Aspirin versus Placebo in Pregnancies at High Risk for Preterm Preeclampsia
New England Journal of Medicine, 2017 · paper · Clinical trials
A double-blind trial comparing daily low-dose aspirin with placebo in women at high risk of preterm preeclampsia. Preterm preeclampsia occurred in 1.6% of the aspirin group versus 4.3% of the placebo group.
1 finding
Substance Screening flow (Results p.4 and Fig. 1) does not reconcile: "2971 (11.0%) were found to be at high risk" minus "332 of these women (11.2%)" excluded gives 2639, but "2641 eligible women" is printed; one of 2971, 332 or 2641 is off by 2.
- p. 4 Results, Trial Participants
332 of these women (11.2%)
- p. 4 Results, Trial Participants
Of the 2641 eligible women, 1776 (67.2%)
- p. 5 Figure 1
332 Were excluded
- p. 5 Figure 1
2641 Were eligible for inclusion
Only the authors can say what is right here.
Judge's note: I checked the p.4 text and the Fig. 1 boxes on p.5 in the page images. The six exclusion reasons add up to exactly 332, and 332/2971 matches the printed 11.2%. 2641 matches 1776 randomised + 865 declined and 1776/2641 = 67.2%. 2971/26941 matches 11.0%. Neither the text nor the figure gives any other route out of the high-risk group (for example, overlapping reasons or a later exclusion step) that would explain the gap of 2. Only the authors can say which count is wrong.
Second check: another model agreed
Report a problem with this finding
3 minor (typos and the like; most checked by the judge only)
Wording The Methods describe the Kaplan–Meier estimates as ones "in which deliveries that were not with preeclampsia were excluded", but the Figure 2 legend says "Data were censored after deliveries not associated with preeclampsia."
- p. 4 Statistical Analysis
in which deliveries that were not with preeclampsia were excluded
- p. 8 Figure 2 legend
Data were censored after deliveries not associated with preeclampsia.
Judge's note: I checked the end of the Statistical Analysis section on p.4 and Figure 2 on p.8. The numbers at risk start at 807 (placebo) and 785 (aspirin). These equal the primary-analysis groups minus losses before 24 weeks (aspirin 798-11-2 = 785), so women with deliveries not associated with preeclampsia were kept in the risk set and censored, not excluded. 'Excluded' in the Methods reads as loose wording that conflicts with the legend. It is a wording inconsistency, not a data error.
Second check: another model agreed
Report a problem with this finding
Wording Table 1 row "Risk of preterm preeclampsia as assessed at screening at 11–13 wk (95% CI) — %" shows "2.3 (1.4–4.8)" and "2.6 (1.5–4.8)". The footnote defines CI as confidence interval, but the bounds look like a spread of individual risks.
- p. 6 Table 1, last row
at 11–13 wk (95% CI) — %§
- p. 6 Table 1, last row values
2.3 (1.4–4.8)
Only the authors can say what is right here.
Judge's note: I checked Table 1 on p.6. The footnote says 'CI denotes confidence interval'. With about 800 women per group, a confidence interval around a central value such as a median would be much narrower than 1.4–4.8, and both groups share the same upper bound. The other continuous rows in the table give medians with interquartile ranges. The authors should say whether this row shows an interquartile range, a 95% range of individual risks, or something else.
Second check: another model agreed
Report a problem with this finding
Typo The Acknowledgments print the given name as "Madgalena Fiolna"; this is probably a transposition of 'Magdalena'.
- p. 9 Acknowledgments
Madgalena Fiolna
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that this person's name is spelled 'Magdalena'. The page image on p.9 shows 'Madgalena' as quoted. This is not a standard variant of the name, so a transposition of letters is likely, but only the authors can confirm the spelling.
Second check: none (judge only)