Anxiety medication vs therapy: a GP's honest take

Somewhere in the first ten minutes of an anxiety appointment, most people ask me a version of the same question. Do I need tablets, or do I need to talk to someone? Usually they have already decided which answer they want. Some would like a prescription and feel embarrassed about wanting one. Others have decided medication is a failure of willpower, and hope I will agree.

My honest answer is duller than both, and I think more freeing. For the commonest kind of anxiety, the national guideline says the two work about as well as each other, and the deciding vote is yours.

The short version

• For generalised anxiety disorder, NICE says to offer either therapy or medication and to base the choice on your preference, because there is no evidence either is better.
• Antidepressants do work. In the largest review to date they clearly beat placebo: for every 7 people treated, one gets a good response they would not otherwise have had.
• They also have a cost. Around 1 in 17 people stops because of side effects, and coming off later needs a slow, planned taper.
• It depends which anxiety you have. For social anxiety, therapy is the first recommendation, not a coin toss.

The line in the guideline that nobody quotes

Anxiety care in England runs on stepped care: the least intrusive thing likely to work, first. Education and monitoring, then guided self-help, then, if that is not enough, step 3.

At step 3 the guidance offers a choice of a high-intensity psychological intervention or a drug treatment. Then it says something surprisingly plain: base the choice on the person's preference, because there is no evidence that either mode of treatment is better.

Two treatments, roughly equal odds, and you are the tie-breaker. That is not a fudge. It is the evidence.

That line takes the moral weight out of a decision people load with far too much of it. Choosing sertraline is not giving up. Choosing CBT is not being precious. They are two reasonable routes to the same place.

What medication actually does

If you choose medication for generalised anxiety, you will usually be offered an SSRI, often sertraline first, largely because it is the most cost-effective option. Worth knowing: for some SSRIs this is off-label use, so your GP should tell you and record your consent.

The evidence is good. A 2025 Cochrane review pooled 37 trials and 12,226 adults and found antidepressants clearly better than placebo at producing a treatment response, with high certainty. The number needed to treat was 7: for every 7 people treated, one gets a good response who would not have got one on a dummy tablet. Real and useful. Also not magic.

Timing matters. Antidepressants take one to two weeks to start doing anything and up to eight weeks to work fully, so the first fortnight is a bad time to judge them. If it works, NICE advises staying on it for at least a year, because relapse is common if you stop early.

The parts of the leaflet people skim

Three things I always say out loud.

First, SSRIs can make anxiety worse before they make it better. NICE warns explicitly about activation: more anxiety, agitation and disturbed sleep in the early days. If nobody tells you, that feels like proof the drug is wrong for you, when often it is just the drug starting.

Second, if you are under 30, these medicines carry a small increased risk of suicidal thinking and self-harm in a minority of people. That is why the guideline says you should be seen within a week of the first prescription and monitored weekly for the first month. If that has not been offered, ask for it.

Third, side effects are the main reason people stop: in the same review, about 1 in 17 discontinued because of unwanted effects. When you do come off, taper gradually with your prescriber over weeks or months. Stopping suddenly is what produces the dizziness, the nausea and the "brain zaps" people describe. That is withdrawal, not failure, and it is largely avoidable.

One thing that should not be on the table: benzodiazepines such as diazepam. NICE says not to use them for generalised anxiety except briefly in a crisis, and not for panic disorder at all, because long-term outcomes are worse.

What therapy actually does

If you choose therapy, CBT for generalised anxiety is typically 12 to 15 weekly sessions of an hour, and applied relaxation is offered on equal terms. For which model suits whom, we went deep on that in CBT vs ACT.

The trade is straightforward. Therapy asks a lot up front: time, homework, and the discomfort of approaching what you have been avoiding. Medication asks little up front and more at the other end, when it is time to stop. What therapy gives back is that the skills stay yours.

It depends which anxiety you have

This is the bit most articles flatten. The free choice at step 3 applies to generalised anxiety disorder. Social anxiety is different: NICE recommends individual CBT developed specifically for it as the first-line offer, and reserves an SSRI (escitalopram or sertraline) for people who decline. So if your anxiety is mainly social, the coin is not evenly weighted, and you can ask for therapy first. If you are not sure what you are dealing with, stress or anxiety is a reasonable place to start.

Can you do both?

Sometimes. If medication has helped only partially, NICE suggests considering therapy alongside it. But for complex cases the guideline notes that the evidence for combining treatments is lacking, and side effects are likelier when you stack things. Both together is a reasonable next step, not an obvious upgrade.

What to do with this

Book the appointment and say which way you are leaning, and why. That one sentence saves ten minutes. If you want therapy, you need not wait for me: in England, if you are 18 or over you can refer yourself directly to NHS talking therapies, no GP referral needed. If you are unsure whether it is time to see anyone, when to see your GP about anxiety sets out the red flags.

The honest snag is the gap between what the guideline offers and what you can actually get this month. Twelve to fifteen weekly hours is a big ask, and the wait for them can be long. Short, focused sessions are one way to start moving in the meantime, which is why brightloaf sessions are 20 minutes rather than fifty. Mel makes that case, with the research, in Can 20 minutes of therapy actually help?. It will not settle the medication question for you, and it is not the right container for a complex, long-standing presentation. But it beats spending the wait doing nothing.

None of this is a substitute for personal medical advice about your own history. What I would like you to take away is smaller than a decision. It is permission to make one, without treating it as a verdict on your character.

If you are in crisis, or thinking about harming yourself, please call 999 or go to A&E, or call Samaritans free on 116 123 at any hour. brightloaf is not a crisis service.