Intensive Blood Glucose Control and Vascular Outcomes in Patients with Type 2 Diabetes
Journal of Vascular Surgery, 2008 · paper · Clinical trials
A randomized trial of 11,140 people with type 2 diabetes comparing intensive glucose control (HbA1c 6.5% or less) with standard control. Intensive control reduced combined vascular events by 10%, mainly through less nephropathy, with no significant effect on macrovascular events and more severe hypoglycemia.
6 findings
Substance Microalbuminuria lower bound is converted as "30 to 300 μg per milligram [0.34 to 33.9 mg per millimole]"; at the paper's own factor (300 → 33.9), 30 μg/mg is 3.39 mg/mmol, not 0.34.
- p. 3 End Points, definition of microalbuminuria
30 to 300 μg per milligram [0.34 to 33.9 mg per millimole]
- p. 3 Secondary outcomes, microalbuminuria definition
30 to 300 μg per milligram [0.34 to 33.9 mg
Judge's note: Page 3 image shows "[0.34 to 33.9 mg per millimole]". The macroalbuminuria definition on p3 and Table 1 footnote † both give 300 μg/mg = 33.9 mg/mmol (factor 0.113), so 30 μg/mg converts to 3.39 mg/mmol. The printed 0.34 is too small by a factor of 10. Both crops show the same passage.
Second check: another model agreed
Report a problem with this finding
Substance Table 1 end-of-follow-up 'None — no. (%)' row reads "42 (1.5)" (intensive, N=4828) and "220 (6.4)" (standard, N=4741); these percentages match no denominator implied by the column or the other drug rows.
- p. 6 Table 1 (continued), Glucose-lowering drug, None row, End of Follow-up columns
42 (1.5)
- p. 6 Table 1 (continued), None row, Standard Control end of follow-up
220 (6.4)
Only the authors can say what is right here.
Judge's note: Page 6 image confirmed. The footnote says denominators are patients who attended and had data. However, the other rows in the same columns imply denominators of about 4650–4830 (4525/0.937≈4829; 1953/0.405≈4822; 4001/0.844≈4740; 1142/0.241≈4739). On those denominators 42 is 0.9% and 220 is 4.6%. Getting 1.5% and 6.4% would need denominators of about 2800 and 3440. Either the counts or the percentages are misprinted; only the authors can say which.
Second check: another model agreed
Report a problem with this finding
Substance Figure 4 prints the new-onset microalbuminuria relative risk reduction as "9 (2 to 5)": the upper bound is below the point estimate, and the text's HR 0.91 (0.85 to 0.98) gives 9 (2 to 15).
- p. 9 Figure 4, New-onset microalbuminuria row, Relative Risk Reduction column
9 (2 to 5)
- p. 8 Results text, renal events
hazard ratio, 0.91; 95% CI, 0.85 to 0.98; P = 0.02
- p. 8 Results, Effects on Death and Other Secondary Outcomes
new-onset microalbuminuria (hazard ratio, 0.91; 95% CI, 0.85 to 0.98; P = 0.02)
- p. 11 Discussion, paragraph 3
a more modest, though significant, reduction in that of new-onset microalbuminuria
Judge's note: Page 9 image shows '9 (2 to 5)'. Page 8 text gives 'hazard ratio, 0.91; 95% CI, 0.85 to 0.98; P=0.02'. Other Fig. 4 rows convert exactly as 1−HR (e.g. nephropathy HR 0.79 (0.66–0.93) → 21 (7 to 34)). The interval cannot exclude its own estimate, so the upper bound should be 15.
Second check: another model agreed
Report a problem with this finding
Substance Figure 5 gives the glycated hemoglobin <7.2% subgroup's relative risk reduction as "10 (3 to 22)", which is inconsistent with its event counts (387 vs 430) and with its plotted interval crossing 1.0.
- p. 10 Figure 5, Glycated hemoglobin <7.2% row, Relative Risk Reduction column
10 (3 to 22)
- p. 11 Discussion, paragraph 1
There was no evidence that the effects of intensive glucose control were dependent on the baseline glycated hemoglobin
Only the authors can say what is right here.
Judge's note: Page 10 image shows '10 (3 to 22)' with no minus sign; other rows show minus signs clearly. The plotted horizontal line for this row extends past the solid 1.0 line, which matches a lower bound below zero. The female row, with similar event counts (374 vs 411), reads 10 (−3 to 22). Variance of log HR ≈ 1/387+1/430 = 0.0049 (SE ≈ 0.07), giving roughly HR 0.79–1.03, i.e. RRR about −3 to 22. A minus sign is probably missing; only the authors can confirm.
Second check: another model agreed
Report a problem with this finding
Substance Figure 5 gives the mean glycated hemoglobin reduction for the <7.2% subgroup as "0.46 (0.43 to 0.55)"; the interval runs 0.03 below and 0.09 above the estimate, whereas every other row is symmetric.
- p. 10 Figure 5, Glycated hemoglobin <7.2% row, Mean Glycated Hemoglobin Reduction column
0.46 (0.43 to 0.55)
Only the authors can say what is right here.
Judge's note: Page 10 image confirms the values. The other rows are symmetric within rounding: 0.86 (0.81 to 0.91), 0.52 (0.48 to 0.55), 0.70 (0.65 to 0.75). A mean-difference CI should be symmetric. The estimate, or more likely the upper bound (about 0.49 expected), appears misprinted. The authors must confirm.
Second check: another model agreed
Report a problem with this finding
Reference or label Table 1 labels the mean±SD triglyceride row "Serum triglycerides — μmol/liter", but its values (1.95±1.29, 1.96±1.29, 1.70±1.06, 1.82±1.15) are on the mmol/liter scale.
- p. 5 Table 1, Serum triglycerides row (mean ± SD)
Serum triglycerides — μmol/liter
Judge's note: Page 5 image: the row directly above, 'Serum triglycerides — mmol/liter', has medians of 1.60/1.64/1.45/1.59. The footnote's conversion factor (0.01129) is for mmol/L. Values near 2 μmol/L are impossible, so the unit label is wrong and the data are fine.
Second check: another model agreed
Report a problem with this finding
4 minor (typos and the like; most checked by the judge only)
Wording Figure 5 prints identical mean HbA1c reductions "0.70 (0.65 to 0.75)" for both the <65 yr and ≥65 yr subgroups, while the overall value is "0.67 (0.64 to 0.70)", below both.
- p. 10 Figure 5, Age rows
0.70 (0.65 to 0.75)
- p. 10 Figure 5, Overall
0.67 (0.64 to 0.70)
Only the authors can say what is right here.
Judge's note: Page 10 image confirms the values. Age partitions the whole cohort, so the overall estimate would be expected to lie between the subgroup estimates. Identical intervals for subgroups of different sizes (about 2280 vs 3290) also suggest a duplicated row. Mixed-model estimates need not be exact weighted averages, so this is an author query rather than a definite error.
Second check: another model agreed
Report a problem with this finding
Typo Results write "(hazard ratio, 0.90; 95% confidence interval (CI), 0.82 to 0.98; P=0.01)" with parentheses nested inside parentheses, where the Fig. 2 legend uses "[CI]".
- p. 7 Effects on Primary Outcomes
interval (CI), 0.82 to 0.98
Judge's note: Page 7 image confirms nested parentheses in the Results text, while the Fig. 2 legend on the same page uses the bracket form "95% confidence interval [CI]". This is a purely cosmetic punctuation inconsistency.
Second check: none (judge only)
Report a problem with this finding
Typo Appendix affiliation reads "University of Queenland, Brisbane, Australia" for "University of Queensland".
- p. 12 Appendix, Writing Committee affiliations (Paul Glasziou)
University of Queenland, Brisbane, Australia
- p. 12 Appendix, affiliation of Paul Glasziou
University of Queenland
Judge's note: Page 12 image shows 'Queenland'. Misspelling of the Australian state name.
Second check: none (judge only)
Report a problem with this finding
Typo Reference 33 title reads "results of prespectively designed overviews of randomized trials"; 'prespectively' should be 'prospectively'.
- p. 13 Reference 33
results of prespectively designed overviews
- p. 13 References, ref. 33 (Turnbull et al.)
results of prespectively designed overviews of randomized trials
- p. 13 Reference list, Turnbull et al.
prespectively designed
Judge's note: Page 13 image shows 'prespectively', which is not a word. The correction is evident without outside knowledge.
Second check: none (judge only)