Read this topic

Most psychiatric drugs are prescribed on dose alone, with no need to check blood levels. Therapeutic drug monitoring (measuring a drug's concentration in blood to guide dosing or check safety) is reserved for a small group of drugs where the effective dose and the toxic dose sit close together, a property called a narrow therapeutic index.

For these drugs, the prescribed dose alone tells you far less than the actual blood level, because how each patient absorbs, metabolises, and clears the drug varies. The main drugs requiring this kind of monitoring in psychiatry are lithium, clozapine, sodium valproate, and carbamazepine, each for a different underlying reason.

For example, imagine two patients on an identical lithium dose: one has a normal level and only a mild tremor, while the other has crept into toxicity with confusion and a coarse tremor. The dose looks the same on paper, but the level tells the real clinical story: never assume a "standard" dose is safe for a narrow-therapeutic-index drug without checking the actual level.

The reason timing matters so much for these levels is straightforward: a drug's concentration rises after each dose and falls until the next one is due. A level taken too soon after dosing catches the peak instead of the true baseline, and falsely overestimates the patient's real steady-state exposure. This is why levels for these drugs are sampled as a trough (the lowest concentration, measured just before the next dose is due) rather than at a random time: a poorly timed sample can look dangerously high or reassuringly low without the dose ever having changed.

Lithium levels are sampled as a strict trough, 12 hours after the last dose. Because lithium clearance depends heavily on kidney function and hydration, levels must be rechecked whenever dose, kidney function, or fluid balance changes. Clozapine levels are used differently: rather than guiding routine dose titration, they mainly confirm adherence, aid investigation of suspected toxicity, and help interpret a specific drug interaction with smoking. This plasma-level monitoring is separate from, and additional to, the mandatory full blood count (FBC, a blood test counting red cells, white cells, and platelets) monitoring required throughout clozapine treatment to screen for agranulocytosis (a dangerous drop in a specific type of white blood cell). Sodium valproate has a quoted reference range, but that range correlates poorly with its mood-stabilising efficacy in psychiatric use, so valproate monitoring is used chiefly to confirm adherence and screen for toxicity, not to fine-tune dosing for effect. Carbamazepine sits in a similar position, with levels used mainly for toxicity and interaction management rather than as a precise efficacy target.

It is easy to assume all four of these drugs' levels work the same way: they don't. Lithium is the odd one out: its level is used to guide dosing directly, because toxicity reliably emerges above roughly 1.5 mmol/L and the therapeutic window is genuinely narrow. Clozapine, valproate, and carbamazepine, by contrast, have levels used mainly to confirm adherence or catch toxicity rather than to hit a precise efficacy target: a valproate level sitting comfortably within the quoted reference range tells you nothing about whether a patient's mood is actually stable. A second look-alike trap is the direction of the clozapine-smoking interaction. Smoking status directly affects clozapine levels through CYP1A2 (a liver enzyme that metabolises clozapine): smoking induces CYP1A2, increasing clozapine clearance and lowering the plasma level, while stopping smoking removes that induction and raises the level again, sometimes into a toxic range, even though the prescribed dose has not changed. Resuming smoking has the opposite effect. Candidates frequently get this direction backwards under exam pressure, so it is worth deliberately memorising which way each change goes.

Diagnosing lithium toxicity relies on the combination of clinical signs: coarse tremor, ataxia (loss of coordinated movement), and confusion, plus a level above the roughly 1.5 mmol/L threshold, not on the level in isolation; these toxic features are distinct from the fine tremor and mild gastrointestinal upset expected at a normal therapeutic level. For clozapine, a level well above what a patient's dose would normally produce is a signal to investigate: possibly toxicity, possibly the smoking interaction, possibly non-adherence followed by a re-titrated restart, rather than a fixed target range to chase the way lithium's is.

Consider a patient established on clozapine who is admitted to a smoke-free ward and stops smoking abruptly. Over the following days, loss of CYP1A2 induction raises their clozapine level, and a check reveals a level above the seizure-risk threshold. The correct management response is a dose reduction and closer monitoring, not an assumption that the patient has taken extra tablets. For lithium, management of a rising or toxic level means addressing the underlying driver: correcting dehydration, reviewing interacting drugs, or reducing the dose, and rechecking the level after any change to kidney function or fluid balance. The BNF and Maudsley Prescribing Guidelines set out the reference ranges and monitoring intervals for each of these drugs, and both emphasise that a trough sample, correctly timed, is what makes any of these levels clinically meaningful.

If a vignette gives you a clozapine patient whose level has unexpectedly risen with no dose change, check for a recent change in smoking status before assuming non-adherence or toxicity from another cause: stopping smoking on a smoke-free ward is one of the most commonly tested triggers in this whole topic.

Recognising which drug's level guides dosing directly (lithium) and which drug's level is used mainly to confirm adherence or catch toxicity (clozapine, valproate, carbamazepine), and always reading that level as a correctly timed trough, is what turns a blood test into a genuinely useful clinical decision rather than a number on a page.