Daily Endocrinology Research Analysis
Large real-world analyses illuminate endocrine care at scale: GLP-1 receptor agonists slightly increase incident diabetic retinopathy but are linked to fewer sight-threatening retinal outcomes, while technology use in type 1 diabetes surged over 15 years with persistent disparities. In hospital settings, clinician-selected SGLT2 inhibitor use showed low rates of ketoacidosis, with most ketosis tied to clear precipitants.
Summary
Large real-world analyses illuminate endocrine care at scale: GLP-1 receptor agonists slightly increase incident diabetic retinopathy but are linked to fewer sight-threatening retinal outcomes, while technology use in type 1 diabetes surged over 15 years with persistent disparities. In hospital settings, clinician-selected SGLT2 inhibitor use showed low rates of ketoacidosis, with most ketosis tied to clear precipitants.
Research Themes
- GLP-1 receptor agonists and ocular safety in diabetes
- Diabetes technology adoption and inequities in type 1 diabetes
- Inpatient safety profile of SGLT2 inhibitors
Selected Articles
1. GLP-1 Receptor Agonists and Sight-Threatening Ophthalmic Complications in Patients With Type 2 Diabetes.
In a propensity score–matched cohort of 185,066 adults with type 2 diabetes, GLP-1 receptor agonist use was linked to a modest increase in incident diabetic retinopathy but no significant increase in NAION. Among patients with preexisting retinopathy, GLP-1 RAs were associated with fewer sight-threatening complications, including lower risks of vitreous hemorrhage, neovascular glaucoma, and blindness.
Impact: This study provides high-quality real-world evidence addressing safety concerns over GLP-1 RAs and ocular events, balancing incident retinopathy risk against reduced sight-threatening outcomes.
Clinical Implications: Do not discontinue GLP-1 RAs solely over retinopathy concerns; instead, ensure baseline and periodic retinal screening for all users, particularly around rapid A1c improvement. Counsel patients on early ocular symptoms and maintain coordinated care with ophthalmology.
Key Findings
- After matching, GLP-1 RA use was associated with higher incident DR (HR 1.07, 95% CI 1.03-1.11).
- No statistically significant increase in NAION risk (HR 1.26, 95% CI 0.94-1.70).
- In patients with preexisting DR, GLP-1 RAs were associated with lower risks of vitreous hemorrhage (HR 0.74), neovascular glaucoma (HR 0.78), and blindness (HR 0.77).
- No association with progression to proliferative DR (HR 1.06) or diabetic macular edema (HR 0.98).
Methodological Strengths
- Very large, multi-center EHR cohort with propensity score matching to balance baseline covariates.
- Time-to-event analysis using Cox models with 2-year follow-up and subgroup analysis in preexisting DR.
Limitations
- Observational design with potential residual confounding and misclassification inherent to EHR data.
- Medication exposure and ophthalmic outcomes rely on coding/prescription records; retinal imaging granularity not available.
Future Directions: Prospective studies linking GLP-1 RA initiation, magnitude/rapidity of glycemic change, and retinal imaging biomarkers could clarify causal pathways and identify high-risk periods.
IMPORTANCE: Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) are associated with increased risk of diabetic retinopathy (DR) and nonarteritic anterior ischemic optic neuropathy (NAION). The risk of sight-threatening complications associated with GLP-1 RAs is underexamined. OBJECTIVE: To investigate whether the use of GLP-1 RAs in patients with T2D is associated with the development of DR, NAION, or DR complications. DESIGN, SETTING, AND PARTICIPANTS: This retrospective cohort study of adults (aged ≥18 years) with T2D and a recent hemoglobin A1c level of 6.5% or higher was conducted between January 1, 2015, and September 30, 2022, using the TriNetX database. The cohort was divided into 2 groups, adjusted for baseline characteristics through propensity score matching (PSM), based on whether the individuals received prescriptions for a GLP-1 RA. The statistical analysis was conducted on October 10, 2024. EXPOSURES: At least 2 prescriptions of a GLP-1 RA given 6 months apart. MAIN OUTCOMES AND MEASURES: Cox proportional hazard regression models were used to evaluate the primary outcome: association between GLP-1 RAs and the risk of incident DR, NAION, or sight-threatening complications over a 2-year follow-up period. RESULTS: After PSM, 185 066 individuals (mean [SD] age, 59.0 [12.5] years; 93 389 females [50.5%]) were prescribed GLP-1 receptor agonists. Use of GLP-1 RAs was associated with an increased incidence of DR (hazard ratio [HR], 1.07; 95% CI, 1.03-1.11), while no statistically significant difference was observed in the risk of NAION (HR, 1.26; 95% CI, 0.94-1.70). In a subgroup analysis of 32 695 patients with preexisting DR, GLP-1 RAs were not associated with progression to proliferative DR (HR, 1.06; 95% CI, 0.97-1.15) or diabetic macular edema (HR, 0.98; 95% CI, 0.95-1.01) but were associated with a lower occurrence of vitreous hemorrhages (HR, 0.74; 95% CI, 0.68-0.80), neovascular glaucoma (HR, 0.78; 95% CI, 0.68-0.88), or blindness (HR, 0.77; 95% CI, 0.73-0.82). CONCLUSIONS AND RELEVANCE: In this cohort study of individuals with T2D, GLP-1 RA use was associated with a modestly increased risk of incident DR; however, fewer patients experienced sight-threatening DR complications, including blindness, even among those with preexisting DR. These findings suggest that all patients with T2D treated with GLP-1 RAs, regardless of preexisting DR, should be regularly screened and monitored for potential complications of T2D.
2. Trends and Disparities in Technology Use and Glycemic Control in Type 1 Diabetes.
Across 186,590 individuals with type 1 diabetes, glycemic control improved and adoption of continuous glucose monitoring and pumps surged between 2009–2011 and 2021–2023. However, Hispanic, non-Hispanic Black, and Medicaid-insured groups consistently had the lowest technology use and glycemic goal attainment, with disparities persisting or widening.
Impact: This nationwide EHR analysis quantifies 15-year improvements and inequities in T1D care, guiding resource allocation and policy to close persistent gaps despite rapid technology diffusion.
Clinical Implications: Scale access to CGM and hybrid closed-loop systems with payer coverage expansion and culturally tailored programs. Address structural barriers for Medicaid-insured and minority populations to translate technology diffusion into equitable glycemic outcomes.
Key Findings
- HbA1c <7% increased from 7% to 19% in youths and 21% to 28% in adults over 15 years (P<.001 for trend).
- CGM use rose from 4% to 82% (youths) and 5% to 57% (adults); pump use from 16% to 50% (youths) and 11% to 29% (adults).
- Racial/ethnic minorities and Medicaid-insured individuals had the lowest technology use and glycemic control, with gaps persisting or growing.
Methodological Strengths
- Very large, national, deidentified EHR dataset spanning 15 years with repeated cross-sections.
- Consistent definitions using laboratory and claims codes; robust trend testing across predefined periods.
Limitations
- Cross-sectional snapshots cannot infer causality; potential misclassification of T1D and device use in EHR.
- Residual confounding from unmeasured social determinants and access factors.
Future Directions: Evaluate interventions that expand reimbursement and reduce logistical barriers in under-resourced communities, and assess their impact on equitable hybrid closed-loop adoption and glycemic outcomes.
IMPORTANCE: Characterizing population-level changes in type 1 diabetes (T1D) management can inform public health policies and interventions. OBJECTIVE: To characterize trends and disparities in glycemic control and use of diabetes technology among US youths and adults with T1D. DESIGN, SETTING, AND PARTICIPANTS: This serial, cross-sectional analysis used the Optum Labs Data Warehouse, a national, deidentified database of electronic health records, to identify US youths (aged <18 years) and adults (aged ≥18 years) with T1D. Data were obtained from records from January 1, 2009, to December 31, 2023. EXPOSURES: Calendar years divided into 3-year study periods from 2009 to 2011 to 2021 to 2023. MAIN OUTCOMES AND MEASURES: Glycemic control (mean hemoglobin A1c level, <7%) and use of diabetes technology (continuous glucose monitoring systems and/or insulin pumps) were defined using laboratory data and prescriptions, procedures, and diagnoses codes from electronic health records. RESULTS: A total of 186 590 participants with T1D was identified (mean [SD] age, 40 [19] years; 96 766 [52%] male; 12 493 [7%] Hispanic, 2819 [2%] non-Hispanic Asian, 21 459 [12%] non-Hispanic Black, and 141 847 [76%] non-Hispanic White). Of these, 26 853 participants were youths (mean [SD] age, 12 [4] years; 14 060 [52%] male; 19 822 [74%] non-Hispanic White) and 159 737 were adults (mean [SD] age, 45 [16] years; 82 706 [52%] male; 122 025 [76%] non-Hispanic White). From the 2009-2011 to 2021-2023 study periods, the prevalence of glycemic control (mean hemoglobin A1c level <7%) increased from 7% (95% CI, 7%-8%) to 19% (95% CI, 19%-20%) in youths (P < .001 for trend) and 21% (95% CI, 21%-22%) to 28% (95% CI, 28%-29%) in adults (P < .001 for trend). During this same period, there was a substantial increase in the percentage of patients using continuous glucose monitoring (4% to 82% for youths; 5% to 57% for adults), insulin pumps (16% to 50% for youths; 11% to 29% for adults), and both devices concurrently (1% to 47% for youths; 1% to 22% for adults) (P < .001 for trend for all). The prevalence of glycemic control and use of diabetes technology were lowest in Hispanic, non-Hispanic Black, and Medicaid-insured youths and adults, and differences persisted or increased over time. CONCLUSIONS AND RELEVANCE: In this cross-sectional study, there was a rapid increase in the use of diabetes technology and notable improvements in glycemic control among youths and adults with T1D during the past 15 years. Nonetheless, the prevalence of glycemic control remained low, and racial, ethnic, and socioeconomic differences grew over time.
3. Ketosis and ketoacidosis in hospitalized patients receiving SGLT2 inhibitor therapy.
In a single-center cohort of 1,511 hospitalized adults with type 2 diabetes receiving SGLT2 inhibitors, post–first-dose ketosis >3 mmol/L occurred in 2.8% and was almost always precipitated by surgery, sepsis, reduced intake, or critical illness. Biochemical ketoacidosis was uncommon after SGLT2i initiation (0.7%), and acidosis of any cause decreased after the first dose.
Impact: The study provides pragmatic inpatient safety data supporting selective continuation/initiation of SGLT2i with attention to peri-procedural and catabolic stressors.
Clinical Implications: When clinically indicated, SGLT2i can be used in hospital with protocols: hold around major surgery, sepsis, and reduced intake; monitor ketones in high-risk settings; resume when stressors resolve and oral intake is adequate.
Key Findings
- After first in-hospital SGLT2i dose, ketones >3 mmol/L occurred in 2.8% and were almost always linked to procedures/surgery, sepsis, reduced intake, or critical illness.
- Biochemical ketoacidosis occurred in 0.7% after first dose vs 1.7% before; acidosis of any cause was 5.5% after vs 16.8% before.
- No heart failure admissions had ketones >2 mmol/L after first SGLT2i dose.
Methodological Strengths
- Clear temporal framing of outcomes relative to first inpatient SGLT2i dose.
- Granular chart review identifying clinical precipitants of ketosis.
Limitations
- Single-center retrospective design; ketones were measured in ~51% only, possibly underestimating events.
- No randomized comparator; clinician selection may bias toward lower-risk patients.
Future Directions: Prospective, protocol-driven trials should test standardized peri-procedural SGLT2i algorithms and ketone monitoring to validate safety across diverse inpatient populations.
AIMS: The use and safety of sodium-glucose co-transporter 2 inhibitors (SGLT2i) in hospital are controversial due to the risk of ketoacidosis. We aimed to describe rates of ketosis, ketoacidosis, and acidosis in patients with type 2 diabetes treated with in-hospital SGLT2i use. MATERIALS AND METHODS: Single-centre retrospective cohort study of adult patients with type 2 diabetes admitted for >24 h. Rates of ketone testing, ketone levels, ketosis precipitants, and ketoacidosis and acidosis frequency were determined from examination of electronic medical records. Outcomes were described relative to the first dose of SGLT2i in hospital. RESULTS: There were 1511 patients who received SGLT2i at any stage of admission; of these, 777 (51.4%) had ketone levels checked. Peak ketone levels after the first dose of SGLT2i were >1 mmol/L in 12.6% and >3 mmol/L in 2.8%. Eleven out of 12 patients with ketones >3 mmol/L post-SGLT2i had ketosis in the setting of a procedure or surgery, sepsis, reduced oral intake, or critical illness. No patient admitted with heart failure had a ketone level >2 mmol/L. Biochemical ketoacidosis occurred in 1.7% of SGLT2i users pre-first dose of SGLT2i and 0.7% after the first dose. Acidosis (irrespective of ketone levels) occurred in 16.8% before the first dose of SGLT2i and 5.5% after the first dose. CONCLUSIONS: Ketosis following inpatient SGLT2i administration was infrequent and generally occurred in the setting of known precipitants. These results suggest that, in hospitalized patients selected by clinicians to receive SGLT2i, clinically significant ketone accumulation is low in prevalence.