8 Minutes
Summary and key finding
Cannabis use has been associated with a substantial increase in the likelihood of developing diabetes in a large retrospective analysis of electronic health records. The study, presented at the European Association for the Study of Diabetes (EASD) Annual Meeting in Vienna in 2025, reports that people with cannabis-related diagnoses were nearly four times more likely to receive a new diabetes diagnosis over a five-year follow-up than matched individuals without substance-use histories or major chronic disease.
Study design and data sources
Researchers led by Dr. Ibrahim Kamel at Boston Medical Center analyzed real-world clinical records aggregated in the TriNetX Research Network, a federated electronic health record (EHR) platform that links data from multiple healthcare organizations across the United States and Europe. The team identified 96,795 outpatients aged 18 to 50 years who had a cannabis-related diagnostic code recorded between 2010 and 2018. Cannabis diagnoses encompassed a range of clinical documentation, from occasional use and intoxication to dependence and withdrawal.
To create a comparison cohort, investigators used propensity score matching to pair the cannabis group with 4,160,998 individuals who had no recorded history of substance use disorder or major chronic disease at baseline. Matching variables included age, sex and baseline comorbidities. All participants were followed for up to five years to ascertain new-onset diabetes diagnoses recorded in the EHRs.
Outcomes, adjustments and statistical findings
During follow-up, 1,937 people (2.2%) in the cannabis-diagnosed cohort developed diabetes, compared with 518 people (0.6%) in the matched control cohort. After adjusting for established cardiometabolic risk factors such as HDL and LDL cholesterol, uncontrolled hypertension, atherosclerotic cardiovascular disease, and for coexisting substance use including cocaine and alcohol, the investigators found that cannabis exposure remained strongly associated with incident diabetes. The adjusted analyses indicated an almost fourfold relative increase in diabetes diagnoses among the cannabis group compared with controls.
The investigators used multiple regression models and propensity score methods to reduce confounding, but they acknowledge that residual confounding is possible and that retrospective EHR studies cannot prove causation. Nevertheless, the magnitude and consistency of the association across adjusted models were notable and raise questions about metabolic effects associated with cannabis use.
Possible biological mechanisms
The authors and independent commentators have suggested several biological and behavioral mechanisms that could link cannabis use to higher diabetes risk. One hypothesis involves changes in insulin sensitivity: cannabinoids interact with the endocannabinoid system, which plays roles in appetite regulation, energy balance and glucose metabolism. Chronic activation or dysregulation of this system may promote insulin resistance in susceptible individuals.
Behavioral and lifestyle pathways may also contribute. Cannabis use has been associated in some studies with altered dietary choices, increased caloric intake, and irregular exercise patterns, each of which can influence weight gain and metabolic health. The current analysis lacked detailed measures of body mass index (BMI) trajectories, diet or physical activity, so disentangling direct physiological effects from lifestyle-mediated effects remains an open question.
Public health implications and clinical recommendations
As cannabis legalization and social acceptance expand worldwide, the study's findings have immediate implications for clinicians, public health practitioners and policymakers. Lead investigator Dr. Kamel emphasized the need to incorporate metabolic risk assessment into care pathways for patients who use cannabis. Routine screening for blood glucose, HbA1c and other metabolic markers may be reasonable when a patient reports regular cannabis use or when a cannabis-related diagnosis appears in the medical record.
Public health messaging should balance known and suspected risks with potential therapeutic applications of cannabis and cannabinoid compounds. The authors call for improved patient education about metabolic monitoring and for integrating diabetes risk awareness into substance use disorder treatment and counseling.
Limitations and unanswered questions
Despite the size and scope of the dataset, the study has multiple limitations that affect interpretation. Key limitations include:
Lack of granular exposure data
Electronic health records typically record diagnostic codes, not precise consumption patterns. The analysis could not reliably differentiate between frequency of use (occasional vs. daily), routes of administration (inhaled smoke/vapor vs. edibles), potency (THC concentration), or duration of use. Those variables could materially affect metabolic outcomes.
Potential misclassification and reporting bias
Cannabis use may be underreported in clinical settings, especially in regions where possession remains illegal. Misclassification of exposure and inconsistent documentation across institutions can introduce bias. The researchers attempted to reduce bias with propensity score matching and adjustment for measured confounders, but unmeasured factors such as socioeconomic status, precise BMI trends, diet quality and physical activity levels may have influenced results.
Retrospective design
Because the study is observational and retrospective, it cannot establish causality. The authors explicitly caution that the observed association is not definitive proof that cannabis causes diabetes; rather, it identifies a clinically relevant correlation that merits prospective investigation.
Expert Insight
Dr. Maria R. Jensen, an endocrinologist and Associate Professor of Medicine at a major academic medical center, offers contextual interpretation: 'This is a robust, large-scale signal that should prompt both clinicians and researchers to pay closer attention to metabolic outcomes in patients who use cannabis. The biologic plausibility is there—endocannabinoid signaling intersects with pathways controlling appetite and insulin sensitivity—but we need prospective cohort studies and mechanistic trials to clarify causation and identify which forms of cannabis, if any, are most risky.' She adds: 'From a practical standpoint, clinicians should ask routinely about cannabis use and consider baseline and periodic metabolic screening when use is reported.'
Next steps for research and monitoring
To translate these findings into safer clinical practice and clearer public health advice, several research priorities emerge:
- Prospective cohort studies that collect detailed exposure measures (frequency, mode, potency) plus objective metabolic endpoints (fasting glucose, HbA1c, insulin resistance indices).
- Controlled mechanistic studies to test how cannabinoids influence insulin signaling, pancreatic beta-cell function and systemic inflammation.
- Stratified analyses to determine whether risk differs by age, sex, baseline BMI, genetic susceptibility and coexisting substance use.
- Health services research to design screening protocols and counseling interventions for healthcare systems in regions with legalized cannabis markets.
The EASD meeting in Vienna provides an international forum for these multidisciplinary conversations. Presentations using large federated EHR networks, like TriNetX, illustrate how real-world evidence can flag emerging public-health concerns rapidly, while also highlighting the need for complementary prospective and experimental work.
Conclusion
This large retrospective analysis linking cannabis-related diagnoses to an increased incidence of diabetes adds an important dimension to ongoing debates about cannabis safety and public policy. The association—nearly a fourfold greater risk in adjusted analyses—does not prove causation but signals a clear need for further investigation, routine metabolic monitoring in clinical practice where appropriate, and balanced public health messaging. Future prospective and mechanistic studies should aim to define which patterns or preparations of cannabis, if any, elevate metabolic risk and to identify effective strategies for prevention and early detection of diabetes among people who use cannabis.
Source: scitechdaily
Comments