Highlights
-
•
Remote pharmacist-led care safely initiated SGLT2i and GLP1 RA in high-risk T2D.
-
•
AE rates matched clinical trials: 29% for SGLT2i, 55% for GLP1 RA initiators.
-
•
Only 10% discontinued due to side effects with structured remote monitoring.
-
•
No severe hypoglycemia, hospitalizations, or ED visits related to medications.
-
•
Pharmacist oversight enables safe remote prescribing of cardiometabolic meds.
ABSTRACT
Background
Sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP1 RA) reduce cardiovascular and kidney risk in patients with type 2 diabetes (T2D), yet real-world use remains suboptimal. Remote care models offer a promising approach to improve access; there is limited data on the safety of and adverse events (AEs) associated with prescribing and titrating these therapies through a protocol-driven remote management program.
Methods
The DRIVE trial was a pragmatic, randomized clinical trial evaluating the safety of a remote disease management program that initiated SGLT2i and/or GLP1 RA in patients with T2D and elevated cardiovascular or kidney risk. Participants were enrolled at a large integrated health care system in Massachusetts between March 2021 and December 2022. Participants were randomized to 1 of 2 implementation strategies: a sequential approach (two months of education first followed by medication initiation) or a bundled approach (simultaneous education and medication initiation). Nonclinical navigators and clinical pharmacists, supervised by physicians, oversaw medication initiation and monitoring under a collaborative drug therapy management agreement and protocol that included algorithms for initiation and titration of SGLT2i and GLP1 RA, as well as adjustment of concurrent glucose-lowering medications to reduce the risk of hypoglycemia Safety outcomes were collected through chart review and patient report, including targeted AEs, hospitalizations, emergency department visits, and urgent care visits. Targeted AEs were prospectively collected over the first 6 months after enrollment: for SGLT2i initiators, symptoms of genital mycotic infection, urinary tract infection, or volume depletion; for GLP 1 RA initiators, nausea, vomiting, diarrhea, and abdominal pain. Hospitalizations, emergency department visits, and urgent care visits over the first 6 months after enrollment were retrospectively collected through electronic health record review.
Results
Of 200 participants enrolled, 106 (53%) initiated SGLT2i (n = 68) or GLP1 RA (n = 40). Among SGLT2i users, 29.4% reported an AE, most commonly genital mycotic infections (10.3%) and symptoms of volume depletion (11.8%); 10.3% discontinued due to AEs. Among GLP1 RA users, 55.0% experienced AEs, predominantly gastrointestinal; 10.0% discontinued due to AEs. No severe hypoglycemia, emergency department visits, or hospitalizations occurred.
Conclusions
A remote, pharmacist- and navigator-led program safely initiated SGLT2i and GLP1 RA in a high-risk T2D population, with AE rates comparable to clinical trials and high persistence. These findings support the feasibility of remote prescribing with appropriate clinical oversight and structured AE management.
Trial Registration
Registry: ClinicalTrials.gov ; URL: https://clinicaltrials.gov/study/NCT06046560 ; unique identifier: NCT06046560.
Background
Sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP1 RA) offer significant cardiovascular (CV) and kidney protection for individuals with type 2 diabetes (T2D). , Clinical guidelines and consensus statements from cardiology, endocrinology, and nephrology consistently recommend both medication classes as guideline-directed medical therapies for patients with T2D and established CV or kidney disease, or those at elevated CV risk. ,,,,,, Despite these robust recommendations, the real-world utilization of these treatments remains inadequate. ,,,,
Multiple implementation studies have explored strategies to increase the uptake and persistence of guideline-directed medical therapies in diabetes care, similar to approaches used for other CV risk factors. Among the most promising methods are remote digital care interventions, which offer scalable solutions for enhancing medication prescribing and adherence across various chronic conditions. , However, alongside other implementation barriers, adverse events (AEs) significantly contribute to the underutilization of these medications, with high discontinuation rates observed in routine clinical practice —approximately 39.4% for SGLT2i and 50.3% for GLP1 RA within 1 year of initiation. The management of side effects associated with remotely prescribed SGLT2i and GLP1 RA may present challenges to broader implementation through remote care models.
Our team, the Accelerator for Clinical Transformation, an academic research group at Mass General Brigham (MGB), has developed a series of remote, algorithm-driven disease management programs. These programs leverage trained patient navigators, pharmacists, nurse practitioners (NPs), and physician oversight to optimize guideline-directed medical therapies for hypertension, hypercholesterolemia, and heart failure (HF) with reduced ejection fraction, demonstrating both safety and efficacy. ,,,,,,, Building on this experience, we established the Diabetes Remote Intervention to improve the use of Evidence-based medications (DRIVE) program.
The DRIVE program systematically identified patients with T2D and established CV or kidney disease or at elevated CV risk who receive longitudinal care in a large healthcare system. The program remotely facilitated initiation, dose titration, ongoing safety monitoring of SGLT2i and GLP1 RA, and management of adverse effects coordinated by pharmacists and navigators under physician supervision. ,, We previously reported the primary, secondary, and exploratory outcomes. In this report, we present AEs and safety outcomes from a prespecified analysis of participants who initiated SGLT2i or GLP1 RA within the DRIVE program.
Methods
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Study design and oversight
This pragmatic, parallel-group, open-label randomized clinical trial was conducted at MGB, a large integrated academic health system in Massachusetts. Full details of the trial design have been published, ,, with a summary of the intervention provided below.
The MGB institutional review board approved the program as posing “no more than minimal risk” and waived the requirement for written informed consent, as this was a clinical disease management program offering guideline-directed care. Participants provided verbal consent. The DRIVE trial enrolled participants between March 2021 and December 2022.
Study population
Eligible patients were adults aged 27 to 79 years with T2D and a history of atherosclerotic CV disease (ASCVD), HF, chronic kidney disease (CKD), or elevated ASCVD risk. Prospective participants were identified via electronic health record searches and clinician referrals. Eligible patients had an established relationship with the MGB healthcare system, defined as having at least 1 outpatient visit within the 12 months preceding enrollment with an MGB primary care, cardiology, or endocrinology provider. Exclusion criteria included current SGLT2i or GLP1 RA use, ongoing end-of-life care, and hemoglobin A1C <6.5% or >9.9% (to exclude patients who likely required intensive glucose-lowering therapies, such as short-acting insulin, beyond the program’s scope). Complete inclusion and exclusion criteria are provided in Table 1 .
Table 1
Inclusion and exclusion criteria.
| Inclusion criteria | Exclusion criteria |
|---|---|
|
|
Study enrollment
Established providers of eligible patients were notified about the program to enhance interdisciplinary collaboration and allow providers to exempt patients they considered inappropriate for participation. A patient navigator then contacted eligible patients to explain the program, verify eligibility, conduct medication reconciliation, and obtain verbal consent.
Study intervention
Participants were randomly assigned to 1 of 2 management strategies to evaluate the optimal implementation approach: (1) a sequential approach with a 2-month education period before initiating SGLT2i or GLP1 RA, or (2) a bundled approach with simultaneous education and medication initiation. The education component comprised of 23 brief videos accessible through an online patient portal, explaining the medical rationale for using SGLT2i and GLP1 RA, reviewing supporting evidence, and addressing common medication-related questions and side effects. Both implementation strategies utilized identical clinical protocols, medication management algorithms, and AE monitoring; they differed only in the timing and sequencing of education delivery relative to medication initiation.
Navigators served as the primary liaison between patients and the DRIVE clinical team. Navigators are college graduates who are nonlicensed but who have received training in chronic disease management and work under the direct supervision of licensed physicians, nurse practitioners, and pharmacists. Navigator training included education on type 2 diabetes, SGLT2i and GLP1 RA mechanisms and side effects, and clinical escalation protocols. Navigator responsibilities included: patient outreach and enrollment, medication reconciliation, delivering educational materials, facilitating medication orders, and conducting structured AE monitoring calls. All clinical decisions were made by supervising pharmacists or physicians, with navigators executing the approved treatment plan under direct oversight.
Medication initiation and titration were overseen by a pharmacist under a collaborative drug therapy management agreement and a program-specific protocol. The collaborative drug therapy management agreement was developed by a team of pharmacists, cardiologists, and endocrinologists and approved by the Brigham and Women’s Hospital Pharmacy and Therapeutics Committee prior to program launch and based on published society guidelines and consensus statements.
Medication selection (SGLT2i vs GLP1RA) was determined using a prespecified, algorithm-driven decision tree . The algorithm prioritized medications based on compelling indications from CV or kidney comorbidities (HF or CKD with eGFR 25-60 mL/min/1.73 m 2 favored SGLT2i; established ASCVD or elevated ASCVD risk without other compelling indications allowed either SGLT2i or GLP1 RA). Additional factors incorporated into the algorithm included baseline eGFR (SGLT2i generally avoided if eGFR <25 mL/min/1.73 m 2) and history of pancreatitis (GLP1 RA contraindicated).
In cases where either SGLT2i or GLP1RA could be appropriately used, navigators engaged in shared decision-making with participants. This discussion incorporated patient-specific factors, including coexisting conditions (eg, overweight/obesity, metabolic associated steatotic liver disease) patient preferences (eg, injection aversion, history of recurrent genital mycotic infections), and tolerability concerns. The final medication selection was reviewed and approved by the program pharmacist under the collaborative drug therapy management agreement.
For GLP1 RA initiation, dulaglutide was started at 0.75 mg weekly and semaglutide at 0.25 mg weekly, with escalation to evidence-based target doses (dulaglutide 1.5 mg weekly, semaglutide 0.5 mg weekly) after 4 weeks. GLP1 RA injection teaching was performed through providing participants with video links to company websites.
Participants graduated from DRIVE 6 months after enrollment or 1 month after initiating SGLT2i or GLP1 RA and completion of laboratory and monitoring follow-up, whichever was later. At graduation, the management of SGLT2i or GLP1 RA was transitioned back to the patients’ care team (primary care provider, cardiologist and/or endocrinologist) through a formal communication through the electronic health record system.
Laboratory monitoring, safety monitoring, and follow-up
Hemoglobin A1C and basic metabolic panel were completed within 3 months of medication initiation. To monitor for side effects, navigators called participants 3 weeks after starting an SGLT2i or GLP1 RA and again at 6 months. Patients were also advised to contact the program with any questions or new side effects. Nonurgent issues were reviewed daily by a multidisciplinary team, including a pharmacist, cardiologist, endocrinologist, and NP. Urgent concerns were escalated promptly to a team cardiologist, endocrinologist, or NP. Follow-up for side effects was determined by the DRIVE team and coordinated over telephone and the electronic patient portal, with updates provided to longitudinal primary and specialty care providers as needed. The duration for the collection of AEs and safety events was defined from enrollment until graduation or 6 months postenrollment, whichever came first.
Targeted AEs were prospectively collected during side effect calls at 3 weeks after medication initiation, at graduation, and at any patient-initiated call following medication initiation. For SGLT2i users, navigators assessed for symptoms of genital mycotic infection, urinary tract infection (UTI), or volume depletion, defined broadly as dizziness, postural dizziness, or lightheadedness. For GLP1 RA users, navigators assessed for nausea, vomiting, diarrhea, and abdominal pain.
To reduce the risk of hypoglycemia when initiating SGLT2i or GLP1 RA in participants taking glucose-lowering medications with hypoglycemia risk (ie, sulfonylureas, glinides, or basal insulin), an algorithm was followed for medication adjustments (Supplementary Tables 1 and 2). All participants on these medications received education on hypoglycemia symptoms and management and agreed to fingerstick blood glucose (FSBG) monitoring in case of hypoglycemia. DRIVE pharmacists prescribed glucometers and testing supplies as needed. Navigators assessed for hypoglycemia in these participants at each follow-up call. Hypoglycemia was defined as a FSBG <70 mg/dL, and severe hypoglycemia as a FSBG <70 mg/dL requiring assistance.
For participants initiating an SGLT2i with a baseline eGFR <45 mL/min/1.73 m 2, a repeat basic metabolic panel was ordered within 3 weeks to assess kidney function. Significant changes were defined as a ≥20% decrease in eGFR, eGFR <30 mL/min/1.73 m 2, or a ≥50% increase in serum creatinine.
Data were prospectively collected for medication discontinuation due to AEs or cost, and death. Any death prompted a clinician review. Medication discontinuation was assessed through patient self-report during structured follow-up calls at 3 weeks after medication initiation and at graduation, as well as through review of medication discontinuation orders in the electronic health record. Participants were asked whether they had stopped taking SGLT2i or GLP1 RA and, if so, the reason for discontinuation (AEs, cost, patient preference, provider recommendation, or other). Discontinuation due to AE was defined as patient-reported medication cessation attributed to intolerance or side effects.
Hospitalizations, emergency department (ED) visits, and urgent care visits within the MGB system during the study period were retrospectively assessed by 2 physicians (LC and SH), with disagreements resolved by consensus.
Statistical analysis
For this safety analysis, we performed a prespecified analysis to identify treatment-related safety events that occurred within 6 months of enrollment. We report the prevalence of safety events by medication class and describe specific serious events at the patient level.
Educational engagement was measured as the number of unique videos and total videos watched (including repeat views) within 6 months of enrollment. The Mann-Whitney U test (Wilcoxon rank-sum test) was used to compare educational engagement between the simultaneous and education-first arms.
Results
Baseline characteristics
Within 6 months of enrollment, 53% (106/200) of participants initiated an SGLT2i or GLP1 RA ( Figure 1 ). The mean (standard deviation [SD]) age of these participants was 65.7 (8.2) years, with a mean (SD) body mass index of 31.8 (6.8) kg/m 2 and a mean (SD) baseline hemoglobin A1C of 7.3 (0.9) %. Of these participants, 38.7% were female and 18.9% were non-White ( Table 2 ). Treatment indications included high ASCVD risk (60.4%), established ASCVD (36.8%), HF (9.4%), and CKD (21.7%). At baseline, 93.4% were on metformin, 23.6% on sulfonylureas, 8.5% on DPP4 inhibitors, 5.7% on basal insulin and none on glinides.
Participants who took SGLT2i or GLP1 RA within 6 months of enrollment (by medication class). *n = 2 who took SGLT2i and GLP1 RA (simultaneously or successively) within 6 months of enrollment; Ɨ A single participant took semaglutide and did not tolerate it due to side effects and was switched to dulaglutide and tolerated it.
Table 2
Baseline characteristics of participants.
| Participants taking SGLT2i (n = 68) | Participants taking GLP1 RA (n = 40) | Participants taking either SGLT2i or GLP1 RA (n = 106) | |
|---|---|---|---|
| Age, years, mean (SD) | 67.1 (8.1) | 63.6 (8.1) | 65.7 (8.2) |
| Female sex, n (%) | 23 (33.8) | 18 (45.0) | 41 (38.7) |
| Race and ethnicity, n (%) | |||
| White | 57 (83.8) | 31 (77.5) | 86 (81.1) |
| Black | 3 (4.4) | 5 (12.5) | 8 (7.5) |
| Asian | 4 (5.9) | 2 (5.0) | 6 (5.7) |
| Other | 2 (2.9) | 1 (2.5) | 3 (2.8) |
| Unknown | 0 (0) | 1 (2.5) | 1 (0.9) |
| Declined to respond | 2 (2.9) | 0 (0) | 2 (1.9) |
| BMI, kg/m 2, mean (SD) | 31.6 (7.0) | 32.6 (6.4) | 31.8 (6.8) |
| Systolic blood pressure, mmHg, mean (SD) | 128.7 (10.6) | 131.9 (10.8) | 129.9 (10.8) |
| Diastolic blood pressure, mmHg, mean (SD) | 73.0 (7.4) | 75.2 (6.0) | 73.8 (7.0) |
| Hemoglobin A1C, %, mean (SD) | 7.3 (0.8) | 7.4 (0.9) | 7.3 (0.9) |
| eGFR, mL/min/1.73 m 2, mean (SD) | 73.6 (19.3) | 84.8 (16.6) | 77.9 (19.2) |
| Urine albumin-to-creatinine ratio, mean (SD) | 54.0 (115.1) | 17.0 (24.4) | 40.3 (95.1) |
| Indication, n (%) | |||
| High ASCVD risk | 38 (55.9) | 26 (65.0) | 64 (60.4) |
| ASCVD | 27 (39.7) | 14 (35.0) | 39 (36.8) |
| Heart failure | 9 (13.2) | 2 (5.0) | 10 (9.4) |
| Chronic kidney disease * | 20 (29.4) | 4 (10.0) | 23 (21.7) |
| Diabetes medication, n (%) | |||
| Metformin | 64 (94.1) | 37 (92.5) | 99 (93.4) |
| Sulfonylurea | 21 (30.9) | 4 (10.0) | 25 (23.6) |
| Glinide | 0 (0) | 0 (0) | 0 (0) |
| DPP4 inhibitor | 6 (8.8) | 3 (7.5) | 9 (8.5) |
| Thiazolidinedione | 0 (0) | 0 (0) | 0 (0) |
| Insulin | 3 (4.4) | 3 (7.5) | 6 (5.7) |
| Other | 0 (0) | 0 (0) | 0 (0) |
SGLT2i or GLP1 RA initiation within 6 months after enrollment
Within 6 months of enrollment, 34% (68/200) of participants started an SGLT2i, and 20% (40/200) started a GLP1 RA, including 2 participants who started both medication classes. Of these, 1 participant started an SGLT2i through the DRIVE program but discontinued it due to increased urination and was subsequently started on a GLP1 RA through the DRIVE program. The other participant started an SGLT2i through DRIVE and later had a GLP1 RA added by his endocrinologist.
Among 68 participants who started an SGLT2i, 62 (91.2%) started empagliflozin, and 6 (8.8%) started dapagliflozin ( Figure 1 ). Among 40 participants who started a GLP1 RA, 32 (80.0%) started dulaglutide, and 8 (20.0%) started semaglutide. One participant switched from semaglutide to dulaglutide due to intolerance ( Figure 1 ).
A greater proportion of participants in the simultaneous education and medication arm started an SGLT2i or GLP1 RA within 6 months of enrollment (59.5%) compared to those randomized to education-first, followed by medication initiation (44.0%) ( Figure 2 ).
Participants who took SGLT2i or GLP1 RA within 6 months of enrollment (by randomization arm).
Changes in hemoglobin A1C and weight
We have previously reported other secondary, exploratory, and post hoc outcomes. Briefly, baseline hemoglobin A1C was 7.3% in the simultaneous arm and 7.2% in the education-first arm, with mean changes of −0.3% in both groups at 6 months. Baseline weight was 93.7 kg in the simultaneous arm and 92.7 kg in the education-first arm, with mean changes at 6 months of −4.2 kg and −2.5 kg in each arm, respectively.
SGLT2i
Among the 68 participants who started an SGLT2i within 6 months of enrollment, 21 (29.4%) experienced an AE potentially related to the SGLT2i (excluding death) during the study period ( Table 3 ). Genital mycotic infections occurred in 10.3% (7/68) of participants, including 6 females and 1 male. The male participant developed a single episode of balanitis; after re-education on genital hygiene, he continued the SGLT2i without recurrence. Two participants (2.8%) developed a UTI—one male, who was treated with antibiotic therapy and continued the SGLT2i, and 1 female, participant discontinued after developing a vaginal ulcer postantibiotic treatment. Symptoms consistent with volume depletion were reported by 11.8% (8/68) of SGLT2i users. These were transient in all cases and did not lead to treatment discontinuation.
Table 3
Adverse events and discontinuation rates among SGLT2i and GLP1 RA initiators (patient-level).
| Education followed by medication initiation * | Simultaneous education and medication initiation * | Total * | |
|---|---|---|---|
| Participants taking SGLT2i | 23 | 45 | 68 |
| Any AE (not including death) | 2 (8.7) | 19 (42.2) | 21 (29.4) |
| Genitourinary AE | 0 (0) | 8 (17.8) | 8 (11.8) |
| Genital mycotic infection | 0 (0) | 6 (13.3) | 6 (8.8) |
| UTI | 0 (0) | 2 (4.4) | 2 (2.9) |
| Vaginal ulcer | 0 (0) | 1 (2.2) | 1 (1.5) |
| Symptoms consistent with possible volume depletion | 1 (4.3) | 7 (15.6) | 8 (11.8) |
| Hypoglycemia (<70 mg/dL) | 0 (0) | 2 (4.4) | 2 (2.9) |
| Severe hypoglycemia | 0 (0) | 0 (0) | 0 (0) |
| Discontinuation (not including due to death) | 1 (4.3) | 8 (17.8) | 9 (13.2) |
| Discontinuation due to AE | 1 (4.3) | 6 (13.3) | 7 (10.3) |
| Genital mycotic infection | 0 (0) | 3 (6.7) | 3 (4.4) |
| Urinary frequency | 1 (4.3) | 1 (2.2) | 2 (2.9) |
| Vaginal ulcer | 0 (0) | 1 (2.2) | 1 (1.5) |
| Other † | 0 (0) | 1 (2.2) | 1 (1.5) |
| Discontinuation due to cost | 0 (0) | 2 (4.4) | 2 (2.9) |
| AE associated with hospitalization, ED visit, or urgent care visit | 0 (0) | 1 (2.2) | 1 (1.5) |
| Death | 0 (0) | 1 (2.2) | 1 (1.5) |
| Participants taking GLP1 RA | 15 | 25 | 40 |
| Any AE (not including death) | 8 (53.3) | 14 (56.0) | 22 (55.0) |
| Gastrointestinal AE | 8 (53.3) | 11 (44.0) | 19 (47.5) |
| Nausea | 6 (40.0) | 7 (28.0) | 13 (32.5) |
| Vomiting | 1 (6.7) | 2 (8.0) | 3 (7.5) |
| Diarrhea | 3 (20.0) | 3 (12.0) | 6 (15.0) |
| Abdominal pain | 2 (13.3) | 4 (16.0) | 6 (15.0) |
| Symptoms consistent with possible volume depletion | 1 (6.7) | 3 (12.0) | 4 (10.0) |
| Hypoglycemia (<70 mg/dL) | 0 (0) | 0 (0) | 0 (0) |
| Severe hypoglycemia | 0 (0) | 0 (0) | 0 (0) |
| Discontinuation | 2 (13.3) | 5 (20.0) | 7 (17.5) |
| Discontinuation due to AE | 1 (6.7) | 3 (12.0) | 4 (10.0) |
| Nausea | 1 (6.7) | 1 (4.0) | 2 (5.0) |
| Vomiting | 0 (0) | 1 (4.0) | 1 (2.5) |
| Other ‡ | 0 (0) | 1 (4.0) | 1 (2.5) |
| Discontinuation due to cost | 1 (6.7) | 2 (8.0) | 3 (7.5) |
| Did not reach or maintain target dose (includes those who had to discontinue GLP1 RA as counted above) | 4 (26.7) | 7 (28.0) | 11 (27.5) |
| Nausea | 1 (6.7) | 2 (8.0) | 3 (7.5) |
| Vomiting | 1 (6.7) | 2 (8.0) | 3 (7.5) |
| Other § | 0 (0) | 1 (4.0) | 1 (2.5) |
| Cost | 2 (13.3) | 2 (8.0) | 4 (10.0) |
| AE associated with hospitalization, ED visit, or urgent care visit | 0 (0) | 1 (4.0) | 1 (2.5) |
| Death | 0 (0) | 0 (0) | 0 (0) |
| Participants taking either SGLT2i or GLP1 RA | 37 | 69 | 106 |
| AE that led to SGLT2i or GLP1 RA discontinuation | 2 (5.4) | 9 (13.0) | 11 (10.4) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
