Ismail 1991.

Methods	Randomized controlled trial without blinding. The method of randomization is not described. Communication with the author indicated a computer‐generated random allocation sequence.
Participants	200 women in Malaysia. Inclusion criteria were healthy women aged 18 to 35 years who were sexually active, were willing to rely exclusively upon the pills as the only method of contraception and had at least one menstrual period since the last pregnancy. Exclusion criteria were contraindications to oral contraceptives, termination of pregnancy less than 42 days prior to admission and breastfeeding. Switchers were included in the study.
Interventions	Triphasic levonorgestrel/ethinylestradiol (50‐75‐125 μg LNG and 30‐40‐30 μg EE in a 6/5/10 days regimen, N=100) versus monophasic desogestrel/ethinylestradiol (150 μg DSG and 30 μg EE for 21 days, N=100) [Marvelon].
Outcomes	Primary outcome measures are: pregnancy; side effects; cycle control; discontinuation and reasons for discontinuation. The method of collecting the data on cycle control and side effects is unclear. The report does not describe the definitions of breakthrough bleeding and spotting.
Notes	The report does not provide an a priori hypothesis or a sample size or power calculation. Study duration: 12 cycles. 41 women in the triphasic group and 33 women in the monophasic group discontinued early. The reasons for discontinuation are described. 2 women in the triphasic group did not start oral contraceptives. 9 women in the triphasic group and 6 women in the monophasic group were lost to follow up. 6 women in the triphasic group and 3 women in the monophasic group were withdrawn because of protocol violations. Analysis not according to intention‐to‐treat principle. The trial was supported by Family Health International.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk	Allocation concealment by use of preprinted sealed envelopes opened at the time of admission.