← Back
NeuroTrials.ai
Neurology Clinical Trial Database

MESS

Immediate Versus Deferred Antiepileptic Drug Treatment for Early Epilepsy and Single Seizures: A Randomised Controlled Trial

Year of Publication: 2005

Authors: Marson A, Jacoby A, Johnson A, ..., Chadwick D; Medical Research Council MESS Study Group

Journal: The Lancet

Citation: Marson A et al. Lancet. 2005;365(9476):2007-2013. DOI: 10.1016/S0140-6736(05)66694-9

Link: https://pubmed.ncbi.nlm.nih.gov/15950714/


Clinical Question

Does immediate antiepileptic drug treatment after a first or early seizures improve long-term seizure outcomes compared to deferring treatment until the patient and clinician agree it is necessary?

Bottom Line

Immediate AED treatment significantly reduced time to first seizure (HR 1.4, p<0.0001) and time to first tonic-clonic seizure (HR 1.5, p<0.0001) compared to deferred treatment. However, this early advantage disappeared by 4-6 years, with no difference in long-term 2-year remission rates (95% vs 96% at 8 years) or seizure-free periods between years 3-5 (76% vs 77%). Immediate treatment increased adverse events (39% vs 31%). These findings support shared decision-making after first seizure rather than mandatory immediate treatment.

Major Points

  • Immediate treatment after first seizure reduces early recurrence (51% vs 39% seizure-free at 2 years) but NOT long-term remission.
  • By 5 years: no significant difference in seizure-free rates between immediate and deferred treatment.
  • 1443 patients randomized. UK multicenter, pragmatic. Published Lancet 2005 (Marson et al.).
  • Immediate: start AED after first/second seizure. Deferred: delay treatment until patient/physician agree.
  • Quality of life: no significant difference at 2 years between groups.
  • Driving: immediate treatment group had earlier return to driving eligibility.
  • Key message: treating after first seizure prevents early recurrence but doesn't alter natural history.
  • Supports individualized approach — not all patients need treatment after a single seizure.
  • Influenced AAN practice parameter on when to start AEDs.
  • Established that antiseizure drugs are symptomatic, not disease-modifying.

Design

Study Type: Pragmatic, multicenter, unmasked (open-label), randomized controlled trial

Randomization: 1

Blinding: Open-label (unmasked)

Enrollment Period: January 1, 1993 to December 31, 2000

Follow-up Duration: Up to 8 years (median not specified)

Centers: Multiple (50% UK, 50% non-UK)

Countries: United Kingdom, Multiple international centers

Sample Size: 1443

Analysis: Intention-to-treat; minimisation randomization balanced by center and number of seizures; target 1,500 (originally 3,000)


Inclusion Criteria

  • Age at least 1 month
  • Adequately documented history of one or more clinically definite, spontaneous, unprovoked epileptic seizures (excluding febrile convulsions or acute symptomatic seizures)
  • Both the clinician AND the patient were in equipoise (uncertain whether to proceed with treatment)

Exclusion Criteria

  • Already treated with antiepileptic drugs (other than short-acting drug for serial seizures/status, or previous prophylactic treatment for acute symptomatic seizures)
  • Progressive disease

Baseline Characteristics

CharacteristicImmediate TreatmentDeferred Treatment
N722721
Male403 (56%)423 (58%)
Age - Median (IQR)23 (17-42) years26 (16-46) years
UK center363 (50%)354 (49%)
Developmental delay34 (5%)23 (3%)
Neurological deficit52 (7%)40 (6%)
Previous febrile seizures53 (7%)52 (7%)
Family history of seizures76 (11%)86 (12%)
EEG focal abnormality184 (25%)200 (28%)
EEG generalized abnormality131 (18%)105 (15%)
Imaging abnormal71 (10%)69 (10%)
Tonic-clonic seizures375 (52%)406 (56%)
Single seizure before randomization404 (56%)408 (57%)
During sleep only140 (19%)125 (17%)

Arms

FieldImmediate TreatmentControl
InterventionClinician-selected AED started immediately. Drugs chosen: carbamazepine 328 (46%), valproate 325 (46%), phenytoin 25 (3%), lamotrigine 19 (3%). Doses: CBZ >=16yr: 400 mg/day (IQR 300-600); VPA >=16yr: 900 mg/day (IQR 500-1000)No drugs until clinician and patient agreed treatment necessary. 332 (46%) eventually started AEDs: CBZ 134 (40%), VPA 142 (43%), PHT 20 (6%), LTG 17 (5%)
DurationContinuous; at 5 years, 60% still receiving AEDsDeferred; at 5 years, 41% had started treatment

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Time to first seizure (all types)PrimaryDeferred: 53% had a seizure (382/721)Immediate: 43% had a seizure (311/722)1.4 (deferred vs immediate)<0.0001
Time to first seizure at 2 years | Result: Immediate 37% vs Deferred 48% (difference 11%, 95% CI 6.2-16.7)Secondary
Time to first seizure at 8 years | Result: Immediate 52% vs Deferred 61% (difference 9%, 95% CI 2.6-15.3)Secondary
Time to first tonic-clonic seizure | Result: HR 1.5 (95% CI 1.2-1.8), p<0.0001; at 2 years: 27% vs 38%Secondary
Time to 2nd seizure | Result: HR 1.3 (95% CI 1.1-1.6), p=0.0025; at 2 years: 24% vs 32%Secondary
Time to 5th seizure | Result: HR 1.2 (95% CI 0.9-1.5), p=0.23 -- NOT significantSecondary
2-year remission at 2 years | Result: 64% immediate vs 52% deferred (p=0.023); difference disappeared by 5-8 years (95% vs 96%)Secondary
Seizure-free between years 3-5 | Result: 76% immediate vs 77% deferred (difference -0.2%, 95% CI -5.8 to 5.5) -- virtually identicalSecondary
Quality of life (change in anxiety, depression, mastery scores) | Result: No significant differences at 2 yearsSecondary
Employment at 2 years | Result: 47% immediate vs 46% deferred -- no differenceSecondary
Any adverse event reportedAdverse214 (31%); difference 8.6% (95% CI 3.6-13.6%)270 (39%)
Depression, anxietyAdverse3140
Dizziness, unsteadinessAdverse3237
GI symptomsAdverse2441
Tiredness, drowsinessAdverse2341
HeadacheAdverse1337
Rash, acneAdverse1431
TremorAdverse617
Deaths during follow-upAdverse23 (6 sudden unexplained deaths: 4 immediate, 2 deferred)31

Subgroup Analysis

Single seizure patients: lower recurrence, NNT 14 at 2 years; Multiple seizure patients: higher recurrence, NNT 5 at 2 years; both subgroups showed long-term remission convergence


Criticisms

  • Open-label (unmasked) design could bias seizure reporting
  • Equipoise-based enrollment selects population with lower seizure burden, may not generalize to high-risk patients
  • No placebo control: 46% of deferred group eventually started treatment, creating significant contamination
  • Predominantly older AEDs (92% CBZ or VPA): may not generalize to newer AEDs with better tolerability
  • QOL assessment limited to UK adults only (527 of 1,443 patients)
  • Under-recruitment: only 48% of planned 3,000 patients enrolled
  • Self-reported seizure outcomes inherently imprecise, subject to recall bias

Funding

UK Medical Research Council (non-industry)

Based on: MESS (The Lancet, 2005)

Authors: Marson A, Jacoby A, Johnson A, ..., Chadwick D; Medical Research Council MESS Study Group

Citation: Marson A et al. Lancet. 2005;365(9476):2007-2013. DOI: 10.1016/S0140-6736(05)66694-9

Content summarized and formatted by NeuroTrials.ai.