A Nature Medicine paper published April 28 by the Johns Hopkins Bloomberg School of Public Health analyzed five years of electronic health records for roughly 175,000 patients with type 1 diabetes who received GLP-1 receptor agonists, and found a 15 percent reduction in major adverse cardiovascular events, a 19 percent reduction in progression to end-stage kidney disease, an 18 percent reduction in hypoglycemia hospitalization, and a 17 percent reduction in diabetic ketoacidosis hospitalization. [1] [2] The paper does not speak in the language of indication. It speaks in the language of outcomes. The outcomes are large and consistent. This is the paper that converts off-label type 1 prescribing from a clinical-judgment call into an actuarial position.

The paper's May 1 piece on GLP-1 class effects now including type 1 diabetes cardiometabolic benefits named the finding and treated it as a class-effect signal — the kidney-cardiovascular benefit pattern semaglutide established for type 2 diabetes appears to extend, on EHR evidence, to type 1. Today's reading carries that further. The payer side of the prescribing equation has had a defensible reason to deny coverage for off-label type 1 use, and that reason was actuarial: insufficient evidence of net benefit at the population level. The Hopkins paper removes that argument.

The methodology is worth describing because the methodology is the actuarial question. The Hopkins group, led by Dr. Yusi Liu and Dr. Jodi Segal, used the OptumLabs Data Warehouse and a propensity-score-matched cohort design — the gold-standard method for observational comparative-effectiveness research when randomized trial data is unavailable. The 175,000-patient cohort was matched on baseline cardiovascular risk, kidney function, diabetes duration, age, sex, and insulin regimen. Five-year outcomes were tracked across patients on GLP-1 agonists (semaglutide, liraglutide, dulaglutide, tirzepatide) versus matched controls on insulin alone. [1]

What the methodology does not provide is causal certainty. EHR studies cannot replace randomized trials. The Hopkins authors are explicit about this; the paper's discussion section names indication bias and survivorship bias as the two unresolved confounders. But the consistency of the effect — across four different GLP-1 molecules, across the cardiovascular and renal endpoints, with a hypoglycemia-reduction finding that runs counter to the prior fear that GLP-1 use in type 1 would worsen hypoglycemia risk — is what the actuarial side of payer policy has been waiting for. [2] [3]

The hypoglycemia finding is particularly important. The pre-paper objection to GLP-1 use in type 1 was that the drug class's appetite-suppressant and gastric-emptying effects would, in a population dependent on exogenous insulin, increase the risk of severe hypoglycemia events. The Hopkins data show the opposite: hypoglycemia hospitalization is down 18 percent in the treated cohort. The mechanism is plausible — better glycemic stability, smaller post-meal insulin requirements, less weight-driven insulin resistance — but the Hopkins paper is the first large-cohort study to put a number on it. [1]

The DKA finding is structurally similar. The pre-paper objection was that GLP-1 use in type 1 might increase ketoacidosis risk by allowing patients to under-dose insulin. The data show a 17 percent reduction. The likely mechanism is similar to the hypoglycemia one: better glycemic stability reduces both excursions. [1]

Off-label prescribing of GLP-1 receptor agonists for type 1 diabetes has been growing rapidly. Endocrinology surveys in 2025 found that approximately 14 percent of US type 1 patients reported having received a prescription for a GLP-1 agonist, mostly through clinician judgment in patients with concurrent obesity or cardiovascular risk. Payer coverage has been inconsistent — some plans require prior authorization, some deny coverage outright, some cover under a weight-management indication, almost none cover under a glycemic-management indication for type 1. [4] The Hopkins paper changes the actuarial input to those decisions.

The American Diabetes Association's Standards of Care, updated annually in January, did not include a GLP-1 recommendation for type 1 in the 2026 edition. The next edition, customarily previewed in November, can now cite the Hopkins paper. [5] The American Society of Nephrology and the American Heart Association have already amplified the renal and cardiovascular findings on professional channels. ASN's account on X called the ESRD reduction "the kind of effect size we don't see often in observational data."

The FDA dimension is more cautious. Novo Nordisk's semaglutide has a kidney-cardiovascular indication for type 2 diabetes; tirzepatide does not yet have a kidney indication for any population; no GLP-1 has a type 1 indication. A formal label-extension trial — randomized, controlled, statistically powered for hard endpoints — would take three to five years even on accelerated review. [6] The Hopkins paper does not deliver a label extension. It delivers something the payer side has not had: a large-cohort actuarial argument for coverage in the absence of one.

The economic consequence is the part the trade press has not yet read. If commercial payers move from prior-authorization denial to coverage under utilization-management protocols, the addressable population for GLP-1 use in type 1 expands by roughly 1.6 million US patients. Even at 30 percent uptake, that is approximately 480,000 additional patients on a class of drugs whose annual cost is roughly $13,000 at list price. The pharmacy-benefit-manager modeling that priced type 2 GLP-1 utilization is now stale on the type 1 side. [7]

Two questions remain open. The first is whether Eli Lilly or Novo Nordisk pursues a formal label-extension trial to lock in the indication, or whether both manufacturers prefer the current off-label market dynamic in which utilization grows without the regulatory cost of a trial. Lilly's earnings call April 30 did not name a type 1 trial in the development pipeline. The second is whether NICE, in the United Kingdom, treats the Hopkins evidence as informative for the cost-effectiveness threshold that has rejected three Alzheimer's drug submissions in the same window. The Hopkins paper produces effect sizes large enough that the threshold question becomes a different question. [8]

What the paper has documented is a familiar structural pattern in modern drug class evolution. A GLP-1 class begins as obesity therapy, acquires diabetes indications, acquires cardiovascular outcomes, acquires kidney outcomes, and acquires off-label populations whose evidence base catches up to the prescribing reality. The Hopkins paper is the catch-up artifact for type 1. Payer policy is the next domino. The label is the one after that.

For now, off-label is the wrong word. The Hopkins data is the institutional cover the off-label use was waiting for. The clinical decision the endocrinologist makes Monday morning has a peer-reviewed citation behind it. That is what changed this week.

-- NORA WHITFIELD, Chicago