de la Rubia Ortí 2018
high risk → dropped
Randomizationexplicit
Blindingopen-label
Analysiswithin-arm test
Serious. Everyone knew their treatment — a cognitive result you can't trust. Removed.
Mutoh 2022
some concerns · 2
Blindingdouble-blind
Analysiswithin-arm test
Missing dataper-protocol
Two minor flags. Strong design, but tested within each arm and per-protocol. Kept.
Baranowski 2025
some concerns · 1 cleanest
Blindingdouble-blind
Analysisbetween-arm
Missing datasome, unclear
The ceiling. Just one concern — a missing-data gap — vs Mutoh's two.
FRESH-Evidence