Do other diabetes drugs stop on a GLP-1?

When an adult with type 2 diabetes starts tirzepatide or a GLP-1 receptor agonist such as semaglutide, a reasonable question follows: does anything else come off the list? A retrospective study of 3,660 Polish outpatients, published in Diabetes Research and Clinical Practice on 10 August 2026, is the closest thing to an answer we have seen for this class. In this cohort, the count of other glucose-lowering drugs on record went down for 17.2% of patients and up for 44.0%^[1].

What the study actually counted

The researchers used anonymised electronic health records from a large Polish private outpatient network. They included adults coded for type 2 diabetes (ICD-10 E11 without E10) who started semaglutide, liraglutide, dulaglutide or tirzepatide, and who already had at least one non-GLP-1 glucose-lowering active substance on record^[1].

Using medication-specific dates, each active substance was classified as baseline (recorded before the index date) or subsequently recorded (on or after it). The outcome was not HbA1c, and not weight. It was a count: how many distinct glucose-lowering substances appeared in the record before the incretin drug arrived, and how many appeared after. Each patient was then scored as a lower count, no change, or a higher count^[1].

The numbers

Among the 3,660 patients (median age 60 years, 53.9% women), the median recorded count rose from 1 (interquartile range 1 to 2) to 2 (1 to 3). The median within-patient change, however, was 0 (0 to 1)^[1].

Split three ways: 17.2% of patients ended with a lower count, 38.8% with no change, and 44.0% with a higher count. The pattern varied by which drug was started. A higher count was most frequent among people starting dulaglutide (52.9%) and least frequent among those starting tirzepatide (20.4%)^[1].

So the headline is not deprescribing. In this cohort, adding an incretin drug was followed more often by a longer medication record than by a shorter one, and the authors note that even the increase applied to fewer than half of patients. The abstract does not report how many people started each of the four drugs, so the per-drug percentages are best read as directional rather than as a ranking.

A bigger count is not the same as more medicine

The authors are unusually blunt about the limits of their own design. Their findings, they write, "describe prescription-record patterns and do not establish concomitant exposure, treatment intensity, or a causal effect"^[1].

Three gaps matter for anyone trying to map this onto their own situation. A record shows that a substance was written down, not that it was taken, and not at what dose. A substance recorded after the index date may have replaced an earlier one rather than joined it. And a count treats a low-dose oral agent and a full basal insulin regimen as one item each, so the number can move without treatment intensity moving at all. A records-based study sees what was entered. It does not see why.

What the labels do say about combining

None of this is a reason to change anything on your own, and the product labels are where the caution actually lives. The US prescribing information for Mounjaro (tirzepatide) states that patients receiving it "in combination with an insulin secretagogue (e.g., sulfonylurea) or insulin may have an increased risk of hypoglycemia, including severe hypoglycemia", and that "the risk of hypoglycemia may be lowered by a reduction in the dose of sulfonylurea (or other concomitantly administered insulin secretagogue) or insulin"^[2]. The Ozempic (semaglutide) label carries the same two statements^[3].

Read those sentences carefully, because they are narrower than they first look. They are addressed to the prescriber. They describe something to consider at initiation. And they name two specific groups, insulin and insulin secretagogues, rather than the whole background regimen. Metformin is not in that sentence. Neither is an SGLT2 inhibitor. That is one plausible reason the Polish records showed reduction in a minority of patients: for a lot of people, there was nothing in the label pointing at the drugs they were already on.

Where this leaves a reader

If you are on an incretin drug plus something else, the useful thing here is a question to raise at your next appointment, not an action to take at home. Lowering or stopping insulin, a sulfonylurea, metformin or anything else without your prescriber carries real risk in both directions, and this study gives no basis for it. It counted what appeared in Polish outpatient records over a defined window. It did not test whether any of those changes helped anyone^[1].

A separate real-world analysis published in Pharmaceuticals in 2022 points the same direction from a different angle. Across 358 adults starting liraglutide or dulaglutide, the mean total daily insulin dose among insulin-treated patients was essentially flat over 12 months (35.0 IU at baseline versus 35.7 IU at 12 months, p = 0.6)^[4]. Adding a GLP-1 drug did not, on average, shrink the insulin sitting behind it.

For what each of these drugs is licensed to do and where, see our regulation sections for tirzepatide and semaglutide. This article is educational journalism, not medical advice. Decisions about any part of a diabetes regimen belong with a healthcare provider who knows your history.