A Randomized Trial of Arthroscopic Surgery for Osteoarthritis of the Knee
New England Journal of Medicine, 2009 · paper · Clinical trials
A randomized trial of arthroscopic surgery plus physical and medical therapy versus physical and medical therapy alone in 178 patients with moderate-to-severe knee osteoarthritis. At two years, symptom scores were similar, so surgery gave no additional benefit.
1 finding
Substance In Figure 2A, "No. of Patients" for Arthroscopic surgery at month 24 is printed as 86, but Panels B + C give 42 + 46 = 88, as do Table 3 ("Surgery (N = 88)") and Figure 1 (83 + 5 completed).
- p. 7 Figure 2A, No. of Patients, Arthroscopic surgery, month 24
86
- p. 7 Figure 2B and 2C, No. of Patients, Arthroscopic surgery, month 24
42
- p. 9 Table 3, Mo 24 surgery column header
Surgery (N = 88)
- p. 9 Table 3, Mo 24 Surgery
(N = 88)
Judge's note: I checked the page 7 image. In every other column, Panel A equals Panel B plus Panel C: surgery 42+50=92, 40+50=90, 41+48=89, 36+42=78, 38+38=76; control 36+49=85, 34+46=80, 29+42=71, 33+44=77, 31+39=70, 34+46=80. The surgery count at month 24 is the only exception: 42+46=88, while Panel A prints 86. Table 3 on p9 and Figure 1 on p4 also give 88.
Second check: another model agreed
Report a problem with this finding
9 minor (typos and the like; most checked by the judge only)
Wording Abstract: "Analyses of WOMAC scores at interim visits ... also failed to show superiority of surgery", yet Figure 2A shows P=0.008 at 3 months, which favours surgery (743 vs 824), and the Results and Discussion describe a greater 3-month improvement with surgery.
- p. 6 Figure 2 legend
P values for the differenc-es in 24-month scores were generated by analysis of co-variance
- p. 7 Figure 2, Panels A–C
P=0.008
- p. 1 Abstract, Results
Analyses of WOMAC scores at interim visits and other secondary outcomes also failed to show superiority of surgery.
- p. 8 Discussion, paragraph 1
Patients assigned to surgery did have a greater improvement in WOMAC scores within the first 3 months
Only the authors can say what is right here.
Judge's note: At 3 months, Figure 2A (p7) prints P=0.008 and Table 3 shows mean WOMAC 743 for surgery against 824 for control. The Results (p6/8) say that at 3 months surgery scores "had improved to a greater extent", and the Discussion (p8) calls this a "transient benefit". The abstract's sentence about interim visits does not describe the 3-month visit accurately. The reviewers' point about the caption covering only the 24-month P values is handled in C007.
Second check: another model agreed
Report a problem with this finding
Wording Methods: "questionnaires and standard-gamble utility scores were obtained at each visit", but Table 3 prints "—" for standard-gamble utility at Mo 18 in both groups, with no footnote.
- p. 3 Methods, Study Treatment
questionnaires and standard-gamble utility scores were obtained at each visit
- p. 9 Table 3, Standard-gamble utility score, Mo 18
—
Only the authors can say what is right here.
Judge's note: The Methods on p3 say all instruments, including the standard gamble, were collected at every visit. In Table 3 (p9, upright image) the Mo 18 utility cells are dashes in both arms, and none of the footnotes explains the missing values. Only the authors can say whether these data were not collected or were collected but not reported.
Second check: another model agreed
Report a problem with this finding
Wording The Methods time visits as "3, 6, 12, 18, and 24 months after the initiation of treatment", but the Figure 2 x-axis reads "Months since Randomization".
- p. 3 Methods
18, and 24 months after the initiation of treat-ment
- p. 7 Figure 2 x-axis
Months since Randomization
Judge's note: On p2-3, treatment initiation is defined as the next available surgery date, within 6 weeks after randomization. The Figure 2 axis on p7 uses randomization as time zero. The two time origins differ by at most a few weeks. This is a labelling inconsistency and either wording could be defended.
Second check: another model agreed
Report a problem with this finding
Wording Figure 2 baseline control counts are 85 (Panel A) and 49 (Panel C), whereas Table 1 ("Control (N = 86)"; grades 3 and 4: 46 + 4 = 50) and Table 3 baseline ("Control (N = 86)") give 86 and 50.
- p. 7 Figure 2A, No. of Patients, Control, month 0
85
- p. 7 Figure 2C, No. of Patients, Control, month 0
49
- p. 9 Table 3, Baseline control header
Control (N = 86)
- p. 5 Table 1, Kellgren–Lawrence grades 3 and 4, control
46 (53)
Only the authors can say what is right here.
Judge's note: Within Figure 2 the counts add up (36+49=85). They do not match Table 1, which has 36 + 46 + 4 = 86 controls with 50 in grade 3 or 4, or Table 3, which reports the same baseline WOMAC mean (1043) with N = 86. One control may have lacked a baseline WOMAC score, but the paper does not say so. The surgery arm agrees across all sources (92 = 42+50).
Second check: another model agreed
Report a problem with this finding
Wording The Figure 2 legend says "P values for the differences in 24-month scores were generated by analysis of covariance", but every panel prints P values at 3, 6, 12, 18 and 24 months (e.g. "P=0.008").
- p. 6 Figure 2 legend
P values for the differenc-es in 24-month scores were generated by analysis of co-variance
- p. 7 Figure 2, Panels A–C
P=0.008
Judge's note: The legend on p6 describes only the 24-month P values. The figure on p7 prints P values at all five follow-up visits in all three panels. The Methods (p3) mention post hoc analyses at 3-18 months, and these were probably done the same way, but the legend does not say so.
Second check: another model agreed
Report a problem with this finding
Wording The Table 3 column Ns (Mo 6 "Surgery (N = 90)"/"Control (N = 73)", Mo 12 surgery 80, Mo 18 surgery 78) differ from the Figure 2A counts for the same total WOMAC data (89/71, 78, 76).
- p. 7 Figure 2A, No. of Patients
89
- p. 9 Table 3, column headers Mo 6, Mo 12, Mo 18
Control (N = 73)
- p. 9 Table 3 column headers, Mo 6, Mo 12 and Mo 18
(N = 73)
- p. 7 Fig 2A, No. of Patients at 6, 12 and 18 mo
71
Only the authors can say what is right here.
Judge's note: I compared the Table 3 headers (p9, upright image) with Figure 2A (p7). At Mo 6 they read 90/73 against 89/71, at Mo 12 80/77 against 78/77, and at Mo 18 78/70 against 76/70. Figure 2A's counts match its subgroup sums, and both sources show the same WOMAC means. The Table 3 footnote says only that "Not all patients attended each follow-up visit". Table 3 may be counting visits attended rather than patients with WOMAC data, but the paper does not state this.
Second check: another model agreed
Report a problem with this finding
Typo The title of reference 21 reads "imaging in ostoearthritis of the knee", a misspelling of "osteoarthritis".
- p. 10 Reference 21
imaging in ostoearthritis of the knee
- p. 10 Reference 21
ostoearthritis
- p. 10
in ostoearthritis of the knee
Judge's note: The crop of reference 21 on p10 shows "ostoearthritis", the same as the text layer.
Second check: none (judge only)
Report a problem with this finding
Typo Reference 26 author is printed as "Verhoeven AC, Boers M, van der Liden S". It is probably "van der Linden S".
- p. 10 Reference 26
Verhoeven AC, Boers M, van der Liden
- p. 10 Reference 26
van der Liden
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: the co-author's surname is van der Linden. The crop of p10 does show "van der Liden".
Second check: none (judge only)
Report a problem with this finding
Typo The title of reference 35 reads "Index questionnnaire and global assessments", with a triple n in "questionnaire".
- p. 11 Reference 35
tionnnaire and global assessments
- p. 11 Reference 35
questionnnaire and global assessments
- p. 11 Reference 35
questionnnaire
- p. 11
questionnnaire and global assessments in pa-
Judge's note: The crop of reference 35 on p11 shows "ques-tionnnaire" with three n's.
Second check: none (judge only)