Advertise - Write For Us

A few years ago, Ozempic was a diabetes injection most people had never heard of. Today it, its sister drug Wegovy and Eli Lilly's Mounjaro are household names, now joined by pill versions. They belong to a class called GLP-1 drugs, and they have changed the treatment of obesity more than anything in a generation. In a large trial of overweight adults with existing cardiovascular disease, semaglutide, the drug in Ozempic and Wegovy, cut major heart attacks and strokes by about 20% (SELECT trial, NEJM).

That matters to every actuary, whatever their practice area. Obesity sits behind heart disease, diabetes, strokes and several cancers. So a drug that treats it could change how long people live, how often they claim, and what their medical care costs. It touches the mortality assumption in a life insurer's pricing, the medical trend in a health plan, the claim rates in disability cover, and the life expectancy in every pension and annuity valuation.

In June 2025 this newsletter asked what actuaries should rethink about these drugs (MyActuary Weekly). Seventeen months on, we have real data. It says the drugs work. It also says something that matters more for our numbers: most people stop taking them.

Prime Therapeutics, which manages prescription benefits for millions of Americans, followed people without diabetes who started a GLP-1 drug for weight loss. After 3 years, only 8% were still taking it, roughly 1 in 12. Even among people who started on Wegovy, the best performer, only about 1 in 7 were still on it (Prime Therapeutics; Becker's).

People stop for ordinary reasons: the cost, the side effects, supply shortages, or simply reaching a target weight. And when they stop, the benefit fades quickly. Researchers at the University of Oxford pooled the evidence and found that people coming off semaglutide or tirzepatide regained about 0.8 kg a month. On average they were projected to be back at their starting weight in about 18 months, with blood pressure, blood sugar and cholesterol back where they began in about 17 months (University of Oxford).

So the real question for an actuary is how many people are still taking them a year, 2 years and 3 years from now. Actuaries call this persistence, and it decides almost every number that follows.

The world's largest reinsurers, the companies that insure insurers and therefore study mortality more closely than almost anyone, have all published estimates. All of them already allow for people stopping, and all of them land on modest numbers.

Munich Re studied 41 million Americans and found lower death rates among GLP-1 users, both with and without diabetes. It estimates the drugs could improve mortality by 0.2% to 0.5% a year over the next 20 years (Munich Re). RGA's central estimate is that US deaths will be 3.5% lower by 2045 than they would otherwise have been, with 2.6% in Canada, 2.0% in the UK and 1.4% in Hong Kong (RGA). Swiss Re's central case is 4.0% lower in the US and 3.2% in the UK (Swiss Re).

Put on the same footing (my arithmetic: a 3.5% to 4% reduction spread over 20 years), RGA's and Swiss Re's figures work out at roughly 0.2% a year, the bottom of Munich Re's range. Real, but small.

And those are figures for the whole population. Insured people tend to be slimmer and wealthier, so they have less to gain, which RGA expects to show up as a smaller effect. When Swiss Re looked at US life insurance applicants, people taking GLP-1s actually scored as higher risk, about 1.5 times average, because the drug is first of all a sign of the condition it treats. Those who took it regularly scored better than those who didn't; simply having been on it for longer made no difference (Swiss Re).

Newsletter continues after job posts…

Persistence figures need careful reading. The Prime group who managed only 8% started treatment in 2021 and early 2022, when the drugs were often out of stock. In the same study, the share of new Wegovy users still on it after one year nearly doubled, from 33% for 2021 starters to 63% for 2024 starters, as supply recovered. Today's patients may well do better. But nobody has yet shown a recent group staying on the drug for 3 years in large numbers, and until someone does, the 18-month bounce-back is the safer assumption.

Life insurance. My view for 2027 is to give no credit for GLP-1s in the main mortality assumption, but to show in every actuarial report what would happen if the reinsurers' central estimates come true. At underwriting, look at how long and how regularly an applicant has filled the prescription, and at their weight history, not just today's weight, which may last only as long as the prescription does.

Health insurance. Here the drugs mostly add cost so far. The 2026 Milliman Medical Index found employer health costs per person up 7.9%, the biggest rise in over a decade, with prescription drug costs up 14.8% and GLP-1s a leading reason (Milliman). Prime found that 3 years in, total medical costs for GLP-1 users were still $4,490 a year higher than for similar people not taking them, with no sign of the savings everyone hoped for (Prime Therapeutics). Stopping makes this worse. Someone who quits after 6 months has cost the full price of the drug and kept none of the health benefit. If savings ever arrive, they will land on whoever insures that person years later.

Disability and critical illness. RGA expects the effect on illness claims to be smaller than on deaths. Better blood sugar or blood pressure may delay a claim more than prevent it, which changes when claims arrive and therefore how much must be reserved.

Property and casualty. There is no claims evidence yet that would justify changing workers' compensation, liability or motor assumptions. If an effect arrives, workers' compensation is where it should show first, because obesity is already linked to longer claim durations and higher medical cost per claim, followed by bodily injury severity on motor and liability lines. Liability actuaries also have an exposure of their own: the drugs themselves are the subject of consolidated product liability litigation in the US federal courts (MDL 3094). Two things are worth doing now: record GLP-1 use in claim files where the medical record allows it, and watch the comorbidity mix of injured workers for shifts in obesity-related conditions, so that any effect can be measured when it appears.

Pensions and annuities. The UK's standard mortality model for pensions, updated in March 2026, raised life expectancy at 65 by about 8 weeks for men and 6 weeks for women because deaths in 2025 hit a record low (IFoA). RGA expects the drugs to cut deaths at ages 85 and over by only 1.8% over 20 years, against 5.1% at ages 45 to 59 (The Actuary). For annuities, the effect is small unless older people stay on the drugs far longer than anyone has yet seen.

Whether people persist depends heavily on who pays and how much. That varies hugely by country, which is why the same science can lead to different assumptions in different markets.

Read across the rows. In the US, prices are halving and a government program pays for older people until the end of 2027, so persistence will follow policy deadlines. In India, a month of generic treatment costs a small fraction of the US price, so cost largely stops being a reason to quit. In the UK, private buyers far outnumber NHS patients, so those who stay on the drug will tend to be those who can keep paying. In the Gulf, the regulator sets the price and the patient pays it, so persistence will track household income more than policy. A life insurer in Mumbai and one in Ohio could reasonably reach different answers from the same research.

Give no credit yet in the main assumption. Add a clear persistence sensitivity to every life, health and annuity report, stating how many people you assume are still on the drug at 1, 2 and 3 years and what happens when they stop. And name the 2 events that could change the view: Lilly's large trial of tirzepatide's effect on heart outcomes in people with obesity, expected to report around late 2027 (ClinicalTrials.gov), and the US price cut on January 1, which will show whether cost was really what made Americans stop.

These drugs may yet change how long people live. If they do, it will be through the people who keep taking them, and for now that is a small group.

If you would like help building a sensible GLP-1 persistence assumption into your 2027 review, for mortality, illness or medical costs, that is a conversation I am having often this year.

Syed Raza, FSA, is an appointed actuary working with life insurers, fraternal benefit societies, and preneed and final expense writers.