One reasonable hope when starting an expensive new drug is that it replaces something you are already paying for. This study tracked what actually happened to the rest of the regimen[1]. The version of that question involving insulin specifically is in what swapping mealtime insulin actually buys.
Mostly, nothing came off. Among 3,660 patients with at least one baseline non-GLP-1 glucose lowering drug, the recorded count went up for 44.0%, stayed the same for 38.8%, and went down for 17.2%. The median count rose from one substance to two, though the median change within any individual patient was zero.
The spread between drugs is the most interesting number and the least explained. Higher counts followed dulaglutide in 52.9% of patients and tirzepatide in 20.4% — a difference of more than two and a half times. Potency is the obvious candidate, since a more effective drug should require less help, but era is another: dulaglutide users started earlier, when prescribing habits and available alternatives were different. Nothing in the abstract separates the two.
For a reader budgeting, the practical reading is that adding one of these drugs is more likely to add cost than to move it. Fewer than one in five patients ended up on fewer medications, and the modal outcome was more. That belongs alongside the price picture in what a GLP-1 actually costs and the gap between advertised and billed amounts in the headline against the bill.
The setting is the same Polish private network behind the PCOS prescribing trends, which is worth knowing: two findings from one data source are not two independent observations, and neither describes a US cash market.