Twenty-six studies (n=8,905 patients; 21,178 patient years of observation) were included in the review. There were nine RCTs (n=3,399 patients), three cohort studies (n=4,122 patients) and 14 case series (n=1,384 patients). Mean follow-up was 74 weeks, and the drop-out rate ranged from 0 to 33%. Sensitivity analysis was reported only for males.
Rate of lymphoma: A statistically significant higher risk of non-Hodgkin's lymphoma was reported in anti-tumour necrosis factor treated patients (SIR 3.23, 95% CI 1.5 to 6.9; 13 cases; absolute rate 6.1 non-Hodgkin's lymphomas per 10,000 patient-years) compared with the expected rate in SEER cancer registry (SIR 1.9 per 10,000 patient-years). The majority of patients with non-Hodgkin's lymphoma had previously been exposed to immunomodulators. The increase in non-Hodgkin's lymphoma was not statistically significant when compared to the meta-analysis of patients treated only with immunomodulators. Men had a consistently higher rate of non-Hodgkin's lymphoma, and this was statistically significant for those aged between 20 and 54 (SIR 5.4, 95% CI 1.3 to 18.1).
Sensitivity and subgroup analyses: When two studies with drop-out rates greater than 15% were excluded, the risk of non-Hodgkin's lymphoma was increased in men of all ages (SIR 9.4, 95% CI 1.8 to 12.3). In specific age categories, this remained statistically significant in those aged between 55 and 64 years (SIR 16.8, 95% CI 2.0 to 64.4; absolute rate 72 per 10,000 patient-years). In sub-group analysis, the rate of non-Hodgkin's lymphoma was significant for case series designs (SIR 9.4, 95% CI 1.35 to 104.0), but this was driven largely by one study.
Significant heterogeneity of the non-Hodgkin's lymphoma rate was reported amongst the studies, with twenty studies reporting no cases.