Table 1. Study characteristics for inception cohorts.
| Study | Objective | Committee approving protocols (country) | Period protocols approved | Date of follow up | Included study designs; Number of studies/Total number of studies (percentage of studies included) | Funding source for all studies | Conclusions |
| Easterbrook, 1991 [26] | Study publication bias: Evidence of publication bias | Central Oxford research Ethics committee (UK) | 1984–1987 | 1990 | Analysed1: RCTs 148/285 (52%), observational 86/285 (30%), non-RCT 51/285 (18%) | 17% unfunded, 20% NHS or department, 13% Government, 38% Pharmaceutical industry, 12% private/charity | Studies with statistically significant results were more likely to be published, also more likely to lead to a greater number of publications and presentations and to be published in journals with a high citation impact factor. |
| Dickersin, 1992 [27] | Study publication bias: To investigate factors associated with the publication of research findings, in particular, the association between ‘significant’ results and publication. | Institutional Review Boards that serve The John Hopkins Health Institutions (USA) | 1980 | 1988 | Completed2: RCTs 168/514 (33%), observational 273/514 (53%), other experimental 73/514 (14%) | 45% NIH, 12% other government, 8% Drug industry, 63% Other, 4% Internal, 18% None. | There is a statistically significant association between significant results and publication. |
| Dickersin, 1993 [3] | Study publication bias: To investigate the association between trial characteristics, findings and publication. | National Institutes of Health (USA) | 1979 | 1988 | RCTs 310/310 (100%) | 50% Grant, 30% Contract, 20% Intramural. | Publication bias is a significant problem |
| Stern, 1997 [4] | Study publication bias and time lag bias: To determine the extent of publication bias and whether publication was delayed for studies with negative results in comparison with those with positive results. | Approved Royal Prince Alfred hospital ethics committee application (Australia) | 1979–1988 | 1992 | Total: RCTs 418/748 (56%), observational 165/748 (22%), non trial experiment 165/748 (22%) | 117/321 Internal, 206/321 External | Confirms the evidence of publication bias found in other studies. Identifies delay in publication as an additional important factor |
| Completed questionnaires: RCTs 277/520 (53%), observational 129/520 (25%), non trial experiment 114/520 (22%) | |||||||
| Analysed3: RCTs 167/321 (52%), observational 90/321 (28%), non trial experiment 64/321 (20%) | |||||||
| Cooper, 1997 [32] | Study publication bias: To determine the fate of studies approved by their departmental human subjects review committee | Department of Psychology Human Subjects Committee or Institutional Review Board, Midwestern, research oriented, state university, USA | 1986–1988 | NI | NI | NI | Significant findings were more likely than non-significant findings to be submitted for meeting presentation or publication. |
| Wormald, 1997 [21] | Study publication bias: To determine the outcome of all randomised controlled trials processed through the pharmacy of Moorfields eye hospital and to determine whether the publication status of these trials is associated with observed effect of treatment. | Trials processed through the pharmacy of Moorfields Eye Hospital (UK) | 1963–1995 | 1997 | RCTs 61/61 (100%) | NI | There was limited evidence of publication bias |
| Ioannidis, 1998 [5] | Study publication bias and time lag bias: To evaluate whether the time to completion and time to publication of randomized phase 2 and phase 3 trials are affected by the statistical significance of results. | Efficacy clinical trials conducted by AIDS Clinical Trials Group and Terry Beirn Community Programs for Clinical Research on AIDS (USA) | 1986–1996 | 1996 | RCTs 109/109 (100%) | Data managed by: 10% Pharmaceutical industry, 90% Other federally sponsored. | There is a time lag in the publication of negative findings that occurs mostly after the completion of the trial follow up. |
| Pich, 2003 [28] | Publication rate: To assess the outcome of protocols submitted to the HCEC. | Hospital Clinic Ethics Committee (Spain) | 1997 | 2001 | RCTs 158/158 (100%) | 89% Pharmaceutical industry, 11% Other. | Only 64% of trials that started were finally implemented and finished in accordance with the original protocol. Only 31% of closed clinical trials were published or in-press in peer reviewed journals. |
| Cronin, 2004 [31] | Study publication bias: Assess the degree to which research project findings were published and explore factors that influenced publication | R&D projects funded by the NHS and commissioned by the North Thames Regional Office (UK) | 1993–1998 | 1995–1998 | NI | 100% government | Funders should consider the significant number of studies that did not result in publication and the higher rate of publication in peer reviewed journals from some programs |
| Decullier, 2005 [29] | Study publication bias and time lag bias: To describe the fate of approved protocols and assess publication bias at a national level. | French Research Ethics Committees (France) | 1994 | 2000–2002 | Total: RCTs 345/649 (53%), descriptive/observational 91/649 (14%), non-randomised 213/649 (33%) | 8% No funding, 73% Private funding, 13% Public, 6% Mixed. | Too many studies are not completed and too many are not published. |
| Completed: RCTs 269/501 (54%), descriptive/observational 66/501 (13%), non-randomised 166/501 (33%) | |||||||
| Decullier, 2006 [30] | Study publication bias: To investigate the fate of protocols submitted for funding, whether they were funded or not. | Greater Lyon regional scientific committee (France) | 1997 | 2003 | RCTs 20/142 (14%), experimental 15/142 (10%), descriptive 45/142 (32%), analytical 27/142 (19%), not clinical 28/142 (20%), not available 7/142 (5%) | 38% committee funded | Some protocols submitted for funding were initiated and completed without any funding declared. To our understanding this means that not all protocols submitted really needed funding and also that health care facilities are unaware that they implicitly financially support and pay for biomedical research. |
| Hahn, 2002 [13] | Outcome reporting bias: To examine the extent of within-study selective reporting in clinical research | Local Research Ethics Committee (UK) | 1994 | 1999 | Of 15 published: RCTs 2/15 (13%), non RCT 2 (13%), uncontrolled trial 2 (13%), case control 1 (7%), survey 2 (13%), cohort and case control 1 (7%), method evaluation study 5 (34%) | Not recorded | Within-study selective reporting may be examined qualitatively by comparing the study report with the protocol. The results suggest that it might well be substantial; the bias could only be broadly identified as protocols were not sufficiently precise. |
| Chan, 2004a [14] | Outcome reporting bias: To determine whether outcome reporting bias would be present in a cohort of government funded trials subjected to rigorous peer review. | Canadian Institutes of Health | 1990–1998 | 2002/2003 | RCTs 108/108 (100%) | 42% jointly funded by industry and CIHR/MRC, 58% no industry funding. | Selective reporting of outcomes frequently occurs in publications of high-quality government-funded trials. |
| Research (Canada) | |||||||
| Chan, 2004b [15] | Outcome reporting bias: To study empirically the extent and nature of outcome reporting bias in a cohort of RCTs | Scientific-Ethical Committees for Copenhagen and Frederiksberge, Denmark | 1994–1995 | 2003 | RCTs 304/304 (100%) | 55% Full industry, 17% Partial industry, 22% Non-industry, 7% non declared. | Reporting of trials is frequently incomplete, biased and inconsistent with protocols. |
| Ghersi, 2006 [17] | Outcome reporting bias: To identify discrepancies in the identity and definition of the primary outcome and to investigate factors associated with the completeness of reporting of the primary outcome. | CSAHS Ethics review committee (Australia) | 1992–1996 | NI | RCTs 318/318 (100%) | 37% commercial funding, 63% no commercial funding. | NI |
| Von Elm, 2008 [18] | Outcome reporting bias: To study trial outcomes specified in protocols of RCTs and reported in subsequent full publications and to estimate publication rate. Investigate whether outcomes are discrepant and to investigate factors that are associated with complete reporting (e.g. statistical significance, funding) | University of Berne/CH ethics committee (Switzerland) | 1988–1998 | 2006 | Total: RCTs 451/1698 (27%) In depth analyses: 451/451 (100%) | 81% industry, 10% other4, | About half of drug trials are not published. A high prevalence of pre-specified outcomes are not reported and discrepancies includes primary outcomes. Completeness of reporting of an outcome is associated with statistical significance. |
1
Easterbrook et al assumed that only studies that had been analysed had the potential for being written up and published, so tests for study publication bias were restricted to these.
2
Studies for which there was a full interview by the researchers of the cohort study and for which information on the nature of results and publication was provided.
3
Of the 520 studies with completed questionnaires, 321 had analysis undertaken with results available and were included in further analysis of the association between study outcome and time to publication.
4
Both groups are not mutually exclusive. 4% had a statement of both sources of funding. In the remainder, the protocols did not include information on how study was funded.
NI No information available.