Herfarth 2006.

Methods	Study design: Multicentre RCT Setting: Not stated (multicentre RCT)
Participants	Inclusion: People with Crohn's who had undergone resective surgery Exclusion: Homozygous TPMT Age: Not reported Sex: Not reported Type of surgery: Not reported Previous surgery: Not reported Start of intervention after surgery: within 2 weeks postoperative Medication use (IG1+ IG2): Smoker (IG1 / IG2): Not reported Number randomised (N = 79): 42/37 Number analysed (N = 37): 18/19 Post‐randomisation exclusion (n = 42)
Interventions	Group 1: 2.0 to 2.5 mg/kg body weight/day azathioprine Group 2: 4 g 5‐ASA/day All participants: Not stated
Outcomes	Duration of study: 1 year (study was discontinued after one year) 1. Treatment failure (due to severe endoscopic recurrence, lack of efficacy and AE related to study drug) 2. Clinical or severe endoscopic relapse 3. Severe endoscopic relapse 4. Clinical relapse (reviewer calculated: clinical or severe endoscopic relapse minus severe endoscopic relapse) 5. Adverse events 6. Withdrawal due to adverse events
Notes	Funding source: Dr. Falk Pharma GmbH, Freiburg, Germany Conflict of interest: Not reported Power calculation: Not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Patients in the present study were assigned to one of the two treatment groups (5‐ASA or azathioprine) at random For creation of the randomisation list the programme "Rancode +" (version 3.6) of IDV, Gauting (Germany) was used. The randomisation into two treatment groups was performed in blocks of four. After voluntary written informed consent was obtained and basic selection criteria were checked, the investigator requested the allocation of a unique patient code number (randomisation number, consecutively allocated to each patient), and received medication packs with the randomisation number for the patient" Comment: Confirmed by correspondence from Muller R (2/5/2012)
Allocation concealment (selection bias)	Low risk	"The randomization code was prepared and stored by a statistician from a CRO, who was not involved in the conduct nor in the analysis of the study. The Qualified Person of the Sponsor and the contract manufacturer responsible for the preparation of the double‐dummy patients sets received a copy of the randomization list, which was safely stored at both sites, without allowing access by other people. Neither the investigator nor the study team from the clinical operation from the sponsor nor the CRO had access to the random list" Comment: Confirmed by correspondence from Muller R (2/5/2012)
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"This was a double‐blind, double‐dummy study. Patients randomized to administer 5‐ASA had to take 5‐ASA VERUM tablets AND azathioprine PLACEBO tablets. Patients randomized to receive azathioprine had to administer azathioprine VERUM tablets AND 5‐ASA PLACEBO tablets Therefore, neither the investigator, nor the patients, nor the sponsor were ware of the TX a patient received until the database was clean, closed, and the code was broken" Comment: Confirmed by correspondence from Muller R (2/5/2012)
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"This was a double‐blind, double‐dummy study. Patients randomized to administer 5‐ASA had to take 5‐ASA VERUM tablets AND azathioprine PLACEBO tablets. Patients randomized to receive azathioprine had to administer azathioprine VERUM tablets AND 5‐ASA PLACEBO tablets Therefore, neither the investigator, nor the patients, nor the sponsor were ware of the TX a patient received until the database was clean, closed, and the code was broken" Comment: Confirmed by correspondence from Muller R (2/5/2012)
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "The study was stopped prematurely after an interim‐analysis due to a high therapy failure rate. 38 patients (AZA 18 patients; 5‐ASA 20 patients) completed the study and could be evaluated regarding the primary endpoint therapy failure. The other patient terminated the trial prematurely due to the study stop, but were also evaluated for adverse events (AE) and adverse drug reactions (ADR)" Comment: 51% of randomised participants discontinued. High risk for primary outcome and low risk for AE and withdrawal due to AE
Selective reporting (reporting bias)	Unclear risk	Insufficient information as trial registration was not available and study was published as abstract
Other bias	Unclear risk	Insufficient information as study was published as abstract