Anxiety Treatment

The public conversation about anxiety treatment and the clinical reality have drifted about as far apart as it’s possible for two conversations on the same topic to drift. People walk in expecting either a Xanax prescription or a magical breathing technique. What they usually need is a stepped approach that takes a few months to build and works in a specific order. Most of the field gets this wrong in at least one direction.

The basic ladder hasn’t moved much in twenty years because it keeps working when people stick with it. We know what helps. The gap is between knowing and doing, and that gap is where most of the people in the chair are living when they finally get around to making the appointment.

Before anything else, the diagnosis matters. Anxiety is a category, not a single disease. Generalized anxiety disorder, social anxiety, panic disorder, and OCD all get treated differently, and “I have anxiety” is shorthand for some specific flavor of one of them. Start there.

Figuring out which anxiety you actually have

GAD looks like worry running in the background all day. The kind of guy who can’t stop running disaster scenarios in his head, about his job, about his kids, about his parents’ health, about the email he sent yesterday, about whether the dog is acting weird, about a tax thing from 2019. There’s no single trigger and the worry doesn’t attach to one thing… it just runs. The sleep is bad, the muscles are tight, and it’s been going on for at least six months.

Social anxiety is specific to being watched or judged. Meetings, dating, eating in front of people, public speaking, ordering coffee. The patient functions fine at home and falls apart at the dinner party. The kind of guy who can code in a room full of strangers all day, even give a code review to four engineers without breaking a sweat, but cannot order at a counter without rehearsing the sentence in his head twice and then getting it wrong anyway. Specific situations, specific dread.

Panic disorder is the one with the discrete attacks. Out of nowhere, heart pounding, can’t breathe, hands tingling, certain you’re dying. The ER visits stack up. The cardiologist clears you and you don’t believe him. What actually drives the diagnosis is the fear of the next attack, which builds the avoidance, which builds the diagnosis.

OCD is the one people miss most often. Intrusive thoughts that feel awful, paired with compulsions that temporarily turn the awful feeling down. Checking the stove. Counting. Mental rituals nobody can see and that the patient has been doing for so long he thinks of it as personality. OCD responds to a specific kind of CBT (cognitive behavioral therapy, the structured worksheet-and-homework kind, not the talk-about-your-mother kind) called ERP, which stands for exposure and response prevention, and to SSRIs at higher doses than you’d use for GAD. Calling it “anxiety” and treating it like GAD costs people years of getting nowhere.

The treatment ladder branches based on which of these you’ve got. So step one is figuring out which one is in the room.

The medication ladder, in order

First line is an SSRI or SNRI. Sertraline, escitalopram, fluoxetine on the SSRI side (SSRIs are the serotonin-focused antidepressants, Zoloft, Lexapro, Prozac). Venlafaxine or duloxetine on the SNRI side (those add a norepinephrine kick, brand names Effexor and Cymbalta). The data on these has been steady since the 90s. They turn the volume down enough that exposure work and behavior change become possible. They aren’t magic and they aren’t supposed to be.

Starting doses for anxiety are lower than for depression. I usually start sertraline at 25 mg for a week before going up to 50, because the first two weeks of an SSRI can spike anxiety before they help it, and patients who get blindsided by that quit on day five with a story they’ll be telling for the rest of their lives about how Lexapro made everything worse. The drug needs four to six weeks at a real dose to do its real work. Most people who say “Lexapro didn’t work for me” quit at week three or never got past 5 mg.

Start there.

Hydroxyzine is the unsexy workhorse for as-needed use. It’s an antihistamine, not addictive, not a controlled substance, sedating enough to take the edge off a bad evening. 25 to 50 mg PRN. It won’t stop a full panic attack in its tracks, but it’ll soften a rough day, and it’s a much better PRN than a benzo for the vast majority of patients.

Propranolol is the beta-blocker I reach for when the anxiety is performance-driven. A specific presentation, an interview, an audition, public speaking. 10 to 20 mg an hour before the event blocks the peripheral adrenaline symptoms (shaky hands, racing heart, the voice that cracks halfway through) without touching cognition. Musicians have been quietly using this for forty years. You use it as a tool for known stressors, not as a daily med.

Buspirone sits in the second tier for GAD specifically. Slow onset, modest effect, safe, non-sedating, worth a try when SSRIs aren’t enough or aren’t tolerated. Nobody loves it. It works for a real subset of people.

The benzodiazepine question, the part where I get opinionated

Xanax, Klonopin, Ativan, Valium. They work too well and too fast, which is exactly the trap. The relief shows up inside thirty minutes, which is why the brain learns the relief, expects it, then requires it. Tolerance builds, doses creep, and withdrawal from chronic daily benzos is genuinely dangerous, more dangerous than alcohol withdrawal in some cases, and the rebound anxiety on the way off can be worse than the anxiety that started the whole thing.

There are appropriate uses. A short bridge while an SSRI is ramping up. A flight twice a year for a guy with a real fear of flying. A bad acute period after a trauma. A surgical procedure. The thread is that they’re time-limited and infrequent.

What I don’t do is daily benzos for chronic anxiety, and the guys I inherit on 2 mg of Xanax three times a day didn’t start there either. They started at 0.5 mg as needed and the dose climbed because the body adapted. Getting them off takes months of slow tapering, often via a switch to a longer-acting agent like clonazepam or diazepam first. People assume Klonopin is the “cleaner benzo” because it lasts longer, and it isn’t really, it’s a slower benzo, which sounds like it should be safer but the longer half-life just means the same problem unfolds at a slower pace, you wind up dependent on it just as completely.

If a clinician is writing you a refillable daily benzo prescription as a first move for chronic worry, get a second opinion. That’s not a fringe take, that’s standard of care, and the providers who haven’t caught up to it are the same ones writing the same scripts for the same patients they’ve been writing them for since 2007.

You don’t think your way out of anxiety.

The benzos work so fast that the brain skips the part where you learn the worry was survivable. That’s not a side effect of benzos, it’s the mechanism.

Where CBT earns its reputation

CBT is the answer most of the time, and the version of CBT that has the data is the one with homework and exposure pieces. Not the chat version where you sit in a chair and talk about your week and your therapist nods. Twelve to sixteen weeks of structured work, assignments between sessions, a therapist who’s pushing you toward the things you’ve been ducking.

For panic disorder, the exposure is interoceptive, which sounds clinical and just means you deliberately reproduce the physical sensations of a panic attack (spinning in a chair, breathing through a coffee straw, running up the stairs) until your brain stops interpreting those sensations as a heart attack. For social anxiety, it’s graded social exposures with somebody coaching you through them. For OCD, ERP, which means provoking the obsession on purpose and then sitting through the discomfort without doing the compulsion you’ve been using to neutralize it. None of this is comfortable. All of it works when patients actually do it. If your therapist uses the word “CBT” but doesn’t give you homework, you got the chat version with the CBT label, find somebody else.

Why insight alone never quite does it

You don’t think your way out of anxiety. The thinking part of the brain sits underneath the alarm circuit, not on top of it. When the amygdala is firing, the prefrontal cortex (the front-of-the-forehead part of the brain that’s supposed to be running the show) is along for the ride, not steering. That’s why “just calm down” is a useless instruction, and why smart self-aware patients are usually the most confused that their insight isn’t fixing anything. The thing you’re supposed to do is approach what you’ve been avoiding, in graded doses, with or without medication softening the volume, which is what teaches the nervous system that the thing is survivable. Insight helps you understand the loop, exposure is what unwinds it.

Lifestyle stuff matters more than people want to believe. Sleep deprivation cranks anxiety. Three cups of coffee at noon will give a healthy person measurable anxiety symptoms by 2pm. Alcohol in the evening sabotages REM sleep and creates a rebound anxiety spike the next afternoon that patients almost never connect to the wine the night before. Stop drinking for two weeks and see what your baseline actually is before you decide your anxiety is a chemistry problem.

What’s nice to hear about the boring approach

Lead with the unpleasant stuff because that’s the field’s default and reverse it now. The patients who do the work walk out feeling a hell of a lot better, and not in a journaling-and-bath-bombs way. The first time the panic attack doesn’t show up at the airport gate, after eight years of it always showing up at the airport gate, lands in a way nothing else in the work has landed. The first time you sit through an OCD obsession without doing the count and notice ninety minutes later that nothing happened. The first time you order at the counter and don’t think about it afterward. That stuff is what makes the slog of the homework feel like it was worth it, and the patients who go all in on the protocol are mostly delighted at the back end. The field doesn’t lead with this because “do twelve weeks of unpleasant exposure exercises” doesn’t sell, but the back-end relief is huge.

Where I land on the medication question, and where you land is yours

I’ll say my view and then back off it. My personal lean on anxiety is that for most of the people who walk in, an SSRI plus CBT with the homework piece does the heavy lifting and they don’t need more than that. A bunch of patients end up doing the work and tapering off medication after a year or two, plenty of others stay on at a stable dose because it’s making their lives appreciably better and there’s no good reason to stop. Both outcomes are fine.

What I won’t do is gatekeep. If you want to start with medication, you get medication. I’m a provider, not a parent, and I hardly ever say no. The most I’ll do is a disapproving yes where you walk out with the script and a clear understanding of what I’d watch for. The same goes the other way, if you don’t want medication and you want to try the therapy and lifestyle pieces first, that’s also a real option, the data on CBT-alone for anxiety is good enough that it isn’t a worse pathway, it’s just a slower one. Either way, the appointment isn’t mine, it’s yours.

The patients who really get nowhere are the ones who half-do one piece for a month and then declare treatment didn’t work for them. An SSRI for three weeks at sub-therapeutic dose isn’t a trial. Four CBT sessions where you didn’t do the homework isn’t a course of CBT. If you’re going to test it, test it.

Sources

1. Bandelow B, Reitt M, Röver C, et al., Efficacy of treatments for anxiety disorders: a meta-analysis, Int Clin Psychopharmacol, 2015;30(4):183-192. PMID 25932596.
2. Carpenter JK, Andrews LA, Witcraft SM, et al., Cognitive behavioral therapy for anxiety and related disorders: A meta-analysis of randomized placebo-controlled trials, Depress Anxiety, 2018;35(6):502-514. PMID 29451967.
3. Slee A, Nazareth I, Bondaronek P, et al., Pharmacological treatments for generalised anxiety disorder: a systematic review and network meta-analysis, Lancet, 2019;393(10173):768-777. PMID 30712879.
4. Kopcalic K, Arcaro J, Pinto A, et al., Antidepressants versus placebo for generalised anxiety disorder (GAD), Cochrane Database Syst Rev, 2025;1(1):CD012942. PMID 39880377.