Documente Academic
Documente Profesional
Documente Cultură
6. Smith R. Opening up BMJ peer review. BMJ. 1999; Rennie D. Does masking author identity improve peer Available at: http://www.research.att.com/~amo/doc
318:4-5. review quality? a randomized controlled trial. JAMA. /tragicloss.txt. Accessed May 13, 2002.
7. Rennie D. Anonymity of reviewers. Cardiovasc Res. 1998;280:240-242. 16. Harnad S. Implementing peer review on the net:
1994;28:1142-1143. 12. van Rooyen S, Godlee F, Evans S, Black N, Smith scientific quality control in scholarly electronic jour-
8. McNutt RA, Evans AT, Fletcher RH, Fletcher SW. R. Effect of open peer review on quality of reviews nals. In: Peek R, Newby G. Scholarly Publishing: the
The effects of blinding on the quality of peer review. and on reviewers’ recommendations: a randomised Electronic Frontier. Cambridge, Mass: MIT Press; 1995.
JAMA. 1990;263:1371-1376. controlled trial. BMJ. 1999;318:23-27. 17. Ginsparg P. Creating a global knowledge net-
9. Godlee F, Gale CR, Martyn CN. Effect on the qual- 13. Walsh E, Rooney M, Appleby L, Wilkinson G. work. Available at: http://www.biomedcentral.com
ity of peer review of blinding reviewers and asking them Open peer review: a randomised controlled trial. Br J /1417-8219/1/9. Accessed May 13, 2002.
to sign their reports: a randomized controlled trial. Psychiatry. 2000;176:47-51. 18. Gura T. Peer review unmasked. Nature. 2002;
JAMA. 1998;280:237-240. 14. van Rooyen S, Black N, Godlee F. Development 416:258-260.
10. van Rooyen S, Godlee F, Evans S, Smith R, Black of the review quality instrument (RQI) for assessing 19. Godlee F, Jefferson T. Peer Review in Health
N. Effect of blinding and unmasking on the quality of peer reviews of manuscripts. J Clin Epidemiol. 1999; Sciences. London, England: BMJ Publishing Group;
peer review: a randomized controlled trial. JAMA. 52:625-629. 1999.
1998;280:234-237. 15. Odlyzko AM. Tragic loss or good riddance? the 20. Altman DG. Poor-quality medical research: what
11. Justice AC, Cho MK, Winker MA, Berlin JA, impending demise of traditional scholarly journals. can journals do? JAMA. 2002;287:2765-2767.
T
HERE IS CONSIDERABLE EVI - directly or indirectly—lead to improvements in the treatment and preven-
dence that many published re-
tion of disease. Each research project should continue systematically from
ports of randomized con-
trolled trials (RCTs) are poor previous research and feed into future research. Each project should con-
or even wrong, despite their clear im- tribute beneficially to a slowly evolving body of research. A study should
portance.1 The results of several re- not mislead; otherwise it could adversely affect clinical practice and future
views of published trials are briefly sum- research. In 1994 I observed that research papers commonly contain meth-
marized in TABLE 1. Poor methodology odological errors, report results selectively, and draw unjustified conclu-
and reporting are widespread. sions. Here I revisit the topic and suggest how journal editors can help.
Similar problems afflict other study JAMA. 2002;287:2765-2767 www.jama.com
types. A review of 308 phase 2 trials in
cancer (295 of which were single-arm
studies) found that 250 (81%) did not re- ity of the individual (primary) studies.6 either identify previous studies or place
port an identifiable statistical design. Fur- Reviewers often conclude that the avail- their findings in the context of those
ther, positive findings were reported in able evidence is of poor scientific qual- previous studies.13
48% of designed studies but 70% of stud- ity,7,8 sometimes leading to heated debate
ies with no reported design (P=.003).3 about interpretation.9 Why Are There So Many
Of 40 molecular genetics articles pub- General reviews also find much to be Errors in Medical Articles?
lishedinleadinggeneralmedicaljournals, concerned about. Serious statistical er- Errors in published research articles in-
15 (38%) failed to meet at least 2 of 7 rors were found in 40% of 164 articles dicate poor research that has survived the
methodological standards. The authors published in a psychiatry journal10 and peer-review process. But the problems
wrote:“Withoutsuitableattentiontofun- in 19% of 145 articles published in an arise earlier, so a more important ques-
damental methodological standards, the obstetrics and gynecology journal.11 I tion is, Why are submitted articles poor?
expected benefits of molecular genetic suspect that many basic errors have be- Much research is done without the
testing may not be achieved.”4 come less common, but statistics has be- benefit of anyone with adequate train-
In recent years, systematic reviews come more complex, and there is evi- ing in quantitative methods.14 Many in-
have become common. In these, all reli- dence of frequent misapplication of
able evidence relating to a clinical ques- newer advanced techniques.12 Author Affiliation: Cancer Research UK/NHS Cen-
tre for Statistics in Medicine, Oxford, England.
tion is sought, systematically appraised, Also, when interpreting a study, read- Corresponding Author and Reprints: Douglas G. Alt-
and, if suitable, combined statistically in ers need to know how it relates to ex- man, DSc, Cancer Research UK/NHS Centre for Sta-
tistics in Medicine, Institute of Health Sciences, Old
a meta-analysis.5 A key component is an isting knowledge. Many authors inter- Road, Headington, Oxford OX3 7LF, England (e-mail:
assessment of the methodological qual- pret their findings narrowly, failing to doug.altman@cancer.org.uk).
©2002 American Medical Association. All rights reserved. (Reprinted) JAMA, June 5, 2002—Vol 287, No. 21 2765
tial postpublication peer review; poten- What Journal Editors Can Do Journals can also help improve the
tial correspondents will surely be deterred Authors and editors should have the literature by requiring the full and trans-
by the unambiguous cutoff. Journals with same goals: the advancement of scien- parent reporting of research. Guide-
such a policy should reconsider. tific understanding and improvement in lines have been developed for RCTs,30
A few journals (eg, BMJ and the CMAJ) the treatment and prevention of dis- systematic reviews and meta-analyses
haverapidpublicationofcorrespondence ease. Poor research is the fault of au- of RCTs31 and observational studies,32
on their Web pages. All (or most) letters thors, not journals. Poor research meth- and studies of diagnostic tests,33 and
are published, and there is no apparent ods, unnecessary research, redundant or other initiatives are under way. Edi-
time limit. Nor is there the same limit on duplicate publication, thinly sliced study tors should continue to be involved in
length as in print journals (Table 2). Elec- results, selective reporting, and scien- the development of reporting recom-
tronic letters are linked to the original tific fraud, as well as a general tendency mendations and explicitly require au-
publication and are relatively easily ac- to inflate the importance of the results, thors to follow them.
cessed. It is remarkable and disappoint- should all be resisted vigorously. All Journals can enable and encourage the
ing that as yet so few journals have such could be less likely if research were not publication of research protocols.34-36
a capability. Restricting the facility to cur- a career necessity for physicians. They can use their Web pages to pub-
rentsubscribers,ascurrentlydonebyNeu- Rather than abandon peer review, as lish extended versions of articles. They
rology and Pediatrics, is inadequate. A some have suggested, journals should should also enable and encourage pub-
weakness yet to be resolved is the absence work to strengthen it. In particular, lication of the raw data used in medical
of pressure on authors to respond to criti- methodological review should be imple- research articles (eg, Clinical Chemistry
cisms.28 For such journals there is uncer- mented much more widely. It will never and Neurology). If journals are willing to
tainty about which version is definitive. be possible to eliminate misleading publish data, they should explicitly sug-
Although the BMJ considers bmj.com to studies, but our imperfect peer-review gest this possibility to authors.
be the definitive version,29 only the let- system is a safeguard without which the
ters that appear in the print journal are quality of published research would be Acknowledgment: I thank Iain Chalmers, DSc, and a
indexed on MEDLINE. lower. reviewer for helpful suggestions.
REFERENCES
1. Altman DG. The scandal of poor medical re- 13. Clarke M, Alderson P, Chalmers I. Discussion sec- tors (WAME): an agenda for the future. Available at:
search. BMJ. 1994;308:283-284. tions in reports of controlled trials published in gen- http://www.wame.org/bellagioreport_1.htm.
2. Altman DG, Schulz KF, Moher D, et al, for the eral medical journals. JAMA. 2002;287:2799-2801. Accessed January 20, 2002.
CONSORT Group. The revised CONSORT state- 14. Altman DG, Goodman SN, Schroter S. How sta- 27. Bland M. Fatigue and psychological distress: sta-
ment for reporting randomized trials: explanation and tistical expertise is used in medical research. JAMA. tistics are improbable. BMJ. 2000;320:515-516.
elaboration. Ann Intern Med. 2001;134:663-694. 2002;287:2817-2820. 28. Rennie D. Freedom and responsibility in medi-
3. Mariani L, Marubini E. Content and quality of cur- 15. Chanter DO. Maintaining the integrity of the sci- cal publication: setting the balance right. JAMA. 1998;
rently published phase II cancer trials. J Clin Oncol. entific record: new policy is unlikely to give investigators 280:300-302.
2000;18:429-436. more control over studies [letter]. BMJ. 2002;324:169. 29. Smith R. The BMJ: moving on. BMJ. 2002;324:
4. Bogardus ST Jr, Concato J, Feinstein AR. Clinical 16. Bland JM, Altman DG. Caveat doctor: a grim tale 5-6.
epidemiological quality in molecular genetic re- of medical statistics textbooks. BMJ. 1987;295:979. 30. Moher D, Schulz KF, Altman D, for the CONSORT
search: the need for methodological standards. JAMA. 17. Sackett DL, Straus S, Richardson WS, Rosenberg Group. The CONSORT statement: revised recom-
1999;281:1919-1926. W, Haynes RB. Evidence-Based Medicine: How to mendations for improving the quality of reports of par-
5. Egger M, Davey Smith G, Altman DG, eds. Sys- Practice and Teach EBM. 2nd ed. Edinburgh, Scot- allel-group randomized trials. JAMA. 2001;285:1987-
tematic Reviews in Health Care: Meta-analysis in Con- land: Churchill-Livingstone; 2000. 1991.
text. 2nd ed. London, England: BMJ Books; 2001. 18. Guyatt G, Rennie D, eds. Users’ Guides to the 31. Moher D, Cook DJ, Eastwood S, Olkin I, Rennie
6. Juni P, Altman DG, Egger M. Assessing the qual- Medical Literature: A Manual for Evidence-Based D, Stroup DF, for the QUOROM Group. Improving
ity of controlled clinical trials. BMJ. 2001;323:42-46. Clinical Practice. Chicago, Ill: AMA Press; 2002. the quality of reports of meta-analyses of ran-
7. Hotopf M, Lewis G, Normand C. Putting trials on 19. Emanuel EJ, Wendler D, Grady C. What makes clini- domised controlled trials: the QUOROM statement.
trial—the costs and consequences of small trials in de- cal research ethical? JAMA. 2000;283:2701-2711. Lancet. 1999;354:1896-1900.
pression: a systematic review of methodology. J Epi- 20. Westgard JO, Hunt MR. Use and interpretation 32. Stroup DF, Berlin JA, Morton SC, et al, for the
demiol Comm Health. 1997;51:354-358. of common statistical tests in method comparison stud- Meta-analysis of Observational Studies in Epidemiol-
8. Lawlor DA, Hopker SW. The effectiveness of exer- ies. Clin Chem. 1973;19:49-57. ogy (MOOSE) Group. Meta-analysis of observa-
cise as an intervention in the management of depres- 21. Rothman KJ. Epidemiologic methods in clinical tri- tional studies in epidemiology: a proposal for report-
sion: systematic review and meta-regression analysis of als. Cancer. 1977;39(4 suppl):1771-1775. ing. JAMA. 2000;283:2008-2012.
randomised controlled trials. BMJ. 2001;322:763-767. 22. Oldham PD. A note on the analysis of repeated 33. The STARD Group. The STARD initiative: to-
9. Olsen O, Gotzsche PC. Cochrane review on screen- measurements of the same subjects. J Chronic Dis. wards complete and accurate reporting of studies on
ing for breast cancer with mammography. Lancet. 1962;15:969-977. diagnostic accuracy. Available at: http://www
2001;358:1340-1342. 23. Donner A, Brown KS, Brasher P. A methodologi- .consort-statement.org/stardstatement.htm. Acces-
10. McGuigan SM. The use of statistics in the Brit- cal review of non-therapeutic intervention trials em- sibility verified May 1, 2002.
ish Journal of Psychiatry. Br J Psychiatry. 1995;167: ploying cluster randomization, 1979-1989. Int J Epi- 34. Horton R. Pardonable revisions and protocol re-
683-688. demiol. 1990;19:795-800. views [commentary]. Lancet. 1997;349:6.
11. Welch GE II, Gabbe SG. Review of statistics usage 24. Goodman SN, Altman DG, George SL. Statisti- 35. Chalmers I, Altman DG. How can medical jour-
in the American Journal of Obstetrics and Gynecol- cal reviewing policies of medical journals: caveat lec- nals help prevent poor medical research? some op-
ogy. Am J Obstet Gynecol. 1996;175:1138-1141. tor? J Gen Intern Med. 1998;13:753-756. portunities presented by electronic publishing. Lan-
12. Schwarzer G, Vach W, Schumacher M. On the 25. Gehlbach SH. Interpreting the Medical Litera- cet. 1999;353:490-493.
misuses of artificial neural networks for prognostic and ture: A Clinician’s Guide. 3rd ed. New York, NY: 36. Godlee F. Publishing study protocols: making them
diagnostic classification in oncology. Stat Med. 2000; McGraw-Hill; 1993. visible will encourage registration, reporting and re-
19:541-561. 26. Report of the World Association of Medical Edi- cruitment [editorial]. BMC News Views. 2001;2:4.
©2002 American Medical Association. All rights reserved. (Reprinted) JAMA, June 5, 2002—Vol 287, No. 21 2767