Lewis 2007.

Study characteristics
Methods	Multicenter, randomized, double‐blind, placebo‐controlled, 2‐way, 2‐attack, cross‐over study of zolmitriptan nasal spray with a single‐blind 'placebo challenge' or 'enrichment' phase
Participants	Adolescents of age 12‐17 years with a diagnosis of migraine with or without aura per the IHS 1988 criteria or revised IHS‐1988 criteria (Winner 1995) with a frequency of 2 or more migraine attacks per month during the school year and < 14 days per month without migraine for the 3 months before screening. The usual duration of untreated migraine was required to be > 2 h. Participants were permitted to use preventive medication that had been stable for 2 months before randomization. Randomized (N = 248); withdrawn (N = 34); placebo responder to both attacks (N = 12); placebo responder to 1 attack (N = 12); missing efficiency data (N = 19); intention‐to‐treat analysis (N = 171)
Interventions	Each participant treated 1 migraine attack with zolmitriptan 5 mg nasal spray and another with matching placebo within a 12‐week period. Each attack was treated initially with placebo normal saline within 30 min after the headache reached moderate or severe intensity. If migraine intensity remained moderate or severe, then participants were randomly assigned to use zolmitriptan or matching placebo (1:1). A second dose of randomized treatment or approved escape medications were permitted 2 h after the first dose if moderate or severe pain persisted.
Outcomes	Headache relief (decrease from moderate or severe to mild or no headache) at 2 h (1 h was used as the primary outcome in the study) Pain‐free at 2 h Use of rescue medication Presence of nausea Use of rescue medications within 2 h Adverse events Other reported outcomes: Presence of photophobia and phonophobia Other time points including 15, 30, 45, 60, and 90 min
Headache severity scale	4‐point scale (none, mild, moderate, severe)
Funding source	AstraZeneca
Publication	Journal and clinical trial registry
Notes	—
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Patients were randomly assigned strictly sequentially"
Allocation concealment (selection bias)	Low risk	Quote: "double‐blind randomization schedule (prepared by the study sponsor)"
Blinding (performance bias and detection bias) All outcomes	Low risk	Quote: "matching placebo" Comment: unique study design with placebo challenge where all subjects initially received placebo in a single‐blind (participant) fashion. Only those who did not respond were randomly assigned to active drug or placebo.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: missing outcome data distributed evenly across intervention groups
Selective reporting (reporting bias)	Low risk	Comment: all expected outcomes reported