Daniel sent us this one about Singulair — montelukast. He's an asthma patient, symptoms not fully managed on a steroid inhaler, and his doctor's suggested adding Singulair. But he's seen the psychiatric warnings, the black box stuff, and he's trying to square the scary headlines with the fact that doctors still routinely prescribe this to patients who have comorbid depression or anxiety. He wants to know what the absolute risk actually looks like, whether pre-existing mild depression puts him in a higher risk bracket, and what a safe monitoring practice would actually entail in real life. And underneath all of it, there's this question of — if not Singulair, then what? Is there a viable replacement yet?
There's a lot here, and I think the place to start is just naming what's unusual about this whole situation. Montelukast was approved in nineteen ninety-eight. It spent over two decades as this almost boringly safe drug in the public imagination — prescribed to millions of kids, available as a chewable, handed out like a slightly more medical vitamin. And then in March twenty twenty, the FDA drops a boxed warning — the strongest one they issue — specifically for serious neuropsychiatric events. That's a dramatic reversal for a drug that had been generic for years at that point.
Which is part of what makes it so disorienting. It's not a new drug where we're still discovering the side effect profile. This thing has been around for more than twenty-five years, and the warning comes that late in the game.
And the warning wasn't based on new clinical trial data. It was based on an accumulation of adverse event reports — the FDA's adverse event reporting system, FAERS — and a growing body of case reports and observational studies that kept pointing in the same direction. Nightmares, insomnia, agitation, aggression, depression, and in rare cases, suicidal ideation and completed suicide. The FDA reviewed eighty-two completed suicides associated with montelukast between nineteen ninety-eight and twenty nineteen. That's over two decades. But each one of those is obviously devastating, and the question becomes — is the drug causal, or is this background noise in a population that already has higher rates of anxiety and depression?
That's the tension the prompt is really getting at. When you hear "eighty-two completed suicides," that sounds terrifying. But you have to put a denominator under it. How many millions of people took this drug over that same period?
The numbers are enormous. There was a study published in twenty twenty-one by researchers in Singapore using a nationwide dataset — they looked at over seventy thousand patients prescribed montelukast and compared them to over two hundred thousand patients using inhaled corticosteroids. They found no significant increase in overall neuropsychiatric events. But they did find a slightly elevated risk of sleep disorders and, in a subgroup analysis, a possible signal for anxiety in patients with pre-existing psychiatric conditions. And that's where things get nuanced — because the signal isn't uniform, and it's not huge, but it keeps popping up in certain subgroups.
Let's get concrete about what the absolute risk actually looks like. If I'm someone like the prompt describes — mild pre-existing depression, adding Singulair to a steroid inhaler — what numbers should I have in my head?
The best data we have comes from a few large observational studies. One of the most cited is a twenty twenty-two study from the University of Chicago that used a national claims database of over three hundred thousand asthma patients. They found that the absolute risk of a new neuropsychiatric diagnosis within ninety days of starting montelukast was about one point seven percent, compared to about one point three percent for patients starting inhaled corticosteroids. So the absolute difference is something like four additional cases per thousand patients. That's the number people should be thinking about — not "this drug will make you depressed," but "for every thousand people who take this, maybe four will experience a neuropsychiatric event they wouldn't have otherwise.
Four per thousand is not nothing, but it's also not the impression you'd get from reading black box warning headlines.
And that's the paradox of the boxed warning. It's there for a reason — the signal is real enough that the FDA felt compelled to act — but the absolute risk is modest enough that the drug remains widely prescribed. The American Academy of Allergy, Asthma and Immunology put out a response to the boxed warning basically saying montelukast remains an important treatment option, and the risk needs to be contextualized. They didn't pull it. They didn't even recommend against it as first-line add-on therapy. They said: counsel patients, monitor them, and make an informed decision.
That "counsel and monitor" piece is, I think, what the prompt is really hungry for. The practical question. If someone with pre-existing depression decides to start this drug, what does that actually look like? What are they watching for?
This is where I think the clinical guidance has actually gotten quite good, even if it's not always communicated well to patients. The first thing to know is that most neuropsychiatric events, when they do occur, happen early. The median time to onset in many of the case reports is within the first two weeks. Some studies suggest the first few days. So the monitoring window is not indefinite — it's front-loaded.
Which is actually reassuring. You're not signing up for a lifetime of wondering whether your mood shift is the drug or just life.
The practical approach that most guidelines suggest is: before starting, do a quick baseline. How's your mood? How's your sleep? Any unusual dreams? And then check in with yourself daily for the first two weeks, and then weekly for the next month or so. The specific things to watch for are: new or worsening depression, agitation, aggressive impulses, trouble sleeping that's new or different from your baseline, nightmares that are vivid or disturbing, and any thoughts of self-harm. And the key instruction is — if any of these appear and they're not clearly attributable to something else, stop the drug and call your doctor.
Does stopping reverse it?
In the vast majority of cases, yes. The effects are generally reversible upon discontinuation, usually within days to weeks. There are case reports of more protracted symptoms, but those are rare. The typical story is: patient starts montelukast, develops nightmares or mood changes within a week or two, stops the drug, and returns to baseline. That's the most common trajectory.
The monitoring protocol is basically: know thyself, pay attention for two weeks, and have a low threshold for calling it quits.
That's it. And for someone with pre-existing depression, the advice gets slightly more specific. The concern with pre-existing psychiatric conditions is twofold. One, you might be more vulnerable to a neuropsychiatric side effect. Two, if your mood worsens, it's harder to attribute — is it the drug, or is it a natural fluctuation of the underlying condition?
That attribution problem seems like the real challenge. If you already have mild depression and you have a bad week, how do you know?
This is where some of the observational data gets tricky. There was a twenty twenty-three study out of Korea that looked specifically at patients with pre-existing depression or anxiety who were prescribed montelukast. They found a modestly elevated risk of psychiatric hospitalization compared to those with pre-existing conditions who got different asthma medications — but the absolute risk was still small, and the authors noted that confounding by indication is a real problem here. People with more severe asthma are more likely to be prescribed montelukast, and people with more severe asthma also have higher rates of depression. Untangling that knot is genuinely difficult.
The data doesn't say "if you have depression, don't take this." It says "if you have depression, be more deliberate about monitoring, and know that the attribution is going to be messier.
And I think this is where the doctor-patient relationship actually matters. A good prescribing conversation should include: here's what we're watching for, here's the plan if something changes, and here's my threshold for saying we should stop. The fact that the prompt mentions his doctor routinely prescribes this to patients with comorbid psychiatric conditions tells me that doctor has probably thought through this calculus and concluded the respiratory benefit outweighs the psychiatric risk for most patients.
Let's talk about that respiratory benefit, because I think it gets lost in the psychiatric conversation. What does Singulair actually do, and why is it still used?
Montelukast is a leukotriene receptor antagonist. Leukotrienes are inflammatory molecules released by mast cells and eosinophils — they're part of the allergic cascade. When you inhale an allergen or trigger, your airways release leukotrienes, which cause bronchoconstriction, mucus secretion, and airway inflammation. Montelukast blocks the receptor that leukotrienes bind to. It's not a bronchodilator — it doesn't open airways in the moment. It's a controller medication that reduces the underlying inflammatory response.
The appeal is that it's oral, right? You just take a pill.
That's a huge part of it. For patients who struggle with inhaler technique — and that's a lot of people, studies suggest something like seventy to eighty percent of patients don't use inhalers correctly — an oral medication is dramatically simpler. It's also once-daily, it's cheap because it's generic, and it has a different mechanism of action than inhaled corticosteroids, which means it can be truly additive. If your steroid inhaler is getting you seventy percent of the way there, adding montelukast might get you to eighty-five or ninety percent. That's meaningful for quality of life.
For a certain subset of patients, it's almost uniquely effective. Exercise-induced bronchoconstriction, aspirin-exacerbated respiratory disease — these are areas where leukotrienes are particularly central to the pathophysiology.
Aspirin-exacerbated respiratory disease is the clearest case. These patients have a triad of asthma, nasal polyps, and aspirin sensitivity, and they overproduce leukotrienes to a dramatic degree. For them, montelukast can be almost transformative. It's not just an add-on — it's targeting a core driver of their disease.
Which brings us to the other half of the prompt's question. Is there a viable replacement yet? If someone reads the black box warning and says "I'm not touching that," what are their options?
This is where the landscape has actually shifted in interesting ways. For a long time, if you had moderate persistent asthma not controlled on an inhaled corticosteroid alone, your add-on options were basically: long-acting beta agonist, or LABA, which is what you get in combination inhalers like Symbicort or Advair; montelukast; or theophylline, which nobody wants to use because it's finicky and has its own side effect issues. LABAs are effective, but they've had their own black box warning saga — there was a whole controversy about whether LABAs increase the risk of severe asthma exacerbations, particularly in certain populations. That warning was actually removed in twenty seventeen after large safety trials showed no increased risk when LABAs are combined with inhaled corticosteroids.
The combination inhaler is the main competitor to montelukast as add-on therapy.
It's the main alternative, yes. But they work differently, and they're not mutually exclusive. You can be on a LABA-ICS combination and still add montelukast. The guidelines from the Global Initiative for Asthma, GINA, recommend a stepwise approach. Step three is low-dose ICS plus LABA. Montelukast is listed as an alternative at step two and step three, but it's generally considered less effective than LABAs for most patients. Where it shines is in patients who have allergic triggers, exercise-induced symptoms, or who can't or won't use inhalers properly.
Then there's the biologics.
The biologics are the big story of the last decade in asthma treatment. We now have five monoclonal antibodies approved for severe asthma: omalizumab, which targets IgE; mepolizumab and reslizumab, which target interleukin-five; benralizumab, which targets the interleukin-five receptor; and dupilumab, which targets the interleukin-four receptor alpha subunit. These are for patients with severe eosinophilic or allergic asthma who aren't controlled on high-dose ICS plus LABA. They're injectable, they're expensive — we're talking tens of thousands of dollars a year — and they're generally reserved for the top of the treatment pyramid.
They're not really a replacement for montelukast. They're for a much sicker population.
The biologics are transformative for severe asthma, but they're not competing with a five-dollar generic pill. Montelukast occupies this interesting niche where it's more effective than nothing, less effective than LABAs for most patients, but uniquely suited to specific phenotypes and practical constraints. And nothing has come along to displace it from that niche.
Which is kind of remarkable. Twenty-five-plus years, and the leukotriene pathway is still only targeted by one drug class, and montelukast is still the only one in that class that's widely used. Zileuton exists but it's a five-lipoxygenase inhibitor with a liver toxicity issue and four-times-daily dosing, so it's barely prescribed.
There was some excitement a few years ago about CRTH2 antagonists — that's the chemoattractant receptor-homologous molecule on Th2 cells, another target in the allergic inflammatory pathway. Fevipiprant was the lead candidate, and Novartis took it through phase three trials. It showed some efficacy in reducing eosinophilic airway inflammation, but the results were modest, and in twenty nineteen Novartis discontinued development. There were a few other CRTH2 antagonists in development — timapiprant, setipiprant — and they've all kind of fizzled.
The pipeline for a direct montelukast replacement is basically empty.
It's empty. And that's worth sitting with. We have a drug that works through a unique mechanism, is orally available, is cheap, and has a concerning but relatively rare side effect profile — and there is nothing in development that would replace it. If you're a patient who benefits from leukotriene receptor antagonism, your options are montelukast or nothing on that pathway.
Which makes the psychiatric question even more loaded. It's not like you can say "I'll just take the other leukotriene receptor antagonist that doesn't have the black box warning." There isn't one.
That's partly a function of drug development economics. Montelukast has been generic since twenty twelve. It costs maybe ten to twenty dollars a month out of pocket. The market for a new branded leukotriene receptor antagonist is essentially nonexistent — no pharmaceutical company is going to spend hundreds of millions of dollars developing a drug to compete with a cheap generic unless they can demonstrate dramatic superiority, and "same efficacy without the rare psychiatric side effect" is a very hard claim to prove in clinical trials.
The market has left us with this one tool, and the psychiatric concern isn't going to be resolved by a newer, safer version. We have to learn to use the tool we have more intelligently.
And I think part of using it more intelligently is understanding what the mechanism of the psychiatric side effects might actually be, because that helps with monitoring and risk stratification. The honest answer is we don't fully know. But there are some compelling hypotheses.
Let's hear them.
The leading hypothesis involves the fact that leukotrienes and their receptors exist in the brain, not just in the airways. Cysteinyl leukotriene receptors — the ones montelukast blocks — are expressed in the hippocampus, the prefrontal cortex, and the amygdala. These are regions involved in mood regulation, fear processing, and memory. There's also evidence that leukotrienes play a role in neuroinflammation and in the regulation of the blood-brain barrier. Montelukast crosses the blood-brain barrier — that's been demonstrated in animal models — and by blocking leukotriene signaling in the brain, it might disrupt normal neurophysiological processes in susceptible individuals.
Some people have actually tried to flip this into a therapeutic mechanism. I've seen papers exploring whether montelukast could be neuroprotective in conditions like Alzheimer's or Parkinson's, precisely because it reduces neuroinflammation.
There was a small phase two trial published in twenty twenty-one looking at montelukast for Parkinson's disease, and the results were mixed but not entirely negative. The idea is that if leukotrienes contribute to neuroinflammation, blocking them might slow neurodegeneration. But the flip side is that if leukotrienes also play a role in normal synaptic plasticity or mood regulation, blocking them might cause problems. It's the same mechanism, just different contexts.
The brain giveth and the brain taketh away.
Another hypothesis is that montelukast might affect neurotransmitter systems indirectly. There's some evidence from animal studies that it can alter serotonin and dopamine levels in certain brain regions, possibly through its effects on inflammatory cytokines, which are known to influence neurotransmitter metabolism. But this is all at the level of mechanistic speculation — we don't have clean human data that says "montelukast causes depression by mechanism X.
Which makes it hard to predict who's at risk.
There have been attempts to identify risk factors. The FDA's analysis suggested that the risk might be higher in children, which is part of why the boxed warning was particularly emphasized for pediatric use. But the data isn't clean enough to say "if you have risk factor Y, your absolute risk is Z percent." The best we can say is: if you have a personal or family history of psychiatric conditions, be more vigilant. If you've had a previous reaction to montelukast or a related drug, don't take it again. And if you're a parent giving this to a child, watch for behavioral changes, especially in the first few weeks.
The pediatric angle is worth pausing on, because Singulair was hugely prescribed to kids. The chewable tablet was a big part of its commercial success. And the idea of giving a child a drug that might cause nightmares or aggression — that's viscerally disturbing to parents in a way that adult side effects aren't always.
The pediatric data is concerning, even if the absolute risk is still low. A twenty twenty-two study from the Netherlands using a national pediatric cohort found a roughly two-fold increased risk of neuropsychiatric events in children prescribed montelukast compared to those prescribed inhaled corticosteroids alone. The most common events were sleep disturbances, nightmares, and behavioral changes like agitation or aggression. The absolute risk was still small — less than two percent — but a two-fold relative risk in children gets your attention.
That's the thing about communicating risk to patients and parents. The relative risk sounds terrifying — "doubles the risk" — but if the baseline is one percent and it goes to two percent, that's a very different picture than if the baseline is twenty percent.
This is one of the great failures of medical risk communication. We lead with relative risks because they sound dramatic, and we bury absolute risks because they sound small. But the absolute risk is what patients actually need to make decisions. "Your child has a ninety-eight percent chance of not experiencing this side effect" is a very different frame than "this drug doubles the risk of psychiatric events.
That's exactly the kind of reframing the prompt seems to be looking for. Not "is this drug safe or dangerous," but "what are the actual numbers so I can make a decision I'm comfortable with.
Let me give you another number that I think helps contextualize this. The number needed to harm — NNH — is a way of expressing how many patients need to be treated before one additional adverse event occurs. Based on that University of Chicago study, the NNH for a new neuropsychiatric event with montelukast versus inhaled corticosteroids is about two hundred and fifty. You need to treat two hundred and fifty people with montelukast instead of an ICS to see one additional neuropsychiatric event. For context, the number needed to treat — NNT — for montelukast to prevent one asthma exacerbation over a year is somewhere between ten and twenty, depending on the population. So the benefit is an order of magnitude more common than the harm, at least in terms of these specific outcomes.
That's a useful ratio to have in your head. You're more than ten times more likely to benefit than to be harmed, in terms of countable events.
That's for the broad neuropsychiatric category. If you look specifically at severe events — suicidal ideation, completed suicide — the NNH is much higher, probably in the thousands. The absolute risk of a severe psychiatric event is very low.
Let's talk about the post-marketing surveillance picture, because the FAERS data is what drove the boxed warning, but FAERS is also notoriously messy. It's voluntary reporting. It's subject to all kinds of biases.
FAERS is a blunt instrument. It's a passive surveillance system — anyone can report an adverse event, and there's no denominator, no control group, and no way to establish causality. The FDA uses it to detect signals, not to quantify risk. The signal for montelukast was strong enough and consistent enough — particularly for neuropsychiatric events in children — that the FDA felt it had to act. But quantifying the risk requires the kind of observational studies we've been discussing, and those have generally found small absolute risks. The gap between the FAERS signal and the observational study findings is part of why this conversation is so difficult. The FAERS data makes it look like a crisis. The cohort studies make it look like a manageable risk.
Both can be true at the same time. A drug can cause real, severe harm in a small number of people and still be net-beneficial for the population. The question is whether we can identify those people in advance.
Which we can't, at least not with any precision. And that's uncomfortable. We're used to the idea that medicine can stratify risk — if you have this genetic variant, don't take this drug; if you're over sixty-five, use a lower dose. With montelukast, we don't have a clean biomarker or demographic predictor. The best we have is: pay attention, especially in the first two weeks, and stop if something feels wrong.
Which brings us back to the practical question. The prompt asks about a safe monitoring practice. We've covered the what — watch for mood changes, sleep disturbances, agitation, self-harm thoughts. But how do you actually operationalize that? Do you keep a journal? Do you ask a family member to check in?
Some clinics have started using a simple daily mood and sleep log for the first two weeks. It doesn't have to be elaborate — a one-to-five scale for mood, a yes-or-no for unusual dreams, a yes-or-no for agitation or irritability. Takes thirty seconds. The point is to create a structured check-in so you're not relying on vague recollection. If your baseline mood is a four and it drops to a two and stays there for three days, that's a signal. If you start having nightmares when you never had them before, that's a signal. The log just makes it harder to dismiss or rationalize.
Having a family member or close friend know you're starting the drug and what to watch for — that's a low-cost intervention that could catch something you might miss yourself.
Agitation and behavioral changes are often more visible to others than to the person experiencing them. A spouse or parent might notice that you're more irritable or withdrawn before you register it yourself. That's standard advice for starting any psychiatric medication, and it applies here too.
There's also the question of what happens if you do have a reaction. If someone stops montelukast because of mood changes, and their asthma was better on it — what then?
That's the hard case. If montelukast was providing meaningful respiratory benefit and you have to stop it, you go back to the treatment algorithm and look at what else you can add. If you're not already on a LABA-ICS combination, that's the obvious next step. If you are, you might look at tiotropium — that's a long-acting muscarinic antagonist, originally for COPD but now approved for asthma. It's an inhaler, but it's once-daily and has a different mechanism. If those aren't enough, you start thinking about biologics, but that requires meeting the severity criteria and going through prior authorization. It's not a quick fix.
For some patients, the answer might be that none of the alternatives work as well, and they have to weigh the respiratory benefit against the psychiatric risk with more granularity.
I've seen case reports where patients and their doctors decided to rechallenge with montelukast at a lower dose — there's a four milligram chewable for kids, and some adults have tried that — or with a very structured monitoring plan. The data on dose reduction as a risk mitigation strategy is basically nonexistent, but it's a logical approach if the side effect might be dose-dependent. We just don't know if it is.
The things we don't know about this drug after twenty-five years are kind of remarkable.
It's a humbling reminder of how much of pharmacovigilance is post-marketing and how slow the signal detection can be. Montelukast was on the market for almost a decade before the first case reports of neuropsychiatric side effects started appearing in significant numbers. The FDA issued a safety communication about it in two thousand eight, then updated it in two thousand nine, then didn't issue the boxed warning until twenty twenty. That's a twelve-year lag between the initial signal and the strongest regulatory action.
In that twelve years, millions of people took the drug, many of them children, without being counseled about psychiatric risks.
Which is why I think the current approach — informed consent, structured monitoring, low threshold for discontinuation — is actually a reasonable equilibrium. We're not pulling the drug. We're not ignoring the risk. We're trying to use it more carefully.
There's one more piece of this I want to touch on. The prompt mentions that doctors routinely prescribe Singulair to asthmatic patients with comorbid psychiatric conditions, and he infers from that that the risk is reasonable. I think that's worth examining. Doctors prescribe a lot of things routinely that carry non-trivial risks. Clinical inertia is real. The fact that something is common practice doesn't necessarily mean it's been carefully optimized.
That's a fair point. There was a survey published in twenty twenty-one of allergists and pulmonologists about their prescribing practices after the boxed warning. About sixty percent said they had changed their practice in some way — more counseling, more monitoring, fewer new prescriptions. But forty percent said they hadn't changed anything. And among those who hadn't changed, the most common reason was "I haven't seen these side effects in my own patients.statistically expected when the absolute risk is one or two percent and your panel might be a few hundred patients. You could easily miss it.
Absence of evidence isn't evidence of absence, and a clinician's personal experience isn't a surveillance system.
The paradox is that the side effect is both rare enough that most individual doctors won't see many cases and common enough that, at a population level, it affects thousands of people. That's the kind of thing that only shows up in large datasets, which is why the FAERS signal and the observational studies are so important even if they're imperfect.
What's your bottom line for someone like the prompt describes? Mild depression, asthma not fully controlled on a steroid inhaler, considering Singulair, wants to feel informed rather than scared.
My bottom line is: the absolute risk of a serious neuropsychiatric event is low — probably on the order of one in a few hundred for anything significant, and much lower for severe events. The risk of milder but still bothersome effects like sleep disturbances or mood changes is higher, maybe one in twenty to one in fifty. Most effects appear early and resolve when the drug is stopped. Having pre-existing depression doesn't dramatically change the absolute risk, but it does make attribution harder, which means monitoring needs to be more structured. If you go into it with a plan — baseline assessment, daily check-in for two weeks, a family member informed, low threshold to stop — you're doing what the best current evidence supports.
If you're still uncomfortable after all of that, the alternatives exist. They're not one-to-one replacements, but a LABA-ICS combination or tiotropium or even stepping up to a biologic if you qualify — there are other paths.
There are other paths. But none of them are as simple as a once-daily pill that costs ten dollars a month. And for the right patient — the one with allergic triggers, exercise-induced symptoms, aspirin-exacerbated respiratory disease, or poor inhaler technique — montelukast might be the best option despite the baggage. The goal isn't to eliminate risk. It's to manage it intelligently.
That, I think, is the reframe the prompt was looking for. Not "is this drug safe," but "how do I use it safely.
That's the whole game with montelukast in twenty twenty-six. The drug hasn't changed. Our understanding of how to use it has.
Now: Hilbert's daily fun fact.
Hilbert: In eighteen eighty-three, the Japanese government issued a sumptuary law limiting commoners to a maximum of one hairpin per person. For context, that's roughly one one-thousandth of the hairpin density of a mid-ranking Edo courtesan, which modern equivalents suggest would be like limiting someone to wearing no more than a single earring stud while the person next to them is allowed a full chandelier.
I have so many questions about hairpin density as a metric.
I'm just impressed Hilbert found a way to turn sumptuary law into a unit conversion problem.
Thanks to Hilbert Flumingtop for that glimpse into the hairpin economy of Meiji-era Japan. This has been My Weird Prompts. If you want more episodes, we're at myweirdprompts.com, and we'd love a review wherever you listen. We'll be back next week.
