Boni 2012.

Methods	Randomized multi‐center phase III trial Eligible patients Histologically or cytologically confirmed NSCLC, stage IIIB (supraclavicular node and/or malignant pleural effusion) or metastatic stage IV (according to sixth TNM classification) Chemotherapy‐naive for advanced disease ≥ 18 years of age PS 0 to 2 Adequate hematological, renal, and hepatic function Exclusion criteria Active infection Severe co‐morbidity History of previous or concomitant neoplasm (except epithelial tumors of the skin or in situ carcinoma of the uterine cervix)
Participants	GP arm: 106 (ITT population)/20 elderly participants GN arm: 106 (ITT population)/29 elderly participants GIP arm: 110 (ITT population)/25 elderly participants GIN arm: 111 (ITT population)/27 elderly participants Median age:72.63 years (range 70 to 79) for entire cohort, 85.15% male, 51.49% PS 0, 42.57% PS 1, 47.52% adenocarcinoma, 26.73% squamous cell carcinoma, 25.74% other histologies
Interventions	GP arm: gemcitabine 1250 mg/m2 on days 1 and 8 plus cisplatin 80 mg/m2 on day 1, every 3 weeks GIP arm: gemcitabine 1000 mg/m2 on days 1 and 8 plus ifosfamide 2 g/m2 with mesna 1200 mg as bolus i.v. infusion before ifosfamide and after 4 hours and 8 hours plus cisplatin 80 mg/m2 on day 1, every 3 weeks GN arm: gemcitabine 1250 mg/m2 on days 1 and 8 plus vinorelbine 25 mg/m2 on days 1 and 8, every 8 weeks GIN arm: gemcitabine 1000 mg/m2 on days 1 and 8 plus ifosfamide 3 g/m2 with mesna 1600 mg as a bolus i.v. infusion before ifosfamide and after 4 hours and 8 hours plus vinorelbine 25 mg/m2 on days 1 and 8, every 3 weeks
Outcomes	Primary outcome Overall survival Secondary outcomes Toxicity Objective response rate Progression‐free survival
Notes	Trial was designed as a factorial trial to compare (1) effectiveness of 2 different treatment strategies, 1 containing cisplatin and 1 containing vinorelbine instead of cisplatin, and (2) 1 regimen with 2 and 1 with 3 drugs for the addition of ifosfamide Subgroup analysis of participants ≥ 70 years of age not planned as part of the original protocol; done at request of study authors
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random assignment centrally performed by fax at Trial Unit of the National Institute for Cancer Reseach of Geneva with use of permuted blocks of variable sizes. Elderly subgroup not planned
Allocation concealment (selection bias)	Unclear risk	No information on allocation concealment analysis
Blinding of outcome assessment (detection bias) OS and 1y OS rate outcome	Unclear risk	Open‐label study considered to have unclear influence on mortality outcomes
Blinding of outcome assessment (detection bias) Other outcomes	High risk	Open‐label study
Incomplete outcome data (attrition bias) All outcomes	Low risk	For elderly population, 4 participants never treated and excluded from safety population; included in ITT population for other efficacy data analyses
Selective reporting (reporting bias)	Low risk	No evidence of selective reporting bias. Participant characteristics and summary data provided by study author after direct contact
Other bias	High risk	Study designed for general population; elderly subgroup analysis not planned. Data for participants ≥ 70 years of ageobtained after request to study authors