Most of the conversations I have about GLP-1 medications start in a fairly predictable place: should we use them at all? How much weight can we expect? What does the monitoring look like?
Those are usually the right questions. Most of the time, that's exactly where the conversation should stay.
But every so often, I realize we need to step back before we even get there. Not because GLP-1s don't work. They clearly do. But because getting to the prescription without a complete picture of someone's history is how you end up in situations that could have been avoided.
It usually comes down to medical history. And not just what's written in the chart, but what actually happened when you ask about it directly.
What follows is basically how I think through that. It's not a perfect checklist. Medicine rarely works that way. It's more of a framework for where I slow down, what I ask about, and where I draw lines — hard or soft — depending on what I find.
Some of these situations are hard contraindications. Others are just "proceed, but carefully." The distinction matters more in practice than it might seem on paper.
The thyroid warning is not just legal language
The boxed warning about medullary thyroid carcinoma gets mentioned in almost every discussion of semaglutide or tirzepatide. It also gets dismissed almost as quickly, either as overly cautious or as a technicality that doesn't really apply in practice.
The truth is somewhere in between.
In rodent studies, these drugs caused C cell tumors at high doses. That finding is real. But rodents and humans don't always translate cleanly, and we don't have strong human data showing the same effect. So the actual risk in people isn't fully characterized.
Still, there are lines I don't cross.
If a patient has a personal history of medullary thyroid carcinoma or Multiple Endocrine Neoplasia type 2, I don't prescribe a GLP-1. That's a hard stop. The signal in animal models is strong enough, and the theoretical mechanism is plausible enough, that I'm not willing to take that risk when there are other options.
Where it gets more gray is family history. I've had patients come in who vaguely remember something thyroid related in a parent or sibling, but can't give specifics. That's where I try to get more information before proceeding. Sometimes the family history is papillary thyroid cancer, which is a completely different thing and not a contraindication. Sometimes it turns out to be MTC, in which case I stop. Sometimes we can't figure it out, and I lean toward caution.
Pancreatitis is one of those histories that needs detail, not assumptions
Pancreatitis gets listed as a caution for GLP-1s, but in real practice, the nuance matters a lot.
Yes, there have been associations in trials and post-marketing reports. But the absolute risk is low, and causality is still debated. What matters more to me is the individual history.
When a patient tells me they've had pancreatitis, I don't stop there. I want to know what actually happened.
Was it from gallstones? Was it alcohol related? Was it a single episode ten years ago that never recurred? Or is this someone with recurrent, unexplained pancreatitis and a still-irritable pancreas?
Those are very different clinical scenarios, even though they get coded the same way.
I had a patient once who had a single episode of pancreatitis after a heavy alcohol binge in college, more than a decade ago, never drank again, no recurrence. I prescribed a GLP-1. She did fine. That history, while technically a flag, didn't change my assessment meaningfully.
Where I draw the line more firmly is active or ongoing pancreatitis, or a pattern of recurrent unexplained episodes. In those cases, I won't start a GLP-1. The mechanism of delayed gastric emptying, combined with potentially elevated triglycerides from weight-related metabolic issues, is enough to make me uncomfortable.
Gastroparesis is easy to miss unless you ask the right questions
GLP-1 medications slow gastric emptying. That's not a side effect. It's part of the mechanism. For most patients, that's what makes them feel full faster and eat less.
But physiology doesn't exist in isolation.
If someone already has delayed gastric emptying, you can run into trouble quickly. Symptoms get worse. Nausea, vomiting, bloating, and early satiety that go from manageable to severe. And in some cases, the oral medications they're taking for other conditions stop absorbing properly.
The challenge is that gastroparesis is often underdiagnosed. Especially in patients with longstanding diabetes, symptoms have sometimes been normalized for years. They don't think to mention it. Or they describe it in ways that don't immediately flag as a gastric issue.
So I try to listen for it indirectly. If a patient tells me they can't finish meals anymore, or that food just sits heavily, or that they feel nauseous after eating, I ask follow-up questions before assuming those are just normal eating pattern variations.
In someone with confirmed gastroparesis, I don't start a GLP-1. In someone with suspected symptoms, I usually want that evaluated first. Starting a medication that further slows gastric emptying in someone with an already-impaired system is not a risk I'm willing to take without a clearer picture.
Pregnancy is straightforward, but the conversation still matters
This is one of the clearer areas clinically, but I still spend time on it because assumptions create problems later.
GLP-1 medications are not used during pregnancy. The animal data raises concern for fetal harm, and we don't have adequate human safety data to say otherwise. That's not a debatable point.
So I tell patients directly: if there is any chance of pregnancy, we need reliable contraception while on this medication. That conversation needs to happen before the prescription, not after.
If someone is planning pregnancy, the medication comes off beforehand. Typically at least two months prior, to allow for clearance and to make sure we're not in a gray zone when they're trying to conceive.
And if pregnancy happens while someone is on therapy, we stop it immediately and coordinate care.
Breastfeeding is similar. The data is sparse, and in medicine, sparse data tends to push you toward caution. So I avoid GLP-1s in that period as well.
Eating disorders are where clinical judgment really matters
This is one of the most important and least cleanly defined parts of prescribing GLP-1s.
We tend to think of these medications in terms of appetite suppression and weight loss. But appetite suppression is not a neutral pharmacological effect in every patient.
In someone with a history of restrictive eating, that effect can reinforce patterns that are already harmful. I've seen patients develop a kind of positive feedback loop where the medication reduces hunger, which reduces eating, which feels good in a way that's disconnected from health, and then the eating restriction continues even when the medication is no longer needed to achieve it.
So I ask about it directly. Not in a checklist way, but in a real conversation.
If there's active anorexia or restrictive pathology, I don't start the medication. Full stop. That needs treatment first, and introducing a strong appetite suppressant into that picture is contraindicated in my judgment even if it's not always listed that way in labeling.
If there's a past history, things get more nuanced. I usually involve behavioral health early, not because I'm refusing to prescribe, but because that partnership matters in these cases. Monitoring is more intensive. The goals are more explicitly framed around metabolic health rather than weight loss as an endpoint.
Binge eating disorder is a different situation. In some patients, GLP-1s actually help reduce binge frequency, likely through the same appetite modulation pathways. But even there, I want a behavioral health provider involved early, because the mechanism behind bingeing isn't just hunger, and medication alone isn't the full answer.
BMI criteria are not arbitrary. They reflect the studied population.
People sometimes think of BMI thresholds as bureaucratic, but they matter because that's the population in which the risk-benefit calculation was actually established.
Approved indications are BMI of 30 or higher, or 27 or higher with at least one weight related comorbidity like hypertension, type 2 diabetes, or dyslipidemia. Those aren't arbitrary cutoffs. They're where the trials were run and where the safety and efficacy data actually applies.
When someone outside that range asks about GLP-1s for modest weight loss, I pause. Not because I don't understand the motivation, but because the benefit-risk calculation genuinely looks different in someone with a lower BMI. The potential benefits are smaller. The side effects are the same. And the downstream effects of significant weight loss in someone who doesn't have a lot of weight to lose can include muscle loss, nutritional deficiencies, and gallstone formation from rapid weight change.
A 15 to 20 percent reduction in body weight is genuinely transformative for someone with obesity. It is not neutral for someone who was already at a healthy weight.
That's where side effects like nutritional deficiencies, muscle loss without resistance training, and gallstones from rapid weight loss become a much harder clinical calculus to justify.
Drug interactions are usually manageable, but not ignorable
Most of the time, medication interactions are not dramatic with GLP-1s, but there are a few that deserve attention.
Patients on insulin or sulfonylureas are the most obvious example. If you add a GLP-1, glucose levels drop, and without adjusting the other medications, hypoglycemia becomes a real risk. That adjustment has to happen proactively, not reactively.
The gastric emptying effect also matters for oral medications that depend on consistent absorption. Thyroid medication is a good example. If the timing of absorption shifts because gastric emptying has slowed, TSH levels can drift unexpectedly. I always flag this and usually recommend monitoring thyroid function more closely after starting a GLP-1 in someone on levothyroxine.
Some conditions don't rule it out, but they change how I approach it
Gallbladder disease is a good example. Rapid weight loss increases the risk of gallstones, regardless of how that weight loss happens. With GLP-1s specifically, there's also been a signal for cholecystitis in some trial data. For someone with a history of gallstones or prior cholecystitis, I don't automatically refuse, but I flag it and watch more closely. If they still have a gallbladder and have had issues before, I want to know if symptoms return as they lose weight.
Kidney disease is another. The medications themselves aren't always the issue. It's the dehydration that can come with nausea and reduced intake early in treatment. Someone with already-compromised kidney function is at higher risk for acute kidney injury in that window. I counsel people about hydration heavily and monitor more frequently if there's baseline renal impairment.
Diabetic retinopathy also deserves a mention. Rapid improvements in blood sugar can temporarily worsen retinopathy in some patients with pre-existing disease. It's a known phenomenon from the diabetes literature. In someone with moderate or severe non-proliferative retinopathy, I'll coordinate with their ophthalmologist before starting and plan for early follow-up.
The part that actually determines outcomes happens before the prescription
At the end of the day, none of this works well if the history isn't complete. If you're looking at something like Wegovy or tirzepatide and framing the conversation purely around anticipated weight loss, you miss the part where someone casually mentions their sister had a thyroid issue, or they had stomach problems a few years ago that they chalked up to stress.
The patients who do best are not necessarily the ones with the simplest profiles. They're the ones where nothing important was left out before we started.
The ones I worry about are patients who minimize or simplify their history because they're focused on getting to treatment. Not out of dishonesty. Usually out of hope. They want this to work, and they're worried that if they mention everything, they'll be told no.
In reality, those details rarely stop treatment. But they almost always change how I approach it. The dose, the monitoring schedule, the conversations about what to watch for and when to call. Getting that right upfront is how you avoid the situations that could have been prevented.
So I usually end the same way: bring everything. The GI history you think is irrelevant, the family history you're unsure about, the eating patterns you've never discussed with a doctor. It doesn't automatically disqualify you. But it helps me take care of you better.
GLP-1s have genuinely changed what we can do for patients. But using them well is less about the prescription itself and more about the conversation that happens before it.
