not demonstrate statistically significant elevations in this analysis. The signal is drug-specific, not class-wide.
Bad-faith take: “This proves patients should avoid GLP-1 therapy.” The paper evaluates one potential adverse effect and provides no assessment of overall risk-benefit balance. The absolute event rates remain modest in the context of demonstrated cardiometabolic benefit.
Why This Matters Clinically
GLP-1 receptor agonists are among the most rapidly expanding drug classes in clinical medicine, with tens of millions of prescriptions written annually in the United States alone. Hair loss, though not life-threatening, significantly affects patient quality of life, treatment satisfaction, and, in some cases, adherence to otherwise effective therapy. Prior studies examining this association were constrained by small sample sizes or inability to control for confounders such as thyroid disease, polycystic ovarian disease, and autoimmune conditions, all of which independently cause nonscarring alopecia.
This analysis, with cohorts exceeding 100,000 propensity-matched pairs for semaglutide and a methodologically sound exclusion of alopecia areata at baseline, represents one of the largest and most rigorously adjusted population-level analyses published to date. The drug-specific rather than class-wide pattern of risk, tracking closely with each drug’s known weight-loss efficacy, has direct implications for how this side effect should be framed in patient education and potentially in future regulatory discussions.
CED Clinical Relevance
At CED Clinic, a substantial proportion of patients seeking evaluation for hair loss are middle-aged adults with concurrent metabolic disease, many of whom are either on or being considered for GLP-1 receptor agonists. This study reinforces the clinical reality that newly presenting nonscarring hair loss in a patient who recently started semaglutide or tirzepatide warrants a careful timeline assessment. If hair shedding began two to four months after drug initiation, the presentation is temporally consistent with telogen effluvium, and the first diagnostic step is a thorough history, not necessarily an immediate trichoscopy workup.
At the same time, clinicians should not reflexively attribute all hair loss in this population to the medication; thyroid disease, iron deficiency, nutritional insufficiency, hormonal changes, autoimmune conditions, and other metabolic contributors must be systematically evaluated. Understanding this association also shapes the conversation with patients who are weighing GLP-1 therapy, allowing for informed, patient-centered decision-making rather than surprise after treatment has begun.
Fits What We Already Know
This study sits within a growing body of literature examining GLP-1 receptor agonists and hair loss. Prior retrospective cohorts, pharmacovigilance analyses, and adverse-event reporting systems have repeatedly raised the possibility that semaglutide and tirzepatide may be associated with increased rates of nonscarring alopecia. Burke et al. (2025, JAAD) conducted the primary methodological comparator. Desai et al. (2024, International Journal of Dermatology) raised the question and proposed hormonal pathway disruption as a potential mechanism. Godfrey et al. (2025, JEADV) used FDA FAERS disproportionality analysis from 2022 to 2023. Kim et al. (2025, Diabetes, Obesity and Metabolism) conducted a global pharmacovigilance analysis. Moiz et al. (2025, Annals of Internal Medicine), cited by the authors, reported semaglutide and tirzepatide as the most efficacious GLP-1RAs for weight loss, providing the pharmacological substrate for the telogen effluvium hypothesis.
The current analysis advances this literature by using large propensity-matched cohorts and systematic adjustment for many known causes of hair loss, while acknowledging that residual confounding remains possible. The consistency of the signal across multiple study types and datasets makes it increasingly difficult to dismiss the association entirely.
What This Study Teaches Us
Among adults beginning semaglutide or tirzepatide, diagnoses of nonscarring hair loss occurred more frequently than among carefully matched adults beginning metformin. The association was statistically significant, consistent across age and sex analyses, and strongest for the medications that generally produce the greatest weight loss. Dulaglutide and liraglutide showed no statistically significant increase, arguing against a uniform GLP-1 class effect and instead supporting the possibility that weight-loss magnitude may be an important mediator.
For patients and clinicians, the broader lesson is that successful weight loss can sometimes create physiologic consequences that are unrelated to medication toxicity. Hair shedding may represent a consequence of the metabolic transition itself rather than direct follicular injury.
What It Does Not Show
- The study cannot establish causation. Association in a matched cohort does not prove that semaglutide, tirzepatide, or weight loss directly caused hair shedding.
- Patient-level weight-loss data were unavailable. The proposed telogen effluvium mechanism is biologically plausible but was not directly tested within this dataset.
- The study cannot determine whether faster weight loss, larger weight loss, nutritional deficiencies, medication exposure, or some combination of those factors drove the observed association.
- The study cannot distinguish with certainty among androgenetic alopecia, telogen effluvium, and other nonscarring alopecia subtypes because outcomes were identified through diagnostic coding.
- The study provides no information about severity, reversibility, treatment response, duration of shedding, or long-term follicular outcomes.
Frequently Asked Questions
Do semaglutide and tirzepatide actually cause hair loss?
No. This study found an association between semaglutide or tirzepatide use and higher rates of nonscarring hair-loss diagnoses, but observational research cannot prove causation. The findings suggest a relationship exists, but the mechanism remains uncertain.
Which GLP-1 medications showed a significant hair-loss signal?
Semaglutide and tirzepatide demonstrated statistically significant elevations in risk compared with matched metformin users. Dulaglutide and liraglutide did not show statistically significant differences in this analysis.
How much higher was the risk with semaglutide?
Semaglutide users experienced a relative risk of 1.43, meaning diagnoses of nonscarring hair loss occurred about 43% more often than among matched metformin users during the one-year follow-up period.
How much higher was the risk with tirzepatide?
Tirzepatide users experienced a relative risk of 1.68, meaning diagnoses occurred approximately 68% more often than among matched metformin users.
What is telogen effluvium?
Telogen effluvium is a common form of diffuse hair shedding that often occurs several weeks to months after a physiologic stressor such as illness, surgery, childbirth, nutritional deficiency, or significant weight loss. It is frequently reversible.
Why do researchers think weight loss may be involved?
The strongest signals appeared with semaglutide and tirzepatide, which generally produce greater weight loss than dulaglutide or liraglutide. This pattern is consistent with weight-loss-induced telogen effluvium, although the study did not directly measure weight change.
Does this mean everyone taking GLP-1 medications will lose hair?
No. Most patients in the study did not receive a diagnosis of hair loss. The observed association increased relative risk, but the absolute number of events remained relatively low.
Should patients stop semaglutide or tirzepatide if shedding develops?
Not necessarily. Hair shedding can result from many factors, including nutritional changes, iron deficiency, thyroid disease, hormonal changes, and weight loss itself. Patients should discuss symptoms with their healthcare team before making treatment decisions.
Can this study determine whether the hair loss is permanent?
No. The study only measured diagnostic coding within one year of medication initiation. It did not evaluate severity, duration, reversibility, treatment response, or long-term outcomes.
What is the most important takeaway from this study?
The study provides strong evidence that semaglutide and tirzepatide are associated with higher rates of nonscarring hair-loss diagnoses than metformin. It does not prove the medications directly damage hair follicles, and the findings are consistent with the possibility that substantial weight loss itself may be contributing to the observed effect.
