Almost 3 decades ago, in an editorial entitled “The Scandal of Poor Medical Research,” Douglas Altman painted a dire image on the misuse of statistical methods in medical research: “(…) seriously flawed [research] through the use of inappropriate designs, unrepresentative samples, small samples, incorrect methods of analysis, and faulty interpretation.”¹ It is hard to know exactly how much things have improved since then. More cynical researchers will claim that very little has changed in terms of the quality of statistical methodology reported in medical journals.^2,3 I tend to have a more optimistic position about the current situation.

The publication year of Altman’s editorial coincided with the beginning of my career as a statistician. At the time, the scientific field of biostatistics was quite unknown to the public, and even among academics from other areas. It was not rare to find inquisitive expressions when the term was used! I recall with amusement being introduced several times as a bio-aesthetician, which I always found very flattering. During my career, I have witnessed the profession being raised from an obscurant activity to one of the “sexiest jobs” in the world⁴ and listed in the top emerging occupations of different countries.^5–7 Biostatistics is nowadays recognized as an integral part of the medical curriculum in most degree programs around the world and has established itself as a foundational science for clinical research and evidence-based medical practice.

The increased prominence of biostatistics is due, in part, to the information and digital data revolution that have affected most human activities, including medicine. As easier access to better and richer data grows, it will be required that more statisticians be involved in medical research and guide the field toward new approaches to tackle novel research problems. Consequently, we have experienced an increased sophistication of statistical methods used in different disciplines of medical research.^8–11 A quick search in PubMed of some terms associated with more complex statistical analyses commonly encountered at Circulation: Cardiovascular Quality and Outcomes highlights this trend (Table).

Table. Number of Articles in PubMed That Included the Respective Statistical Expressions in the Abstract, for 2010 and 2020

To each one of the expressions, “& year [dc]” was added, where year was replaced by 2010 or 2020, to search within the respective year. The first row indicates the total number of articles in 2010 and 2020 as a baseline comparison.

With the rise in sophistication of methods, many medical journals nowadays include a statistical reviewer in their process of peer review, and it is becoming more common for journals to have statisticians on their core editorial team. Given the strong quantitative focus of Circulation: Cardiovascular Quality and Outcomes, this has been our practice. Our Editorial Board includes a strong team of 7 academic biostatisticians, who are involved in the discussion and evaluation of nearly all manuscripts that are submitted to our journal. Other journals of the American Heart Association have adopted a similar strategy for their Editorial Board arrangements.

In late 2018, the American Heart Association pursued another important initiative in a continuous effort to improve the scientific standards of its publications. Several statisticians across the editorial boards of the different American Heart Association journals formed a task force with the purpose of developing standard recommendations and guidelines for reporting statistical methods and results in manuscripts submitted for publication to the American Heart Association journals. Our goal was to propose more specific recommendations for different types of statistical methods used often in medical research and when appropriate, integrate these recommendations with existing guidelines. This document was not only meant as a list of best practices for authors but importantly, as a guide for statistical reviewers in determining whether the data have been analyzed and reported appropriately. Statistical reviewers (like all peer reviewers) are likely to benefit from specific guidance on how to evaluate and report on concerns that might be raised in the assessment of research in a manuscript.

After 2 years of meetings, discussions, reviews and compromises, the task force completed this challenge, and the “Recommendations for Statistical Reporting in Cardiovascular Medicine” is being finally published in Circulation.¹² It was challenging, frustrating, intellectually stimulating, and quite enjoyable—but most of all, it was a privilege to have worked closely with such a remarkable group of colleagues. I am confident that this document will be a significant tool for authors and a major enhancement in the peer review process and help reduce the heterogeneity of the statistical reviews, within and between journals! And, hopefully, a modest but additional step away from the “scandal of poor medical research.”

The author thanks Dr Brahmajee Nallamothu and Dr Michael Ho for the valuable comments.

Disclosures Disclosures provided by Dr Teixeira-Pinto in compliance with the American Heart Association annual Journal Editor Disclosure Questionnaire are available at https://www.ahajournals.org/pb-assets/policies/COI_09_2020-1600719273583.pdf.

The opinions expressed in this article are not necessarily those of the American Heart Association.

For Disclosures, see page 756.

大约 3 年前，在一篇题为“不良医学研究的丑闻”的社论中，道格拉斯·奥特曼描绘了医学研究中滥用统计方法的可怕形象：“（……）通过使用不适当的设计，[研究] 存在严重缺陷，不具有代表性的样本、小样本、不正确的分析方法和错误的解释。” ¹很难确切地知道从那时起情况有多大改善。更多愤世嫉俗的研究人员会声称，医学期刊报道的统计方法的质量几乎没有变化。^2,3我倾向于对当前形势持更乐观的态度。

Altman 社论的出版年份恰逢我开始了统计学家的职业生涯。当时，生物统计学这个科学领域对公众来说是相当陌生的，甚至在其他领域的学者中也是如此。使用这个词时，发现好奇的表达并不罕见！我记得作为生物美学家被介绍过好几次，我总是觉得这很讨人喜欢。在我的职业生涯中，我见证了这个职业从一个默默无闻的活动上升到世界上“最性感的工作”之一⁴并被列入不同国家的顶级新兴职业。^5–7如今，生物统计学被认为是世界上大多数学位课程中医学课程的一个组成部分，并已成为临床研究和循证医学实践的基础科学。

生物统计学日益突出的部分原因是信息和数字数据革命影响了包括医学在内的大多数人类活动。随着越来越容易获得更好、更丰富的数据，需要更多的统计学家参与医学研究，并引导该领域采用新方法来解决新的研究问题。因此，我们经历了不同医学研究学科中使用的统计方法的日益复杂。^8-11在 PubMed 中快速搜索一些与循环中常见的更复杂统计分析相关的术语：心血管质量和结果突出了这一趋势（表）。

桌子。2010 年和 2020 年 PubMed 中在摘要中包含各自统计表达的文章数量

在每个表达式中都添加了“& year [dc]”，其中年份被 2010 或 2020 替换，以在相应年份内进行搜索。第一行表示作为基线比较的 2010 年和 2020 年的文章总数。

随着方法的复杂化，如今许多医学期刊在同行评审过程中都包括一名统计审稿人，并且期刊在其核心编辑团队中拥有统计员变得越来越普遍。鉴于循环的强大定量重点：心血管质量和结果，这一直是我们的做法。我们的编辑委员会包括一个由 7 名学术生物统计学家组成的强大团队，他们参与了几乎所有提交给我们期刊的手稿的讨论和评估。美国心脏协会的其他期刊对其编委安排也采用了类似的策略。

2018 年底，美国心脏协会采取了另一项重要举措，不断努力提高其出版物的科学标准。来自不同美国心脏协会期刊编辑委员会的几位统计学家组成了一个工作组，目的是制定标准建议和指南，以在提交给美国心脏协会期刊的手稿中报告统计方法和结果。我们的目标是为医学研究中经常使用的不同类型的统计方法提出更具体的建议，并在适当的时候将这些建议与现有指南相结合。本文档不仅是作者的最佳实践列表，而且重要的是，作为统计审查人员确定数据是否经过适当分析和报告的指南。统计审稿人（与所有同行审稿人一样）可能会受益于有关如何评估和报告手稿研究评估中可能提出的问题的具体指导。

经过 2 年的会议、讨论、审查和妥协，工作组完成了这一挑战，“心血管医学统计报告建议”终于在Circulation 上发表。¹²这是具有挑战性的、令人沮丧的、智力上的刺激，而且相当令人愉快——但最重要的是，能与这样一群出色的同事密切合作是一种荣幸。我相信这份文件将成为作者的重要工具和同行评审过程的重大改进，并有助于减少期刊内部和期刊之间统计评论的异质性！并且，希望是远离“医学研究不佳的丑闻”的适度但额外的一步。

作者感谢 Brahmajee Nallamothu 博士和 Michael Ho 博士的宝贵意见。

披露Teixeira-Pinto 博士根据美国心脏协会年度期刊编辑披露问卷提供的披露可在 https://www.ahajournals.org/pb-assets/policies/COI_09_2020-1600719273583.pdf 上获得。

本文中表达的观点不一定是美国心脏协会的观点。

有关披露，请参见第 756 页。