Effect of Intensive Blood-Glucose Control with Metformin on Complications in Overweight Patients with Type 2 Diabetes (UKPDS 34)

• p. 3 Methods, Trial in non-overweight and overweight sulphonylurea-treated patients

  early addition of metformin to the sulphonylurea (n=269) or continued sulphonylurea alone (n=268; figure 2)

• p. 3 Figure 2

  269 sulphonylurea alone

• p. 1 Abstract, Methods

  sulphonylurea therapy alone (n=269) or addition of metformin (n=268)

• p. 9 Table 2 header

  Sulphonylurea alone (n=269)

Judge's note: Checked on the page images. Page 3 Methods text gives metformin addition n=269 and SU alone n=268. Figure 2 on the same page shows 269 SU alone and 268 SU+metformin. The Abstract (p1) and Table 2 headers agree with Figure 2, and so does Figure 10 (events/268 for SU+metformin, /269 for SU). The Methods sentence has the arm sizes swapped.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Results, Combined analysis of both trials

  with a 12% reduced risk for any diabetes-related endpoint (p=0·033)

• p. 1 Abstract, Findings

  risk reduction 19% [2-33], p=0·033

• p. 10 Discussion and Figure 10, Combined groups

  net reduction of 19% in any diabetes-related endpoint (p=0·033)

• p. 10 Figure 10, Combined groups, any diabetes-related endpoint

  0·81 (0·67–0·98)

Judge's note: Page 9 image shows '12% reduced risk ... (p=0·033)', with the text pointing to Figure 10. In Figure 10 (page 10), the combined-groups row for any diabetes-related endpoint reads 179/610 vs 242/680, 0·033, 0·81 (0·67–0·98). The Abstract (page 1) says 'risk reduction 19% [2–33], p=0·033' and the Discussion (page 10) says 'net reduction of 19% ... (p=0·033)'. This is one analysis reported two ways. The meta-analysis in the next sentence has a different p (0·026), so it does not explain the 12%.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Results, Epidemiological analysis

  The 4417 patients had 45 527 person-years of follow-up

• p. 4 Methods, Epidemiological assessment

  The aim of the epidemiological assessment in 4416 participants

• p. 1 Abstract

  concurrent therapy of diabetes in 4416 patients

• p. 9 Results, Epidemiological analysis

  The 4417 patients had 45 527 person-years

Only the authors can say what is right here.

Judge's note: The images confirm 4416 in the Abstract (page 1, 'in 4416 patients') and in Methods (page 4, 'epidemiological assessment in 4416 participants'), and 4417 in Results (page 9). I found no exclusion or addition that would explain a difference of one patient. Only the authors can say which figure is correct.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 3 Figure 2 trial profile

  411 not eligible—FPG >15·0 mmol/L or taking metformin or insulin

• p. 3 Figure 2

  1148 attending UKPDS clinics in 1990

• p. 3 Figure 2

  411 not eligible-FPG >15·0 mmol/L

• p. 3 Figure 2

  211 not eligible

Only the authors can say what is right here.

Judge's note: Figure 2 on page 3 shows the parent box 1148 (1234-86=1148, consistent) with three mutually exclusive branches. 211+537+411=1159, which is 11 more than 1148. The 537 is confirmed by 269+268. Either 211 or 411 is off by 11; only the authors can say which. No overlap rule is stated.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 4 Table 1, Smoking (%) never/ex/current, Insulin column

  37/34/39

Judge's note: Page 4 image checked. Every other column sums to 100: 39/36/25, 43/32/25, 38/30/32, 34/35/31, 38/34/28. Weighting current smokers by group n reproduces the All-patients value of 28% only if the insulin value is 29, so 39 is evidently a misprint for 29.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 4 Table 1, M/F row

  784 (46%)/920

Only the authors can say what is right here.

Judge's note: Page 4 image checked. Within each group M+F equals n (411, 342, 409, 265, 277), and the groups partition the 1704 patients. Male total 788 ≠ 784; female total 916 ≠ 920. The 46% fits either male count.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 6 Results, paragraph on Figure 3

  cohort of 482 patients with data available studied over 10 years

• p. 5 Figure 3 legend, Patients followed for 10 years

  n=200

• p. 10 Figure 9 legend

  Conventional (n=129)

• p. 5 Figure 3 legend

  n=200 Conventional

Only the authors can say what is right here.

Judge's note: Figure 3 (p5): each 10-year cohort n equals the cross-sectional n at 10 years, and they total 857; conventional plus metformin is 381. No combination gives 482. Figure 9 (p10) describes the 10-year cohorts from primary randomisation in the diet vs metformin study. Its metformin n=181 matches Figure 3, but its conventional n=129 differs from Figure 3's 200 (129 is the chlorpropamide cohort). No subset definition explains either number.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 6 Results, paragraph on person-years

  there were 3557 (83%) of person-years with crossover to metformin therapy

• p. 6 Results, same paragraph

  2395 (56%) were treated by diet. The remaining 44% of person-years required

• p. 6 Results

  3557 (83%)

• p. 6 Results

  The remaining 44% of person-

Only the authors can say what is right here.

Judge's note: Page 6 image checked. 3557/4292 = 82.9%, but only 44% of conventional person-years involved any added drug, so 83% cannot be person-years with metformin. The parallel sentence for the sulphonylurea trial reads '75% of person-years without metformin therapy', which suggests 'without' was meant. Only the authors can confirm.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Table 2, M/F row

  164 (61%)/108

• p. 9 Table 2, M/F row

  158 (59%)/118

• p. 9 Table 2, M/F row

  322 (60%)/226

Only the authors can say what is right here.

Judge's note: The Table 2 image (page 9) confirms the values. The male percentages fit the stated n, but male plus female totals are 272, 276 and 548, each larger than its column n. Missing data could only make the totals smaller. The female counts implied by n are 105, 110 and 215.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Table 2, Total cholesterol row

  5·9 (1·0)

• p. 9 Table 2, Alcohol row

  34/52/13/0·8

Only the authors can say what is right here.

Judge's note: The Table 2 image (page 9) confirms 5·9, 5·6 and 5·6. With arms of 269 and 268, the pooled mean is about 5·75, and rounding limits keep it at or above about 5·70. The LDL, HDL and triglyceride rows pool correctly. Only the authors can say which cholesterol value is wrong. The alcohol row raised in reviewer note 3 is judged under C022.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Table 2, Alcohol (%) none/social/regular/dependent

  34/52/13/0·8

Only the authors can say what is right here.

Judge's note: The Table 2 image (page 9) confirms these values. The two arms are nearly equal in size, so a pooled percentage must lie between the arm values. Pooling gives about 47·5% social and 17% regular. The smoking and exercise rows pool correctly.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 9 Table 2, Ethnicity row, All patients column

  77/11/11/1

Only the authors can say what is right here.

Judge's note: The Table 2 image (page 9) confirms these values. 'Other' is 0% in both arms, so each arm has at most 1 patient (under 0·5%), at most 2 overall, which is 0·4% and rounds to 0, not 1. For Indian Asian, rounding allows 34–36 and 29–30 patients, a pooled range of about 11·7–12·3%, which rounds to 12, not 11.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 10 Figure 10, Peripheral vascular disease, Combined groups, Added metformin absolute risk

  1·7

Only the authors can say what is right here.

Judge's note: Figure 10 (page 10 image). The component rows imply about 3750 and 1540 patient-years, so 8 events give about 1·5 per 1000. Even at the rounding limits the maximum is about 1·56. The other combined rows reconcile: any endpoint 38·7, stroke 5·2, microvascular 7·7, and the PVD conventional column 1·7.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 10 Figure 9 legend

  Conventional (n=129)

• p. 5 Figure 3 legend

  n=200

• p. 5 Figure 3 key

  Conventional (411, 309, 200, 22)

Only the authors can say what is right here.

Judge's note: Page 10 image: the Figure 9 legend reads Conventional (n=129) and Metformin (n=181). Its caption says these are cohorts followed to 10 years from primary randomisation. Page 5 image: Figure 3 lists the 10-year cohorts as n=200 Conventional, n=129 Chlorpropamide and n=181 Metformin, and the conventional row reads (411, 309, 200, 22). The metformin counts match but the conventional counts do not.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 11 Figure 11 column header

  RR (95%CI)

• p. 5 Methods, Statistical analysis

  For single endpoints 99% CIs are quoted

Judge's note: Header seen on the p11 image; Methods on the p5 crop. Check on fatal MI, 1·79 (0·64–4·99) with 17 vs 10 events: half-width on the log scale is ln(4.99/0.64)/2 ≈ 1.027, and SE ≈ sqrt(1/17+1/10) ≈ 0.398, so z ≈ 2.58, which is 99%. Non-fatal MI, 0·62 (0·26–1·48) with 14 vs 23 events, also gives z ≈ 2.57. So the label, not the data, is wrong.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 7 Figure 5, x-axis of diabetes-related death panel

  16

Judge's note: Page 7 image: the lower panel of Figure 5 has ticks labelled 0, 3, 6, 9, 12, 16, equally spaced, with patients-at-risk columns beneath them. Figure 7 (page 8) and Figure 3 (page 5) end at 15. The 3-year spacing means the last tick is 15, so this is a mislabelled axis tick.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send

• p. 11 Discussion, Clinical use of metformin in overweight patients

  lessen the likelihood of extension of a thrombolysis

• p. 11 Discussion, Clinical use of metformin in overweight patients

  the likelihood of extension of a thrombolysis

• p. 11 Discussion

  likelihood of extension of a thrombolysis

Judge's note: Read on the page 11 image, left column. The sentence argues that lower PAI-1 increases fibrinolysis. That would limit extension of a thrombus, not of thrombolysis, so the wrong word inverts the logic. The fix is evident.

Second check: another model agreed

Report a problem with this finding

This finding is

wrong right unclear or overstated already corrected (erratum, later version)

I am an author of this paper Send