Tofacitinib or Adalimumab versus Placebo in Rheumatoid Arthritis
New England Journal of Medicine, 2013 · paper · Clinical trials
Reports patient-reported outcomes from a 12-month phase 3 trial of 717 rheumatoid arthritis patients taking methotrexate, given tofacitinib, adalimumab or placebo. Both drugs improved pain, function, fatigue, sleep and quality of life more than placebo.
Checked: PDF from europepmc.org, fetched 6 Oct 2026. Page numbers are those of that PDF.
3 findings
Reference or label In Table 2 the adalimumab General Health value '39.04**' carries no 'b' (n = 187) footnote. The percentage is consistent only with n = 187, and Table 1 flags the same cell 'j' (n = 187).
- p. 7 Table 2, General Health, adalimumab
39.04**
- p. 3 Table 1, General Health, adalimumab
4.93 (0.51)j,**
- p. 3 Table 1 footnotes
jn = 187
Judge's note: Table 2 image on p7: PCS 66.31*b and MCS 51.87b carry b, but General Health 39.04** does not. No integer count over 188 gives 39.04%, while 73/187 does. The reviewer's second point, that '**' cannot be reached, is not confirmed. A pooled Z gives P≈0.002, but an unpooled Z gives z≈3.33 (P<0.001), and the Methods do not specify which normal approximation was used. So the marker is not shown to be inconsistent.
Second check: another model agreed
Report a problem with this finding
Wording Abstract: changes with tofacitinib 5 mg BID and adalimumab were significant 'across most PROs, excluding SF-36 Mental Component Score and Social Functioning, Role Emotional, and Mental Health domains'. The list leaves out MOS-Sleep, which was not significant for adalimumab (Table 1: −4.49 (1.05)j, no asterisk).
- p. 1 Abstract, Results
statistically significant vs placebo across most PROs, excluding SF-36 Mental Component Score and Social Functioning, Role Emotional, and Mental Health domains
- p. 3 Table 1, MOS-Sleep, adalimumab
−4.49 (1.05)j
- p. 5 Results, FACIT-F and MOS-Sleep
LSM changes in FACIT-F, but not MOS-Sleep scores, were significant vs placebo with adalimumab
Judge's note: Table 1 on p3 (image) shows the adalimumab MOS-Sleep change −4.49 (1.05) with footnote j only and no significance marker. The p5 Results say 'FACIT-F, but not MOS-Sleep scores, were significant vs placebo with adalimumab'. The abstract's exclusion list is given jointly for both arms, so it is incomplete for adalimumab.
Second check: another model agreed
Report a problem with this finding
Wording Discussion: 'The low NNTs ... observed with both tofacitinib doses (⩽5) for HAQ-DI, PtGA and Pain'. Table 2 and the Results give an HAQ-DI NNT of 5.5 for tofacitinib 5 mg BID ('ranged from 2.9 to 5.5').
- p. 5 Discussion, para 3
with both tofacitinib doses (⩽5) for HAQ-DI, PtGA and Pain
- p. 7 Table 2, HAQ-DI NNT, tofacitinib 5 mg BID
5.5
- p. 5 Results, HAQ-DI/PtGA/Pain
ranged from 2.9 to 5.5 for tofacitinib 5 mg BID
- p. 5 Discussion, para 3
observed with both tofacitinib doses (≤5) for HAQ-DI, PtGA and Pain
Judge's note: The p5 image shows '(⩽5)'. In the Table 2 image on p7 the HAQ-DI NNT for tofacitinib 5 mg BID is 5.5, which exceeds 5. The 10 mg values (3.0–4.6) fit the bound. The Results text on p5 states the range correctly.
Second check: another model agreed
Report a problem with this finding
7 minor (typos and the like; most checked by the judge only)
Wording Abstract: 'Numbers Needed to Treat were lowest for tofacitinib 10 mg BID'. In Table 2 the PtGA NNT is 2.9 for tofacitinib 5 mg BID and 3.1 for 10 mg BID.
- p. 1 Abstract, Results
Numbers Needed to Treat were lowest for tofacitinib 10 mg BID
- p. 7 Table 2, PtGA NNT
2.9
Judge's note: Table 2 image on p7: PtGA NNTs are 2.9 / 3.1 / 3.5. For every other PRO the 10 mg NNT is the lowest, so PtGA is the only exception. The abstract statement has no qualifier. These are minor wording points.
Second check: another model agreed
Report a problem with this finding
Wording Abstract: tofacitinib 10 mg BID showed 'the highest number of patients reporting improvements ⩾minimum clinically important differences vs placebo'. For PtGA, however, Table 2 gives 70.27% for 5 mg BID vs 67.76% for 10 mg BID.
- p. 1 Abstract, Results
with the highest number of patients reporting improve-ments ≥minimum clinically important differences vs placebo
- p. 7 Table 2, PtGA row
67.76***
Only the authors can say what is right here.
Judge's note: Table 2 image on p7: PtGA 70.27*** (5 mg) and 67.76*** (10 mg). The p3/p5 Results say 'the highest with tofacitinib 5 mg BID by PtGA (70.27%)'. The abstract claim is unqualified. It holds for the other PROs. The Discussion statement is handled separately in C010.
Second check: another model agreed
Report a problem with this finding
Wording Results: 'Randomized patients (n = 717, full analysis set)'. Methods: 'one of its study sites (nine patients randomized) ... those patients have been removed from the analyses presented here'. The analysed N is never stated, and the Fig. 1 baseline Ns sum to 707.
- p. 2 Methods, para 1
one of its study sites (nine patients randomized) was found to be non-compliant to study procedures, and those patients have been removed from the analyses presented here
- p. 3 Results, Patients
Randomized patients (n = 717, full analysis set)
- p. 1 Abstract, Methods
patients (n = 717)
- p. 4 Fig. 1 legend
Placebo, n = 106 at baseline
Only the authors can say what is right here.
Judge's note: On p2 (Methods) the nine patients are removed. On p3 the full analysis set is n=717 with 556 (77.5%) completers, both cited to [7]. Fig. 1 on p4 gives baseline Ns 106+201+199+201=707. The 717 is probably the original trial total from [7], but labelling it 'full analysis set' conflicts with the exclusion. Only the authors can give the analysed N.
Second check: another model agreed
Report a problem with this finding
Wording Discussion: 'Significantly greater proportions of patients reported improvements ⩾MCID in these outcomes ..., the highest being with tofacitinib 10 mg BID'. For PtGA, Table 2 gives 70.27% (5 mg) vs 67.76% (10 mg), and the Results say 'the highest with tofacitinib 5 mg BID by PtGA'.
- p. 5 Discussion, para 3
the highest being with tofacitinib 10 mg BID
- p. 1 Abstract, Results
with the highest number of patients reporting improvements ⩾minimum clinically important differences vs placebo
- p. 7 Table 2, PtGA
70.27***
- p. 3 Results, HAQ-DI, PtGA and Pain
the highest with tofacitinib 5 mg BID by PtGA
Only the authors can say what is right here.
Judge's note: Checked the p5 Discussion, the Table 2 image on p7 and the Results on p3/p5. PtGA is the one exception to the blanket statement, and the paper's own Results single it out. This is the Discussion counterpart of the abstract issue in C004.
Second check: another model agreed
Report a problem with this finding
Typo The abstract expands MCS as 'SF-36 Mental Component Score'. The Methods and the table footnotes use 'Mental Component Summary (MCS)'.
- p. 1 Abstract, Results
SF-36 Mental Component Score
- p. 2 Methods, Assessment of PROs
Mental Component Summary (MCS) scores
Judge's note: Checked p1 abstract and p2 Methods, plus the Table 1 and 2 abbreviation lists ('MCS: Mental Component Summary'). The terminology is inconsistent but the meaning is unaffected.
Second check: none (judge only)
Report a problem with this finding
Typo The single-asterisk threshold is '*P < 0.05' in the Table 1 and 2 footnotes but '*P ⩽ 0.05' in the Fig. 2–4 legends and in the Methods ('declared at P ⩽ 0.05').
- p. 3 Table 1 footnote
*P < 0.05
- p. 6 Fig. 2 footnote
*P ⩽ 0.05
- p. 2 Methods, Statistical analysis
Statistical significance was declared at P ⩽ 0.05
Judge's note: The images show '<' in the Table 1 (p3) and Table 2 (p7) footnotes and '⩽' in the Fig. 2 legend (p6) and the Methods (p2). This is a minor inconsistency with negligible practical impact.
Second check: none (judge only)
Report a problem with this finding
Typo 'eight vertices (representing the eight SF-36 domains) radiate for the origin in a circular plot' should read 'radiate from the origin'.
- p. 3 Methods, Statistical analysis (cont.)
radiate for the origin in a circular plot
- p. 3 Methods, Spydergram description
radiate for the
Judge's note: The p3 page image shows 'radiate for the origin', so this is not an extraction artefact.
Second check: none (judge only)