Schriger David L, Sinha Reshmi, Schroter Sara, Liu Pamela Y, Altman Douglas G
University of California, Los Angeles Emergency Medicine Center, Los Angeles, CA; University of California, Los Angeles School of Medicine, Los Angeles, CA, USA.
Ann Emerg Med. 2006 Dec;48(6):750-6, 756.e1-21. doi: 10.1016/j.annemergmed.2006.06.017. Epub 2006 Sep 15.
We characterize the quantity and quality of data tables and figures in reports of randomized controlled trials (RCTs) submitted to the British Medical Journal (BMJ) and published in peer-reviewed journals. We investigate how the peer review process affected table and figure quality.
We reviewed 62 consecutive reports of RCTs submitted to the BMJ in 2001 that were later published in the BMJ (n=12) or elsewhere. We counted and categorized the tables and figures in both the initial submissions and published articles. Using standardized instruments and procedures, we analyzed the quality of these tables and figures and checked BMJ editorial documents to see whether changes were triggered by their review process.
The numbers of tables and figures did not change markedly between submission and publication. Five percent of publications had no data tables; 56% had no data figures. Data density was low for published tables and figures. Tables seldom showed data stratified on important covariates; 88% of published tables were simple lists or were stratified on only 1 variable. Less than half the figures met their data presentation potential, with most failing to portray by-subject data and few displaying advanced features such as pairing, symbolic dimensionality, or small multiples. BMJ external peer reviewers seldom commented on tables or figures.
Tables and figures can convey details and complex relationships not easily described in text. Although tables are included in most submitted and published articles, they are not presented optimally; figures are used sparingly and are also of suboptimal quality. Journals should consider improving their table and figure quality in the hope that improved graphics will empower readers to scrutinize the data, thereby dissuading authors from presenting biased analyses and misrepresented conclusions.
我们对提交给《英国医学杂志》（BMJ）并发表在同行评审期刊上的随机对照试验（RCT）报告中的数据表和图表的数量及质量进行了描述。我们调查了同行评审过程如何影响表格和图表的质量。
我们回顾了2001年连续提交给BMJ的62份RCT报告，这些报告后来发表在BMJ（n = 12）或其他地方。我们对初始投稿和发表文章中的表格和图表进行了计数和分类。使用标准化工具和程序，我们分析了这些表格和图表的质量，并查阅了BMJ编辑文件，以查看其评审过程是否引发了变化。
从投稿到发表，表格和图表的数量没有明显变化。5%的出版物没有数据表；56%没有数据图表。已发表的表格和图表的数据密度较低。表格很少显示按重要协变量分层的数据；88%已发表的表格是简单列表或仅按1个变量分层。不到一半的图表发挥了其数据呈现潜力，大多数未能描绘个体数据，很少有图表展示诸如配对、符号维度或小倍数等高级特征。BMJ外部同行评审员很少对表格或图表发表评论。
表格和图表可以传达文本中不易描述的细节和复杂关系。虽然大多数投稿和发表的文章都包含表格，但它们的呈现方式并非最佳；图表使用较少，质量也欠佳。期刊应考虑提高其表格和图表的质量，希望改进后的图表能使读者更好地审视数据，从而促使作者避免呈现带有偏见的分析和错误表述的结论。
