Dulaglutide and Cardiovascular Outcomes in Type 2 Diabetes (REWIND)
2019 · paper · Clinical trials
A double-blind randomised trial of weekly dulaglutide versus placebo in 9,901 people with type 2 diabetes with prior cardiovascular disease or risk factors. Major cardiovascular events were lower (12.0% versus 13.4%), and gastrointestinal side effects were more common.
1 finding
Substance Median follow-up of the other GLP-1 RA outcome trials is given as 'ranged from 1·6 years to 3·8 years' (Research in context, p2) but 'ranged from 1·5 years to 3·8 years,5,7' (Discussion, p7).
- p. 2 Research in context, Evidence before this study
durations ranged from 1·6 years to 3·8 years
- p. 7 Discussion, paragraph 2
in which median follow-up ranged from 1·5 years to 3·8 years
- p. 2 Research in context, Evidence before this study
Median follow-up durations ranged from 1·6 years to 3·8 years
- p. 2 Research in context, Evidence before this study
Median follow-up durations ranged from 1·6 years to 3·8 years.
Only the authors can say what is right here.
Judge's note: Both images show the different lower bounds. The Discussion cites refs 5 (LEADER) and 7 (Harmony Outcomes), and both trials are among the five in the panel. The shortest median follow-up among these trials therefore cannot be 1·6 years in one place and 1·5 years in the other. Nothing in the paper explains the difference. Only the authors can say which value is right.
Second check: another model agreed
Report a problem with this finding
6 minor (typos and the like; most checked by the judge only)
Substance Figure 3, history of CVD 'No', dulaglutide: '277/3093 (8·9%)'; 277/3093 = 8·96%, which rounds to 9·0%.
- p. 7 Figure 3, History of cardiovascular disease, No, dulaglutide events/patients
277/3093 (8·9%)
- p. 7 Figure 3, History of cardiovascular disease, No row (placebo)
317/3128 (10·1%)
- p. 4 Table 1, Cardiovascular disease
1560 (31·5%)
- p. 2 Research in context, Added value
men and women with or without previous cardiovascular disease
Judge's note: The Figure 3 crop shows 277/3093 (8·9%). Other cells in the figure are rounded conventionally (for example 218/2306 is printed 9·5% and 263/2635 is printed 10·0%), so truncation does not explain the value. The point that the Yes/No rows do not sum to the arm totals is explained by the Figure 3 footnote (see C012), so it is not confirmed here.
Second check: another model agreed, code check
Report a problem with this finding
Wording Introduction says the other trials were in people with 'a 4% or higher annual risk of cardiovascular events', whereas the Discussion says 'annual placebo incidence rates for the same composite outcome of 3·9% or higher in the other trials'.
- p. 2 Introduction, para 2
a 4% or higher annual risk of cardiovascular events
- p. 7 Discussion, para 2
annual placebo incidence rates for the same composite outcome of 3·9% or higher in the other trials
Judge's note: Found via the text layer at p2 lines 198-199 and confirmed on the p7 image. The Introduction's 4% looks like a rounded form of 3·9%. Read literally, though, the two lower limits disagree. This is a minor wording inconsistency.
Second check: another model agreed
Report a problem with this finding
Wording Discussion (p8) lists among the strengths "investigator freedom to use any non-GLP-1 receptor agonist drug", but Methods (p3) say investigators "were free to add any glucose-lowering drug apart from another GLP-1 receptor agonist or pramlintide". The Discussion leaves out the pramlintide restriction.
- p. 8 Discussion, strengths paragraph
investigator freedom to use any non-GLP-1 receptor agonist drug
- p. 3 Methods, Procedures
free to add any glucose-lowering drug apart from another GLP-1 receptor agonist or pramlintide
Judge's note: I checked both crops against the page images. Page 3 (Methods, Procedures) names two exclusions: another GLP-1 receptor agonist and pramlintide. It also limits the freedom to glucose-lowering drugs. Page 8 (Discussion, strengths paragraph) mentions only the GLP-1 receptor agonist exclusion. A search of the text finds no other mention of pramlintide or of this wording. Read literally, the Discussion describes the protocol a little more broadly than the Methods do. This is a small wording inconsistency, so it is filed as an observation: the Discussion could add "or pramlintide". No results or conclusions are affected.
Second check: another model agreed
Report a problem with this finding
Typo Table 1 diastolic blood pressure: '78·4 (9.8)' and '78·5 (9.9)' use a full stop as the decimal point, whereas the rest of the table uses the mid-dot.
- p. 4 Table 1, Diastolic blood pressure row
78·4 (9.8)
Judge's note: In the crop, the SD decimals in this row sit on the baseline, while the means in the same cells, and the other rows, use mid-dots. The text layer agrees. Cosmetic.
Second check: code check
Report a problem with this finding
Typo Figure 2B annotation reads 'HR=0·91 (95% CI 0·78–1·06)' while panels A, C, D read 'HR 0·88', 'HR 0·96', 'HR 0·76'.
- p. 5 Figure 2, panel B annotation
HR=0·91 (95% CI 0·78–1·06)
Judge's note: Confirmed on the p5 image: only panel B uses the equals sign. Cosmetic only.
Second check: none (judge only)
Report a problem with this finding
Typo Results: 'whose HbA1c was less than 7·2% or greater-than or equal to 7·2%' – awkward mixed hyphenation of the comparator.
- p. 6 Results, subgroups paragraph
less than 7·2% or greater-than or equal to 7·2%
Judge's note: Seen in the p6 crop. A copy-editing slip; the meaning is unchanged.
Second check: none (judge only)