Major Outcomes in High-Risk Hypertensive Patients Randomized to Angiotensin-Converting Enzyme Inhibitor or Calcium Channel Blocker vs Diuretic
2002 · paper · Clinical trials
A randomized double-blind trial of 33,357 older, high-risk hypertensive people comparing amlodipine or lisinopril with the diuretic chlorthalidone. Coronary heart disease outcomes did not differ between drugs; heart failure was more frequent with amlodipine.
6 findings
Substance Chlorthalidone 5-year medication use: the identical value "5.8% [n=399]" is given both for "a diuretic with a CCB" and for "a CCB ... without a diuretic", and the component n/percentage pairs imply a denominator of about 6880, while the totals (67.5% [n=4387], 13.2%, 9.0%; Table 2) imply about 6500.
- p. 4 Results, Visit and Medication Adherence, chlorthalidone crossovers
13.2% were taking a diuretic with a CCB (5.8% [n=399])
- p. 4 same paragraph
Only 9.0% were tak-ing either a CCB (5.8% [n=399])
- p. 4 Visit and Medication Adherence, chlorthalidone group
13.2% were taking a diuretic with a CCB (5.8% [n=399]) or an ACE inhibitor (9.3% [n=641])
- p. 6 Table 2, chlorthalidone partial crossovers year 5
860 (13.2)
Only the authors can say what is right here.
Judge's note: Checked the p4 text (crop C003-c, full page) and Table 2 on p6. The same 5.8% [n=399] appears for two mutually exclusive categories (with and without a diuretic), which suggests a copying slip. The reviewers' argument that the components must add up to 13.2% or 9.0% does not hold on its own. The Table 2 footnotes define partial and full crossovers as being on 'CCB or ACE inhibitor', so someone taking both could be counted in each component, and the union can be smaller than the sum. However, 399/0.058, 641/0.093 and 385/0.056 each imply a denominator of about 6880. The paragraph's own 4387/0.675 and Table 2 (860 = 13.2%, 583 = 9.0%, 5247 = 80.5%) imply about 6500, and 399/6499 = 6.1%, not 5.8%. The paper states no alternative denominator. Only the authors can say which values are correct.
Second check: another model agreed
Report a problem with this finding
Substance Table 4, fasting glucose among nondiabetics, number of participants at 4 years, chlorthalidone: "2606 (40.3)" — 2606/6766 = 38.5%.
- p. 8 Table 4, Fasting Glucose Among Nondiabetics, No. of participants, 4 Years, chlorthalidone
2606 (40.3)
Judge's note: Crop C014-a confirms the printed values. All the sibling cells use the baseline n: 3074/6766 = 45.4, 1787/3954 = 45.2, 1567/3954 = 39.6 and 1464/4096 = 35.7. The count 2606 is supported by 302/2606 = 11.6%, which matches both the table and the text. The printed percentage is therefore the value that is off; it should be 38.5.
Second check: another model agreed
Report a problem with this finding
Reference or label Figure 1 legend: "which captures more information than that reported previously.24" — ref 24 is Levey et al, the MDRD GFR estimation equation, not the earlier doxazosin report.
- p. 5 Figure 1 footnote, last sentence
information than that reported previously.24
- p. 17 Reference 24
Levey AS, Bosch JP, Lewis JB, et al. A more ac-curate method to estimate glomerular filtration rate
- p. 5 Figure 1 legend, last sentence
which captures more information than that reported previously.24
- p. 17 Reference 24
A more accurate method to estimate glomerular filtration rate from serum creatinine
Judge's note: The p5 legend cites 24. The p17 reference list gives ref 24 as Levey AS et al, 'A more accurate method to estimate glomerular filtration rate from serum creatinine', Ann Intern Med 1999. The Introduction (p2) and the Comment cite the earlier doxazosin report as refs 19 and 20. Ref 24 is cited correctly for eGFR in the Methods.
Second check: another model agreed
Report a problem with this finding
Reference or label Figure 1 legend: "On January 24, 2000, the NHLBI decided to discontinue the doxazosin group and report results.18" — ref 18 is the 1996 rationale-and-design paper.
- p. 5 Figure 1 footnote
decided to discontinue the doxazosin group and report results.18
- p. 17 Reference 18
Rationale and design for the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT). Am J Hypertens. 1996
Judge's note: Ref 18 on p17 is Davis BR et al, 'Rationale and design for ... ALLHAT', Am J Hypertens 1996, so it cannot document a 2000 decision. The Introduction cites refs 19 and 20 for the early termination of the doxazosin arm. Ref 19 is the JAMA 2000 doxazosin report.
Second check: another model agreed
Report a problem with this finding
Reference or label 'trial results in isolated systolic hypertension suggest that a 3-mm Hg higher systolic BP could explain a 10% to 20% increase in risk.8,47' cites ref 47, 'Antihypertensive treatment based on conventional or ambulatory blood pressure measurement: a randomized controlled trial'.
- p. 13 Comment, column 1
trial results in isolated systolic hyper- tension suggest that a 3-mm Hg higher systolic BP could explain a 10% to 20% increase in risk.8,47
- p. 17 Reference 47
Antihypertensive treatment based on conventional or ambulatory blood pressure measurement: a randomized controlled trial
- p. 13 Discussion
suggest that a 3-mm Hg higher systolic BP could explain a 10% to 20% increaseinrisk.8,47
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that the Staessen 1997 ambulatory BP monitoring trial does not report isolated-systolic-hypertension heart failure outcomes. The page image of p17 confirms the title of ref 47, which is about conventional vs ambulatory BP-guided treatment. Ref 47 is first cited at this point, consistent with citation-order numbering, but its title does not match the claim it supports. A different Staessen paper may have been intended; only the authors can confirm.
Second check: another model agreed
Report a problem with this finding
Wording Comment: angioedema "occurred 4 times more frequently in participants randomized to lisinopril", but the reported counts (38/9054 vs 8/15255) give about 8 times.
- p. 9 Comment, paragraph on lisinopril vs chlorthalidone
occurred 4 times more frequently
- p. 7 Primary Safety Outcomes
8 of 15255 (0.1%)
- p. 9 Comment, column 1, paragraph on lisinopril vs chlorthalidone
occurred 4 times more frequently in participants randomized to lisinopril
- p. 7 Results, Primary Safety Outcomes
Angioedema occurred in 8 of 15255 (0.1%), 3 of 9048 (<0.1%), and 38 of 9054 (0.4%)
Judge's note: The p7 image shows 8 of 15255 (0.1%), 3 of 9048 (<0.1%) and 38 of 9054 (0.4%). The 3/9048 sub-point is a text-layer artefact: the '<' is present in the image (see C012). The '4 times' figure only follows from the rounded 0.4%/0.1%. The unrounded ratio is 8.0, and follow-up was similar across arms. Table 5 gives no angioedema RR that would support 4.
Second check: another model agreed, code check
Report a problem with this finding
14 minor (typos and the like; most checked by the judge only)
Wording Figure 1 amlodipine Year 5 "1052 Discontinued Study Drug" does not match the text percentages for reasons ("40.5% [n=443]", "16.4% [n=180]"), which imply a denominator of about 1094, the same as Table 2 (3920 - 2826).
- p. 5 Figure 1, amlodipine box, Year 5
1052 Discontinued Study Drug
- p. 4 Results, Visit and Medication Adherence
40.5% [n=443]
Only the authors can say what is right here.
Judge's note: Checked Figure 1 (p5) and the p4 text. For chlorthalidone, the text percentages use the Figure 1 total of 1873 exactly (775/1873 = 41.4%, 282/1873 = 15.1%). For amlodipine, 443/1052 = 42.1%, not 40.5%. The denominator implied by the text matches Table 2 (5-year denominator 3151/0.804 = 3919, minus 2826 on blinded drug = about 1094). Either the figure count or the percentages use a different base, and the paper does not say which. Only the authors can resolve this.
Second check: another model agreed
Report a problem with this finding
Wording Figure 1 lisinopril Year 5 "1399 Discontinued Study Drug" does not match the text percentages ("37.9%[n=552]", "18.1% [n=264]", "9.0% [n=131]"), which imply about 1456 to 1463, as does Table 2 (2736/0.726 - 2307).
- p. 5 Figure 1, lisinopril box, Year 5
1399 Discontinued Study Drug
- p. 4 Results, Visit and Medication Adherence
37.9%[n=552]
Only the authors can say what is right here.
Judge's note: Checked the same pages as C007. 552/1399 = 39.5%, 264/1399 = 18.9% and 131/1399 = 9.4%. The text percentages instead imply about 1456. Table 2 gives 3769 - 2307 = about 1462. For chlorthalidone, the figure count and the text percentages agree, so the bases are inconsistent. Only the authors can resolve this.
Second check: another model agreed
Report a problem with this finding
Wording Results (p6) say 'Cause-specific mortality rates (except for unintentional injuries/suicides/homicides ...) were similar for the 2 groups (TABLE 6)', but the Comment (p9) says 'the mortality from noncardiovascular causes was significantly lower in the CCB group (Table 6)', and Table 6 gives P = .02 for noncardiovascular deaths (.04 in the corrected Table 6).
- p. 9 Comment, column 1
the mortality from noncardiovascular causes was significantly lower in the CCB group (Table 6)
- p. 6 Results, Amlodipine vs Chlorthalidone
Cause- specific mortality rates (except for unintentional injuries/suicides/ homicides in amlodipine compared with chlorthalidone, not a prespeci- fied hypothesis) were similar for the 2 groups (TABLE 6).
- p. 12 Table 6, Noncardiovascular row
1058 (8.9)
Only the authors can say what is right here.
Judge's note: Checked Results p6, Comment p9, the original Table 6 on p12 (Noncardiovascular amlodipine vs chlorthalidone P = .02) and the replacement Table 6 in the Corrections notice on p18 (P = .04). The correction does not resolve the problem, because the noncardiovascular P value stays below .05. The Results sentence names only the injury/suicide/homicide category as an exception, while the Discussion relies on the noncardiovascular aggregate being significant. This is a wording inconsistency between the two sections, and only the authors can say which statement they intend.
Second check: another model agreed
Report a problem with this finding
Wording Comment says 'In the lisinopril vs chlorthalidone comparisons, there were 2 outcomes with significant interactions', while Results also report a significant age interaction for combined CHD ('P=.01 for interaction').
- p. 13 Comment, subgroup paragraph
there were 2 outcomes with significant interactions
- p. 6 Results, Lisinopril vs Chlorthalidone
differential effect by age (P=.01 for in-teraction)
- p. 13 Comment, column 3
In the lisinopril vs chlorthalidone comparisons, there were 2 outcomes with significant interactions.
- p. 6 Results, Lisinopril vs Chlorthalidone
For com- bined CHD, there was a significant differential effect by age (P=.01 for in- teraction)
Judge's note: Results (p6) report three significant interactions for lisinopril vs chlorthalidone: combined CHD by age (P = .01), stroke by race (P = .01) and combined CVD by race (P = .04). The Results add that with age modelled as a continuous variable the age interaction was not significant, which is probably why the Discussion (p13) counts only 2. The Discussion does not state that qualification, so the count is ambiguous as written. Minor consistency point.
Second check: another model agreed
Report a problem with this finding
Wording 'mean follow-up BPs were well below 140/90 mm Hg in all treatment groups' overstates Table 3, where lisinopril mean systolic BP is 140.0 (18.5) at 1 year and 138.4 at 2 years.
- p. 14 Comment, limitations paragraph
mean fol- low-up BPs were well below 140/90 mm Hg in all treatment groups
- p. 7 Table 3, Systolic BP, 1 Year, Lisinopril
140.0 (18.5)
Only the authors can say what is right here.
Judge's note: Table 3 (p7) gives lisinopril mean SBP of 140.0, 138.4, 136.7, 135.5 and 135.9 mm Hg at years 1-5, and amlodipine 138.5 at year 1. The 1-year lisinopril value equals the 140 mm Hg threshold, so 'well below' overstates the data for systolic BP at least early in follow-up.
Second check: another model agreed
Report a problem with this finding
Wording Conclusion 'They are unsurpassed in lowering BP' is not qualified, although Table 3 shows lower diastolic BP with amlodipine than chlorthalidone (Δ −0.6 to −0.8 mm Hg, P < .001 at years 1-5).
- p. 14 Comment, concluding paragraph
They are unsurpassed in low- ering BP, reducing clinical events, and tolerability
- p. 7 Table 3, Diastolic BP, Δ mm Hg, amlodipine
−0.8
Judge's note: Table 3 (p7) shows the diastolic Δ for amlodipine vs chlorthalidone as −0.6, −0.6, −0.8, −0.8, −0.8, with P < .001 at every year. Chlorthalidone did achieve the lowest systolic BP, so the claim holds for systolic BP. The unqualified wording sits uneasily with the paper's own diastolic data. Minor wording point.
Second check: another model agreed
Report a problem with this finding
Wording 'chlorthalidone was superior to lisinopril for several CVD outcomes ... in both diabetic and nondiabetic participants' is not clearly supported for diabetic participants in Figure 6.
- p. 13 Comment, column 2
chlorthali- done was superior to lisinopril for sev- eral CVD outcomes and superior to am- lodipine for HF in both diabetic and nondiabetic participants
- p. 13 Figure 6, Diabetic rows
1.08 (1.00-1.17)
Only the authors can say what is right here.
Judge's note: Figure 6 (p13), diabetic rows for lisinopril vs chlorthalidone: HF 1.22 (1.05-1.42); combined CVD 1.08 (1.00-1.17), whose lower bound touches 1; stroke 1.07 (0.90-1.28); combined CHD 1.03 (0.93-1.14); CHD 1.00 (0.87-1.14). Only HF clearly excludes 1. The nondiabetic rows do support 'several' outcomes (stroke 1.23, CVD 1.12, HF 1.20). The wording overstates the diabetic-subgroup data.
Second check: another model agreed
Report a problem with this finding
Typo "258 (2.8%) of the amlodipine group" had unknown vital status; 258/9048 = 2.85%, which rounds to 2.9%.
- p. 3 Results, Visit and Medication Adherence
258 (2.8%) of the amlodipine
Judge's note: The p3 image shows 258 (2.8%). Figure 1 (p5) gives 200 lost plus 58 refused = 258. The neighbouring values in the same sentence are rounded correctly: 419/15255 = 2.7% and 276/9054 = 3.0%. The slip is a 0.1-point rounding error.
Second check: none (judge only)
Report a problem with this finding
Typo Table 1 footnote ‡ reads "stress thalium or dipyridamole thalium"; the correct spelling is "thallium".
- p. 4 Table 1 footnote ‡
stress thalium or dipyridamole thalium
Judge's note: The p4 page image shows 'thalium' twice in the footnote. This is a spelling error.
Second check: none (judge only)
Report a problem with this finding
Typo Table 1, LVH by echocardiogram: amlodipine "411 (4.6)" (411/9048 = 4.5%), and lisinopril "402 (4.5)" (402/9054 = 4.4%).
- p. 4 Table 1
411 (4.6)
Judge's note: Crop C006-a confirms the printed values. Other Table 1 rows on the arm totals reproduce exactly (for example LVH by ECG and HDL-C). In this row, two cells are 0.1 above the arm-total percentage. This may reflect a denominator of participants with echo data, but no such denominator is stated.
Second check: none (judge only)
Report a problem with this finding
Typo Author Contributions, Drafting of the manuscript: 'Leenan, Oparil, Probstfield' misspells Leenen.
- p. 14 Author Contributions, Drafting of the manuscript
Leenan, Oparil, Probstfield
- p. 14 Drafting of the manuscript
Leenan
- p. 14 ALLHAT Authors list
Frans Leenen
- p. 14
Leenan, Oparil, Probstfield, Whelton, Nwachuku,
Judge's note: The page image of p14 shows 'Leenan' in the Drafting line. The author list reads 'Frans Leenen', and the other contribution lines read 'Leenen'.
Second check: none (judge only)
Report a problem with this finding
Typo Financial Disclosures list 'Forrest Labs' (several times).
- p. 14 Financial Disclosures
Forrest Labs
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: the company's name is Forest Laboratories. The page image of p14 shows 'Forrest Labs' under Wright, Davis, Cushman and Oparil. Cosmetic spelling point.
Second check: none (judge only)
Report a problem with this finding
Typo Calhoun disclosure 'Novartis, GlaxoSmithKline; Solvay; McDowell' has a stray semicolon before Solvay.
- p. 14 Financial Disclosures
Novartis, GlaxoSmithKline; Solvay; McDowell
Judge's note: The page image of p14 shows 'Calhoun: AstraZeneca, Aventis, Merck, Novartis, GlaxoSmithKline; Solvay; McDowell:'. Elsewhere in the list, semicolons separate authors and commas separate companies, so this semicolon is a punctuation slip.
Second check: none (judge only)
Report a problem with this finding
Typo Ref 29 gives the subtitle as 'Survival Analysis: Techniques for Censored and Truncated Regression' (the book's subtitle ends 'Data').
- p. 17
Techniques for Censored and Truncated Regression.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: the subtitle of Klein & Moeschberger (Springer, 1997) is 'Techniques for Censored and Truncated Data'. The page image of p17 shows 'Regression'. The book is still identifiable, so this is cosmetic.
Second check: none (judge only)