During some recent new analysis of the data for this study by the CARE Study Group, the authors noticed an inconsistency in the data recording for caffeine intake (BMJ 2008;337:a2332, doi:10.1136/bmj.a2332). This meant that a small proportion of women had been wrongly allocated zero caffeine intake for instant coffee even though they might have consumed instant coffee. The authors have recalculated the corrected total caffeine values, resulting in small changes to the results. In table 3 in the full online version (and the single table in the print version), the adjusted odds ratios (95% confidence intervals) for the risk of fetal growth restriction according to caffeine intake averaged over pregnancy should read: for caffeine intake 100-199 mg/day, 1.1 (0.8 to 1.5) [not 1.2 (0.9 to 1.6)] ; 200-299 mg/day, 1.3 (0.9 to 1.8) [not 1.5 (1.1 to 2.1)]; ≥300 mg/day, 1.5 (1.0 to 2.1) [not 1.4 (1.1 to 2.0)]; and the P value for the test for trend (0.02) remains the same. At specific time points in table 3 and in the analysis reported in table 4, the new analysis resulted in only modest changes. In the figure the “best-fitting” curve remains closely similar to that in the published version. And the mean percentage of caffeine is now 21% from coffee and 58% from tea (not 14% and 62% respectively, as given in the main text). These revised results do not change the interpretation of the findings or the strength of association. Readers requiring more detailed information should contact the authors (Professor Janet Cade, j.e.cade@leeds.ac.uk).
Cite this as: BMJ 2010;340:c2331