Abstract
Objective:
Glucagon-like peptide-1 receptor agonists (GLP-1RAs), approved for type 2 diabetes (T2D) and obesity, may modulate reward pathways and reduce substance-related behaviors. This study assessed the association between GLP-1RA use and substance use-related hospitalization in older adults with coexisting T2D and substance use disorders (SUD).
Methods:
A retrospective cohort study using 2016–2020 Medicare data compared GLP-1RA initiators with new users of SGLT2 inhibitors (SGLT2is) or DPP-4 inhibitors (DPP4is) among adults aged ≥ 65. The primary outcome was hospitalization for any SUD; secondary outcomes included hospitalizations for alcohol use disorder (AUD) and opioid use disorder (OUD). Propensity score 1:1 matching and Cox proportional hazards models were used under an intention-to-treat approach.
Results:
In the GLP-1RA versus DPP4i cohort (n = 4920), GLP-1RA users had a lower risk of SUD hospitalization (HR 0.76; 95% CI, 0.67–0.86) and OUD hospitalization (HR 0.64; 95% CI, 0.43–0.96), with a nonsignificant trend for AUD (HR 0.76; 95% CI, 0.53–1.08). No significant differences were observed compared to SGLT2is (n = 4620).
Conclusions:
Among older adults with T2D, GLP-1RA use was associated with reduced SUD-related hospitalization versus DPP4i use, suggesting potential repurposing for SUD management.
Keywords: glucagon-like peptide 1 (GLP-1), older adults, subtance use disorder, type 2 diabetes
1 |. Introduction
Substance use disorders (SUD), including alcohol use disorder (AUD) and opioid use disorder (OUD), are a growing public health crisis, contributing significantly to morbidity, mortality, and health care resource utilization [1]. In 2020, an estimated 40.3 million individuals aged 12 or older in the United States (US) had SUD [2]. Between 2018–2019 and 2020–2021, emergency department (ED) visits with a primary diagnosis of SUD among adults increased by 40%, from 74.4 to 103.8 per 10,000 population [3]. Individuals with SUD are at increased risk for homelessness, co-occurring mental illness, and frequent hospitalizations, often requiring intensive care [4]. The rising burden of SUD-related hospitalizations underscores the urgent need for more effective prevention and treatment strategies.
Although pharmacological treatments exist for some SUD, their effectiveness remains limited. Many patients experience barriers to access, high costs, stigma, or suboptimal treatment outcomes [5]. Repurposing medications already approved by the US Food and Drug Administration (FDA) offers a potential strategy to address these challenges by leveraging existing drugs with well-characterized safety profiles. One promising class of medications is glucagon-like peptide-1 receptor agonists (GLP-1RAs), which are primarily used for type 2 diabetes (T2D) and obesity management but have shown emerging potential in modulating addiction-related behaviors [5].
GLP-1 receptors are expressed in key brain regions involved in reward and addiction, raising the possibility that GLP-1RAs may help reduce substance-related behaviors and associated hospitalizations [6]. Preclinical studies support this hypothesis. In rodent and nonhuman primate models, treatment with GLP-1RAs—such as exendin-4, liraglutide, dulaglutide, exenatide, and semaglutide—has been shown to reduce alcohol intake, decrease preference for alcohol, and lower motivation to obtain it, including in binge-drinking and dependence models [7–11]. When delivered directly into reward-related brain areas, GLP-1RAs also suppress alcohol- and cocaine-seeking behaviors [12, 13]. Similar effects have been observed with psychostimulants, where GLP-1RAs reduce cocaine- and amphetamine-induced reward responses and locomotor activity, and with opioids such as heroin, fentanyl, and oxycodone [7, 13–16].
In humans, a recent systematic review of five randomized controlled trials evaluated the effects of GLP-1RAs on alcohol, tobacco, and cocaine use [17]. Two studies reported significant reductions in substance use: exenatide improved smoking abstinence [18], and dulaglutide reduced alcohol consumption [19]. Another trial found no overall effect of exenatide on alcohol use but reported a significant reduction in heavy drinking days and total alcohol intake among participants with obesity [20]. The remaining studies showed no significant effects of GLP-1RAs on cocaine use or smoking abstinence [21, 22].
A growing body of pharmacoepidemiologic research provides additional evidence for the potential of GLP-1RAs to reduce the risk of substance-related disorders. In a U.S. Veterans cohort, GLP-1RA use was associated with greater reductions in alcohol consumption compared to both nonusers and recipients of dipeptidyl peptidase-4 inhibitors (DPP4is), particularly in individuals with preexisting alcohol use disorder or hazardous drinking [23]. Another large-scale analysis from the same health system, examining 175 clinical outcomes, found lower rates of diagnosed SUD among GLP-1RA users relative to those prescribed other antihyperglycemics, including sulfonylureas, DPP4is, and SGLT2 inhibitors (SGLT2is) [24]. Additional observational studies add further support: one self-controlled analysis reported lower rates of alcohol intoxication and opioid overdose during periods of GLP-1RA use, while another found fewer AUD-related hospitalizations, particularly among individuals with comorbid obesity or T2D [25, 26]. However, one Danish cohort study comparing GLP-1RAs with DPP4is found only short-term reductions in alcohol-related hospital contacts, with no sustained benefit over time [27].
Despite encouraging evidence, older adults with T2D remain underrepresented in studies evaluating GLP-1RAs for SUD. This population is particularly important to study, as aging is a major risk factor for T2D, and older adults often experience multiple chronic conditions—including cardiovascular, renal, and neurocognitive diseases—that may influence substance use patterns and treatment outcomes [28, 29]. GLP-1RAs have shown potential benefits across many of these conditions, raising the question of whether their observed effects on substance-related outcomes extend to older populations. Given age-related changes in drug metabolism and heightened vulnerability to hospitalizations, it is essential to evaluate the real-world effectiveness of GLP-1RAs in this group [30]. To address this gap, the present study uses Medicare data to examine the association between GLP-1RA use and substance-related hospitalizations in older adults with T2D.
2 |. Methods
2.1 |. Study Design and Data Source
We conducted a retrospective cohort study with a new-user, active-comparator design to investigate the association between GLP-1RAs and SUD among older patients (≥ 65 years old) with T2D, compared to SGLT2is or DPP4is.
Our study utilized a 15% random sample of Medicare administrative claims data from January 2013 to December 2020. Medicare is a federal health insurance program in the US that primarily serves individuals aged 65 and older, as well as certain younger individuals with specific disabilities or conditions. The Medicare administrative claims data include Parts A (inpatient), B (outpatient physician services), and D (dispensed prescription drugs), providing a comprehensive longitudinal record of healthcare utilization. These data encompass demographic characteristics, diagnoses, procedures, medication prescriptions, and services delivered across various care settings.
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines and was approved by the University of Florida Institutional Review Board.
2.2 |. Study Population
Eligible patients were Medicare beneficiaries aged ≥ 65 years coexisting with T2D and SUD, who initiated treatment with either GLP-1RAs, SGLT2is, or DPP4is between January 1, 2016 and December 31, 2020. The date of treatment initiation (index date) was defined as the date of first prescription for any treatment. To ensure a comprehensive assessment of patients’ baseline characteristics, we required a minimum of 1 year of continuous enrollment in Medicare Parts A, B, and D prior to the index date.
To focus exclusively on new users of the study’s medications, we excluded individuals who had received prescriptions for either GLP-1RAs or the comparator drugs in 1 year preceding the index date, had end-stage renal disease, or initiated treatment with both GLP-1RAs and a comparator drug on the cohort entry date. Detailed information regarding the exposures of interest and the specific codes used for the inclusion and exclusion criteria is provided in Tables S1 and S2.
2.3 |. Treatment Strategy
We conducted two pairwise comparisons: patients initiating GLP-1RA treatment versus those starting SGLT2i therapy and GLP-1RA treatment versus DPP4i therapy. All three drug classes—GLP-1RAs, SGLT2is, and DPP4is—are approved for the treatment of T2D. Although certain GLP-1RAs (specifically liraglutide, semaglutide, and tirzepatide at higher doses) are also FDA-approved for obesity, their use in our cohort was for T2D management.
The selection of SGLT2is and DPP4is as active positive comparators was based on their common use as second-line therapies for T2D, in accordance with clinical practice guidelines. SGLT2is share similar indications with GLP-1RAs, particularly for patients with established cardiovascular disease or high cardiovascular risk. DPP4is exhibit mechanisms of action analogous to GLP-1RAs in terms of glucose-lowering effects.
2.4 |. Study Outcomes and Follow-Up
The primary outcome was hospitalization for SUD. The AUD and OUD hospitalizations were evaluated as secondary outcomes. We focused on hospitalization as a clinically meaningful indicator of substance-related harm, while the primary SUD hospitalization outcome encompassed all types of SUD, including both AUD and OUD. SUD were evaluated as a composite to capture a broad range of substances, while AUD and OUD hospitalizations were examined separately to explore potential differences by substance type. The outcomes were identified by an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) or Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis code recorded in any diagnostic position on a Medicare claim (Table S1).
We performed an intention-to-treat analysis, following patients from the index date until the earliest of the following events: occurrence of hospitalization due to SUD, AUD, or OUD; death; disenrollment from Medicare Parts A, B, or D; or the end of the study period on December 31, 2020.
2.5 |. Covariates
Baseline covariates included demographic characteristics (e.g., age, sex, race/ethnicity), socioeconomic factors (e.g., region, low-income subsidy), comorbidities (e.g., obesity, chronic kidney disease), other glucose-lowering drugs (e.g., insulin, metformin), and non-diabetes-related medications (e.g., beta blockers, statins), as shown in Table 1. These covariates were collected during the year before or on the index date.
TABLE 1 |.
Baseline characteristics of GLP-1RA versus SGLT2i cohort and GLP-1RA versus DPP4i cohort after 1:1 propensity score matching.
| GLP-1RA versus SGLT2i |
GLP-1RA versus DPP4i |
|||||
|---|---|---|---|---|---|---|
| Characteristic | GLP-1RA, no (%) (n = 2310) | SGLT2i, no (%) (n = 2310) | SMD | GLP-1RA, no (%) (n = 2460) | DPP4i, no (%) (n = 2460) | SMD |
|
| ||||||
| Age, years, mean (SD) | 71.7 (4.9) | 71.6 (5.0) | −0.01 | 71.8 (5.0) | 71.9 (5.2) | −0.01 |
| Race/ethnicity | ||||||
| NHW | 1788 (77.4%) | 1797 (77.8%) | 0.04 | 1862 (75.7%) | 1870 (76.0%) | 0 |
| NHB | 213 (9.2%) | 213 (9.2%) | 282 (11.5%) | 266 (10.8%) | ||
| Hispanic | 196 (8.5%) | 182 (7.9%) | 198 (8.0%) | 211 (8.6%) | ||
| Other | 113 (4.9%) | 118 (5.1%) | 118 (4.8%) | 113 (4.6%) | ||
| Female | 1005 (43.5%) | 1012 (43.8%) | 0.006 | 1197 (48.7%) | 1229 (50.0%) | −0.01 |
| Year of enrollment | ||||||
| 2017 | 520 (22.5%) | 521 (22.6%) | 0.03 | 602 (24.5%) | 608 (24.7%) | 0.03 |
| 2018 | 519 (22.5%) | 521 (22.6%) | 636 (25.9%) | 651 (26.5%) | ||
| 2019 | 648 (28.1%) | 658 (28.5%) | 661 (26.9%) | 677 (27.5%) | ||
| 2020 | 623 (27.0%) | 610 (26.4%) | 561 (22.8%) | 524 (21.3%) | ||
| Medicare and medicaid dual eligibility | 750 (32.5%) | 743 (32.2%) | −0.01 | 878 (35.7%) | 913 (37.1%) | −0.002 |
| Low-income subsidy | 852 (36.9%) | 840 (36.4%) | −0.01 | 995 (40.4%) | 1034 (42.0%) | −0.01 |
| Region | ||||||
| West | 269 (11.6%) | 291 (12.6%) | 0.04 | 285 (11.6%) | 278 (11.3%) | 0 |
| Midwest | 389 (16.8%) | 402 (17.4%) | 443 (18.0%) | 446 (18.1%) | ||
| Northeast | 357 (15.5%) | 346 (15.0%) | 363 (14.8%) | 340 (13.8%) | ||
| Southwest | 203 (8.8%) | 196 (8.5%) | 217 (8.8%) | 219 (8.9%) | ||
| Southeast | 1092 (47.3%) | 1075 (46.5%) | 1152 (46.8%) | 1177 (47.8%) | ||
| Diabetes-related conditions | ||||||
| Diabetes retinopathy | 236 (10.2%) | 239 (10.3%) | 0.004 | 255 (10.4%) | 263 (10.7%) | 0 |
| Diabetic neuropathy | 875 (37.9%) | 880 (38.1%) | 0.005 | 989 (40.2%) | 1016 (41.3%) | 0.03 |
| Peripheral vascular disease | 705 (30.5%) | 700 (30.3%) | −0.005 | 782 (31.8%) | 812 (33.0%) | −0.002 |
| Hypoglycemia | 56 (2.4%) | 55 (2.4%) | −0.003 | 89 (3.6%) | 93 (3.8%) | −0.006 |
| Hyperglycemic emergency | 17 (0.7%) | 16 (0.7%) | −0.005 | 28 (1.1%) | 23 (0.9%) | −0.01 |
| GLD use at baseline | ||||||
| Insulin | 774 (33.5%) | 770 (33.3%) | −0.004 | 953 (38.7%) | 942 (38.3%) | 0.006 |
| Metformin | 1535 (66.5%) | 1540 (66.7%) | 0.002 | 1575 (64.0%) | 1579 (64.2%) | 0.005 |
| Sulfonylureas | 877 (38.0%) | 879 (38.1%) | 0.002 | 911 (37.0%) | 950 (38.6%) | −0.01 |
| DPP4i | 163 (7.1%) | 146 (6.3%) | −0.03 | 12 (0.5%) | 13 (0.5%) | −29.17 |
| Thiazolidinediones | 190 (8.2%) | 188 (8.1%) | −0.003 | 164 (6.7%) | 184 (7.5%) | 0.004 |
| Meglitinides | 30 (1.3%) | 34 (1.5%) | 0.01 | 35 (1.4%) | 36 (1.5%) | −0.004 |
| Alpha-glucosidase inhibitors | < 11 | < 11 | 0 | < 11 | < 11 | 0 |
| Comorbidities | ||||||
| Acute myocardial infarction | 211 (9.1%) | 196 (8.5%) | −0.03 | 203 (8.3%) | 207 (8.4%) | 0.002 |
| ADRD | 321 (13.9%) | 321 (13.9%) | 0 | 440 (17.9%) | 442 (18.0%) | 0.01 |
| Atrial fibrillation | 399 (17.3%) | 383 (16.6%) | −0.02 | 437 (17.8%) | 442 (18.0%) | 0 |
| Cataract | 1272 (55.1%) | 1266 (54.8%) | −0.01 | 1428 (58.0%) | 1445 (58.7%) | −0.002 |
| Chronic kidney disease | 1717 (74.3%) | 1723 (74.6%) | 0.006 | 1900 (77.2%) | 1898 (77.2%) | 0.003 |
| COPD | 1177 (51.0%) | 1164 (50.4%) | −0.01 | 1307 (53.1%) | 1328 (54.0%) | 0.003 |
| Chronic heart failure | 893 (38.7%) | 871 (37.7%) | −0.02 | 1024 (41.6%) | 1033 (42.0%) | 0.01 |
| Glaucoma | 461 (20.0%) | 453 (19.6%) | −0.01 | 486 (19.8%) | 495 (20.1%) | 0.006 |
| Hip or pelvic fracture | 46 (2.0%) | 42 (1.8%) | −0.01 | 58 (2.4%) | 64 (2.6%) | −0.01 |
| Ischemic heart disease | 1557 (67.4%) | 1553 (67.2%) | −0.004 | 1673 (68.0%) | 1666 (67.7%) | 0.01 |
| Osteoporosis | 327 (14.2%) | 323 (14.0%) | −0.005 | 395 (16.1%) | 398 (16.2%) | 0.01 |
| Rheumatoid arthritis/osteoarthritis | 1910 (82.7%) | 1916 (82.9%) | 0.01 | 1743 (70.9%) | 1729 (70.3%) | 0.01 |
| Stroke/TIA | 400 (17.3%) | 394 (17.1%) | −0.007 | 474 (19.3%) | 485 (19.7%) | 0.01 |
| Anemia | 1380 (59.7%) | 1365 (59.1%) | −0.01 | 1546 (62.8%) | 1533 (62.3%) | 0.01 |
| Asthma | 480 (20.8%) | 483 (20.9%) | 0.003 | 562 (22.8%) | 584 (23.7%) | 0.002 |
| Hyperlipidemia | 2181 (94.4%) | 2182 (94.5%) | 0.002 | 2329 (94.7%) | 2327 (94.6%) | −0.002 |
| Hypertension | 2217 (96.0%) | 2213 (95.8%) | −0.01 | 2360 (95.9%) | 2363 (96.1%) | 0.002 |
| Obesity | 958 (41.5%) | 960 (41.6%) | 0.002 | 1084 (44.1%) | 1068 (43.4%) | 0.004 |
| Depression | 1189 (51.5%) | 1210 (52.4%) | 0.02 | 1376 (55.9%) | 1388 (56.4%) | −0.001 |
| Schizophrenia | 67 (2.9%) | 67 (2.9%) | 0 | 90 (3.7%) | 93 (3.8%) | −0.003 |
| Anxiety | 757 (32.8%) | 751 (32.5%) | −0.01 | 864 (35.1%) | 869 (35.3%) | −0.003 |
| Breast cancer | 101 (4.4%) | 108 (4.7%) | 0.01 | 132 (5.4%) | 134 (5.4%) | 0.002 |
| Colorectal cancer | 58 (2.5%) | 60 (2.6%) | 0.01 | 64 (2.6%) | 65 (2.6%) | −0.01 |
| Prostate cancer | 146 (6.3%) | 141 (6.1%) | −0.01 | 138 (5.6%) | 125 (5.1%) | 0.01 |
| Lung cancer | 57 (2.5%) | 58 (2.5%) | 0.003 | 54 (2.2%) | 58 (2.4%) | −0.003 |
| Endometrial cancer | 20 (0.9%) | 17 (0.7%) | −0.01 | 22 (0.9%) | 26 (1.1%) | −0.02 |
| Comedications | ||||||
| ACEI | 965 (41.8%) | 937 (40.6%) | −0.02 | 979 (39.8%) | 994 (40.4%) | −0.01 |
| ARB | 775 (33.5%) | 766 (33.2%) | −0.01 | 797 (32.4%) | 801 (32.6%) | 0.003 |
| Beta blockers | 1316 (57.0%) | 1315 (56.9%) | −0.001 | 1389 (56.5%) | 1412 (57.4%) | 0.003 |
| Calcium channel blockers | 840 (36.4%) | 847 (36.7%) | 0.006 | 919 (37.4%) | 933 (37.9%) | 0.002 |
| Diuretics | 983 (42.6%) | 1003 (43.4%) | 0.02 | 1165 (47.4%) | 1182 (48.0%) | 0.001 |
| Antibiotics | 427 (18.5%) | 437 (18.9%) | 0.01 | 464 (18.9%) | 465 (18.9%) | −0.004 |
| Statins | 1886 (81.6%) | 1865 (80.7%) | −0.02 | 1989 (80.9%) | 1991 (80.9%) | −0.01 |
| Antipsychotics | 81 (3.5%) | 84 (3.6%) | 0.007 | 123 (5.0%) | 126 (5.1%) | −0.01 |
| NSAIDs | 636 (27.5%) | 623 (27.0%) | −0.01 | 707 (28.7%) | 690 (28.0%) | 0.01 |
| Oral steroids | 1131 (49.0%) | 1134 (49.1%) | 0.003 | 1227 (49.9%) | 1230 (50.0%) | 0.01 |
| Opioids | 1067 (46.2%) | 1072 (46.4%) | 0.0043 | 1198 (48.7%) | 1214 (49.3%) | −0.013 |
| Antiplatelets | 98 (4.2%) | 92 (4.0%) | −0.01 | 93 (3.8%) | 106 (4.3%) | −0.005 |
| Aldosterone receptor antagonists | 183 (7.9%) | 192 (8.3%) | 0.01 | 217 (8.8%) | 214 (8.7%) | 0.002 |
| Anticoagulants | 402 (17.4%) | 377 (16.3%) | −0.03 | 421 (17.1%) | 430 (17.5%) | −0.01 |
| Immunosuppressants | 12 (0.5%) | 16 (0.7%) | 0.02 | 18 (0.7%) | 20 (0.8%) | 0.02 |
| Antidepressants | 1120 (48.5%) | 1117 (48.4%) | −0.003 | 1245 (50.6%) | 1258 (51.1%) | −0.001 |
| Anti-AUD | < 11 | < 11 | 0.0 | < 11 | < 11 | 0.0 |
| Anti-OUD | < 11 | < 11 | 0.0 | 12 | < 11 | 0.0 |
Note: Anti-AUD, “ACAMPROSATE CALCIUM,” “DISULFIRAM,” “NALTREXONE HYDROCHLORIDE”; Anti-OUD, “BUPRENORPHINE,” “BUPRENORPHINE HYDROCHLORIDE,” “BUPRENORPHINE/NALOXONE,” “BUPRENORPHINE HYDROCHLORIDE/NALOXONE HYDROCHLORIDE,” “METHADONE HYDROCHLORIDE,” “NALTREXONE HYDROCHLORIDE,” “LOFEXIDINE.”
Abbreviations: ACEI, angiotensin-converting-enzyme inhibitors; ADRD, Alzheimer’s disease and related dementias; ARB, angiotensin II receptor blockers; COPD, chronic obstructive pulmonary disease; DPP4i, dipeptidyl peptidase 4 inhibitors; GLD, glucose-lowering drug; GLP-1RA, glucagon-like peptide-1 receptor agonists; NSAIDs, nonsteroidal anti-inflammatory drugs; SGLT2i, sodium-glucose cotransporter 2 inhibitors; SMD, standardized mean difference; TIA, transient ischemic attack; TNF inhibitors, tumor necrosis factor inhibitors.
2.6 |. Statistical Analysis
To minimize potential confounding effects, we implemented 1:1 time-dependent propensity score (PS) matching [31]. PS was calculated using multivariable logistic regression incorporating the baseline covariates. Matched cohorts were constructed using a nearest-neighbor matching algorithm without replacement, with a maximum caliper width of 0.05 [32]. The balance of baseline covariates was evaluated both before and after matching by standardized mean differences (SMD), with SMD < 0.1 indicating negligible imbalance between the two comparison groups [33].
We calculated the incidence rate (IR) of each of the study outcomes and employed Kaplan–Meier survival analysis to illustrate the cumulative incidence over time between the matched groups. Additionally, Cox proportional hazards regression models were used to estimate the hazard ratio (HR) and 95% confidence interval (CI) by comparing patients prescribed GLP-1RAs to those prescribed SGLT2is or DPP4is.
To assess the robustness of the study findings, we employed the per-protocol principle [34] to further mitigate bias related to deviations from initial treatment regimens. This approach involved censoring patients at the point they discontinued their index treatment, which was defined as a 60-day gap following the end of the last prescription’s supply, in the absence of a subsequent refill. Additionally, we utilized Fine-Gray subdistribution hazard models to account for the competing risk of mortality. This method provided a more accurate estimation of the cumulative incidence of the primary outcome, accounting for the fact that death may preclude its occurrence.
Subgroup analyses were performed to assess potential effect modifications by age (≥ 75 vs. < 75 years), sex (female vs. male), race/ethnicity (non-Hispanic White vs. non-Hispanic Black vs. Hispanic vs. others), obesity (obesity vs. nonobesity), insulin (use vs. non-use), and specific GLPs (specific GLP vs. active comparators, in terms of exenatide, dulaglutide, liraglutide and semaglutide). For each subgroup analysis, we applied a Cox proportional hazards method to estimate the HR within each newly PS-matched subgroup. Statistical significance was set at p < 0.05. All analyses were conducted using SAS, version 9.4 (SAS Institute Inc.), and survival curves were produced using R software with the “survminer” package.
3 |. Results
3.1 |. Study Population
The flowchart of patient selection is presented in Figure 1. The study population comprised 12,875 patients coexisting with T2D and SUD who initiated GLP-1RAs, SGLT2is, or DPP4is, including: 6388 in the GLP-1RA versus SGLT2i cohort and 9374 in the GLP-1RA versus DPP4i cohort. After 1:1 PS matching, we retained 4620 patients (2310 per group) in the GLP-1RA versus SGLT2i cohort and 4920 patients (2460 per group) in the GLP-1RA versus DPP4i cohort. Baseline characteristics before and after 1:1 PS matching are presented in Tables S3 and S4. In the matched GLP-1RA versus SGLT2i cohort, patients had a mean age of 71.6 years, with 43.6% women and 77.6% non-Hispanic White. At baseline, 33.4% used insulin, and 41.5% had a diagnosis of obesity. The matched GLP-1RA versus DPP4i cohort had a mean age of 71.8 years, with 49.4% women and 75.8% non-Hispanic White. In this cohort, 38.5% used insulin at baseline, and 43.8% had an obesity diagnosis.
FIGURE 1 |.
Patient attrition flowchart.
3.2 |. Risk of Hospitalizations for SUD, OUD, and AUD
Table 2 presents the number of events, follow-up time, IR, and HR for all outcomes. Figure 2 shows Kaplan–Meier plots of the cumulative incidence of hospitalization related to SUD during follow-up, comparing GLP-1RA users with SGLT2i and DPP4i users. Patients were followed for up to 7 years. The mean follow-up time for SUD-related hospitalization was 1.40 years (SD = 1.04) in the GLP-1RA group and 1.38 years (SD = 1.06) in the SGLT2i group. In the GLP-1RA versus DPP4i cohort, the mean follow-up duration was 1.45 years (SD = 1.05) for GLP-1RA users and 1.37 years (SD = 1.06) for DPP4i users. Hospitalizations for AUD and OUD were also evaluated. For AUD, the average follow-up was 1.63 years (SD = 1.08) for GLP-1RA users and 1.60 years (SD = 1.08) for DPP4i users. For OUD, the mean follow-up time was 1.57 years (SD = 1.07) in the GLP-1RA group and 1.55 years (SD = 1.08) in the SGLT2i group; in comparison with DPP4is, the mean follow-up was 1.63 years versus 1.60 years, respectively.
TABLE 2 |.
Primary and secondary outcomes in 1:1 propensity score matched GLP-1RA versus SGLT2i and GLP-1RA versus DPP4i cohorts.
| GLP-1RA versus SGLT2i |
GLP-1RA versus DPP4i |
|||
|---|---|---|---|---|
| Outcomes | GLP-1RA | SGLT2i | GLP-1RA | DPP4i |
|
| ||||
| Hospitalization due to substance use disorder | ||||
| No. of events/no. of patients at risk | 370/2310 | 373/2310 | 433/2460 | 544/2460 |
| Follow-up, years, mean (SD) | 1.40 (1.04) | 1.38 (1.06) | 1.45 (1.05) | 1.37 (1.06) |
| Incidence rate, 1000 person-years | 114.41 | 116.76 | 121.56 | 161.19 |
| HR (95% CI) | 0.98 (0.85, 1.13) | 1 (ref) | 0.76 (0.67, 0.86) | 1 (ref) |
| Hospitalization due to alcohol use disorder | ||||
| No. of cases/no. of patients at risk | 46/2310 | 39/2310 | 53/2460 | 69/2460 |
| Follow-up, years, mean (SD) | 1.56 (1.07) | 1.55 (1.08) | 1.63 (1.08) | 1.60 (1.08) |
| Incidence rate, 1000 person-years | 19.91 | 16.88 | 21.54 | 28.05 |
| HR (95% CI) | 1.17 (0.76, 1.80) | 1 (ref) | 0.76 (0.53, 1.08) | 1 (ref) |
| Hospitalization opioid use disorder | ||||
| No. of cases/no. of patients at risk | 32/2310 | 33/2310 | 41/2460 | 63/2460 |
| Follow-up, years, mean, (SD) | 1.57 (1.07) | 1.55 (1.08) | 1.63 (1.08) | 1.60 (1.08) |
| Incidence rate, 1000 person-years | 13.85 | 14.29 | 16.67 | 25.61 |
| HR (95% CI) | 0.96 (0.60, 1.53) | 1 (ref) | 0.64 (0.43, 0.96) | 1 (ref) |
Abbreviations: DPP4i, dipeptidyl peptidase 4 inhibitors; GLP-1RA, glucagon-like peptide-1 receptor agonists; HR, hazard ratio; IR, incidence rate; SGLT2i, sodium-glucose cotransporter 2 inhibitors.
FIGURE 2 |.
Cumulative incidence curves for AUD, OUD, and SUD among patients prescribed (A) GLP-1RA versus DPP-4i or (B) SGLT2i.
In the GLP-1RA versus SGLT2i cohort, the IR of SUD-related hospitalization was 114.41 per 1000 person-years for GLP-1RA users and 116.76 per 1000 person-years for SGLT2i users. In the GLP-1RA versus DPP4i cohort, the IR of SUD-related hospitalization was 121.56 per 1000 person-years for GLP-1RA users and 161.19 per 1000 person-years for DPP4i users.
In the GLP-1RA versus SGLT2i cohort, the HR for hospitalization due to any SUD was 0.98 (95% CI, 0.85–1.13), suggesting no significant difference between the two groups. Similarly, no differences were observed for specific substance-related hospitalizations. The HR for OUD-related hospitalization was 1.17 (95% CI, 0.76–1.80); for OUD, the HR was 0.96 (95% CI, 0.60–1.53).
In the GLP-1RA versus DPP4i cohort, GLP-1RA use was associated with a statistically significant lower risk of hospitalization for any SUD (HR, 0.76 [95% CI, 0.67 to 0.86]) compared to DPP4i use. This trend was also observed in OUD (HR, 0.64 [95% CI, 0.43 to 0.96]). Although the risk of AUD-related hospitalization was also lower in GLP-1RA users (HR, 0.76 [95% CI, 0.53–1.08]), the difference was not statistically significant.
3.3 |. Subgroup Analyses
Subgroup analyses by age, sex, race/ethnicity, obesity, insulin use, and specific GLP-1RAs are presented in Table S5, and Figure 3 presents subgroup forest plots. In subgroup analyses comparing GLP-1RAs to DPP4is, several statistically significant reductions in the risk of SUD-, AUD-, and OUD-related hospitalizations were observed. Consistent with the main analysis, no statistically significant subgroup differences were observed in the comparisons between GLP-1RAs and SGLT2is for SUD- and AUD-related hospitalization outcomes. However, GLP-1RA use was associated with a significantly increased risk of OUD-related hospitalization among females (HR, 2.01 [95% CI, 1.05–3.87]).
FIGURE 3 |.
Subgroup forest plots for hospitalization risk for AUD, OUD, and SUD comparing GLP-1RA versus DPP-4i or SGLT2i.
3.4 |. Sensitivity Analyses
Sensitivity analyses were conducted using a per-protocol approach, as well as a competing risk model based on the Fine-Gray subdistribution hazard model.
In the per-protocol analysis, comparing GLP-1RAs to SGLT2is, there was no statistically significant difference in the risk of SUD-related hospitalization (HR, 0.98 [95% CI, 0.85–1.13]), AUD-related hospitalization (HR, 1.16 [95% CI, 0.75–1.78]), or OUD-related hospitalization (HR, 0.95 [95% CI, 0.60–1.52]). In contrast, when comparing GLP-1RAs to DPP4is, GLP-1RA use was associated with a significantly lower risk of SUD-related hospitalization (HR, 0.77 [95% CI, 0.68–0.87]) and OUD-related hospitalization (HR, 0.65 [95% CI, 0.44–0.96]). Although the risk of AUD-related hospitalization was also lower (HR, 0.76 [95% CI, 0.54–1.09]), this difference did not reach statistical significance.
In the competing risk analysis accounting for death as a competing event, GLP-1RA use showed no significant differences compared to SGLT2i use for SUD-related hospitalization (HR, 0.98 [95% CI, 0.85–1.12]), OUD-related hospitalization (HR, 0.95 [95% CI, 0.60–1.52]), and AUD-related hospitalization (HR, 1.16 [95% CI, 0.76–1.79]), respectively. Regarding GLP-1RAs versus DPP4is, the HR for the risk of SUD-related hospitalization was 0.78 (95% CI, 0.69–0.88), indicating a significantly reduced risk. A similarly significant reduction was observed for OUD-related hospitalization (HR, 0.66 [95% CI, 0.45–0.98]). In contrast, the HR for AUD-related hospitalization was 0.78 (95% CI, 0.54–1.12), showing no statistically significant difference. This shares similar findings compared to intent-to-treat analysis.
4 |. Discussion
In this real-world retrospective cohort study, we examined the association between GLP-1RA use and the risk of hospitalizations for SUD, OUD, and AUD among older adults with T2D using national Medicare claims data from 2013 to 2020. GLP-1RA use was associated with a significantly lower risk of SUD-related hospitalization compared to DPP4i use, including a reduced risk of OUD. No significant differences were observed in comparisons between GLP-1RA and SGLT2i users for any substance-related outcome.
The significant reduction in SUD- and OUD-related hospitalizations observed in our study may be attributed to the action of GLP-1RAs on central GLP-1 receptors in brain regions involved in reward processing, such as the ventral tegmental area and nucleus accumbens [35]. GLP-1RAs have been shown to modulate dopamine release in response to addictive substances, potentially reducing their reinforcing effects and decreasing substance-seeking behavior [36]. This aligns with preclinical evidence demonstrating that GLP-1 signaling influences both food- and drug-related reward pathways [37]. Our findings are consistent with those of Qeadan et al., who conducted a real-world retrospective study using Oracle Cerner Real-World Data to evaluate the association between GLP-1RA and/or glucose-dependent insulinotropic polypeptide (GIP) prescriptions and the risk of opioid overdose. That study reported significantly lower overdose rates among patients prescribed GIP/GLP-1RAs compared to nonusers (adjusted incidence rate ratio [aIRR] = 0.60; 95% CI: 0.43–0.83) [25]. Unlike our analysis, which focused specifically on hospitalization events and employed an active comparator design, Qeadan et al. examined overdose events in a broader population and did not differentiate between GLP-1RA and dual GIP/GLP-1RA agents. Our study provides complementary evidence using a distinct outcome and an older adult population with T2D—a group especially vulnerable to substance-related harm. Although our primary mechanistic interpretation centers on GLP-1 receptor signaling, the absence of significant differences between GLP-1RA and SGLT2i users may reflect comparable outcomes between the two classes rather than a true lack of effect. SGLT2is may influence substance-related outcomes through alternative pathways, including modulation of neuroinflammation, cerebrovascular function, or systemic metabolic regulation [38, 39]. This potential warrants further investigation.
At the drug-specific level, our results add important nuance to prior findings. Wang et al. utilized a target trial emulation design with TriNetX data and reported a significantly reduced risk of opioid overdose with semaglutide compared to other antidiabetic agents (hazard ratios ranging from 0.32 to 0.58), including other GLP-1RAs [40]. In contrast, our subgroup analyses found that both dulaglutide and semaglutide were associated with significantly lower risk of SUD-related hospitalization (HR, 0.74 [95% CI, 0.61–0.89] and HR, 0.63 [95% CI, 0.43–0.94], respectively). Importantly, a significant reduction in OUD-related hospitalization was observed only among dulaglutide users (HR, 0.48 [95% CI, 0.26–0.90]), suggesting potential heterogeneity in effectiveness across individual agents. These findings underscore the importance of evaluating GLP-1RAs not only at the class level but also through a drug-specific lens. Variability in pharmacokinetic profiles may contribute to differential outcomes in the context of SUD. Further research is warranted to clarify whether certain GLP-1RAs confer greater benefit in specific patient populations or in relation to particular substance use outcomes.
In our study comparing GLP-1RA to DPP4i use, we observed a lower risk of AUD-related hospitalization, although the associations did not reach statistical significance. These findings differ somewhat from previous real-world studies evaluating alcohol-related outcomes during GLP-1RA treatment. A Swedish registry study using a within-person design found significantly lower rates of AUD-related hospitalization during periods of active semaglutide or liraglutide use compared to periods of nonuse within the same individuals, thus eliminating between-person confounding but lacking an external comparator [26]. In contrast, a Danish registry study employed both self-controlled and new-user comparative designs and reported a significant reduction in alcohol-related events among GLP-1RA users compared to DPP4i users, but only during the first 3 months after initiation. No long-term difference was observed thereafter, suggesting a transient benefit [27]. Similarly, Qeadan et al. used US claims data and compared GIP/GLP-1RA users to nonusers, finding a 50% reduction in alcohol intoxication episodes among incretin users [25]. While these studies differ in outcome definitions and comparator selection (nonuse, DPP4is, or general population), they consistently suggest a potential benefit associated with active GLP-1RA treatment, particularly in the short term. While our results suggest a directionally similar effect, the lack of statistical significance warrants cautious interpretation. This may reflect differences in study population, outcome definition, or follow-up time. Further research is needed to clarify the potential impact of GLP-1RAs on alcohol-related outcomes, particularly with respect to duration of treatment, severity of substance use, and timing of effect.
Notably, in subgroup analyses of the primary (intention-to-treat) cohort, significant associations were observed among insulin users and semaglutide users, suggesting that certain patient populations may derive greater benefit. It is also important to consider that while preclinical studies consistently show that GLP-1RAs, such as exendin-4, semaglutide, and liraglutide, reduce alcohol intake, especially in high alcohol-consuming phenotypes, these effects may not fully translate to human populations [41]. Human studies remain limited and mixed, with some showing reductions in alcohol use that appear transient or restricted to certain subgroups, such as individuals with high BMI [20, 41]. These observations highlight the need for further research to clarify which patients may benefit most from GLP-1RAs in the context of alcohol-related outcomes, and whether effects are sustained over time.
Our findings carry important public health and clinical implications, particularly for older adults with T2D—a population at increased risk for both metabolic disorders and SUD. The observed association between GLP-1RA use and reduced risk of hospitalization for SUD suggests that these agents may offer therapeutic benefit beyond glycemic control. Given the established role of GLP-1 signaling in reward processing, our results align with emerging evidence that GLP-1RAs may modulate addictive behaviors through central mechanisms. Clinically, these findings underscore the need for further investigation into the potential of GLP-1RAs as adjunctive treatments for older adults with coexisting metabolic and substance use conditions. If validated in prospective trials, the integration of GLP-1RAs into treatment paradigms could offer a metabolically informed approach to reducing addiction-related hospitalization in this high-risk population. From a public health perspective, our findings highlight the interconnectedness of metabolic and neuropsychiatric health in aging populations, suggesting a need to reevaluate traditional treatment frameworks for SUD. As GLP-1RAs gain widespread use in the management of obesity and T2D, their potential to impact addiction-related outcomes warrants further exploration in older adults, who are often underrepresented in clinical trials. Future research should assess real-world accessibility, cost-effectiveness, and health care system integration in geriatric populations to determine whether GLP-1RAs can be leveraged to improve both metabolic and behavioral health outcomes. A multidisciplinary approach bridging geriatrics, endocrinology, addiction medicine, and public health policy will be essential to optimizing the use of GLP-1RAs in clinical practice for older adults.
This study has several important strengths. We evaluated three clinically meaningful outcomes, in terms of SUD, OUD, and AUD, within a large, nationally representative Medicare cohort of older adults with T2D. By using an active-comparator, new-user design and applying time-dependent PS matching, we minimized confounding by indication and immortal time bias. Our analytic approach included a range of sensitivity analyses, such as per-protocol and competing risk models, to examine the consistency of findings. Subgroup analyses further reinforced the robustness and potential generalizability of the observed associations.
However, several limitations warrant consideration. The retrospective observational design precludes causal inference, as unmeasured confounders may influence the observed associations. While we adjusted for key demographic and clinical variables, factors such as behavioral influences and homelessness were not captured in claims data and may have contributed to hospitalization risk. Additionally, treatment adherence in real-world settings may differ from clinical trial conditions, potentially attenuating or amplifying observed associations. Our focus on older adults, while a strength in terms of clinical applicability, may limit generalizability to younger populations with different patterns of substance use and treatment response. The study also lacked data on the duration and severity of AUD, SUD, and OUD, which prevented stratified analyses that could have provided more insight into the heterogeneity of treatment effects. Lastly, our findings are exploratory and should be interpreted with caution. Future studies incorporating longitudinal data with richer behavioral and clinical measures are needed to further elucidate these associations.
5 |. Conclusion
This study provides real-world evidence that GLP-1RA use is associated with a reduced risk of hospitalization for SUD compared to DPP4i use among older adults coexisting with T2D and SUD. In addition to metabolic control, GLP-1RAs may confer neurobehavioral benefits, positioning them as a promising option for patients with overlapping metabolic and behavioral health needs. These findings may help guide treatment selection when substance use risk is a concern. As the burden of coexisting diabetes and substance use rises in aging populations, further research is needed to elucidate underlying mechanisms and confirm these treatment effects in clinical trials.
Supplementary Material
Supporting Information
Additional supporting information can be found online in the Supporting Information section. Data S1: oby70024-sup-0001-Supinfo.docx.
Funding:
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases, R01DK133465.
Footnotes
Conflicts of Interest
The authors declare no conflicts of interest.
Data Availability Statement
The data that support the findings of this study are available from Centers for Medicare & Medicaid Services (CMS). Restrictions apply to the availability of these data, which were used under license for this study. Data are available from https://data.cms.gov/ with the permission of CMS.
References
- 1.Hernandez A, Lan M, MacKinnon NJ, Branscum AJ, and Cuadros DF, “‘Know Your Epidemic, Know Your Response’: Epidemiological Assessment of the Substance Use Disorder Crisis in the United States,” PLoS One 16, no. 5 (2021): e0251502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Substance Abuse and Mental Health Services Administration, Key Substance Use and Mental Health Indicators in the United States: Results from the National Survey on Drug Use and Health, NSDUH Series H-56 (SAMSA, 2021), https://www.samhsa.gov/data/sites/default/files/reports/rpt35325/NSDUHFFRPDFWHTMLFiles2020/2020NSDUHFFR102121.htm#:~:text=In%202020%2C%2040.3%20million%20people,an%20illicit%20drug%20use%20disorder. [Google Scholar]
- 3.O’Jiaku-Okorie A, Yin X, and Lucas C, “QuickStats: Rate of Emergency Department Visits for Substance Use Disorders Among Adults Aged ≥18 Years, by Age Group — National Hospital Ambulatory Medical Care Survey, United States, 2018–2019 and 2020–2021,” Morbidity and Mortality Weekly Report (MMWR) 72, no. 39 (2023): 1073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zhang X, Wang N, Hou F, et al. , “Emergency Department Visits by Patients With Substance Use Disorder in the United States,” Western Journal of Emergency Medicine 22, no. 5 (2021): 1076–1085. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ozburn AR and Spencer SM, “Repurposing Anti-Inflammatory Medications for Alcohol and Substance Use Disorders,” Neuropsychopharmacology 49, no. 1 (2024): 317–318. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Brunchmann A, Thomsen M, and Fink-Jensen A, “The Effect of Glucagon-Like Peptide-1 (GLP-1) Receptor Agonists on Substance Use Disorder (SUD)-Related Behavioural Effects of Drugs and Alcohol: A Systematic Review,” Physiology & Behavior 206 (2019): 232–242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Shirazi RH, Dickson SL, and Skibicka KP, “Gut Peptide GLP-1 and Its Analogue, Exendin-4, Decrease Alcohol Intake and Reward,” PLoS One 8, no. 4 (2013): e61965. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Vallof D, Kalafateli AL, and Jerlhag E, “Brain Region Specific Glucagon-Like Peptide-1 Receptors Regulate Alcohol-Induced Behaviors in Rodents,” Psychoneuroendocrinology 103 (2019): 284–295. [DOI] [PubMed] [Google Scholar]
- 9.Vallof D, Vestlund J, and Jerlhag E, “Glucagon-Like Peptide-1 Receptors Within the Nucleus of the Solitary Tract Regulate Alcohol-Mediated Behaviors in Rodents,” Neuropharmacology 149 (2019): 124–132. [DOI] [PubMed] [Google Scholar]
- 10.Sorensen G, Caine SB, and Thomsen M, “Effects of the GLP-1 Agonist Exendin-4 on Intravenous Ethanol Self-Administration in Mice,” Alcoholism, Clinical and Experimental Research 40, no. 10 (2016): 2247–2252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Allingbjerg ML, Hansen SN, Secher A, and Thomsen M, “Glucagon-Like Peptide-1 Receptors in Nucleus Accumbens, Ventral Hippocampus, and Lateral Septum Reduce Alcohol Reinforcement in Mice,” Experimental and Clinical Psychopharmacology 31, no. 3 (2023): 612–620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bruns Vi N, Tressler EH, Vendruscolo LF, Leggio L, and Farokhnia M, “IUPHAR Review–Glucagon-Like Peptide-1 (GLP-1) and Substance Use Disorders: An Emerging Pharmacotherapeutic Target,” Pharmacological Research 207 (2024): 107312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Egecioglu E, Engel JA, and Jerlhag E, “The Glucagon-Like Peptide 1 Analogue, Exendin-4, Attenuates the Rewarding Properties of Psychostimulant Drugs in Mice,” PLoS One 8, no. 7 (2013): e69010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hernandez NS, Ige KY, Mietlicki-Baase EG, et al. , “Glucagon-Like Peptide-1 Receptor Activation in the Ventral Tegmental Area Attenuates Cocaine Seeking in Rats,” Neuropsychopharmacology 43, no. 10 (2018): 2000–2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hernandez NS, Weir VR, Ragnini K, et al. , “GLP-1 Receptor Signaling in the Laterodorsal Tegmental Nucleus Attenuates Cocaine Seeking by Activating GABAergic Circuits That Project to the VTA,” Molecular Psychiatry 26, no. 8 (2021): 4394–4408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Schmidt HD, Mietlicki-Baase EG, Ige KY, et al. , “Glucagon-Like Peptide-1 Receptor Activation in the Ventral Tegmental Area Decreases the Reinforcing Efficacy of Cocaine,” Neuropsychopharmacology 41, no. 7 (2016): 1917–1928. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Martinelli S, Mazzotta A, Longaroni M, and Petrucciani N, “Potential Role of Glucagon-Like Peptide-1 (GLP-1) Receptor Agonists in Substance Use Disorder: A Systematic Review of Randomized Trials,” Drug and Alcohol Dependence 264 (2024): 112424. [DOI] [PubMed] [Google Scholar]
- 18.Yammine L, Green CE, Kosten TR, et al. , “Exenatide Adjunct to Nicotine Patch Facilitates Smoking Cessation and May Reduce Post-Cessation Weight Gain: A Pilot Randomized Controlled Trial,” Nicotine & Tobacco Research 23, no. 10 (2021): 1682–1690. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Probst L, Monnerat S, Vogt DR, et al. , “Effects of Dulaglutide on Alcohol Consumption During Smoking Cessation,” JCI Insight 8, no. 22 (2023): e170419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Klausen MK, Jensen ME, Moller M, et al. , “Exenatide Once Weekly for Alcohol Use Disorder Investigated in a Randomized, Placebo-Controlled Clinical Trial,” JCI Insight 7, no. 19 (2022): e159863. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Angarita GA, Matuskey D, Pittman B, et al. , “Testing the Effects of the GLP-1 Receptor Agonist Exenatide on Cocaine Self-Administration and Subjective Responses in Humans With Cocaine Use Disorder,” Drug and Alcohol Dependence 221 (2021): 108614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lengsfeld S, Burkard T, Meienberg A, et al. , “Effect of Dulaglutide in Promoting Abstinence During Smoking Cessation: A Single-Centre, Randomized, Double-Blind, Placebo-Controlled, Parallel Group Trial,” EClinicalMedicine 57 (2023): 101865. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Farokhnia M, Tazare J, Pince CL, et al. , “Glucagon-Like Peptide-1 Receptor Agonists, but Not Dipeptidyl Peptidase-4 Inhibitors, Reduce Alcohol Intake,” Journal of Clinical Investigation 135, no. 9 (2025): e188314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Xie Y, Choi T, and Al-Aly Z, “Mapping the Effectiveness and Risks of GLP-1 Receptor Agonists,” Nature Medicine 31, no. 3 (2025): 951–962. [DOI] [PubMed] [Google Scholar]
- 25.Qeadan F, McCunn A, and Tingey B, “The Association Between Glucose-Dependent Insulinotropic Polypeptide and/or Glucagon-Like Peptide-1 Receptor Agonist Prescriptions and Substance-Related Outcomes in Patients With Opioid and Alcohol Use Disorders: A Real-World Data Analysis,” Addiction 120, no. 2 (2025): 236–250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Lahteenvuo M, Tiihonen J, Solismaa A, Tanskanen A, Mittendorfer-Rutz E, and Taipale H, “Repurposing Semaglutide and Liraglutide for Alcohol Use Disorder,” JAMA Psychiatry 82, no. 1 (2025): 94–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wium-Andersen IK, Wium-Andersen MK, Fink-Jensen A, Rungby J, Jorgensen MB, and Osler M, “Use of GLP-1 Receptor Agonists and Subsequent Risk of Alcohol-Related Events. A Nationwide Register-Based Cohort and Self-Controlled Case Series Study,” Basic & Clinical Pharmacology & Toxicology 131, no. 5 (2022): 372–379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.American Diabetes Association Professional Practice C, “13. Older Adults: Standards of Medical Care in Diabetes-2022,” Diabetes Care 45, no. S1 (2022): S195–S207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Hashemi R, Rabizadeh S, Yadegar A, et al. , “High Prevalence of Comorbidities in Older Adult Patients With Type 2 Diabetes: A Cross-Sectional Survey,” BMC Geriatrics 24, no. 1 (2024): 873. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Mangoni AA and Jackson SH, “Age-Related Changes in Pharmacokinetics and Pharmacodynamics: Basic Principles and Practical Applications,” British Journal of Clinical Pharmacology 57, no. 1 (2004): 6–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Zhang Z, Li X, Wu X, Qiu H, Shi H, and AMEB-DCTCG, “Propensity Score Analysis for Time-Dependent Exposure,” Annals of Translational Medicine 8, no. 5 (2020): 246. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Tang H, Lu Y, Donahoo WT, et al. , “Glucagon-Like Peptide-1 Receptor Agonists and Risk for Depression in Older Adults With Type 2 Diabetes: A Target Trial Emulation Study,” Annals of Internal Medicine 178, no. 3 (2025): 315–326. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Franklin JM, Rassen JA, Ackermann D, Bartels DB, and Schneeweiss S, “Metrics for Covariate Balance in Cohort Studies of Causal Effects,” Statistics in Medicine 33, no. 10 (2014): 1685–1699. [DOI] [PubMed] [Google Scholar]
- 34.Hernan MA and Robins JM, “Per-Protocol Analyses of Pragmatic Trials,” New England Journal of Medicine 377, no. 14 (2017): 1391–1398. [DOI] [PubMed] [Google Scholar]
- 35.Merchenthaler I, Lane M, and Shughrue P, “Distribution of PrePro-Glucagon and Glucagon-Like Peptide-1 Receptor Messenger RNAs in the Rat Central Nervous System,” Journal of Comparative Neurology 403, no. 2 (1999): 261–280. [DOI] [PubMed] [Google Scholar]
- 36.Sorensen G, Reddy IA, Weikop P, et al. , “The Glucagon-Like Peptide 1 (GLP-1) Receptor Agonist Exendin-4 Reduces Cocaine Self-Administration in Mice,” Physiology & Behavior 149 (2015): 262–268. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Hayes MR and Schmidt HD, “GLP-1 Influences Food and Drug Reward,” Current Opinion in Behavioral Sciences 9 (2016): 66–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Alami M, Zerif E, Khalil A, et al. , “Neuroprotective Effects of SGLT2 Inhibitors Empagliflozin and Dapagliflozin on Abeta(1–42)-Induced Neurotoxicity and Neuroinflammation in Cellular Models of Alzheimer’s Disease,” Journal of Alzheimer’s Disease 105, no. 2 (2025): 464–480. [DOI] [PubMed] [Google Scholar]
- 39.Mei J, Li Y, Niu L, et al. , “SGLT2 Inhibitors: A Novel Therapy for Cognitive Impairment via Multifaceted Effects on the Nervous System,” Translational Neurodegeneration 13, no. 1 (2024): 41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Wang W, Volkow ND, Wang Q, et al. , “Semaglutide and Opioid Overdose Risk in Patients With Type 2 Diabetes and Opioid Use Disorder,” JAMA Network Open 7, no. 9 (2024): e2435247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Zheng YJ, Soegiharto C, Au HCT, et al. , “A Systematic Review on the Role of Glucagon-Like Peptide-1 Receptor Agonists on Alcohol-Related Behaviors: Potential Therapeutic Strategy for Alcohol Use Disorder,” Acta Neuropsychiatrica 37 (2025): e51. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from Centers for Medicare & Medicaid Services (CMS). Restrictions apply to the availability of these data, which were used under license for this study. Data are available from https://data.cms.gov/ with the permission of CMS.