Lamotrigine Is Not a Diagnosis

Editor’s note: This piece is commentary, not a sourced brief. Unlike the standing series, it is not built on adjudicated citations or a companion worksheet. It describes an off-label prescribing pattern for which the closest large randomized trial was negative. It is my clinical opinion, informed by the literature where the literature exists and by pattern recognition where it does not. I have tried to say plainly which is which.

There is a patient I see every week, under different names.

She describes moods that turn on a dime. A text goes unanswered and the afternoon collapses. A minor criticism at work produces two days of misery. She is irritable in ways she hates, tearful in ways she can't predict, and fine, genuinely, fully fine in between. She has usually tried two or three SSRIs. One did nothing. One made her "wired and weird." She has never had a manic episode, never a sustained period of decreased need for sleep accompanied by increased energy or activity, never anything a careful history can shape into mania or hypomania.

And, increasingly, she arrives with a diagnosis already in hand: "I have ADHD."

She is a fictional composite, not a patient case.

I want to make two arguments about her. The first is about the label. The second is about what I actually do, which is, more often than my residency training would have predicted, prescribe lamotrigine. Both arguments are contestable. That is why this is an editor's note and not a brief. They are also separate arguments: nothing in the first one tells anyone to prescribe anything.

Readers of this publication will recognize the move. "I have ADHD" is an utterance about status. The acts it can displace are the differential itself: the longitudinal history, the collateral, the question of whether anything was present in childhood, and the inventory of what actually happened on each prior antidepressant.

Let me be fair to the patient, because the easy version of this complaint, the eye-roll at self-diagnosis, is both ungenerous and clinically lazy. The patient is not wrong that something is disordered. Emotional dysregulation is a real, well-documented, and under-recognized feature of adult ADHD. The popular discourse around rejection sensitivity has handed patients a vocabulary that maps closely onto their experience, and our own taxonomy has given them almost nothing else to reach for. The DSM has no comfortable home for affective instability that does not reach bipolar disorder and does not sit squarely within borderline personality disorder. Into that vacuum rushed the one culturally available word. The failure is upstream of the patient.

But a vocabulary is not a mechanism, and a self-report screener is not a differential. One question I find clinically useful, though not diagnostic, is whether the dysregulation is predominantly cue-driven and interpersonal, predominantly embedded in a lifelong executive-attentional syndrome, or both. "Or both" is doing real work in that sentence. Emotional lability is among the strongest emotional dysregulation dimensions associated with adult ADHD, and ADHD-related dysregulation can itself be triggered, interpersonal, and storm-shaped. The question organizes a history. It does not settle one.

What settles nothing at all is the current label lottery. The same surface presentation, including reactivity, rejection sensitivity, irritability, and poor frustration tolerance, collects at least three competing labels depending on who is doing the labeling: the patient says ADHD, the spectrum-minded psychiatrist says soft bipolarity, and the criteria-faithful colleague says borderline traits. Those are hardly the only possibilities. Trauma, anxiety and depressive illness, hormonal patterning, substance effects, sleep, autism, and other neurodevelopmental conditions belong in the room too. But they are the three labels I most often watch compete for this particular presentation.

The label chosen too often determines whether this person walks out with a stimulant, a mood stabilizer, or a therapy referral. When one phenotype yields three treatments, the honest conclusion is that the taxonomy is failing, not that two of the three clinicians are fools.

Nothing in that differential tells me to prescribe lamotrigine. What follows is a separate claim.

Here is the opinion: for the patient with mood reactivity that does not meet criteria for bipolar disorder, particularly when it is cue-driven, SSRI-unresponsive or SSRI-worsened, and storming through her life, I like lamotrigine. I reach for it earlier than the evidence would justify as a general recommendation. I have watched it work often enough that I no longer consider it a last resort, and I suspect I am not the only psychiatrist doing this, though I have no evidence for how common the practice is. That silence is part of why I am writing this.

The case for it is comparative, not absolute. Lamotrigine is generally weight-neutral. It is usually less sedating than the alternatives I would otherwise consider for mood instability, though somnolence occurs. It does not usually carry the same sexual-side-effect burden I expect from serotonergic antidepressants, and it does not bring the routine metabolic-monitoring burden of an atypical antipsychotic. For a population that has often already been punished by medication side effects, that profile is not a footnote. It is half the argument.

The other half is the clinical signal, and I will describe it while being honest about what it is worth. Patients who respond describe it less as feeling medicated than as the storms getting smaller. The trigger still registers. The disappointment still hurts. The day is no longer automatically forfeited.

That impression is why I keep reaching for the drug. It is not a reason anyone else should believe the drug works. Repeated clinical impressions generate hypotheses; they also generate highly polished errors. Lamotrigine's slow titration gives regression to the mean, expectancy, concurrent therapy, and the ordinary passage of six weeks every opportunity to impersonate a medication response.

Now the caveats, at full strength, because a case that cannot survive its own caveats was never a case.

This use is off-label, and "approved" is a status word too. Lamotrigine's FDA approvals are maintenance treatment of bipolar I disorder and epilepsy. Full stop. Nothing I describe here carries an indication, and I tell patients this in those words.

The best trial in the adjacent population was negative, and it was not small. The hypothesis was once plausible on trial evidence. A 12-week placebo-controlled study of 28 patients with borderline personality disorder, designed specifically around affective instability and excluding bipolar disorder, reported benefit (Reich et al., 2009). It was small enough to generate a hypothesis rather than settle one.

LABILE was the much larger test. That study (Crawford et al., American Journal of Psychiatry, 2018) randomized 276 adults with borderline personality disorder, a population that, like mine, explicitly excluded bipolar disorder, to placebo or lamotrigine. Dosing was generally up to 200 mg daily and up to 400 mg for participants taking combined oral contraceptives. At 52 weeks, the mean ZAN-BPD scores were essentially identical: 11.3 versus 11.5, an adjusted difference of 0.1. No statistically significant differences in secondary outcomes were seen at any time point.

Adherence was poor. Thirty-six percent of the lamotrigine group was still taking it at a year. That is a real limitation but also a convenient one for me to cite, so I hold it loosely. My honest reconciliation is that the trial enrolled by categorical BPD diagnosis rather than by the narrower affective-reactivity phenotype I select for.

But that distinction can become too convenient, so let me say the quiet part: I do not have evidence that selecting for cue-driven affective instability identifies a lamotrigine-responsive subgroup. If such evidence existed, it would materially change this argument. That is my clinical hypothesis, not an established finding.

Readers should weight LABILE heavily. I do. It is why this is a judgment call and not a recommendation.

Six weeks of titration is also six weeks of time. Nobody gets to a target dose quickly, and a great deal of affective misery improves over six weeks regardless of what is prescribed. Every apparent response carries this confound.

The rash counseling follows the label, not my intuitions. Serious rash, including Stevens-Johnson syndrome, was uncommon in the adult mood-disorder trial data, on the order of one per thousand. The label states that risk may be increased by valproate coadministration, exceeding the recommended initial dose, or exceeding the recommended escalation schedule. It describes these as suggestions yet to be proven. Nearly all life-threatening cases have occurred in the first two to eight weeks, but later cases occur. The instruction patients receive is therefore the label's, not mine: stop the drug and call at the first sign of rash, whenever it appears, unless it is clearly unrelated.

The predictable price is anxious phone calls about unrelated skin findings. I consider that a fair price.

A pill can become a way of not doing the other work. The cue-driven, interpersonal character of this dysregulation is precisely the argument for skills-based therapy, and the fair objection is that lamotrigine medicates what needs DBT. I concede most of it.

My answer is sequencing, not substitution. Many of these patients cannot yet use therapy well while the storms are at their worst. My working hypothesis, and I emphasize hypothesis, is that a modest pharmacological floor makes the skills learnable rather than replacing them. Therapy is raised at the start, re-raised at the checkpoint, and the chart says so.

There is a maneuver adjacent to this practice that I have seen and consider corrosive, so I want to name it and refuse it.

I do not prescribe lamotrigine because I have secretly diagnosed bipolar disorder. I do not relabel the patient as bipolar to make the prescription look conventional, convert "rule out bipolar disorder" into a permanent diagnosis because a mood stabilizer was selected, or tell the patient she is "on the spectrum" when the history does not support it.

A diagnosis should explain the history; it should not be reverse-engineered from the medication.

The patient does not need a fictional bipolar diagnosis to receive an honest off-label trial. What she needs is a clinician willing to say what the trial is, what evidence is missing, and how failure will be recognized. That conversation does not make weak evidence strong. It makes the decision legible to the patient and to whoever reads the chart after me.

One more refusal, this time of a shortcut I find tempting myself.

It would be tidy to treat the medication sequence as a diagnostic assay: stimulant sharpened the irritability, therefore it wasn't ADHD; lamotrigine quieted the storms, therefore the attention complaint was weather all along.

Neither inference holds.

A randomized adult trial showed that stimulant treatment can improve emotional dysregulation in established ADHD, not just worsen it, so an affective response in either direction proves nothing about the underlying syndrome.

Medication response is data. It is not a diagnostic test.

What sequencing can honestly do is reduce noise. Where affective reactivity dominates the presentation, I sometimes address it first and then return to the question of what attentional impairment remains. Some patients, storms quieted, still cannot organize a morning or finish a page. Those patients get a proper ADHD workup and, where warranted, a stimulant, from me, without argument.

Others no longer describe attention as the problem they came in with.

Neither outcome, by itself, diagnoses anything. No patient has to fail lamotrigine to earn an ADHD diagnosis, and no patient loses one because the storms improved.

Opinion without stopping rules is just enthusiasm. Off-label prescribing without stopping rules is maintenance treatment by inertia. Mine:

These rules have external company. In the adjacent borderline-personality-disorder literature, the current APA practice guideline recommends that any psychotropic treatment be time-limited, directed at a specific measurable target, and adjunctive to psychotherapy. It also finds no pharmacotherapy established for the disorder's core symptoms.

That does not validate the drug choice described here. It is a way to contain an uncertain trial, and I borrow it.

I am not asking readers to prescribe lamotrigine the way I do. I am asking for two smaller things.

First, treat "I have ADHD" as the beginning of a differential rather than the end of one, with the generosity of recognizing that the patient reached for the only word we left on the table.

Second, when the differential lands on the phenotype our taxonomy does not comfortably name, say out loud what you actually do about it: the formulation, the evidence against it, and the rules that would make you stop.

A symptom cluster is not a diagnosis, and my treatment preference is not evidence that my formulation is correct. Neither a patient's favored explanation nor a clinician's favored prescription should be allowed to complete the differential.

The honest response is not to hide that tension. It is to practice inside it, in public.

This note is my attempt to go first.

Consistent with the genre, these are informing sources rather than adjudicated citations. Each was opened before publication.

Conflicts: none relevant. I have no financial relationship with a manufacturer of lamotrigine or any competing product. A sufficiently powered trial enrolling patients on the transdiagnostic, non-bipolar affective-reactivity phenotype described here, rather than on categorical borderline personality disorder, has not, to my knowledge, been done.

Educational disclaimer: The Psychiatric Record provides general educational information for psychiatric and mental-health professionals. This piece is clinical commentary describing an off-label prescribing pattern. It does not constitute medical advice, does not establish a standard of care, is not individualized treatment advice, and does not recommend lamotrigine, stimulant treatment, or any other intervention for a particular patient.