Published by Unseen Progress, an independent publisher of caregiver research. Last reviewed 2026-05-10. Part of the epilepsy caregiver research overview.
Short answer. ASM side effects are "too much" when they meet either of two thresholds: (1) the impairment in daily functioning is comparable to or greater than the impairment the seizures themselves caused, or (2) the side-effect burden is producing secondary harms — academic decline, social withdrawal, depressed mood, refusal to take the medication — that compound across months. The literature consistently treats this as a legitimate clinical data point, not a soft preference, and recommends caregivers track side effects with the same structure as seizures (French et al., 2004; Wilner, 2008).
The AAN/AES treatment guidelines for new antiepileptic drugs (French et al., 2004) and the comparative tolerability data from large cohort studies converge on a few stable findings:
These are not edge cases. They are part of the expected profile of effective ASMs, and the clinical decision is rarely "any side effect = wrong drug" but rather "is the trade-off acceptable for this patient at this dose?"
A child who was having two seizures a month and is now having none but is also failing a grade, withdrawing from friends, or visibly miserable has not necessarily improved. The literature on epilepsy outcomes — particularly the long-term cohort work — emphasises that seizure control is a means, not an end. The end is functional life. A regimen that controls seizures while destroying school performance, social participation, or mental health is not a successful regimen.
Caregivers often feel they cannot raise side effects without seeming to undervalue seizure control. The research is explicit that this framing is wrong: tolerability is one of the four criteria for an adequate trial in the ILAE consensus (Kwan et al., 2010), and a drug that is intolerable has not been adequately tried — it has been intolerable. Reporting intolerability is part of clinical decision-making, not a complaint about clinical decision-making.
The structured way the literature recommends operationalising side effects is with daily ratings on the 2–4 dimensions most relevant to the patient. For most paediatric patients, these are:
The data this produces makes the neurology conversation specific. Instead of "he seems off," the conversation becomes "frequency is down 70% from baseline, but alertness is down two points and his English grade has dropped from B to D." That is a clinical data point a neurologist can act on. Most will adjust dose, switch drugs, or accept the trade-off based on family priorities — but they need the data to make that call.
Across the clinical guidance, certain side effects warrant prompt contact with the prescribing neurologist regardless of seizure benefit:
Distinguishing these from tolerable side effects is the clinical conversation. The point is not to abandon a drug at the first sign of any effect — most settle — but to treat persistent, functionally limiting effects as data the regimen needs to respond to.
Children and adults with epilepsy have higher baseline rates of attention, mood, and learning difficulties than the general population, independent of medication. When a child on a new ASM is struggling at school, the question of whether the struggle predates the drug, was caused by it, or reflects the underlying neurology is genuinely hard. The research-backed approach is to establish a pre-medication baseline on the same dimensions before starting the drug, so that subsequent changes have something to compare against. This is rarely possible at diagnosis (when starting medication is urgent), but it is possible before any subsequent dose change or switch.
Where a baseline is unavailable, the literature suggests using the temporal pattern: side effects tied to dose changes or new drug starts are usually drug-related; ones that pre-existed the change are usually not.
1. Pick two to four side-effect dimensions relevant to this patient (alertness, mood, school, appetite are common starting points) and rate each daily on a 1–5 anchored scale. 2. Distinguish initiation/titration effects (often resolve in 4–6 weeks) from steady-state effects (do not resolve and represent the long-term burden). 3. Bring the side-effect data to every neurology appointment with the same structure as the seizure data. Lead with the comparison: "frequency change = X, alertness change = Y, mood change = Z." 4. Treat any new suicidal ideation, rash on lamotrigine, or severe behavioural change as a same-day call to the prescriber — not a wait-for-the-next-appointment item. 5. If side effects are persistent and functionally limiting at maximum tolerated dose, raise the trade-off explicitly. The ILAE definition of an adequate trial includes tolerability; an intolerable drug is not a failed drug, it is an untried one, and switching is appropriate.
---
Unseen Progress publishes long-form caregiver research and builds research-backed daily trackers for the families covered. See the full epilepsy caregiver research overview for the complete framework.