Phillips 1998b.

Methods	Same as for Phillips 1998a
Participants
Interventions
Outcomes
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	Block randomisation: blocks of 12 to one of three treatment groups, according to a predefined randomisation code. ensuring that an equal number of patients were allocated to each treatment group
Allocation concealment?	Unclear risk	Not reported
Blinding? All outcomes	Low risk	Double‐blind: participant and investigator blinded. "Active and placebo forms provided by Jansen Research Foundation and were blinded so that neither the investigators or the patients were aware of their contents."