Lutein + Zeaxanthin and Omega-3 Fatty Acids for Age-Related Macular Degeneration: The AREDS2 Randomized Clinical Trial
2013 · paper · Clinical trials
A randomized trial of 4203 people at risk of advanced age-related macular degeneration, testing added lutein/zeaxanthin and omega-3 fatty acids in a supplement. Over a median five years, neither addition produced a statistically significant reduction in progression to advanced disease.
5 findings
Substance Table 1, Smoking status, Former, Lutein + Zeaxanthin: "510 (48.0)". The column sums to 469+510+74=1053, more than n=1044, and 510/1044 is 48.9%. A count of 501 would fit both the total and 48.0%.
- p. 5 Table 1, Smoking status, Lutein + Zeaxanthin column
510 (48.0)
Judge's note: Checked the page image. The other columns sum exactly to their n (1012, 1068, 1079). A missing-data explanation is ruled out because the categories add up to more than n.
Second check: another model agreed
Report a problem with this finding
Reference or label Discussion p10: 'Two randomized controlled clinical trials of beta carotene reported an increase in lung cancer ... in cigarette smokers assigned to receive beta carotene.21,22'. Refs 21 (Wei et al, regression of multivariate failure time data) and 22 (Willett et al, diet questionnaire) are not beta carotene trials.
- p. 10 Discussion, para 3
signed to receive beta carotene.21,22
- p. 11 References 21, 22
Wei LJ, Jin DY, Weissfeld L. Regression analysis
- p. 11 References 21 and 22
Wei LJ, Jin DY, Weissfeld L. Regression analysis of multivariate incomplete failure time data
- p. 2 Interventions, column 3
rette smokers,15,16
Judge's note: The superscript '21,22' is clear on the page 10 image. The reference list on p11 gives ref 21 as Wei LJ, Jin DY, Weissfeld L (J Am Stat Assoc 1989) and ref 22 as Willett WC et al (Am J Epidemiol 1988). Both are cited correctly elsewhere: 21 in Statistical Analyses and 22 for the food frequency questionnaire. Refs 15 (ATBC) and 16 (CARET) are the beta carotene trials, and the Methods cite them as '15,16' for the same claim on p2.
Second check: another model agreed
Report a problem with this finding
Reference or label Results (p4): "(1680 [55%] women vs 708 [61%] for men; P=.002)". The 708 (61%) are the women among the 1167 who refused, not men.
- p. 4 Results, para 2
(1680 [55%] women vs 708 [61%] for men; P=.002)
Judge's note: 1680+708=2388, the total number of women (Table 1: 548+596+603+641). 1680/3036=55% and 708/1167=61% are the shares of women among those who joined and those who refused the secondary randomization. The label 'for men' is wrong.
Second check: another model agreed
Report a problem with this finding
Reference or label Results (p4): "589 [19%] for post-graduate education vs 275 [24%] for bachelor's degree or less". The 275 (24%) are postgraduates among the refusers, not people with a bachelor's degree or less.
- p. 4 Results, para 2
589 [19%] for post-graduate education vs 275 [24%] for bachelor’s degree or less
Judge's note: Table 1 postgraduate total is 219+222+215+208=864, which equals 589+275. 589/3036=19% and 275/1167=24% are the shares of postgraduates among those who joined and those who refused. The label 'bachelor's degree or less' does not fit.
Second check: another model agreed
Report a problem with this finding
Reference or label Reference 18 (Hambidge, J Nutr) is printed as "2003;113(5)(suppl 1):1438S-1442S". The volume for J Nutr in 2003 appears to be 133, not 113.
- p. 11 Reference 18
2003;113(5)(suppl 1):1438S-1442S
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: J Nutr volume 133 corresponds to 2003, and volume 113 to 1983. The page image (crop C031-a, page 11) clearly shows '113' alongside the year 2003, so this is not an extraction artefact. The files contain no correction notice. The 'Corrected on May 14, 2013' footer is only a page stamp and does not list any corrections.
Second check: another model agreed
Report a problem with this finding
20 minor (typos and the like; most checked by the judge only)
Wording Figure 1 footnote c gives smokers as "n=181, AREDS with no beta carotene; n=166, AREDS with no beta carotene and with low-dose zinc" (347 in total), but the Results say "(277 [9%] vs 5 [0.4%]; P<.001)" smokers took part in the secondary randomization. The paper does not say which definition of smoker each count uses.
- p. 3 Figure 1 footnote c
n=181, AREDS with no beta carotene; n=166, AREDS with no beta carotene and with low-dose zinc
- p. 4 Results, para 2
(277 [9%] vs 5 [0.4%]; P<.001)
- p. 3 Figure 1 footnote c
n=181, AREDS with no beta carotene; n =166
- p. 4 Results, para 3
277 [9%] vs 5
Only the authors can say what is right here.
Judge's note: Footnote counts total 347. The text gives 277 smokers among the 3036 in the secondary randomization, and 277+5=282 equals the Table 1 'Current' smokers total (73+74+69+66). The methods (p2) restrict 'current or former smokers within the past year', so the footnote may also count recent quitters. Neither the footnote nor the text says which definition it uses, so this goes to the authors as a query.
Second check: another model agreed
Report a problem with this finding
Wording Statistical Analyses (p4): "level of .013, Bonferroni-adjusted for 3 treatments vs placebo comparisons". .05/3 is .0167, while .013 (and the 98.7% CIs) matches .05/4.
- p. 4 Statistical Analyses
level of .013, Bonferroni-adjusted for 3 treatments vs placebo comparisons
- p. 4 Statistical Analyses
level of .013, Bonferroni-adjusted for 3
Only the authors can say what is right here.
Judge's note: Checked p4. The paper describes only 3 active-vs-placebo comparisons and gives no reason for a fourth divisor. The 98.7% CIs used throughout match alpha .013. The authors should say how .013 was derived.
Second check: another model agreed
Report a problem with this finding
Wording Adherence (p4-5): "3420 (84%) of the participants in each treatment group", "2431 (83%)", "989 (87%)". These percentages do not match the cohort sizes 4203, 3036 and 1148, and no other denominator is given.
- p. 4 Results, adherence paragraph
3420 (84%) of the participants in each treat-ment group
- p. 5 Results, top of col 1
(87%) took at least 75% of the AREDS
- p. 4 Results, adherence
3420 (84%)
Only the authors can say what is right here.
Judge's note: Recomputed: 3420/4203=81.4%, 2431/3036=80.1%, 989/1148=86.1%. The denominators may be participants with pill-count data, but the paper does not say so. 'In each treatment group' is attached to a pooled count.
Second check: another model agreed
Report a problem with this finding
Wording Results p6 says 'lowering zinc dose and eliminating beta carotene had no statistically significant effect ... (HR, 1.06 ... and HR, 1.07 [95% CI, 0.94-1.20]...)', but in Figure 4 the 1.07 row has 'Beta carotene' as the treatment arm (2221 eyes), so the HR is for including beta carotene, not eliminating it.
- p. 6 Results, Secondary Randomization
lowering zinc dose and eliminating beta carotene had no statistically significant effect
- p. 7 Figure 4, Beta carotene row
1.07 (0.94-1.20)
Only the authors can say what is right here.
Judge's note: In Figure 4 the treatment arm of the beta carotene row is 2221 eyes. That equals 1096+1125, the two Figure 1 groups that received beta carotene (original AREDS and low-dose zinc). So the HR is beta carotene vs no beta carotene. The zinc row's treatment is low-dose zinc, which matches 'lowering zinc dose'. The 'no significant effect' conclusion holds either way. Only the direction attached to the beta carotene HR is reversed in the wording. The authors should confirm which way the comparison runs.
Second check: another model agreed
Report a problem with this finding
Wording The lutein + zeaxanthin main effect on advanced AMD, with identical eyes and events (3451/940 vs 3440/1000), is given as '0.91 (0.82-1.00)', P=.05 in Figure 4 and the text, but as '0.90 (0.82-0.99)', P=.04 in Figure 6. Which model (stratified or unstratified by the secondary interventions) does each come from?
- p. 7 Figure 4, Lutein + zeaxanthin row
0.91 (0.82-1.00)
- p. 6 Results, Main Effects
0.91 (95% CI, 0.82-1.00; P=.05)
- p. 8 Figure 6, Advanced AMD, Lutein + zeaxanthin row
0.90 (0.82-0.99)
Only the authors can say what is right here.
Judge's note: Both figures were checked on the images. The methods (p4 text) say the models were run 'with and without stratification by the secondary interventions'. The p6 text ties Figure 4 to the stratified analysis. Figure 6 does not say which model it uses, so the two sets of numbers may come from different models. The paper never says so, and the two P values fall either side of .05. This result is not in the abstract, so it is a query to the authors rather than a correction.
Second check: another model agreed
Report a problem with this finding
Wording Discussion p10: 'AREDS2 participants who were not current smokers ..., and those assigned to one of the 2 groups receiving the AREDS formulation with beta carotene, showed an increased incidence of lung cancers (23 [2.0%] and 11 [0.9%], respectively...)'. 'Respectively' maps 23 to the non-smokers and 11 to the beta carotene group, the reverse of the Results (p7). The percentages also cannot be reproduced from the stated group sizes.
- p. 10 Discussion, para 3
showed an increased incidence of lung cancers (23 [2.0%] and 11 [0.9%], respectively
- p. 10 Discussion, column 1
cers (23 [2.0%] and 11 [0.9%], respec-
- p. 9 Table 4, Incident lung neoplasm row
9 (1.4)
- p. 7 Safety Outcomes, Serious Adverse Events
(23 [2.0%] vs 11 [0.9%])
Only the authors can say what is right here.
Judge's note: The Results on p7 read 'beta carotene group than in the no beta carotene group (23 [2.0%] vs 11 [0.9%])', so the Discussion sentence is garbled and never names the comparison group. On denominators: the beta carotene groups have 659+689 = 1348 participants, none of them smokers. Table 4 gives 9+14 = 23 lung neoplasms, and 23/1348 = 1.7%, not 2.0%. The no-beta-carotene groups minus smokers number 863+825-181-166 = 1341, and 11/1341 = 0.8%, not 0.9%. The estimator or denominator behind the printed percentages is not stated, so this part is a query to the authors.
Second check: another model agreed
Report a problem with this finding
Wording Mortality p8: 'The HR for mortality comparing lutein + zeaxanthin vs no lutein + zeaxanthin was 1.06 (95% CI, 0.87-1.31; P=.56) for lutein + zeaxanthin vs no lutein + zeaxanthin and 1.16 ... for DHA + EPA'. The comparison is repeated, and as written the DHA + EPA HR falls under the lutein comparison.
- p. 8 Results, Mortality
(95% CI, 0.87-1.31; P=.56) for lutein + zeaxanthin vs no lutein + zeaxanthin and 1.16
Judge's note: Read on the page 8 image (column 1). The comparison label appears twice, and the sentence subject grammatically covers the DHA + EPA HR as well. The intended meaning can be recovered.
Second check: another model agreed
Report a problem with this finding
Typo Interventions (p2): "DHA + EPA in ethyl ester form as ROPUFA 75 n-33 EE"; 'n-33' is presumably meant to be 'n-3'.
- p. 2 Interventions
ROPUFA 75 n-33 EE
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: the DSM product is named 'ROPUFA 75 n-3 EE'. The page image clearly prints 'n-33'. Omega-3 fatty acids are n-3, and n-33 is not a fatty-acid class.
Second check: none (judge only)
Report a problem with this finding
Typo Results (p4): "4058 (96%) white". 4058/4203 = 96.55%, which rounds to 97%.
- p. 4 Results, para 1
4058 (96%) white
Judge's note: Table 1 white counts 973+1014+1022+1049=4058, so the count is right. The other percentages in the same passage are rounded normally (57%, 65%, 35%). 96.55 is past the .5 midpoint, so this is not a rounding tie.
Second check: none (judge only)
Report a problem with this finding
Typo Results (p4): "took at least 75% the study medications" is missing 'of'.
- p. 4 Results, adherence paragraph
took at least 75% the study medications
Judge's note: The page image shows the missing 'of'. The parallel clause reads '75% of the variations'.
Second check: none (judge only)
Report a problem with this finding
Typo Table 1, Women, Lutein + Zeaxanthin: "596 (57.0)". 596/1044 = 57.09%, which rounds to 57.1.
- p. 5 Table 1, Women row, Lutein + Zeaxanthin
596 (57.0)
Judge's note: The count is consistent: the women total 548+596+603+641=2388. The other percentages in the row are rounded correctly. The error is a 0.1-point rounding slip.
Second check: none (judge only)
Report a problem with this finding
Typo Table 1, Placebo, Education 'Some college or associate's degree': '255 (25.6)'. Using the column's education denominator of 994, 255/994 = 25.65%, which rounds to 25.7.
- p. 5 Table 1, Education, Some college, Placebo
255 (25.6)
Judge's note: Checked on the page 5 image. The placebo education counts are 61+268+255+191+219 = 994. The other placebo percentages match 994: 61/994 = 6.1, 268/994 = 27.0, 191/994 = 19.2 and 219/994 = 22.0. The missing counts in the four columns are 18+19+18+23 = 78, which matches footnote a ('Seventy-eight refused'). 25.654 is not a rounding tie. This is a 0.1-point rounding slip.
Second check: none (judge only)
Report a problem with this finding
Typo Table 2, Placebo (n = 1012), 'No AREDS supplement': '4 (0.3)'. 4/1012 = 0.40%, which rounds to 0.4.
- p. 6 Table 2, No AREDS supplement, Placebo
4 (0.3)
Judge's note: Checked on the page 6 image. The placebo column sums to 1012 (168+201+184+190+4+265). The other cells in the row round correctly: 8/1044 = 0.8, 5/1068 = 0.5, 2/1079 = 0.2. Only the placebo cell is off.
Second check: none (judge only)
Report a problem with this finding
Typo Subgroup Analyses p7: '0.89 (95% CI, 0.79-1.00]; P=.05)' and '0.92 (95% CI, 0.78-1.07]; P=.27)' each open with a parenthesis and close with a square bracket.
- p. 7 Results, Subgroup Analyses para 2
0.89 (95% CI, 0.79-1.00]; P=.05)
- p. 7 Subgroup Analyses paragraph
0.92 (95% CI, 0.78-1.07]; P=.27)
Judge's note: The crop image shows the ']' after both upper bounds, with no '[' opened before them.
Second check: none (judge only)
Report a problem with this finding
Typo Figure 5 column header reads 'Lutein + Zeathanxin Dietary Intake, µg'; it should be 'Zeaxanthin'.
- p. 8 Figure 5 table header
Lutein + Zeathanxin Dietary Intake, µg
- p. 8 Figure 5, top-left column header
Lutein + Zeathanxin Dietary Intake
Judge's note: The misspelling is visible in the figure image on p8, so it is not an extraction artefact.
Second check: none (judge only)
Report a problem with this finding
Typo The Figure 6 footnote defines 'DHA indicates docosahexaenoic acid; EPA, eicosapentaenoic acid', but neither abbreviation appears in Figure 6.
- p. 8 Figure 6 footnote
DHA indicates docosahexaenoic acid; EPA, eicosapentaenoic acid.
Judge's note: The page 8 image confirms that Figure 6 (and Figure 5) contain only lutein + zeaxanthin and beta carotene comparisons. The footnote appears to be boilerplate carried over. It is harmless.
Second check: none (judge only)
Report a problem with this finding
Typo Discussion p9: '(HR, 0.74 [95% CI, 0.59-0.94; P=.01)'. The square bracket is never closed.
- p. 9 Discussion, para 2
(HR, 0.74 [95% CI, 0.59-0.94; P=.01)
Judge's note: The page 9 image shows '[95% CI, 0.59-' at the end of column 2 and '0.94; P=.01),' at the start of column 3, with no ']'.
Second check: none (judge only)
Report a problem with this finding
Typo Table 3 column header reads 'Lutein + Zeaxanthin and EPA + DHA', while the same table's 'DHA + EPA' column and the rest of the paper use 'DHA + EPA'.
- p. 9 Table 3 column header
and EPA + DHA
- p. 9 Table 3 column header
and EPA DHA
- p. 9 Table 3 column header
and EPA ⫹ DHA
Judge's note: Seen on the page 9 image. The order inconsistency is cosmetic.
Second check: none (judge only)
Report a problem with this finding
Typo The system organ class is labelled 'Infections' in Table 3 but 'Infections and infestations' in Table 4.
- p. 9 Table 3 vs Table 4 row labels
Infections and infestations
Judge's note: Both labels are visible on the page 9 image. The labelling is inconsistent, but either form could be acceptable as a house abbreviation.
Second check: none (judge only)
Report a problem with this finding
Typo The title of reference 32 misspells 'carotenoids' as 'caretenoids': "outcomes in a clinical trial of caretenoids with coan-tioxidants".
- p. 11 Reference 32
clinical trial of caretenoids with coan-
- p. 11 Reference 32
outcomes in a clinical trial of caretenoids with coan-
- p. 11 Reference 32
trial of caretenoids with coan-
Judge's note: The page 11 image (crop C032-a) shows 'caretenoids' in the title of ref 32. Everywhere else, the article and its references spell it 'carotenoids' (e.g., abstract and refs 26-28). This is a spelling error on the page, not an extraction artefact.
Second check: none (judge only)