Editor—In the context of an evaluation of the CONSORT guidelines for the reporting of randomised controlled trials, we searched each issue published in 1998 of Annals of Internal Medicine, BMJ, JAMA, Lancet, and the New England Journal of Medicine for reports of randomised controlled trials.1,2 The hand search identified 290 articles. We excluded one quasi-randomised trial and nine reports whose focus was not on randomised comparisons. Our study sample thus consisted of 280 reports (table).
The number of trials published in 1998 ranged from 83 in the New England Journal of Medicine to 20 in Annals of Internal Medicine. Most trials were of parallel group design (265/280, 94.6%), with individuals as the randomisation unit (270/280, 96.4%). Most trials evaluated pharmacological interventions (174/280, 62.1%). Trials published in the BMJ differed from trials published in the other journals in two respects. Firstly, randomisation was more likely to be at an aggregate level—for example, practices in general practice—rather than at the individual level (odds ratio 9.86, 95% confidence interval 2.21 to 49.0, comparing BMJ reports with other reports). Secondly, trials were less likely to test drug interventions (0.35, 0.17 to 0.69).
Trials published in general medicine journals will to some extent reflect the agenda in clinical trial research, which has been shown to be at odds with the needs of consumers and planners of health services.3,4 Our results are in line with previous studies showing that drug interventions dominate the literature. For example, 380 (82.6%) of 460 trials of osteoarthritis of the knee evaluated drugs whereas the evidence on the effectiveness of other interventions was inadequate or absent.3 Commercial interests of the pharmaceutical industry, which funds many drug trials, vested interests of researchers, and lack of involvement of healthcare consumers may contribute to the dominance of drug trials. The low proportion of drug trials and the large proportion of cluster trials, many from health services research, indicate that the material in the BMJ may be less affected by the biases distorting the research agenda. Further research is required to refute or confirm this hypothesis.
Table.
Characteristics of reports of controlled trials published in five general medicine journals in 1998
| Characteristic of trial reports | Annals of Internal Medicine (n=20) | BMJ (n=48) | JAMA (n=47) | Lancet (n=82) | New England Journal of Medicine (n=83) |
|---|---|---|---|---|---|
| Parallel group trials | 18 (90.0) | 46 (95.8) | 45 (95.7) | 78 (95.1) | 78 (93.9) |
| Randomised individuals | 19 (95.0) | 42 (87.5) | 45 (95.7) | 81 (98.8) | 83 (100) |
| Tested pharmacological intervention | 13 (65.0) | 15 (31.3) | 30 (63.8) | 55 (67.1) | 61 (73.5) |
| Median total sample size (range) | 88 (35 to 1283) | 214 (11 to 17 187) | 197 (24 to 6605) | 382 (15 to 19 193) | 400 (20 to 10 948) |
Differences were significant for unit of randomisation (individuals v other, P=0.003), type of intervention (drugs v other, P<0.001), and sample size (P=0.004). Probability from χ2 and Kruskal-Wallis tests.
References
- 1.Egger M, Jüni P, Bartlett C. Value of flow diagrams in reports of randomized controlled trials. JAMA. 2001;285:1996–1999. doi: 10.1001/jama.285.15.1996. [DOI] [PubMed] [Google Scholar]
- 2.Altman DG, Schulz KF, Moher D, Egger M, Davidoff F, Elbourne D, et al. The revised CONSORT statement for reporting randomized trials: explanation and elaboration. Ann Intern Med. 2001;134:663–694. doi: 10.7326/0003-4819-134-8-200104170-00012. [DOI] [PubMed] [Google Scholar]
- 3.Tallon D, Chard J, Dieppe P. Relation between agendas of the research community and the research consumer. Lancet. 2000;355:2037–2040. doi: 10.1016/S0140-6736(00)02351-5. [DOI] [PubMed] [Google Scholar]
- 4.Chalmers I. What do I want from health research and researchers when I am a patient? BMJ. 1995;310:1315–1318. doi: 10.1136/bmj.310.6990.1315. [DOI] [PMC free article] [PubMed] [Google Scholar]