European Heart Journal Advance Access published online on February 8, 2007
European Heart Journal, doi:10.1093/eurheartj/ehl502
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Keeping a high standard in quantitative analyses, meta-analyses, and systematic reviews
Interventional Cardiology
Division of Cardiology
University of Turin
Turin, Italy
Tel: +39 (0) 3408626829
Fax: +39 (0) 184502244
E-mail address: gbiondizoccai{at}gmail.com
Department of Internal Medicine
Virginia Commonwealth University
Richmond
VA, USA
Institute of Cardiology
University of Ferrara
Ferrara and Cardiovascular Research Center
Salvatore Maugeri Foundation
IRCCS, Gussago
Italy
Antwerp Cardiovascular Institute Middelheim
AZ Middelheim
Antwerp
Belgium
We are frankly disappointed by the recent article by Holmes et al.1 appraising the risk of death in patients randomized to sirolimus-eluting (Cypher) vs. bare-metal stents, which amounts to a missed opportunity. Indeed, analysis and reporting are clearly inadequate to provide a reliable answer to the scientific question posed.
First, given the focus on randomized trials with sirolimus-eluting vs. bare-metal stents and the likely endorsement by Cordis (manufacturer of Cypher), it is unclear why other potentially eligible trials were not included (e.g. the SES-SMART or the DIABETES studies). This type of methodological error, potential harbinger of publication bias, is well known to authors and readers of systematic reviews (and indeed the work by Holmes et al. should be appraised as an individual patient data quantitative review).2
Secondly, the study was designed as a meta-analysis pooling separate data sets, but analysis is inappropriately conducted on a single data set (e.g. with Kaplan–Meier curves). Whether this choice was made by the authors or by an anonymous statistical reviewer is unclear. However, pooling all patients into a large group is incorrect and may lead to biased estimates. Other more valid approaches should have been used, which separately maintain the data sets of each individual study before pooling them.3
Thirdly, no data on conflicts of interest are reported. Several authors have received research grants, organized sponsored congresses, and/or are involved in contract research organizations for revenues likely amounting to millions of euros.4 Unfortunately, a laconic ‘Conflict of interest: none declared’ is posted at the end of the article, but this all-to-common practice is best seen as involuntary malpractice and should be discouraged.
We conversely praise Wijns and Krucoff for detailing in their accompanying editorial their conflicts of interest and the potential amount of money that is at the base of such conflicts.5 As even a layman can understand, we should always declare the rough estimate of a conflict of interest, as a 5000
consultancy is much different from a 10 000 000 contract with a related research company.
The implication of the previous point are all evident if we keep in mind that reviews from people strongly linked to involved companies are more likely to be biased.6,7 A famous example of such biased reviews increasing the ‘smoke’ and hiding the ‘fire’, to use Wijns and Krucoff's metaphor, are actually the many dozens produced in the past by authors affiliated with tobacco companies in support of cigarette smoking.6,8 These issues are also relevant to the recent so-called ‘Independent physician-led patient-level analyses’ on sirolimus- and paclitaxel-eluting stents presented at the 2006 Transcatheter Cardiovascular Therapeutics, whose findings still await peer-review.9,10
In conclusion, as enthusiastic supporters and practitioners of systematic reviews and meta-analyses, we are well aware of their inherent limitations and potential for biased conclusions.11,12 It is thus pivotal to follow strict quality criteria whenever systematic reviews and meta-analysis are performed, reported, peer-reviewed, and read.13 Otherwise the ‘smoke’ from conflicting meta-analysis on drug-eluting stents will only increase and might become overwhelming.
Conflicts of interest: Dr Biondi-Zoccai has consulted for Cordis, Italy and Boston Scientific Italy either directly or through a Milan-based contract research organization for a total amount of <20 000 in the last 5 years.
References
- Holmes DR Jr, Moses JW, Schofer J, Morice MC, Schampaert E, Leon MB. (2006) Cause of death with bare metal sirolimus-eluting stents. Eur Heart J 27:2815–2822.
[Abstract/Free Full Text] - Begg CB and Berlin JA. (1988) Publication bias: a problem in interpreting medical data. J R Statist Soc A 151:419–463.[CrossRef]
- Sweeting MJ, Sutton AJ, Lambert PC. (2004) What to add to nothing? Use and avoidance of continuity corrections in meta-analysis of sparse data. Stat Med 23:1351–1375.[CrossRef][ISI][Medline]
- Medicine In Conflict. There is more concern than ever that doctors are blurring the lines between objective science and financial gain. Business Week, 23 October 2006. Available at: http://www.businessweek.com/magazine/content/06_43/b4006081.htm (last accessed 23 November 2006).
- Wijns WC and Krucoff MW. (2006) Increased mortality after implantation of first generation drug-eluting stents: seeing the smoke, where is the fire? Eur Heart J 27:2737–2739.
[Free Full Text] - Barnes DE and Bero LA. (1998) Why review articles on the health effects of passive smoking reach different conclusions. JAMA 279:1566–1570.
[Abstract/Free Full Text] - Lexchin J, Bero LA, Djulbegovic B, Clark O. (2003) Pharmaceutical industry sponsorship and research outcome and quality: systematic review. BMJ 326:1167–1170.
[Abstract/Free Full Text] - Bero LA, Glantz S, Hong MK. (2005) The limits of competing interest disclosures. Tob Contr 14:118–126.
- Stone GW. Independent physician-led patient-level analysis: CYPHER randomized trials. Presented at the 2006 Transcatheter Cardiovascular Therapeutics , Washington, DC, USA Available at: http://www.tctmd.com/csportal/appmanager/tctmd/tct2006?_nfpb=true&_pageLabel=TCT2006Content&hdCon=1460968&srcId=1&destId=78 (last accessed 23 November 2006).
- Leon MB. Independent physician-led patient-level analysis: TAXUS randomized trials. Presented at the 2006 Transcatheter Cardiovascular Therapeutics , Washington, DC, USA Available at: http://www.tctmd.com/csportal/appmanager/tctmd/tct2006?_nfpb=true&_pageLabel=TCT2006Content&hdCon=1460971&srcId=1&destId=78 (last accessed 23 November 2006).
- Biondi-Zoccai GG, Agostoni P, Abbate A. (2003) Parallel hierarchy of scientific studies in cardiovascular medicine. Ital Heart J 4:819–820.[Medline]
- Biondi-Zoccai GG, Lotrionte M, Abbate A, Testa L, Remigi E, Burzotta F, Valgimigli M, Romagnoli E, Crea F, Agostoni P. (2006) Compliance with QUOROM and quality of reporting of overlapping meta-analyses on the role of acetylcysteine in the prevention of contrast associated nephropathy: case study. BMJ 332:202–209.
[Abstract/Free Full Text] - Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. (1999) Improving the quality of reports of meta-analyses of randomised controlled trials: the QUOROM statement. Quality of Reporting of Meta-analyses. Lancet 354:1896–1900.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||