Ardizzone 2004.

Methods	Study design: RCT, single centre Setting: University “L.Sacco” Hospital (Milan, Italy), 1994 to 2001
Participants	Inclusion: Adult (18 to 70 years) participants who underwent surgery for symptomatic intestinal stenosis or occlusion, which is clinically quiescent (CDAI ≤ 150) able to start oral nutrition and oral medication within the first 2 postoperative weeks Exclusion: Contraindications for use of mesalamine or AZA and pre‐existing hepatic disease, renal dysfunction, clinically important lung disease, systemic infection, short‐bowel syndrome, presence of alcoholic stoma, history of cancer, hypersensitivity to mesalamine or AZA, erythrocyte macrocytosis, use of immunosuppressive drugs in the past 3 months; participants who had received treatment with anti–TNF‐α within the 6 months before surgery; pregnancy/breastfeeding; participants who had undergone surgical procedures other than conservative surgery or for perianal disease only; history of corticosteroid‐dependent disease Age (IG1 / IG2) mean: 38.4 years Sex (M:F): 95: 52 overall; (45:26) versus (50:26) Type of surgery: Stricturoplasty‐ 36; Minimal bowel resection‐ 70; Minimal bowel resection stricturoplasty‐36 Previous surgery (IG1+IG2): 69/142 overall (38/71) versus (31/71) Start of intervention after surgery: < 2 weeks Medication use (IG1+ IG2): Mesalamine or sulphasalazine 62; Corticosteroids 41; Immunosuppressants 9; None 30 Smoker (IG1 / IG2): (28/71) versus (36/71) Number randomised (N = 142): 71 versus 71 Number analysed (N = 138): (69/71) versus (69/71) ‐ ITT; 50/71 versus 61/71 ‐ per protocol Post‐randomisation exclusion (n = 11): (6/71) versus (5/71) (did not start the treatment ‐3 (2 versus 1); lost to follow‐up ‐8 (4 versus 4))
Interventions	Group 1: Azathioprine administered at a dosage of 2 mg/kg/day Group 2: Mesalamine was administered at a dosage of 3 g/day divided into 3 doses All participants: treatment with aminosalicylates, metronidazole, and any other CD‐specific treatment had to be discontinued. Corticosteroids were allowed to be tapered by standardized stepwise dose reductions within 6 weeks after surgery at the latest. Symptomatic treatment with antacids, antidiarrhoeal agents, or spasmolytic agents was allowed but had to be scrupulously recorded. Compliance with treatment was evaluated by a simple questionnaire in which adverse events were also recorded. Participants receiving AZA were regularly assessed by total blood cell count and serum transaminase values to monitor any myelotoxicity and hepatotoxicity of the treatment. Participants were seen at baseline and every 6 months
Outcomes	Duration of study: 24 months 1. Clinical relapse defined as the presence of symptoms related to CD, variably associated with radiologic, endoscopic, and laboratory findings, with a CDAI score > 200, which is considered severe enough to warrant treatment with a systemic corticosteroid at a medium‐high dose 2. Surgical Relapse defined as the presence of symptoms refractory to medical treatment or complications requiring another surgical procedure (e.g., occlusive disease, intra‐abdominal abscesses, or high‐flow fistulas) 3. Adverse events
Notes	Funding source: Not reported Conflict of interest: Not reported Power calculation: Based on a maximum relapse rate at 2 years of 45% mesalamine, 62 participants per treatment group was considered sufficient to detect a difference of ≥ 25% for the AZA treatment group (type 1 error of 5%). The number of participants in each group was increased to 68 to compensate for an anticipated drop out rate of 10%
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “After surgery, patients who met the inclusion criteria and who agreed to enter the study were randomised to receive mesalamine or AZA by a computer‐generated list” and ”Randomization was performed in blocks of 10” Comment: computer generated block randomisation
Allocation concealment (selection bias)	Unclear risk	Insufficient information to make judgment
Blinding of participants and personnel (performance bias) All outcomes	High risk	Comment: the study is open‐label and blinding was not performed
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information to make judgment, however it is unlikely
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quotes: “In the intention‐to‐treat analysis, all randomised patients who received at least one dose of the study drug and were subjected to the baseline evaluation were considered for the analysis.”and “Outcome measures were analysed in all randomised patients who had taken at least one dose of the study medication (intention‐to‐treat population)…” Comment: ITT analysis applied, all withdrawals were low and balanced across groups
Selective reporting (reporting bias)	Low risk	Trial registration not available, however all outcomes stated in the method section assessed and reported
Other bias	Low risk	Quote: “No significant differences were observed between the 2 treatment groups regarding age, sex, duration of disease, location of disease, fistula and abscess at surgery, surgical procedure, previous operations, and CD therapy during the previous 6 months” Comment: baseline characteristics well balanced across groups