Eighteen relevant RCTs (n=8,808) were identified and over 7,846 years of follow-up were included in the analysis. There was no evidence for increased mortality, serious adverse effects, serious infections, lymphomas, non-cutaneous cancer/melanoma or non-melanoma skin cancer with any anti-TNF used at recommended doses compared to controls.
Higher than recommended doses were given to 15% of patients who received adalimumab, 54% of patients who received infliximab and none of the patients who received etanercept. The unadjusted meta-analyses identified an increased risk of serious infection with high dose anti-TNF therapy, OR 2.07 (95% CI 1.31 to 3.26) and RR 1.83 (95% CI 1.18 to 2.85). Adjustment for exposure found evidence of heterogeneity and a random-effects analysis was no longer significant, RR 1.99 (95% CI 0.90 to 4.37).
Risk of death, serious adverse effects, all malignancies, lymphomas and non-melanoma skin cancer did not increase with high dose anti-TNF therapy; however, risk of non-cutaneous cancer/melanoma combined showed a trend towards significance in the unadjusted meta-analyses, which was not significant after adjustment for exposure.
The duration of the individual trials did not affect the OR for death, serious adverse effects, non-cutaneous cancer/melanoma or non-melanoma skin cancer. The OR for serious infection decreased as the length of trials increased, from 2.08 after 12 weeks duration to 0.97 for trials of 104 weeks duration. A meta-regression showed the risk of serious infection with anti-TNF therapy decreased with increasing trial duration (p=0.04). The effect of time on lymphoma risk could not be determined due to the low levels of lymphoma.
Details of the previous treatment for arthritis and concomitant treatment with methotrexate in the patients were provided.
The results of sensitivity analyses were reported.