Field notes: A mild and yet potentially deadly side effect.

Every effective medication has benefits and potential side effects. Most side effects are mild and short-lived, and are experienced by a subset of people. A few patients are allergic or have extreme reactions to certain therapies. It’s often pretty obvious when the risks and burdens of the drug surpass any benefits, and patients need to stop taking a medication. For example, if sulfa medications cause you to develop anaphylactic shock, your doctor will choose another class of antibiotic when needed.

But these decisions are often less straightforward and highly situational. What happens when a side effect is bothersome but not dangerous—that is, a drug is causing some type of discomfort, but the benefits are so substantial that the unpleasant side effects have to be mitigated or tolerated in service of a greater goal?

Lately, I’ve been seeing a particular situation with increasing frequency. This has been leading to a problem…and possibly a major one we won’t be able to detect.

Miracle drugs are not perfect.

Since I became aware of both the amazing results and increasingly common use of GLP-1 drugs like Ozempic, Wegovy, Mounjaro, and Zepbound, I’ve also noticed how much concern there is about their side effects. The rate of side effects (and their severity) looks to be similar to other important medications in our toolkit. But because the GLP-1s are delivering such incredible results—for weight loss, diabetes, cardiovascular disease, liver disease, kidney disease, etc.—there seem to be fears that “this can’t be real” or “this can’t be free.” As in, the outcomes are so good, they must be too good to be true.

These are good instincts and reasonable concerns. The problem is that side effects around GLP-1s are getting disproportionate and possibly unwarranted levels of special scrutiny, simply because of the many incredible results we keep hearing about and the growing popularity of the drugs.

Cases from my practice.

In the last year or two, I’ve seen some patients in the ER with a chief complaint of nausea and/or vomiting who had recently started taking a GLP-1. In each case, the patients’ symptoms were mild (or briefly a bit worse), and were easily controlled with a single dose of either oral or intravenous medications that treat their symptoms.

In a couple of instances, the patients understood that their symptoms were not entirely unexpected and that in most people, the symptoms go away or become far less bothersome. But in others, I’ve heard patients say things like, “My doctor says that I should no longer take this drug.”

This is highly problematic and premature in many instances. For example, if a person if taking semaglutide (Wegovy) because they have obesity and a history of a heart attack, the drug could eventually prevent another heart attack, a new stroke, or even save their life (see: the SELECT trial). So, if a person stops taking their GLP-1 because of a few bad days (over several weeks) during the initiation phase of the drug, this could literally be a deadly choice for some of them years down the road.

Look, I get it. At first, patients are experiencing only side effects, long before seeing any benefits. That can’t be encouraging. However, I’m sensing that the bar for my physician colleagues to recommend discontinuing these medications altogether often appears too easy to clear. We know that side effects abate over time, and many patients eventually have no side effects, despite a rough start.

Instead of stopping the GLP-1 so easily, doctors should consider trying an alternative approach. Indeed, there are many GLP-1 brands on the market already. Lower doses may also be appropriate, or spacing injections out a bit from the recommended weekly schedule. This type of drug management requires attention and care from clinicians, but it may be worth it.

Imagine if we gave up as easily on giving cancer patients chemotherapy because of their side effects. The reason we don’t (so soon) is that the stakes are much easier to understand. But I’d argue that in many cases, the stakes here are similar. It’s just that they play out over a longer period.

The medical literature is not helping.

One problem is that the medical literature is confusing even experts. A highly cited study from 2023 in the Journal of the American Medical Association of post-marketing observational data on GLP-1 adverse effects found increased rates of some gastrointestinal conditions that sound scary. I’ve had GLP-1 patients referred into the ER by their primary care doctor who were told to ask me to check for pancreatitis, stomach emptying problems, or intestinal blockages. That’s right out of the JAMA study.

Two problems. First, while these sound scary, like all medical problems, there’s a spectrum. Most cases of pancreatitis are “uncomplicated” and resolve quickly. Intestinal blockages sound really scary, but in reality what’s being described in GLP-1 patients is a “physiologic” block, that is, the usual forward progression of food and liquid in the GI tract grinds to a temporary halt. Unlike physical blockages, these resolve on their own. In both cases, some IV fluids may be useful to give the gut rest. Importantly, the JAMA paper did not distinguish between mild and more severe cases of these adverse effects. I’ve seen and heard of instances where GLP-1 recipients have developed these conditions, but I’ve never heard of a serious one.

The second issue is that large (double-blinded, randomized, placebo-controlled) clinical trials did not show anywhere near the increased risk of these conditions described in the JAMA paper. For example, the JAMA paper found a 10-fold increase in pancreatitis compared to controls taking different drugs for weight loss. But in the SELECT trial, there were no differences in rates, whether the patients were on the GLP-1 or placebo (and there were more than enough cases for the data to be convincing.) Meanwhile there was a tiny difference in gallbladder conditions (2.8% in the semaglutide group vs 2.3% in the placebo, which was statistically significant). Interestingly, that 0.5% absolute (or 22% relative) increase was not replicated in the JAMA study, which found no statistical difference. This just demonstrates that when you have observational data and randomized, blinded, placebo-controlled trial data, it’s wise to remember that the latter is much more likely to be devoid of bias and confounding variables.

So I’ve got doctors telling their patients that GLP-1s cause an increase in pancreatitis based on observational data, while ignoring the far-more impressive randomized trial data that refute it. Also, the observational findings in the JAMA study had fewer patients in the dataset than the SELECT trial. Usually, it’s the other way around; observational datasets have far more patients than the trials. That’s generally the whole point of post-marketing observational studies: to gather data from more patients than could be enrolled in the trials. Here, it was backwards.

This is all being compounded by the fact that the FDA requires drug makers to list side effects that occurred in the trials, even if there were no statistical increases compared to placebo. Doing this is often silly and it just scares people. It’s one thing when a side effect or adverse event is too rare for researchers to statistically evaluate. But in the case of the adverse events we’ve been discussing here, that’s simply not the case.

Side effects and adverse events are not fun.

I recently posted on social media about how, in my opinion, nausea and vomiting were not worse than death. I didn’t think that was controversial, although in some circumstances (end-stage cancer), there comes a point where quality of life is more important than quantity. This was intended as a commentary on how people’s brains are being broken by the GLP-1 story. But I got some blowback, because people thought I was downplaying the misery of nausea and vomiting. That was not my intention, which I clarified. My message was not to patients but to doctors. We need to manage expectations correctly. And we need to treat symptoms as they come. We also need to be nuanced and thorough in our evaluations. Patients who had bad bouts of nausea and vomiting deserve immediate treatment. But patients also deserve careful consideration—more careful than, “Oh, you had a bad episode of nausea and vomiting? Stop taking your GLP-1 forever.”

This move seems like a doctor protecting the patient, but I worry it is something else. I worry it is more, “Ugh, I don’t have time to deal with this. Let’s get them off the drug and go back to the previous plan,” than it is actually helping the patient’s overall care. The problem is that the particular GLP-1 and dose were not being well-tolerated. That requires a response, not a full-fledged retreat. I’m starting to sense that doctors’ thresholds to stop GLP-1s are driven by the desire for fewer inbox messages than what would be best for the patient—things like trying a different version of a drug, or changing the usual dosing regimen.

Of course, the patients are going to go along with what the doctors say. If the doctor says, “I’m worried you’ll get pancreatitis,” the patient hears, “I’m worried you’ll have a dangerous problem with your pancreas.” That’s not even accurate most of the time.

And again, nausea and vomiting may be “not dangerous,” most of the time, but they certainly are awful. Ever seen a patient on certain chemotherapy regimens try to eat? Eating is one of life’s great pleasures. It should tell us something when people don’t want to eat. Many chemo regimens give patients nausea, vomiting, or pain with eating. Some take away taste and smell. Eating when you’re not hungry or can’t taste anything—or worse, when it causes pain—is a total chore. I’ve seen it a thousand times.

It’s obvious that we all need calories and hydration to live. Nausea and vomiting prevent that. Unresolved nausea and vomiting are unsustainable. When that point is reached, the risks get very real. But I’m worried that doctors are seeing treatable and temporary nausea and vomiting in GLP-1 patients, getting spooked (or lazy), and calling the whole thing off too soon.

We need to be more thoughtful and guide our patients through the GLP-1 dose escalation process more carefully. Many patients on GLP-1s have what could eventually become deadly conditions. These drugs can help—but we need to remember that they are imperfect and come with trade-offs. Our job is to help manage these complexities, not run away from them.