Rosell 2008.

Methods	1. Design: A multicentre, open‐label, parallel group, randomised phase II trial 2. Centres: 16 centres in 6 European countries 3. Randomisation: No details were provided
Participants	Inclusion criteria: Chemotherapy‐naive patients with histologically or cytologically proven NSCLC, stage IV or stage IIIB with documented malignant pleural effusion, according to American Joint Committee on Cancer criteria, and immunohistochemical evidence of EGFR expression in the primary tumour and/or metastases 1. Female, n (%): 22 (25.6) 2. Age in years, median (range): 58 (33‐74) 3. White people, n (%): 86 (100.0) 4. ECOG PS 0‐1, n (%): NA (Karnofsky performance status 80‐100: 78 (92.9) 5. Tumour stage IIIB/IV: 86 (100.0) 6. Adenocarcinoma, n (%): 37 (43.0) 7. Never smoked, n (%): NA 8. EGFR expression, n (%): 86 (100.0) 9. KRAS mutations, n (%): NA 10. EGFR mutations, n (%): NA
Interventions	1. Arm A (n = 43): cisplatin + vinorelbine + cetuximab 2. Arm B (n = 43): cisplatin + vinorelbine
Outcomes	1. Primary: Overall response rate 2. Secondary: Overall survival (including data on one‐year survival rate), progression‐free survival, time to treatment failure, duration of response, safety
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No details on the procedure were provided
Allocation concealment (selection bias)	Unclear risk	There was no mention of allocation concealment
Blinding of participants and personnel (performance bias) overall survival and one‐year survival rate	Low risk	No blinding. However, the results on the two outcomes were mainly determined by the biological, objective effect of treatments and unlikely to be affected by the participants' and personnels' knowledge of the assignment status
Blinding of participants and personnel (performance bias) progression‐free survival, objective response rate, quality of life, and serious adverse events	High risk	No blinding. Progression‐free survival, objective response rate, and serious adverse events are not objective outcomes and could be affected by participants' and/or personnels' knowledge of the assignment status. The study did not use quality of life as an outcome
Blinding of outcome assessment (detection bias) overall survival and one‐year survival rate	Low risk	No blinding. However, overall survival and one‐year survival rate were objective, "hard" outcomes and unlikely to have been affected by the subjective judgement of assessors
Blinding of outcome assessment (detection bias) progression‐free survival, objective response rate, quality of life, and serious adverse events	High risk	No blinding. The assessments of these outcomes involved subjective judgements and were vulnerable to the performance of assessors who were aware of the assignment status
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "The efficacy analysis was based on the intent to treat population defined as all randomized patients. The safety analysis was based on all patients who had received any dose of study treatment." Almost all (85, 98.8%) patients were available for safety analysis
Selective reporting (reporting bias)	Unclear risk	Data on quality of life was not reported in the paper. As no protocol or registration can be found for this trial, it is difficult to say whether quality of life was a pre‐specified outcome. Thus, the risk for selective reporting bias was considered unclear
Other bias	Low risk	No evidence about other bias was found

ECOG PS ‐ Eastern Cooperative Oncology Group performance status
 EGFR ‐ epidermal growth factor receptor

NA ‐ not available

NSCLC ‐ non‐small cell lung cancer