| Biological agent | Trial | Group | Recruited | ACR50 responders | Percentage of responders (%) | NNT |
| Infliximab | Maini et al (1999)11 | 3 mg/kg per 8 weeks | 86 | 22 | 27 | 5 (4–9) |
| 3 mg/kg per 4 weeks | 86 | 25 | 29 | 4 (3–7) | ||
| Placebo | 88 | 4 | 5 | – | ||
| Etanercept | Klareskog et al (2004)12 | 2 × 25 mg/week | 231 | 136 | 59 | 3 (3–4) |
| Placebo | 228 | 93 | 41 | – | ||
| Adalimumab | Keystone et al (2004)13 | 40 mg/2 weeks | 207 | 81 | 39 | 3 (3–5) |
| Placebo | 200 | 19 | 10 | – | ||
| Golimumab | Keystone et al (2009)2 | 50 mg/4 weeks | 89 | 33 | 37 | 4 (3–8) |
| 100 mg/4 weeks | 89 | 29 | 33 | 5 (3–13) | ||
| Placebo | 133 | 18 | 14 | – | ||
| Certolizumab pegol | Smolen et al (2009)1 | 200 mg/4 weeks | 246 | 80 | 33 | 3 (3–4) |
| 400 mg/4 weeks | 246 | 81 | 33 | 3 (3–4) | ||
| Placebo | 127 | 4 | 3 | – | ||
| Keystone et al (2009)9 | 200 mg/4 weeks | 393 | 146 | 37 | 3 (3–4) | |
| 400 mg/4 weeks | 390 | 156 | 40 | 3 (3–4) | ||
| Placebo | 199 | 15 | 8 |
These trials differed in the type of patients enrolled, their initial disease activities and the clinical situation from which they were drawn (treatment in the 1990s was different to the 2005–8 era). These differences explain variations in the absolute numbers of active and placebo responders. However, the relationship between active and placebo treatments shown by the number needed to treat (NNT) should not be affected by these variations. ACR50, American College of Rheumatology 50% improvement; RCT, randomised controlled trial; TNF, tumour necrosis factor.