Look for CONSORT and QUORUM

Home , Clinical research

Ian Needleman Specialist in Periodontics, Department of Periodontology, Eastman Dental Institute for Oral Health Care Sciences, University College London, London, UK

these guidelines9 and the Journal of The quality of clinical research publications is variable and poor quality may Orthodontics has followed suit.10 invalidate the findings of a study. Guidelines have now been published that CONSORT is a checklist of items that should improve the reporting of clinical trials (CONSORT) and systematic should be included in the publication of reviews (QUORUM). These guidelines make it easier to judge research quality a randomised-controlled trial. The and should improve peer review: conclusions from clinical trials or systematic checklist items require clear and thor- ough descriptions of hypothesis, pro- reviews using these guidelines are more likely to be reliable. tocol, randomisation, blinding, follow- up and analysis; full details are available One of the key aspects of improving the treatment effect may be overestimated at www.consort-statement.org. As far care you offer to your patients is in basing by a substantial 30±40%.3±5 Evidence as possible, items are evidence-based, ie, your practice on the best available about study quality in dentistry is included if evidence supports their use. evidence. But with so many journal lacking, but is slowly starting to appear, With more complete reporting, the publications, how do you sift the good confirming many of these concerns.6,7 whole process of evaluating the quality research from the bad? Of course, when I defined good of research should be easier. I believe In this case I will define good research research, what many of you will have that such transparency should also help as research that is of high enough realised is that I was really referring to referees and editors to be more objec- quality to support the conclusions the reporting of the research, since this tive in assessing the quality of new reached by the authors (whether a is all we usually have to go on. The research, which should lead to better positive or a negative finding). This is problem that I often encounter is that research in print. Early results show that crucial, since as the reader of a study, I research articles lack many of the this is starting to happen (D Moher et need to have the confidence to accept features that help me to evaluate their al, 2000, unpublished data (www.con- the findings, at least for the patients in quality. The same difficulty may be sort-statement.org)), but more com- that study. faced by journal referees and editors. prehensive evaluation will appear once Research is very hard to do well and As an aid to researchers and to sufficient studies have been published there are many reasons why an author's improve the quality of reporting of after the adoption of the guidelines. As conclusions may not be valid. For research, two guidelines have been an added benefit, CONSORT provides instance, a finding of no difference proposed that should make the task of clear guidance on constructing better between treatments may simply indi- assessing publications much easier. trials, which should also improve future cate that the study was too small to These are CONSORT and, more re- research quality. demonstrate a difference: if more pa- cently, QUORUM. QUORUM (Quality Of Reporting Of tients had been recruited one therapy CONSORT (Consolidated Standards Meta-analyses) is a more recent guide- could have proved superior (a type-II for Reporting of Trials) was developed line11 and sets out to achieve the same error). If you are interested in clinical by a group of journal editors, clinical improvement in the quality of report- trial design I would recommend an researchers and biostatisticians.8 Since ing of systematic reviews as CONSORT excellent short book by Jadad1 and a 1996 it has been adopted by more than is attempting to do for clinical trials. recent review article by Harrison.2 70 biomedical journals as a requirement Confusingly, in North America sys- Investigations into the effect of re- for publication of randomised con- tematic reviews are termed meta-ana- search quality have also shown that, if trolled trials. The British Dental Journal lyses, whereas in Europe the term meta- incorrectly conducted, the size of the was the first dental journal to adopt analysis is used for the statistical 62 combination of studies in a systematic ously the author has conducted the 279:1489±1491. review. review. In order to make the process 5. Moher D, Cook DJ, Jadad AR, Tugwell P, Moher M, Jones A, et al. Assessing the A good systematic review reduces the more transparent, the QUORUM quality of reports of randomised trials; bias that may enter a traditional, so- guidelines, like CONSORT, set out a implications for conduct of meta-analyses. called narrative review, and therefore checklist that should ensure the inclu- Health Technol Assess 1999a; 3:1±98 (see should give the reader more confidence sion of information needed to evaluate also http://www.hta.nhsweb.nhs.uk and in the conclusions.12±14 Bias is mini- the quality of the systematic review. search on first author's name). 6. Antczak-Bouckoms A. Meta-analysis of mised by having a clear prestated Research into the impact on the clinical trials in periodontal research protocol (or systematic approach) de- quality of clinical trial and systematic Periodontology 2000; 2:140±149. scribing: review results is at a comparatively early 7. Knoll-Kohler E. Metronidazole dental gel as . the question that will be reviewed stage. CONSORT and QUORUM an alternative to scaling and root planing in . how the reviewers will attempt to should be thought of as dynamic and the treatment of localized adult periodontitis. Is its efficacy proved? Eur J locate all relevant research able to change in the light of good new Oral Sci 1999; 107:415±421. . how data will be included or evidence. There can be little doubt, 8. Begg C, Cho M, Eastwood S, Horton R, excluded form the review however, that both provide us with the Moher D, Olkin I, et al. Improving the . how quality of the research will be best currently-available benchmarks quality of reporting of randomized examined with which to assess the quality of controlled trials. JAMA 1996; 276:637±639. 9. Needleman IG. CONSORT. BDJ 1999; . how the data may be combined in a reporting of research. So, if you are 186:207. meta-analysis. looking for indications of quality on 10. Newcombe RG. Reporting of clinical trials I believe that it is not unreasonable to which to base your evidence-based in the JO ± the CONSORT guidelines. J expect the systematic approach to practice, look for CONSORT and Orthodont 2000; 27:69±70. reduce the potential for reviewers to QUORUM. 11. Moher D, Cook JC, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality manipulate, unwittingly or otherwise, of reports of meta-analyses of randomised the results of the review. Similarly, by 1. Jadad A. Randomised Controlled Trials. A controlled trials: the QUORUM statement. attempting to find all the research that User's Guide. London: BMJ Books, 1998. Lancet 1999b; 354:1896±1900. has been conducted on the topic (both 2. Harrison JE. Evidence-based orthodontics ± 12. Mulrow CD. Rationale for systematic how do I assess the evidence? J Orthodont published and unpublished) as well as reviews. BMJ 1994; 309:597±599. 2000; 27:189±197. 13. Deeks JJ. Systematic reviews of published non-English language papers, reviews 3. Schulz KF, Chalmers I, Hayes RJ, Altman evidence: miracles or minefields? Ann are less likely to be prone to either DG. Empirical evidence of bias: dimensions Oncol 1998; 9:703±709. selective reporting or publication bias of methodological quality associated with 14. McAlister FA, Clark HD, van-Walraven C, (which can and should be tested for). estimates of treatment effects on controlled Strauss SE, Lawson FM, Moher D, et al. The trials. JAMA 1995; 273:408±412. As with any research, however, the medical review article revisited: has the 4. Moher D. Consort: an evolving tool to help science improved? Ann Intern Med 1999; quality of systematic reviews is likely to improve the quality of reports of 131:947±951. be variable depending on how rigor- randomized controlled trials. JAMA 1998;

Evidence-Based Dentistry