BMJ No 7109 Volume 315
Papers Saturday 13 September 1997
Impact of covert duplicate publication on meta-analysis: a case study
Martin R Tramer, D John M Reynolds, R Andrew Moore, Henry J McQuay
| Ondansetron for prevention and treatment of postoperative nausea and vomiting. Duplication of data from full reports of published randomised controlled trials | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Reports (reference No)
|
Evidence of duplication of data
|
Number of patients
|
|||||||||
|
|
Problems which made detection of duplicate data difficult (ITT=intention-to-treat) |
Original
|
|
||||||||
|
|
Original | Duplicate | Cross reference | Confirmation (A=author M=manufacturer)* | Patient characteristics=number, mean age and weight, sex, type of surgery, etc | Common authorship** |
Total | Taking drug | Total | Taking drug | |
|
Trial 1: ondansetron 1 mg IV v 4 mg IV v 8 mg IV v placebo (treatment of established postoperative nausea and vomiting) |
|||||||||||
| 1a | 9 | No | A confirmed | Identical: patient characteristics (table 1), adverse events (table 2), efficacy data (abstract) as 1b (table 1 and 3, abstract) | Yes | 500 | 371 | ||||
| 1b | 10 | No | A confirmed | Identical: patient characteristics (table 1), adverse events (table 3), efficacy data (abstract) as 1a (table 1 and 2, abstract) | Yes | 500 | 371 | ||||
| 1c | 11 study I | No | A confirmed | No identical data | Reported together with another trial. Patient characteristics not reported. Similar but not identical efficacy data because different number of patients | Yes | 465 | 348 | |||
|
Trial 2: ondansetron 8 mg by mouth x 3 (8 hourly) vplacebo (prevention of postoperative nausea and vomiting) |
|||||||||||
| 2a | 12 | No | A confirmedM confirmed | Identical: number of patients per group, incidence of nausea and/or vomiting 0-24 h (fig 1), absence of vomiting (table 4) | Patient characteristics on ITT data (table 1) | No | 237 | 117 | |||
| 2b | 13 study II | No | A not confirmedM confirmed | Identical: number of patients per group, incidence of nausea and/or vomiting 0-24 h (table 4), percentage of vomiting patients (table 4) | Patient characteristics on efficacy data (table 2) | No | 237 | 117 | |||
|
Trial 3: ondansetron 5 mg/m2 IV vmetoclopramide 0.12 mg/kg IV v placebo (prevention of postoperative nausea and vomiting) |
|||||||||||
| 3a | 14 | No | A confirmed | Identical: patient characteristics (results section), efficacy data (table 2) | In French | Yes | 60 | 20 | |||
| 3b | 15 | No | Not applicable | Identical: patient characteristics (results section), efficacy data (fig 1) | In Italian | Yes | 60 | 20 | |||
|
Trial 4: ondansetron 8 mg IV x 2 (8 hourly) v placebo (prevention of postoperative nausea and vomiting) |
|||||||||||
| 4a | 16 | No | Not sought | Full report | No identical data | Yes | 163 | 82 | |||
| 4b | 17 | Yes, as footnote | Not sought | Preliminary report | No identical data | Yes | 50 | 25 | |||
|
Trial 5: ondansetron 1 mg by mouth v 8 mg by mouth v16 mg by mouth vplacebo; each dose x 3 (prevention of postoperative nausea and vomiting) |
|||||||||||
| 5a | 18 | No | A not confirmedM confirmed | Identical: patient characteristics (table 1) as 5b (table 1) | 13 of 995 patients excluded from efficacy analysis, but only data from 973 analysed (table 5) | No | 973 | 726 | |||
| 5b | 13 study I | No | A not confirmedM confirmed | Identical: patient characteristics (table 1) as 5a (table 1) | 13 of 995 patients excluded from efficacy analysis. Efficacy data (table 3) assumed to be ITT analysis; only percentages reported | No | 982 | 733 | |||
|
Trial 6: ondansetron 8 mg IV v droperidol 1.25 mg IV v placebo (prevention of postoperative nausea and vomiting) |
|||||||||||
| 6a | 19 | No | A no reply | Identical: efficacy data (fig 1), surgery, number of patients, anaesthesia | 26 male and 34 female patients | No | 60 | 20 | |||
| 6b | 20 | Yes without comment | A no reply | Identical: efficacy data (graph), surgery, number of patients, anaesthesia | 30 male and 30 female patients | No | 60 | 20 | |||
|
Trial 7: ondansetron 1 mg IV v 4 mg IV v 8 mg IV v placebo (prevention of postoperative nausea and vomiting) |
|||||||||||
| 7a | 21 | No | A confirmed | Identical: time to recovery (table 2) as 7c (table 1), but not as 7b; efficacy results (table 3) as 7c (fig 1), adverse events (table 4) as 7b (table 3) | Exact number of patients per group not stated (|mK36 of 580 excluded from analysis|mK). Efficacy data reported only as percentages | Yes | 544 | 405 | |||
| 7b | 22 | No | A no reply | Identical: number of patients per group (table 2) as 7c (Fig 1), efficacy results (Fig 1a) as 7a (Fig 2), adverse events (table 3) as 7a (table 4) | Patient characteristics given on efficacy data rather than on ITT data as in 7a and 7c | Yes | 544 | 405 | |||
| 7c | 23 study I | No | A no reply | Identical: time to recovery (table 1) as 7a (table 2), but not as 7b; efficacy results (fig 1) as 7a (fig 1) | Reported together with two other trials. Efficacy data reported only as percentages | No | 544 | 405 | |||
|
Trial 8: ondansetron 1 mg IV v 4 mg IV v 8 mg IV v placebo (prevention of postoperative nausea and vomiting) |
|||||||||||
| 8a | 24 | No | A no reply | Identical: awakening times (table 1) and efficacy data (fig 1) as 8b (table 1 and fig 1) | Efficacy data only as percentages in graphs | No | 539 | 437 | |||
| 8b | 23 study II | No | A no reply | Identical: awakening times (table 1) and efficacy data (fig 1) as 8a (table 1 and fig 1) | Reported together with two other trials | No | 539 | 437 | |||
|
Trial 9: ondansetron 4 mg IV vplacebo (prevention of postoperative nausea and vomiting) |
|||||||||||
| 9a | 25 | No | A confirmed | Full report | Premedication with diazepam 10 mg + atropine 0.5 mg im 1 hour before induction. Induction with thiopentone. Maintenance with isoflurane and nitrous oxide in oxygen. If rescue medication fails: metoclopramide 10 mg iv | Yes | 928 | 464 | |||
| 9b | 26 | No | A confirmed | Identical: numbers of ophthalmic subgroup, and rescue medication as 9a | Different premedication (diazepam 2-5 mg iv), different induction (propofol), and different maintenance (nitrous oxide in oxygen) as 9a, but identical with 9c | Yes | 260 | 130 | |||
| 9c | 27 | to 9b without comment* | A confirmed | Identical: anaesthetic technique, rescue medication as 9a. Similar: ophthalmic subgroup as 9a and 9b, but here 140 more; surgical subgroups (table 1); adverse effects (table 3) as 9b - here all numbers are higher as in 9b | Anaesthetic technique different from 9a but identical with 9b | Yes | 260 | 130 | |||
| 9d | 28 | No | A confirmed | No similarity of data | Additional ondansetron 8 mg arm. Different induction method (propofol) than 9a | Yes | 145 | 74 | |||
| 9e | 29 | No | A confirmed | Identical: premedication, anaesthetic technique, and rescue medication as 9a | Efficacy data reported as scores | Yes | 240 | 120 | |||
| 4004 | 2642 | 4886 | 3335 | ||||||||
*Without comment: matching report was mentioned but duplication of data not acknowledged. * Not confirmed means that author replied but neither confirmed nor denied duplication. ** At least one author common to all reports.
| Treatment efficacy in 16 original trials which were not duplicated, three original trials which were duplicated, six duplicates of these three trials, and the combination of originals and duplicates. Number needed to treat (with 95% confidence intervals) to prevent vomiting up to the 24th postoperative hour with intravenous ondansetron 4 mg compared with placebo. The numbers above the symbols are the numbers of reports. |
Impact of covert duplicate publication on meta-analysis: a case study
Home | Current contents | Past issues | Classified ads | Career Focus | Feedback
Collections | About this site | About the BMJ | BMA | Medline