Ko Clifford Y, Sack Jonathan, Chang John T, Fink Arlene
Department of Medicine, UCLA School of Medicine and Public Health, Los Angeles, California 90095-6904, USA.
Dis Colon Rectum. 2002 Apr;45(4):443-7. doi: 10.1007/s10350-004-6217-x.
Evidence-based medicine relies on reproducible, high-quality reporting in the literature. Previous evaluations, which have assessed 11 basic elements of design and analysis in top impact clinical journals (both nonsurgical and surgical), have demonstrated that the reporting quality is less than perfect, although improving. The current study evaluates the quality of reporting in Diseases of the Colon and Rectum and other clinically related journals to identify specific areas where future improvements may be made.
Two independent evaluators assessed all randomized, controlled trials published in Diseases of the Colon and Rectum in the years 1990, 1995, and 2000. Additional assessments for 2000 were performed on all randomized, controlled trials published in Annals of Surgery, Archives of Surgery, and Gastroenterology. The frequency of reporting of 11 explicitly defined, traditionally important, basic elements of design and analysis were determined. These elements included reporting of eligibility criteria, admission before allocation, randomization (and method), blinded assessment (patient and observer), complications, loss to follow-up, statistical approach and tests, and power calculation.
Interobserver reliability was strong (kappa, 0.76). The number of randomized, controlled trials published in Diseases of the Colon and Rectum increased from 5 (in 1990) to 13 (in 1995) to 17 (in 2000). Of the 1990 randomized, controlled trials, an average of 60 percent of the 11 basic elements were reported. Of the 1995 randomized, controlled trials, 72 percent of the items were reported (P = 0.05), whereas of the 2000 randomized, controlled trials, 77 percent of the 11 items were reported (P < 0.002 vs. 1990). The best-reported items were eligibility criteria, discussion of statistical tests, and accounting for all patients lost to follow-up. Only 11 percent of the 2000 randomized, controlled trials reported statistical power calculations. For the other journals that were evaluated, 72 to 88 percent of items were reported, with eligibility criterion being the best consistently reported item and power calculation being the worst.
For Diseases of the Colon and Rectum, the number of randomized, controlled trials and the quality of reporting is improving. However, although certain research standards are reported adequately, others are not. The calculation of statistical power is clearly important when interpreting randomized, controlled trial results (whether differences are reported or not), yet only 11 percent of studies contained this information. Improving the reporting of this single item would likely lead to improving the overall quality of clinical studies in colorectal surgery. Improved reporting might be best facilitated by having authors adhere to a list of explicitly determined elements that should be included.
循证医学依赖于文献中可重复的高质量报告。以往对顶级影响力临床期刊（包括非手术和手术领域）设计与分析的11个基本要素进行评估的研究表明，报告质量虽在提高，但仍不尽人意。本研究评估《结肠与直肠疾病》及其他临床相关期刊的报告质量，以确定未来可改进的具体领域。
两名独立评估者对1990年、1995年和2000年发表在《结肠与直肠疾病》上的所有随机对照试验进行评估。对2000年发表在《外科学年鉴》《外科学文献》和《胃肠病学》上的所有随机对照试验也进行了额外评估。确定了11个明确界定的、传统上重要的设计与分析基本要素的报告频率。这些要素包括纳入标准的报告、分配前的入组情况、随机化（及方法）、盲法评估（患者和观察者）、并发症、失访情况、统计方法和检验以及效能计算。
评估者间信度较强（kappa值为0.76）。《结肠与直肠疾病》上发表的随机对照试验数量从1990年的5项增加到1995年的13项，再到2000年的17项。在1990年的随机对照试验中，11个基本要素平均报告率为60%。1995年的随机对照试验中，各项报告率为72%（P = 0.05），而2000年的随机对照试验中，11项要素的报告率为77%（与1990年相比，P < 0.002）。报告最好的项目是纳入标准、统计检验的讨论以及对所有失访患者的记录。2000年的随机对照试验中只有11%报告了统计效能计算。对于其他评估的期刊，各项报告率为72%至88%，纳入标准是一直报告最好的项目，效能计算是报告最差的项目。
对于《结肠与直肠疾病》，随机对照试验的数量和报告质量在提高。然而，尽管某些研究标准报告充分，但其他标准并非如此。在解释随机对照试验结果时（无论是否报告了差异），统计效能的计算显然很重要，但只有１１％的研究包含此信息。改进这一单项的报告可能会提高结直肠手术临床研究的整体质量。让作者遵循一份明确确定应包含的要素清单可能最有助于改进报告。
