The authors concluded that azathioprine showed a modest efficacy in the treatment of ulcerative colitis, but the quality of evidence was not high. This was a generally well-conducted study, but the lack of available high-quality studies and the small number of participants meant that the reliability of the authors' conclusions was unclear.

To evaluate the effectiveness of azathioprine (AZA) and 6-mercaptopurine (6-MP) in the maintenance of ulcerative colitis (UC).

MEDLINE, EMBASE, Cochrane Database of Systematic Reviews, American College of Physicians Journal Club, DARE and Cochrane Central Register of Controlled Trials were searched. Search dates varied across sources, spanning 1966 to June 2006. Search terms were reported. Four clinical trial registries were searched to identify ongoing trials. Experts in the field were contacted and the reference lists of retrieved articles were handsearched. No language restrictions were applied to the search.

Randomised controlled trials (RCT), with a minimum six months follow up, of AZA 2mg/kg to 2.5mg/kg daily or of 6-MP 1mg/kg to 1.5mg/kg daily compared to placebo or 5- aminosalicyclic acid (5-ASA ) in patients of any age with moderate to severe UC measuring treatment success were eligible for inclusion. Studies of AZA or 6-MP at lower doses were eligible for inclusion in the systematic review, but were not included in the meta analysis. Success of treatment was defined as clinical remission, significant clinical response or failure to relapse as measured by a validated and commonly used outcome measure. Studies were excluded if participants in the intervention arms received concomitant methotrexate, cyclosporine or biologic agents.

Included studies were of steroid dependent or newly diagnosed UC patients treated with AZA at doses of 2mg/kg or 2.5 mg/kg daily compared to 5-ASA, sulfasalazine or placebo. One study included patients with mild UC. In one study, the dose was decreased in 50 per cent of the patients after three months. The duration of the studies ranged from six to 18 months. The definition of treatment success varied between studies and included: scores of 0 on the Powell-Tuck Index; scores of 0 to 1 on the Baron Index; a score of less than 150 on the UC activity index; cessation of steroids; absence of symptoms; normal histology; or the absence of blood and no inflammation on sigmoidoscopy.

Two reviewers independently selected the studies for review with decisions made by consensus and disagreements arbitrated by a third reviewer.

Methodological quality was assessed using the Jadad scale with additional information recorded on baseline variables, inclusion and exclusion criteria, treatment allocation concealment and intention-to-treat analysis in line with the Chalmers quality assessment scale. Validity was assessed independently by two reviewers with disagreements resolved by consensus.

The number of treatment successes for each group were extracted and used to calculated relative risks (RR) with 95% confidence intervals (CI) for each study. Data were extracted independently by two reviewers with disagreements resolved by consensus.

Pooled RRs with 95% CIs were calculated using both fixed-effects and random-effects models. Stratified analysis was performed to investigate the effect of the following factors on the RR: steroid dependency; having an active comparator arm, including objective endoscopic outcome data; and study quality. Sensitivity analyses were performed according to the following factors: studies with steroid dependent patients; studies with a 5-ASA control arm; and studies with a Jadad score greater than 2. Statistical heterogeneity was assessed using the Χ² statistic. Publication bias was investigated using Begg's test for asymmetry and Egger's test.

Five RCTs were included for review (n=262). On the Jadad scale, one RCT scored 4, one scored 3 and three scored 2. Intention to treat analysis was used in three trials.

AZA was not associated with significantly greater treatment success in patients with UC when all five studies were included. Significant statistical heterogeneity was found (p=0.048). Sensitivity analyses of steroid dependent versus not steroid dependent patients and of studies with an active versus placebo control arm did not alter the results. There was evidence of significant statistical heterogeneity for these outcomes (p<0.05). But, when only the two high-quality studies were considered, AZA was associated with significantly increased treatment success compared to placebo or 5-ASA (n= 107, RR 2.05, 95% CI: 1.30, 3.23, p=0.002). There was no evidence of statistical heterogeneity.

Stratified analysis showed that there was greater efficacy of AZA in maintaining clinical remission of UC in steroid dependent patients (RR 1.57, 95%CI: 1.08, 2.28, p=0.0185), in the high quality studies (RR 2.17, 95% CI: 1.39, 3.41, p=0.0007), in studies with a placebo control (RR 1.41, 95% CI: 1.02, 1.95, p=0.0357) and in studies with an endoscopic score (RR 1.43, 95% CI: 1.07, 1.90, p=0.0146).

There was no evidence of publication bias from inspection of the funnel plot or using Begg's statistic.

AZA shows a modest efficacy in the treatment of UC, but the quality of evidence is not high.

The review addressed a clear question. Inclusion criteria were well-defined. Several relevant databases were searched. Attempts were made to identify unpublished data and publication bias was assessed with no evidence of bias was found, but the assessment was of limited value due to the small number of studies. No language restrictions were applied to the search, reducing the possibilty of language bias. Three of the five studies included in the meta-analysis were conducted by the same group of researchers. Appropriate steps were taken in the study selection, data extraction and validity assessment processes to minimise reviewer error and bias. A suitable tool was used to measure methodological quality and was used to inform the meta-analysis. Given the presence of statistical and clinical heterogeneity, a narrative synthesis may have been more appropriate. This was a generally well-conducted study. However, the lack of available high quality studies, differences between the studies and the small number of participants meant that the reliability of the authors' conclusions was unclear.

Practice: The authors stated that the current use of AZA in clinical practice was not supported by high-quality data.

Research:The authors stated that a well-designed and adequately powered RCT with stratified randomisation by steroid dependency, disease duration and disease severity was needed.

Not stated.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.