Effect of Laparoscopic Sleeve Gastrectomy vs Laparoscopic Roux-en-Y Gastric Bypass on Weight Loss at 5 Years
2018 · paper · Clinical trials
A randomized trial of 240 morbidly obese patients in Finland, comparing sleeve gastrectomy with gastric bypass for weight loss at five years. Excess weight loss was 49% versus 57%; the two operations did not meet the prespecified criteria for equivalence.
10 findings
Substance Abstract and Conclusions say "the difference was not statistically significant, based on the prespecified equivalence margins", but the Results (p4) say "Gastric bypass resulted in statistically greater weight loss than sleeve gastrectomy at 5 years" (difference 8.2 [95% CI, 3.2%-13.2%]).
- p. 1 Abstract, Conclusions and Relevance
the difference was not statistically significant, based on the prespecified equivalence margins
- p. 13 Conclusions
ence was not statistically significant, based on the prespeci-
- p. 4 Results, Primary End Point
Gastric bypass re-sulted in statistically greater weight loss than sleeve gastrec-tomy at 5 years
- p. 13 Conclusions
the differ-ence was not statistically significant, based on the prespeci-fied equivalence margins
Judge's note: Images: abstract crop (p1), Results crop (p4) and Conclusions crop (p13). The Results call the 5-year difference statistically greater and describe it as not clinically significant against the margin. The Abstract and Conclusions use 'not statistically significant' for the same comparison. The two sections describe the same 5-year comparison in contradictory terms. The Methods use the equivalence margins for clinical significance or equivalence, not for statistical significance. No erratum is among the files.
Second check: another model agreed
Report a problem with this finding
Substance Baseline dyslipidemia: text "At baseline, 83 patients (35%) had dyslipidemia" and Abstract "34.6% dyslipidemia", but Table 1 gives 39 (32.2) + 45 (37.8) = 84.
- p. 5 Results, Dyslipidemia
At baseline, 83 patients (35%) had dyslipidemia.
- p. 1 Abstract, Results
34.6% dyslipidemia
- p. 4 Table 1, Dyslipidemia row
39 (32.2)
- p. 5 Dyslipidemia
At baseline, 83 patients (35%) had dyslipidemia
Only the authors can say what is right here.
Judge's note: Table 1 image (p4) shows 39 (32.2) and 45 (37.8), and both percentages agree with those counts. 84/240 = 35.0%, while 83/240 = 34.6% matches the abstract. The other baseline figures agree (diabetes 101 = 52 + 49; hypertension 170 = 83 + 87). No different population is stated for dyslipidemia.
Second check: another model agreed
Report a problem with this finding
Substance Sleeve vitamin B12 median printed as '54 (range, 79-1999) pg/mL (409 [range, 58-1475] pmol/L)'; the median lies below its own range and 409 pmol/L corresponds to about 554 pg/mL.
- p. 8 Post Hoc Outcomes, micronutrients paragraph
54 (range, 79-1999) pg/mL (409 [range, 58-1475] pmol/L)
- p. 8 Post Hoc Outcomes, micronutrients
54(range,79-1999) pg/mL (409
Judge's note: The p8 image shows '54'. The range bounds convert with the same factor (1 pg/mL = 0.738 pmol/L) as the SI values. 409 pmol/L converts to about 554 pg/mL, so a digit appears to have been dropped from the median.
Second check: another model agreed
Report a problem with this finding
Substance Bypass folate median '384 (range, 98-840) ng/mL (801 [range, 221-1904] nmol/L)': 384 ng/mL and 801 nmol/L do not match under the factor that the other folate values follow.
- p. 8 Post Hoc Outcomes, folate
384 (range, 98-840) ng/mL (801 [range, 221-1904] nmol/L)
- p. 8 micronutrients, folate
384 (range, 98-840) ng/mL
Only the authors can say what is right here.
Judge's note: Checked on the p8 image. The sleeve median and range, and the bypass range, all follow a factor of about 2.266. The bypass median does not: 384 ng/mL converts to about 870 nmol/L, and 801 nmol/L back-converts to about 353 ng/mL. One of the two values is misprinted, and only the authors can say which.
Second check: another model agreed
Report a problem with this finding
Substance Table 3 HbA1c model-based mean at 0.5 y printed as '6.6 (6.2 to 6.5)': the estimate lies above its upper bound.
- p. 10 Table 3, HbA1c model-based means in time points
6.6 (6.2 to 6.5)
Only the authors can say what is right here.
Judge's note: The p10 upright image shows '6.6 (6.2 to 6.5)'. No minus-sign loss is involved. Even on a back-transformed log scale, a point estimate cannot fall outside its own interval. Either the estimate or a bound is misprinted, and only the authors can say which.
Second check: another model agreed, code check
Report a problem with this finding
Reference or label Figure 1, bypass arm: "119 Randomized" → "117 Received intervention as randomized" plus "3 Did not receive intervention as randomized" (2 no surgery, 1 converted to sleeve) sums to 120, not 119.
- p. 5 Figure 1, gastric bypass arm
117 Received intervention as randomized
- p. 5 Figure 1, gastric bypass arm
3 Did not receive intervention as randomized
Only the authors can say what is right here.
Judge's note: The p5 crop confirms the numbers. 117 matches the number operated (119 − 2), and the next box follows from it: 117 − 6 lost = 111 at 6 months. The patient converted to sleeve is counted both in the 117 and among the 3. Either the label 'Received intervention as randomized' or the count (116) is inconsistent.
Second check: another model agreed
Report a problem with this finding
Reference or label Figure 3 number-at-risk rows are swapped: 'Gastrectomy 119 111 108 100 95' / 'Bypass 121 119 111 108 98' (and 49/44/43/42/41 vs 52/50/50/45/41 for diabetes), in both panels A and B.
- p. 8 Figure 3, numbers at risk under panels A and B (all patients and patients with diabetes)
Gastrectomy 119 111 108 100 95
- p. 6 Table 2, No. rows
121 119 111 108 98
- p. 8 Figure 3, number-at-risk rows
Gastrectomy 119 111 108
- p. 6 Table 2 / Figure 1
Sleeve gastrectomy No. 121 119 111 108 98
Judge's note: On the p8 image, the Figure 3 row labelled Gastrectomy shows 119/111/108/100/95 and the Bypass row shows 121/119/111/108/98. Table 2 (p6 upright) gives sleeve 121/119/111/108/98 and bypass 119/111/108/100/95; for the diabetes subgroup it gives sleeve 52/50/50/45/41 and bypass 49/44/43/42/41. Figure 1 has 121 randomized to sleeve. The row labels are reversed in all four sets of rows.
Second check: another model agreed
Report a problem with this finding
Wording Results: "the difference was not clinically significant, as the minimal clinically important difference of 9 is within the confidence interval" conflicts with the Discussion: "no conclusion can be drawn about whether gastric bypass is clinically superior".
- p. 4 Results, Primary End Point
the difference was not clinically signifi-cant, as the minimal clinically important difference of 9 is within the confidence interval
- p. 9 Discussion, first paragraph
no conclu-sion can be drawn about whether gastric bypass is clinically superior
- p. 9 Discussion, paragraph 1
no conclu-sion can be drawn about whether gastric bypass is clinically superior to sleeve gastrectomy
Judge's note: p4 Results crop and p9 Discussion text. The interval 3.2 to 13.2 contains 9, so a difference of 9 or more is neither established nor excluded. The Discussion states this. The Results sentence uses the same fact to say the difference was 'not clinically significant'. The two sections give inconsistent wording about the same interval.
Second check: another model agreed
Report a problem with this finding
Wording Discussion says 'gastric bypass showed significantly better weight loss than sleeve gastrectomy at each time point ... (Figure 4)', but the 6-month difference is 4.5 (−0.2 to 9.3).
- p. 9 Discussion, second paragraph
gastric bypass showed significantly better weight loss than sleeve gastrec-tomy at each time point, evaluated by percentage excess weight loss (Figure 4)
- p. 6 Table 2, EWL difference, all patients, 6 mo
4.5 (−0.2 to 9.3)
- p. 9 Discussion, paragraph 2
showed significantly better weight loss than sleeve gastrec-tomy at each time point
- p. 9 Discussion para 2
gastric bypass showed significantly better weight loss than sleeve gastrec-tomy at each time point
Judge's note: Table 2 (p6 upright) gives the 6-month %EWL difference as 4.5 (−0.2 to 9.3), and Figure 4 (p9) shows the 0.5-y interval crossing 0. Lower bounds at 1, 3 and 5 y are positive. The 'each time point' wording is therefore broader than the data shown.
Second check: another model agreed
Report a problem with this finding
Wording Albumin result gives no bypass median: '37 (range, 30-45) g/L and (range, 31-71) g/L (P = .28) for albumin'.
- p. 8 Post Hoc Outcomes, micronutrients paragraph
37 (range, 30-45) g/L and (range, 31-71) g/L (P = .28) for albumin
- p. 8 Post hoc outcomes, albumin
37 (range, 30-45) g/L and (range, 31-71) g/L
Only the authors can say what is right here.
Judge's note: Crop C019-a confirms the image has no number before '(range, 31-71)'. Every other micronutrient has a median for both groups. Only the authors can supply the missing value.
Second check: another model agreed
Report a problem with this finding
12 minor (typos and the like; most checked by the judge only)
Reference or label Table 4 'Triglycerides, mmol/La,b' lacks footnote c (log transformation), although the Methods list triglycerides among the log-transformed variables.
- p. 11 Table 4, Triglycerides row label
Triglycerides, mmol/La,b
- p. 4 Methods, Statistical Analysis
variables (HbA1C, fasting plasma glucose, HDL-C, and triglyc-erides), logarithmic transformation was used
Judge's note: The p11 upright image shows HDL-C with footnotes b,c and triglycerides with a,b only. The Methods (crop C027-b, p4) list HDL-C and triglycerides as log-transformed. The footnote marker is missing or misassigned.
Second check: another model agreed
Report a problem with this finding
Reference or label Table 5 header 'Gastric Bypass (n = 119)', but the early-complication percentages for bypass (e.g. '20 (17.1)', '11 (9.4)') use a denominator of 117; the late rows use 119.
- p. 12 Table 5, Minor/Major Early, Gastric Bypass column
20 (17.1)
- p. 12 Table 5, column header
Gastric Bypass (n = 119)
Only the authors can say what is right here.
Judge's note: Checked the Table 5 image on p12. Early bypass percentages match 117, which is the number who received bypass per Figure 1. Late percentages match 119. Neither the table nor its footnotes says the denominator differs. The text mentions 238 operated patients (p5) but does not explain this. The counts look right; the stated denominator is the problem.
Second check: another model agreed
Report a problem with this finding
Wording Follow-Up says outcomes "were assessed at 1, 2, 3, and 5 years", but analyses, Tables 2-4 and Figure 1 cover only 0.5, 1, 3 and 5 years, and no 2-year data are reported or explained.
- p. 3 Follow-Up
assessed at 1, 2, 3, and 5 years
- p. 3 Statistical Analysis
(0.5, 1, 3, and 5
Judge's note: The p3 crops show the Follow-Up schedule includes 2 years, while the Statistical Analysis evaluates '4 points (0.5, 1, 3, and 5 years)'. The tables and the flow diagram have no 2-year column or step. This is not a contradiction as such, because a 2-year visit may have taken place without being analysed. The omission is unexplained.
Second check: another model agreed
Report a problem with this finding
Typo Abstract Results: "mean baseline body mass index, 45.9ʼ [SD, 6.0]" has a stray mark after 45.9; the main text has "45.9 [SD, 6.0]".
- p. 1 Abstract, Results
45.9, [SD, 6.0]
- p. 1 Abstract, Results
index, 45.9, [SD, 6.0]
Judge's note: The page-1 crop shows a small stray apostrophe or comma-like mark right after 45.9 and before the bracket. The p4 Results text has no such mark.
Second check: none (judge only)
Report a problem with this finding
Typo Outcome Measures: the parenthesis opened at "(complete remission defined as glycated hemoglobin [HbA1c]..." is never closed before "...or ongoing procedures. For dyslipidemia".
- p. 3 Outcome Measures
(complete remission defined as glycated hemoglobin [HbA1c]
Judge's note: Checked in the p3 crop. The inner parentheses (42 mmol/mol), (48 mmol/mol) and (5.6-6.9 mmol/L) all close. The outer one opened before 'complete remission' has no closing parenthesis.
Second check: none (judge only)
Report a problem with this finding
Typo Type 2 Diabetes text: bypass fasting glucose "120.7 (95% CI, 109.9-131.56.7) mg/dL" has a garbled upper bound.
- p. 5 Results, Type 2 Diabetes, fasting glucose
109.9-131.56.7) mg/dL
Judge's note: The p5 image shows '131.56.7', the same as the text layer. From the mmol/L bound, 7.3/0.0555 ≈ 131.5, so a stray '6.7' (the mmol/L point estimate) appears to be fused on.
Second check: none (judge only)
Report a problem with this finding
Typo Triglycerides: "109.7 (95% CI, 102.7-116.8] mg/dL" opens with a parenthesis and closes with a bracket.
- p. 5 Results, Dyslipidemia, triglycerides
CI, 102.7-116.8] mg/dL
Judge's note: The p5 image confirms the '(' … ']' mismatch.
Second check: none (judge only)
Report a problem with this finding
Typo SADI is expanded as "single duodenoileal bypass" in the Figure 1 legend but as "single-anastomosis duodeno-ileal bypass" in the Discussion.
- p. 5 Figure 1 legend
SADI, single duodenoileal bypass
- p. 12 Discussion
single-anastomosis duodeno-ileal bypass
Judge's note: The p5 legend image shows 'SADI, single duodenoileal bypass'. The p12 Discussion text reads 'single-anastomosis duodeno-ileal bypass'. The legend leaves out 'anastomosis'.
Second check: none (judge only)
Report a problem with this finding
Typo Inconsistent spelling: 'Laparoscopic Roux-en-y gastric bypass' (Figure 3 legend) and 'Roux-en Y gastric bypass' (Figure 1 box), against 'Roux-en-Y' elsewhere.
- p. 8 Figure 3 legend key
Laparoscopic Roux-en-y gastric bypass
- p. 5 Figure 1
laparoscopic Roux-en Y gastric
Judge's note: Both forms appear on the images: p8 Figure 3 legend and crop C022-b (Figure 1). This is cosmetic.
Second check: none (judge only)
Report a problem with this finding
Typo Figure 3 caption: 'Lower and upper borders of boxes indicate 25th and 75th quartiles' should say percentiles.
- p. 8 Figure 3 caption
Lower and upper borders of boxes indicate 25th and 75th quartiles
Judge's note: Confirmed in the p8 caption. The wording is wrong but the meaning is recoverable.
Second check: none (judge only)
Report a problem with this finding
Typo Table 3 footnote b: 'less than 6.0% 42 mmol/mol)' is missing its opening parenthesis.
- p. 10 Table 3, footnote b
less than 6.0% 42 mmol/mol)
Judge's note: Visible in the p10 upright image. Footnote c has the correct form '6.5% (48 mmol/mol)'.
Second check: none (judge only)
Report a problem with this finding
Typo Table 5: 'Troacar site pain' should be 'Trocar site pain'.
- p. 12 Table 5, Minor Early complications
Troacar site pain
Judge's note: The misspelling is visible in Table 5 on the p12 image.
Second check: none (judge only)