Beebe 1999.

Study characteristics
Methods	Study design: multicentre, randomised, double‐blind, placebo‐controlled Intention‐to‐treat: yes Country: USA
Participants	Number randomised: 516 (cilostazol 100 mg, n = 175; cilostazol 50 mg, n = 171; placebo, n = 170) Age (mean years ± SE): cilostazol 100 mg = 64.3 ± 8.5; cilostazol 50 mg = 64.5 ± 9.9; placebo = 65.1 ± 9.3 Sex M/F: cilostazol 100 mg 130/45; cilostazol 50 mg = 131/40; placebo = 131/39 Inclusion criteria: ≥ 40 years of age; ≥ 6 months history of stable symptomatic IC secondary to lower extremity arterial occlusive disease; reproducible walking distances on screening treadmill tests; treadmill tests terminated solely because of claudication pain; ICD in screening period between 30 and 200 m on two consecutive tests; resting ABI of 0.90 or less and a 10 mmHg or more decrease in ankle artery blood pressure following the onset of ACD Exclusion criteria: ischaemic pain at rest; gross obesity; childbearing potential; hypertension; current metastatic malignant neoplasm; exercise‐limiting cardiac disease; history of bleeding tendencies; or concomitant use of antiplatelet, anticoagulant, vasoactive or NSAIDs
Interventions	Treatment 1: cilostazol 100 mg, twice daily, orally Treatment 2: cilostazol 50 mg twice daily, orally Control: placebo Duration: 24 weeks
Outcomes	PFWD and MWD by treadmill testing, Doppler‐measured bilateral peripheral limb pressures, patient‐based QoL questionnaires (SF‐36, WIQ, COM), patient and physician end‐of‐treatment global therapeutic assessments, cardiovascular morbidity, all‐cause mortality, amputation and adverse events. Outcomes evaluated at baseline (three times), 4, 8, 16, 20 and 24 weeks
Funding	Otsuka America Pharmaceutical Inc.
Declaration of interests	Not reported
Notes	"The COM questionnaire was developed by the study sponsor and has not been independently validated." Constant‐rate, constant‐grade treadmill test design, with 12.5% incline and speed of 3.2 km/h Minimum three‐week screening period
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization of eligible patients was stratified by each clinical center. A master randomization list of patient code assignments to the test medication… was developed using a permuted‐block design".
Allocation concealment (selection bias)	Unclear risk	Insufficient description of allocation concealment methods
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"The master list was forwarded to the drug packaging company, where separate medication supply was prepared for each unique patient code. All 3 test medications had a similar appearance".
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"For all‐cause mortality and cardiovascular morbidity assessment, an independent study committee, blinded to treatment assignment, adjudicated all patient deaths and serious adverse event…". Although it was not directly addressed for other outcomes, it was assumed blinding was adequate.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Table 1 describes patient flow with all participants included in safety outcomes, and participants excluded for efficacy endpoints were similar across treatment groups.
Selective reporting (reporting bias)	Low risk	Although no protocol was available, all outcomes included in description of methods were reported on.
Other bias	Low risk	No evidence of other bias