What happens to the heart when people stop GLP-1 drugs?
The short answer: nothing good.
New from our team: a study of 330,000+ people in @BMJMedicine
🧵
About 1 in 8 U.S. adults has taken a GLP-1 drug. But 36–81% of users stop within the first year. Most studies have focused on weight regain after stopping. We focused on what happens to the heart.
We followed 333,000+ adults with type 2 diabetes for 3 years, emulating a randomized trial across 16 treatment scenarios with different durations of use, discontinuation, and interruption.
The core finding: Cardiovascular protection builds slowly. It took 3 years of continuous GLP-1 use to achieve an 18% reduction in heart attack, stroke, and death.
But that protection erodes fast.
Stopping for 6 months: 4% increased risk
Stopping for 1 year: 14% increased risk
Stopping for 2 years: 22% increased risk
— all compared to staying on.
Here is the most striking finding:
What 3 years of continuous treatment builds, a year and a half off the drug can undo.
Protection accumulates slowly. It unravels quickly. That asymmetry may be the single most important clinical implication of this study.
Using the drug for 0.5, 1, or even 1.5 years and then stopping produced no significant cardiovascular benefit at 3 years.
The risk ratios were close to 1.0 — as if the drug had never been taken.
Restarting helped — but didn’t fully restore the protection of uninterrupted use.
Discontinuation leaves a lasting scar.
This happened in the VA system, where copays are capped at $11/month. Over a quarter of users still discontinued entirely. Nearly another quarter had interruptions.
Cost matters — but it is far from the only reason people stop.
When people stop GLP-1 drugs, it’s not just weight that comes back. They experience a resurgence in inflammation, blood pressure, cholesterol, and insulin resistance.
Weight regain is visible. The metabolic reversal is not.
We think of this as a form of metabolic whiplash.
GLP-1 drugs are not short-term treatments. They are long-term treatments, and patients, providers, and insurers all need to reckon with that. Stopping can undo years of accrued heart protection.
Weight regain is visible. The metabolic reversal is not.
Stopping can undo years of heart protection.
The heavy lifting here was done by the amazing @biostayan
📢NEW: In a study of 600,000 people published today in @bmj_latest, we found GLP-1 drugs (Ozempic, Wegovy, Mounjaro) associated with reduced risk of addiction across alcohol, opioids, cocaine, cannabis, nicotine—and 50% fewer substance-related deaths.
🧵 bmj.com/content/392/bm…
We used VA data—one of the largest healthcare systems in the world—to emulate 8 randomized trials.
600,000 people. 3 years of follow-up. Over 1.5 million person-years.
Two questions: Do GLP-1 drugs prevent addiction? And do they reduce harm in people already struggling with it?
Among people with no prior addiction, GLP-1 drugs were associated with lower risk of addiction across substances:
▸ Opioids: 25%
▸ Cocaine: 20%
▸ Nicotine: 20%
▸ Alcohol: 18%
▸ Cannabis: 14%
Consistent across every substance we tested.
Our new study in @naturemedicine comprehensively maps all the benefits and risks of GLP-1 receptor agonists (GLP-1) (Ozempic, Wegovy, Mounjaro, Zepbound).
We saw that GLP1 use has substantially increased, and reports emerged about some off-target effects. But we realized that no one had comprehensively investigated effectiveness & risks of GLP-1 across all possible health outcomes nature.com/articles/s4159…
3. So, we decided to do the comprehensive study!
We did a discovery analysis which comprehensively examined 175 possible health outcomes (leaving no stone unturned) in 215,970 people on GLP1 and more than 2 million individuals not on GLP1
🚨 Long COVID is a significant health crisis in China too!
My commentary on the largest Chinese study to date, involving 74,075 participants, which found that 10-30% of them reported long COVID symptoms such as fatigue and brain fog #LongCovid
People who were randomly assigned to molnupiravir during the acute phase of COVID-19 had better outcomes at 3 months and 6 months than those randomly assigned to usual care