First-Episode Psychosis: Which Second-Generation Antipsychotic First-Line?
A 19-year-old with his first psychotic episode needs an antipsychotic started this week. Head-to-head trials have never shown one second-generation agent reliably outperforming another — so what is the decision actually resting on?
J.T., a 19-year-old man, withdrew from his second semester of community college six weeks ago after his roommates began noticing he was talking to someone who wasn't there. Before this, by his mother's account, he was an outgoing kid who played intramural soccer, worked weekends at a hardware store, and had never given anyone reason to worry — the kind of ordinary, unremarkable stability that makes the past six weeks feel to her like watching a stranger move into her son's life. He now spends most nights awake, convinced the university's wifi network is transmitting instructions meant only for him, and has stopped returning his roommates' calls.
His mother brought him in after he accused her of being replaced by an impostor, an accusation that visibly frightened her more than any prior symptom had. He has no prior psychiatric contact of any kind, and no substance use beyond occasional weekend drinking with friends before this began; family history hadn't come up in a chaotic intake focused mostly on getting him stabilized — by both their accounts otherwise, a genuinely healthy, previously unremarkable baseline with nothing in his history that would have predicted this.
There is no dispute that an antipsychotic is indicated; the dispute is which one. The large first-episode comparative trials — across olanzapine, risperidone, quetiapine, and aripiprazole — have not established that one agent reliably out-performs another on symptom response in this population, even where they have separated on other endpoints such as all-cause discontinuation. What the same body of trials has shown consistently is that first-episode patients are unusually sensitive to both extrapyramidal and metabolic effects at doses chronic patients tolerate without difficulty, likely reflecting a drug-naive, more sensitized D2 system. That reframes the question: with efficacy roughly equivalent across agents, the real first-line decision is being made almost entirely on which side-effect profile this particular patient can least afford, at the lowest effective starting dose rather than the dose a textbook lists for established illness.
Choosing the first agent
I'd start aripiprazole. The only large head-to-head trial we actually have in exactly this population — Robinson and colleagues' 2015 randomized comparison in Schizophrenia Bulletin, first-episode patients randomized to aripiprazole or risperidone and followed for twelve weeks — found no difference in positive-symptom response between them, but a real separation on the metabolic side: the aripiprazole arm came out ahead on total and LDL cholesterol, fasting glucose, and prolactin. He's nineteen. If he's on an antipsychotic for the next fifty years, that's not a rounding-error difference.
You're right about the metabolic numbers — that's the same trial I'd cite, and I'm not disputing that half of it. But the same study found the reverse on the other side of the ledger: aripiprazole produced significantly more akathisia than risperidone.
The mechanism is why it's not symmetric: aripiprazole is a partial D2 agonist, so even at the doses we'd use, it still produces some dopaminergic activation rather than clean blockade. Risperidone doesn't carry that same profile, which is a real part of why the trial found more restlessness on aripiprazole and not the reverse.
You called the metabolic gap "not a rounding-error difference" — I'd say the same about akathisia in someone who already believes the wifi is targeting him. Restlessness that reads to him as something being done to him, rather than a drug side effect, is a real path to him deciding on day four that we can't be trusted and stopping the medication himself. That's not a smaller risk than a lipid panel just because it shows up faster.
You're both citing the same trial, and I don't think it decides this on its own — no positive-symptom difference, and the metabolic and akathisia findings genuinely cut in opposite directions, which is exactly why the trial's own authors frame their conclusion by scenario rather than declaring a winner: risperidone when akathisia risk is the dominant concern, aripiprazole in most other situations. What actually tips it for me isn't the trial average, it's this patient specifically. His mother mentioned, once I asked directly, a maternal uncle with schizophrenia who was also diagnosed with diabetes within a few years of starting treatment. We don't know if that was illness-related or drug-related, but it's a specific, individual reason — not just the general first-episode argument — to weight the metabolic side for him in particular.
That doesn't mean dismissing the akathisia risk — it means managing it instead of trading it away for a different drug. We start well below standard dose, titrate slowly, and tell him and his mother in plain terms what early restlessness would mean, so if it happens, it registers as a medication effect to watch, not as one more thing being done to him.
Agreed: aripiprazole, started at a low first-episode dose with a slow titration schedule and metabolic baseline labs drawn before the first dose, not after.
Not fully settled: how long to hold at the starting dose before considering an increase if response is partial rather than absent.
The team's shared position, stated by the attending directly to J.T.'s mother: this choice reflects his specific risk profile, not a claim that aripiprazole is simply the better drug — a distinction worth keeping honest as the family researches the diagnosis online.