Clinical Psychopharmacology · Paper A
Mood stabiliser choice in renal impairment
Every mood stabiliser leaves the body through one of two exits: the kidney or the liver. Renal impairment only matters for the ones that use the kidney door. Lithium is cleared almost entirely by the kidney, so a falling eGFR raises its level at an unchanged dose. Valproate, lamotrigine and carbamazepine leave mainly through the liver, so they barely notice renal decline directly, but each carries its own separate renal-adjacent trap worth knowing by name. The whole decision is a severity ladder, not a switch: mild impairment means reduce and monitor, severe impairment (or dialysis) means genuinely switch.
A patient on lithium for fifteen years walks into clinic, and her routine bloods show her kidneys are failing. Nothing about her mood has changed. Everything about her prescription now has to be re-thought, because the drug that has kept her well leaves the body through the exact organ that is now struggling. This is the moment every psychiatrist eventually meets: choosing, or keeping, a mood stabiliser once the kidneys can no longer be trusted to clear it the way they used to. Get the reasoning right and a patient keeps the best drug for as long as it's safe; get it wrong and you either poison someone with a drug they didn't need to stop, or abandon a highly effective treatment too early.
A mood stabiliser is a long-term medicine that keeps someone with bipolar disorder from swinging back into or depression. Renal impairment simply means the kidneys are filtering blood less well than they should — measured with a number called eGFR (estimated glomerular filtration rate, a lab-calculated score for how many millilitres of blood the kidneys clean per minute). The four main mood stabilisers — lithium, sodium valproate, lamotrigine and carbamazepine — leave the body by two completely different exit routes: the kidney, or the liver. That single fact is the whole engine of this topic.
The mistake almost everyone makes first is treating renal impairment as a single on/off switch: kidneys are bad, so stop lithium. It isn't a switch, it's a dial. Mild-to-moderate impairment usually means keep lithium, just at a lower dose with closer monitoring, because nothing else prevents relapse and reduces suicide risk as well as lithium does. Only once impairment becomes severe, or the patient reaches end-stage renal failure without dialysis, does the balance genuinely tip towards switching to a drug that doesn't rely on the kidney at all. Read an exam stem for the actual severity described, not just the phrase renal impairment on its own.
Lithium is a small, simple ion, chemically closer to sodium than to any other psychiatric drug, and the body has no way to break it down. It leaves entirely unchanged, filtered out at the glomerulus (the tiny cup-shaped filter inside each of the kidney's million or so nephrons) and then partly reabsorbed back into the blood in the proximal tubule, riding along the same transporter the kidney uses to reclaim sodium. When someone is dehydrated or sodium-depleted, the kidney reabsorbs more of both, which is exactly why diarrhoea, vomiting, or a new diuretic can push a stable lithium level into toxicity. When eGFR itself falls, the whole system slows down, so a dose that was safe for years starts producing a higher steady-state level.
Valproate, lamotrigine and carbamazepine take an entirely different route out: the liver breaks each of them down chemically first. Valproate is cleared by beta-oxidation (a liver process that chops fatty-acid-like molecules into pieces small enough to excrete); lamotrigine and carbamazepine are cleared mainly by glucuronidation, where the liver bolts on a sugar-like tag that makes a drug water-soluble enough to remove. Because none of that chemistry happens in the kidney, their own barely notices a falling eGFR.
Here's a worked example to make the numbers concrete. Imagine a patient stable on lithium 800 mg at night for years, with a steady level of 0.7 mmol/L and an eGFR of 75 mL/min. Her eGFR drops to 55, still only mild impairment. Because the kidney is filtering roughly a quarter less blood per minute than before, the same 800 mg dose will settle at a noticeably higher level, maybe 0.9 to 1.0 mmol/L, dangerously close to the toxic range above about 1.2 mmol/L. The correct response is not to switch her drug; it's to drop the dose, say to 600 mg, and recheck the level and eGFR again in a few weeks, then every few months instead of annually.
Now imagine a second patient whose eGFR has fallen to 20 over several years, partly because lithium itself has scarred the kidney's tubules, a recognised long-term complication. At that point, dose-juggling stops being enough: the kidney's to clear lithium safely at any dose that still controls mood is disappearing, so guidance shifts to switching her onto valproate or lamotrigine instead. There is one further branch: a patient already established on regular haemodialysis is not automatically barred from lithium, because dialysis itself removes lithium efficiently. Under specialist supervision, a single dose can be timed right after each session, using post-dialysis levels to set the next one.
Several traps look like renal reasoning but aren't. Valproate isn't cleared by the kidney, but uraemia (the build-up of waste toxins in the blood once the kidneys fail to clear them) still matters to it indirectly: it lowers how tightly valproate sticks to albumin in the blood, so more of it circulates free and active even when the routine total level looks perfectly normal, a genuinely different mechanism from lithium's problem and easy to mix up with it.
Carbamazepine isn't renally cleared either, but it independently causes low sodium through an SIADH-like effect. SIADH stands for syndrome of inappropriate antidiuretic hormone, meaning the body holds onto water it shouldn't, which dilutes the blood's sodium; that matters more, not less, in someone whose kidneys are already under strain. And two anticonvulsants that sound like safe cousins of valproate are traps in the opposite direction: topiramate is itself renally cleared and causes kidney stones, and gabapentin, also renally cleared, simply doesn't work as a mood stabiliser at all, regardless of anyone's renal function.
Renal impairment itself is staged by eGFR, not by symptoms; a patient can feel entirely well with an eGFR of 25. Chronic kidney disease is generally graded from mild (eGFR 60 to 89, often with some other marker of kidney damage) through moderate (30 to 59) to severe (15 to 29) and established renal failure (below 15, or on dialysis). On the drug side, lithium toxicity is a clinical diagnosis backed by a level: coarse tremor, (unsteady, uncoordinated movement), confusion and vomiting alongside a level above the therapeutic range point to toxicity, and the trigger is very often a recent fall in effective renal , such as dehydration, a new ACE inhibitor or diuretic, or a genuine drop in eGFR, rather than a deliberate dose increase.
The actual plan follows the severity ladder above: mild-to-moderate impairment means reduce the lithium dose and monitor levels and renal function more often. NICE CG185, the national guideline on bipolar disorder, recommends checking renal function before starting lithium and at least every six months during treatment, more often if impairment is found. Severe impairment or progressive decline means switching to sodium valproate, lamotrigine or carbamazepine, chosen on their own separate merits and cautions, and dialysis-dependent patients are a specialist exception where lithium dosed after dialysis remains an option.
The BNF's renal-impairment tables and the Maudsley Prescribing Guidelines in Psychiatry both set out the practical dose-reduction bands and monitoring intervals doctors actually use at the bedside, and any switch away from lithium still has to weigh valproate's teratogenicity, lamotrigine's rash risk, or carbamazepine's interactions and hyponatraemia against the renal problem being solved.
When a stem gives you a falling eGFR and a normal-looking valproate level in a symptomatic patient, ask yourself: total level, or free level? The routine assay measures total valproate, and renal impairment, through uraemia's effect on , can hide a genuinely higher active exposure behind a normal number. Ask yourself a second retrieval question here: why would starting an ACE inhibitor in a patient already on lithium demand a lithium level check, even with no dose change and no new symptoms? Because it reduces renal lithium through a completely separate mechanism, stacking on top of whatever the patient's baseline eGFR is already doing.
Lithium leaves through the kidney and needs the dose turned down as renal function falls, only crossing over to a full switch once impairment is severe or dialysis-dependent, while valproate, lamotrigine and carbamazepine leave mainly through the liver but each carry their own separate renal-adjacent trap worth knowing by name.