Prime's 8.1% three-year GLP-1 persistence figure comes from patients who started during shortages

In headcount, the 8.1% who persisted all three years ^[3] is about 468 of the 5,780 members followed ^[13], drawn from a pool of 16.5 million commercially insured Prime members ^[1] and already filtered to adults without diabetes who stayed continuously enrolled ^[2]. The average proportion of days covered was 37.5% ^[5], and if that is measured over the full 1,095-day window, it comes to about 411 days of drug per member in three years ^[14]. More than one-third of users switched products during the period ^[6], so a drugmaker's hold on its own patients and the class's hold on patients are different numbers. The published account does not include prices, spend, or whether a switch counts as persistence, so any revenue figure built on it is our extrapolation.

The earlier Prime studies sketch the shape of the loss. Divide the 15% who persisted for two years ^[8] by the 32% at one year ^[7] and 47% of one-year persisters reached a second year ^[15]; divide the 8.1% by the 15% and 54% of two-year persisters reached a third ^[16]. Attrition slows a little in year three, if the figures are comparable. They may not be: these are separate studies, and the 8.1% cohort was selected for three years of continuous enrollment ^[2].

We think the 8.1% overstates how fast patients starting now will leave (or rather, it does if supply was the driver, as Gleason suggests). Its members began therapy between January 2021 and March 2022 ^[2]. Prime's Patrick Gleason, an assistant vice president of health outcomes ^[12], said of the period: "During this time, there was a lot going on in the marketplace, with challenges to obtaining products and shortages," ^[21]. He suggested that shortages were largely resolved in 2024 and likely explain the better numbers in the year-by-year series ^[18]. On prescribers, he said: "This data was from claims data, so I can't quantify this, but I suspect that prescribers of these products have become better at managing their patients and understanding the side effects." ^[20] Prime did not assess whether anti-emetics were being prescribed more often ^[19].

The counter-thesis is that supply explained the first year and little after it. Apply the 47% second-year ratio from the earlier studies to the 62.7% one-year rate for 2024 starters ^[9] and two-year persistence would be about 29%, roughly double the 15% ^[24]. That is a cross-study extrapolation, and nothing in Prime's data measures it. Three outcomes are plausible: better care carries through and two-year persistence beats 29%; the old year-two ratio holds and it lands near 29%; or the first-year gain washes out and it lands near 15%. We are wrong if 2024 starters' two-year figure lands near 15%.