Skip to main content

Two Second-Line Diabetes Drugs Tied to Higher Heart Disease Risk

December 26, 2018

By Marilynn Larkin

NEW YORK (Reuters Health) - Basal insulin and sulfonylureas, commonly prescribed to treat type-2 diabetes after metformin, are associated with a higher risk of heart disease compared to newer second-line medications, researchers say.

"We think our study should be considered practice-changing," Dr. Matthew O'Brien of Northwestern University Feinberg School of Medicine in Chicago told Reuters Health by email. "Most providers currently focus medication treatment decisions on lowering blood sugar. Our findings suggest that providers should also consider cardiovascular effects of the medications (and) that we are harming many patients with diabetes by prescribing these two medication classes so commonly."

Dr. O'Brien and colleagues analyzed data from 2011-2015 on 132,737 insured U.S. adults who received a second-line antidiabetic medication after taking either metformin alone or no drug. Fifty-eight percent of participants were ages 45-64; 55% were men; and 63%, white.

Participants received either dipeptidyl peptidase 4 (DPP-4) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, sodium-glucose cotransporter 2 (SGLT-2) inhibitors, thiazolidinediones (TZDs), basal insulin, or sulfonylureas (including meglitinides).

DPP-4 inhibitors were the comparator group in all analyses.

As reported online December 21 in JAMA Network Open, 3,480 patients had a first cardiovascular event during 169,384 person-years of follow-up and were censored. Patients were also censored after insurance discontinuation (86,411), transition from International Classification of Diseases, Ninth Revision (ICD-9) to ICD-10 (7,697) or two years of follow-up (35,116) without an event.

After adjustment for patient, prescriber, and health plan characteristics, the risk of composite cardiovascular events after starting GLP-1 receptor agonists was lower than for DPP-4 inhibitors (hazard ratio, 0.78); however, this finding was not significant in all sensitivity analyses.

Cardiovascular event rates after starting treatment with SGLT-2 inhibitors (HR, 0.81) and TZDs (HR, 0.92) were not statistically different from DPP-4 inhibitors.

By contrast, the comparative risk of cardiovascular events was higher after starting treatment with sulfonylureas (HR, 1.36) or basal insulin (HR, 2.03).

Summing up, the authors state, "short-term cardiovascular outcomes of GLP-1 receptor agonists, SGLT-2 inhibitors, and DPP-4 inhibitors were similar. Higher cardiovascular risk was associated with use of sulfonylureas or basal insulin compared with newer antidiabetic medication classes. Therefore, clinicians may consider prescribing GLP-1 receptor agonists, DPP-4 inhibitors, or SGLT-2 inhibitors more routinely after metformin rather than sulfonylureas or basal insulin."

"There are likely individual factors that influence (patient) response to second-line therapy," Dr. O'Brien acknowledged. "We adjusted for all of them that we could observe in the available data (and) conducted a number of additional analyses that help account for whether unobserved patient factors may have contributed to the findings."

"However, based on our experience as clinicians, we do believe that patients who receive basal insulin may be different from those who receive the other medication classes," he said. "For example, insulin is more often prescribed in the hospital when patients are admitted for high blood sugar. Therefore . . . unobserved factors may have influenced our findings on basal insulin."

Dr. Nigam Shah of Stanford University in California, coauthor of a related editorial, commented to Reuters Health on the study's "ability to produce evidence at scale."

"There is increasing interest in the ability to use observational health data to answer questions of clinical interest," he wrote in an email. "This study uses several good practices that we as a community should adopt, which are also summarized in the editorial."

"For example," he said, "this study performs sensitivity analyses, after which the seemingly lower risk of composite cardiovascular events with GLP-1 receptor agonists was no longer significant."

"As a community, via the Observational Health Data Science and Informatics collaborative ( and studies such as (this one), we have advanced our understanding of essential properties of rigorous and meaningful observational studies," Dr. Shah concluded.


JAMA Netw Open 2018.

(c) Copyright Thomson Reuters 2018. Click For Restrictions -

For more Pharmacy Learning Network articles, visit the homepage

Back to Top