Cosmos Study

GLP-1 Therapy Associated with Lower Fracture Risk in Patients with Bone Density Disorders

December 16, 2025

Team A:Kersten Bartelt, RNNicholas Volker

Team B:Tony Dunnigan, MDEric Barkley

Key Findings

Among patients with type 2 diabetes, GLP-1 use was associated with a 32% lower fracture risk for those with either osteopenia or osteoporosis.
Among patients without type 2 diabetes, GLP-1 use was associated with a 34% lower fracture risk for those with osteopenia and 38% lower risk for those with osteoporosis.

Glucagon-like peptide-1 receptor agonists (GLP-1s) are commonly prescribed to manage type 2 diabetes and obesity. Prior research has shown that weight loss can affect bone density,¹ so the potential effect of GLP-1s is of growing clinical interest. Early research results on bone health and the use of GLP-1s are mixed, with some randomized trials and meta-analyses suggesting neutral or modestly protective effects on bone density and fracture risk.^2,3 Understanding the real-world relationship between GLP-1 exposure, weight change, and fracture risk can provide additional insight for patients with existing bone density disorders.

We studied adult patients first diagnosed with osteopenia or osteoporosis between 2017 and 2024. Patients were grouped into nondiabetic or type 2 diabetic cohorts, then each patient on a GLP-1 was matched to four patients who were not prescribed a GLP-1 based on age at bone density disorder diagnosis, total observed time with the condition, and number of prior fractures. Patients were excluded if they had GLP-1 exposure of less than six months. We additionally accounted for sex, race, ethnicity, BMI prior to starting the GLP-1, change in BMI upon starting the GLP-1 as well as throughout their usage, use of osteoporosis medications, steroid use, smoking history, and comorbidities including hyperparathyroidism, chronic falls, cancer, CKD, malabsorption, IBD, and eating disorders in our analysis.

Across both diabetic and nondiabetic cohorts, GLP-1s were associated with a meaningfully lower fracture risk. Among patients with type 2 diabetes, GLP-1 use was associated with a 32% lower risk of fracture for those with osteopenia and for those with osteoporosis, as seen in Figure 1. Among patients without diabetes, GLP-1 use was associated with a 34% lower risk of fracture for those with osteopenia and a 38% lower risk for those with osteoporosis.

Figure 1

Fracture Risk by GLP-1 Usage

Figure 1. The risk of a patient with a pre-existing bone density disorder experiencing a fracture by whether they were prescribed a GLP-1.

The magnitude of risk reduction was slightly greater among non-diabetic patients, suggesting that the difference in fracture risk might be independent of glycemic control. However, this study did not assess bone mineral density (DEXA) results or lifestyle factors such as physical activity and nutrition, which might moderate fracture risk.

These data come from Cosmos, a dataset created in collaboration with a community of Epic health systems representing more than 300 million patient records from 1,800 hospitals and more than 41,000 clinics from all 50 U.S. states, Canada, Lebanon, and Saudi Arabia. This study was completed by two teams that worked independently, each composed of a clinician and research scientists. The two teams came to similar conclusions. Graphics by Brian Olson.

References

Paccou J, Compston JE. Bone health in adults with obesity before and after interventions to promote weight loss. Lancet Diabetes Endocrinol. 2024;12(10):748-760. doi:10.1016/S2213-8587(24)00163-3
Tan Y, Liu S, Tang Q. Effect of GLP-1 receptor agonists on bone mineral density, bone metabolism markers, and fracture risk in type 2 diabetes: a systematic review and meta-analysis. Acta Diabetol. 2025;62(5):589-606. doi:10.1007/s00592-025-02468-5
Su B, Sheng H, Zhang M, et al. Risk of bone fractures associated with glucagon-like peptide-1 receptor agonists’ treatment: a meta-analysis of randomized controlled trials. Endocrine. 2015;48(1):107-115. doi:10.1007/s12020-014-0361-4

Data Definitions

Study period: 2017 to 2024

Study population: inclusion

Adult patients with:

A diagnosis of osteopenia or osteoporosis
At least two outpatient face to face visits prior to the onset of osteopenia or osteoporosis

Study population: exclusion: Prescription for a GLP-1 for less than six months
History of GLP-1 use before the inclusion diagnosis occurred

Censoring

Date of death
Last face-to-face visit
Patients at the point of the onset of following conditions:

Osteopenia: Censor at fracture, osteoporosis, or other bone conditions
Osteoporosis: Censor at fracture or other

Osteopenia: A billing, encounter, or problem list diagnosis with ICD-10-CM M85.8* or SNOMED code 312894000

Osteoporosis: A billing, encounter, or problem list diagnosis with ICD-10-CM M81* or SNOMED code 64859006

Other bone conditions: A diagnosis with any of:
Osteomalacia: A billing, encounter, or problem list diagnosis with ICD-10-CM M83*, SNOMED CT 4598005
Paget’s: A billing, encounter, or problem list diagnosis with ICD-10-CM M88*, SNOMED CT 2089002
Osteogenesis Imperfecta: A billing, encounter, or problem list diagnosis with ICD-10-CM Q78.0, SNOMED CT 78314001
Osteomyelitis: A billing, encounter, or problem list diagnosis with ICD-10-CM M86*, SNOMED CT 60168000
Bone Cancer: A billing, encounter, or problem list diagnosis with ICD-10-CM C41*, V79.5*, SNOMED CT 428281000

Face-to-face visit: An encounter of type “Office Visit,” “Follow-up,” “Telemedicine,” “Walk-in,” “Routine Prenatal,” “Postpartum Visit,” or “Fetal Care Consult”

Type 2 diabetes: A billing, encounter, or problem list diagnosis with ICD-10-CM code E11*

GLP-1: A medication order with a type of prescription, historical, or administered and a pharmaceutical class of “ANTI-OBESITY GLUCAGON-LIKE PEPTIDE-1 RECEP AGONIST,” “ANTIHYPERGLYCEMIC – INCRETIN MIMETICS COMBINATION,” or containing “GLP-1”

Fracture: A billing, encounter, or problem list diagnosis with ICD-10-CM S[0-9]2* or M80* or SNOMED code 125605004

Confounders

Evaluated sex
Age on index date: <50, 50-64, 65-79, 80+
Race: Asian, Black, Hispanic, other, White
BMI nearest index date: <18.5, 18.5-25, 25-30, 30-40, 40+
Osteoporosis medication: A medication order that has a type of prescription, historical, or administered in the study period or prior (simple generic name):

Bisphosphonates
Alendronate
Risedronate
Ibandronate
Zoledronic acid
Denosumab
Romosozumab
Teriparatide
Abaloparatide

Steroid use greater than 30 days
History of smoking
Comorbidities:

Hyperparathyroidism and related: A billing, encounter, or problem list diagnosis with ICD-10-CM code E21* or SNOMED code 66999008
Chronic falls: A billing, encounter, or problem list diagnosis with ICD-10-CM code R29.6 or SNOMED code 279992002
Cancer: A billing, encounter, or problem list diagnosis with ICD-10-CM code C* (excluding C41* and C79.5*) or SNOMED code 363346000 (excluding 428281000)
CKD: A billing, encounter, or problem list diagnosis with ICD-10-CM code N18* or SNOMED code 90708001
Intestinal malabsorption: ICD-10-CM code K90* or SNOMED code 197476001
IBD: A billing, encounter, or problem list diagnosis with ICD-10-CM code K50*-K52* or SNOMED code 24526004
Eating disorder: A billing, encounter, or problem list diagnosis with ICD-10-CM code F50* or SNOMED code 72366004

Race and ethnicity: Patients were classified into exclusive groups by self-reported race and ethnicity values mapped to standards: Hispanic, single race Asian, single race Black, single race White, or multiracial/another race

Steroid: A medication order that has a type of prescription, historical, or administration and a pharmaceutical class or pharmaceutical subclass of “GLUCOCORTICOIDS” and a route of “injection,” “intra-articular,” “intramuscular,” “intravenous,” or “oral”

Model specifications

Matching 1:4 on:

Duration between condition start and censoring (6-month bands)
Age at condition start (5-year bands)
Number of fractures documented on the patient as of 60 days after the first bone condition diagnosis

Observation period starts 6 months after the start of the GLP-1 in the exposure patients, which was the same observation window applied to the matched non-exposure patients.
We conducted a negative-control analysis using benzonatate and ibuprofen, medications not expected to influence fracture risk. When we applied the same methods used in the GLP-1 analysis, these drugs showed no difference in fracture risk between the exposure and non-exposed patients. This suggests that the GLP-1 findings are unlikely to be an artifact of the study’s design.