Lithium, Valproate, or an Atypical: First-Line Maintenance for Bipolar I
A single patient, three days from his first manic episode. No guideline ranks lithium, valproate, or an atypical antipsychotic above the others for maintenance — the choice turns on which risk profile fits this particular 24-year-old, not on which drug is theoretically strongest.
D.M., a 24-year-old man who works as a line cook and shares an apartment with two roommates, is being discharged today after a nine-day hospitalization for his first manic episode. He'd gone nearly two weeks on three or four hours of sleep before his roommates called an ambulance, convinced he could cover his restaurant's entire supply order out of his own checking account and annoyed that anyone doubted him. He has no psychiatric history before this, no substance use his family or roommates are aware of, and no other medical problems — a genuinely clean baseline for a 24-year-old, consistent with a first presentation this age. His mother has bipolar I disorder herself, diagnosed in her twenties, and has done well for over a decade on valproate; she is the one who recognized what was happening and got him to the hospital.
The acute episode is resolved — he has been euthymic on olanzapine and lorazepam for the last four days and both are being tapered off before discharge. What remains is the maintenance question, and it does not have a clean answer. Lithium carries an evidence base for reducing suicide and completed self-harm that no other mood stabilizer has replicated, but its therapeutic window is narrow enough that toxicity is a real risk if he misses labs or gets dehydrated on a hot line during a dinner rush. Valproate is easier to dose and, if his mother's response predicts anything about his own pharmacogenomic makeup, may work unusually well for him specifically — but it carries its own hepatic and pancreatic monitoring burden. An atypical antipsychotic would cover him for both poles of the illness with one prescription, which matters because his most likely future episode is depressive, not manic — but it would mean starting a 24-year-old on a drug class whose metabolic cost compounds over decades of exposure, not weeks.
At discharge planning
Lithium is the only one of the three with trial evidence specifically for reduced suicide mortality, and he is 24 with a fifty-plus-year exposure horizon ahead of him — that argues for starting with the agent that has the strongest long-term relapse-prevention data, not the one that is easiest to start this week.
I recognize the monitoring burden is real. My answer to that is logistics, not a reason to avoid the drug — a standing lab order tied to his existing work schedule solves more of this than switching agents does.
I would weight his mother's response more heavily than a population-level comparison between agents. Family response to a specific mood stabilizer is an imperfect but real signal, and it points to valproate — plus he will have someone in his own household who has already lived with this drug's routine and can help him stay on it.
The suicide-mortality data is real, but it is drawn overwhelmingly from cohorts followed for years on a stable maintenance dose. It says less about a 24-year-old three days out from his first manic episode who has never had to remember a medication refill in his life.
He does not currently have an established primary care relationship, which is exactly where lithium's labs and valproate's labs both tend to fail in practice — not from the drug, from the follow-up. An atypical would let us defer that infrastructure question by a few months while he settles into outpatient care, at the cost of starting a metabolic-risk drug he may not end up needing long-term.
I am not proposing this as the theoretically best choice — I think it is the choice most likely to still be happening in six months, which is not nothing.
Divalproex sodium ER started at 500mg twice daily, with a plan to check a trough level and hepatic panel in two weeks and titrate toward a therapeutic range. Lithium was not ruled out permanently — the team documented it explicitly as the first drug to revisit if valproate is poorly tolerated or if he establishes a primary care relationship stable enough to support the monitoring lithium requires.
The disagreement was not fully resolved, only deferred: the pharmacologist's view that lithium's outcome data should have carried more weight regardless of logistics was noted in his chart as a standing minority opinion, to be revisited at his three-month follow-up rather than argued to a forced consensus today.