Importance of competing risks in the analysis of anti-epileptic drug failure

Table 2 Classification of reasons for treatment change used in the SANAD trial

Reason for withdrawal from drug or other drug added (earliest event)	Categorised as event (type classified as ISC or UAE) or censored in 'time to treatment failure' analysis	Number of patients (CBZ, LTG, TPM and GBP)
Inadequate seizure control	Event (ISC)	348
Unacceptable adverse events	Event (UAE)	320
Perceived risk of adverse effect	Event (UAE)	8
Remission of epilepsy categorised by patient¹(less than 12 months' remission from seizures)	Event (UAE)	46
Remission of epilepsy categorised by clinician (any length) or patient (more than 12 months' remission from seizures)	Censored	88
Study withdrawal – consent withdrawn²	Censored	23
Death (unrelated to epilepsy)³	Censored	32
Death (related to epilepsy)³	Event (ISC)	3
Moved from area	Censored	0
Patient non-compliant or patient decision⁴	Censored	12
Pregnant or planning pregnancy	Censored	6
Other	Censored	12
Still on drug at last follow-up	Censored	597
Not epilepsy	Excluded	16

¹Patients decision to withdraw before 12 months' freedom from seizures is likely to be highly influenced by side effects or perception of side effects
²Study withdrawals are automatically checked to ensure that the patient wants to withdraw from study rather than from drug only
³Blinded assessment to identify whether death is related or unrelated to epilepsy/AED to take place prior to analysis
⁴Further information is sought if patient withdraws from drug due to "non-compliance" as the underlying reason could be unacceptable adverse events, inadequate seizure control or remission of epilepsy.

ISSN: 1745-6215