Phillips 1998a.

Methods	Multicentre, multi‐country study conducted at 25 centres in seven countries, i.e. Austria, Belgium, Canada, Germany, the Netherlands, Spain and the United Kingdom, from June 1993 to August 1993. Loss to follow‐up: Inctraconacole BD ‐ 21/79 (27%) Ictraconazole daily ‐ 19/79 (24%) Fluconazole ‐ 18/86 (21%) Analysis: no ITT
Participants	Inclusion criteria: HIV positive adults, at least 19 yrs old, CD4 cell count < 400/mm3 previous month, had OPC Exclusion criteria: inability to take oral medication; systemic antifungal treatment within previous 2 weeks / intraoral topical antifungal treatment within 1 week before the trial; hypersensitivity to azoles; non‐responsive candidiasis to fluconazole or itraconazole, suspicion of candidal esophagitis, liver dysfunction or estimated creatinine clearance < 50ml/min, concurrent use of terfenadine, astemizole, phenytoin, carbamazepine, phenobarbital, rifampicin, oral anticoagulants or sulfonylureas not permitted during trial. Diagnosis confirmed: microscopy with KOH and culture Enrolled 244 79 Ictraconazole BD 79 Ictraconazole daily 86 Fluconazole
Interventions	Fluconazole capsules (100mg once daily for 14 days ) with placebo oral solution Itraconazole oral solution 100mg once daily for 14 days plus placebo capsules Itraconazole oral solution 100mg bd for 7 days plus placebo capsules
Outcomes	Clinical response ‐ based upon changes in investigator's rating of signs and symptoms. Severity of symptoms and signs were scored on 3 point scale. Extent of lesions was also scored. Responses classified as complete response or markedly improved, moderately improved, unchanged or deteriorated condition. Mycological efficacy ‐ based on presence or absence of fungal forms consistent with candidiasis Clinical relapse during follow up (at 1 and 2 weeks) Adverse events: GIT (nausea, vomiting, abdominal pain, anorexia and liver enzyme abnormalities), rash, fever, neurological (headache, coma, convulsions and hemiparesis) , hypotension and 1 death in the fluconazole arm.
Notes	Ethics: approval obtained from each participating centre and informed consent signed
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."