Mora Ripoll R, Ascaso Terrén C, Sentís Vilalta J
Departament de Salut Pública i Legislació Sanitària, Facultat de Medicina, Universitat de Barcelona.
Med Clin (Barc). 1996 Mar 30;106(12):451-6.
The use of statistical techniques has become strongly consolidated in biomedical investigation. The present study analyzes the statistical repertoire of the original articles published in 1993 in four general medicine journals: Med Clin (Barc), Rev Clin Esp, Lancet and N Engl J Med.
A single reviewer examined 100 original articles from Med Clin (Barc), 42 from Rev Clin Esp, 105 from Lancet and 116 from N Engl J Med, studying the use of 18 categories of statistical analysis and the statistical accessibility of the reader. The criteria for assigning techniques to a specific category were those described by Emerson and Colditz. The knowledge of bivariable techniques (standard reader) was established as the reference for the study of statistical access. Statistical use consisted in the tabulation of frequencies, graphic representations and chi square tests.
The five most used categories were: Contingency tables, t and z tests, epidemiologic statistics, survival analysis (in both English journals) and analysis of variance (in both Spanish journals). The percentage of articles without statistics or with only descriptive statistics was 16% in Med Clin (Barc), 29% in Rev Clin Esp, 18% for in Lancet and 23% in the N Engl J Med. A reader familiarized with bivariable techniques has statistical access to 58% for the originals of Med Clin (Barc), 62% of Rev Clin Esp, 35% of the Lancet and 42% of the N Engl J Med.
In the four journals selected, the use of bivariable techniques is still frequent although the growing use of multivariant analysis is of note. The study of the current profile of the standard reader in Spain is the aim of priority.
统计技术在生物医学研究中的应用已得到广泛巩固。本研究分析了1993年发表在四种普通医学期刊上的原创文章的统计方法:《医学临床(巴塞罗那)》(Med Clin (Barc))、《西班牙临床评论》(Rev Clin Esp)、《柳叶刀》(Lancet)和《新英格兰医学杂志》(N Engl J Med)。
一名评审员检查了《医学临床(巴塞罗那)》的100篇原创文章、《西班牙临床评论》的42篇、《柳叶刀》的105篇和《新英格兰医学杂志》的116篇,研究了18类统计分析方法的使用情况以及读者对统计内容的可理解性。将技术分配到特定类别的标准是由艾默生和科尔迪茨描述的那些标准。将双变量技术知识(标准读者)确立为统计可理解性研究的参考标准。统计使用包括频率列表、图表表示和卡方检验。
使用最多的五个类别是:列联表、t检验和z检验、流行病学统计、生存分析(在两种英文期刊中)和方差分析(在两种西班牙文期刊中)。没有统计学内容或只有描述性统计的文章比例在《医学临床(巴塞罗那)》中为16%,在《西班牙临床评论》中为29%,在《柳叶刀》中为18%,在《新英格兰医学杂志》中为23%。熟悉双变量技术的读者对《医学临床(巴塞罗那)》原创文章的统计可理解率为58%,对《西班牙临床评论》为62%,对《柳叶刀》为35%,对《新英格兰医学杂志》为42%。
在所选的四种期刊中,双变量技术的使用仍然很频繁,尽管多变量分析的使用日益增加值得关注。研究西班牙标准读者的当前概况是当务之急。