Ghosh 2003.

Methods	Randomized, double‐blind, placebo‐controlled, multi‐center, 12 week induction trial Computer generated, site stratified, block randomization schedule
Participants	248 adult patients (> 18 years) with moderate to severe Crohn's disease (CDAI > 220 to < 450) Stable dose of AZA or 6‐MP required for at least 4 months Exclusion criteria: MTX, CYA, or ID within 3 months prior to entry, prior treatment with antibody agent, oral prednisolone (> 25 mg/day) or equivalent, elemental or parenteral nutrition, infectious or neoplastic bowel disease, bowel surgery within 3 months, ostomy, fibrotic strictures, need for abdominal surgery
Interventions	Each group received an infusion at week zero and week four Treatment groups: two infusions of placebo (n = 63), one infusion of natalizumab (3 mg/kg) and one infusion of placebo (n = 68), two infusions of natalizumab 3 mg/kg; n = 66) and two infusions of natalizumab (6 mg/kg; n = 51)
Outcomes	The primary outcome variable was remission at week six (CDAI < 150) Secondary outcomes: clinical response (> 70 point decrease in CDAI from baseline), serum level of CRP, absolute neutrophil counts, serum antibodies against natalizumab and IBDQ Patients were evaluated at baseline and weeks two, four, six, eight, and twelve
Notes	Industry‐involvement: funded by Elan pharmaceuticals, some authors were employees or paid consultants for Elan, some owned company equity
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated, site‐stratified, block randomization schedule
Allocation concealment (selection bias)	Low risk	Centrally randomized
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blind ‐ neither the study personnel nor the patients were aware of treatment assignments
Incomplete outcome data (attrition bias) All outcomes	Low risk	Patient withdrawals were described and distributed evenly across treatment groups
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported
Other bias	Low risk	The study appears to be free of other sources of bias