Fifteen RCTs with a total of 2,097 participants (1,049 received mesalamine and 1,048 were controls) were included: there were 13 full papers and 2 abstracts.
Methodological quality: the scores were variable across the studies (range: 45 to 100%).
The pooled estimate of the treatment effect was significant: the overall risk difference was -6.3% (95% confidence interval, CI: -10.4, -2.1, p=0.0028). The number-needed-to-treat was 16. There was no significant heterogeneity among the studies (Qw=12.9, p=0.53).
The overall rate of adverse events was 13.7% in the mesalamine group and 14.9% in the control group (not significant). The most frequent adverse effects were of a gastrointestinal nature.
When separate meta-analyses were performed for surgically-induced remission (4 RCTs) and medically-induced remission (10 RCTs), the pooled risk difference was significant in the postsurgical setting (overall risk difference -13.1%, 95% CI: -21.8, -4.5) but not in the medical setting (overall risk difference -4.7%, 95% CI: -9.6, 2.8).
The multivariate model (12 RCTs, 1,543 participants) predicted that the probability of clinical relapse was significantly lowered by mesalamine treatment, with increasing proportions of patients with ileal disease, surgically-induced remission, or prolonged disease duration. No significant relation was found between the treatment effect and the drug dosage used.
Sensitivity analysis based on methodological quality: 4 RCTs with a quality score of less than 66% were excluded; this had no significant effect on the results.