*DRAFT Ethical Peer Reviewing*

INITIAL DRAFT by Jonathan Shuster and Robert Oster

EDITS BY WORKING GROUP MEMBERS ARE WELCOME AND ENCOURAGED

(To edit this page: Click on the edit button at the top of this page and add your edits. (You need to be a registered user of CTSpedia.) To see the original document and each set of changes, go to the bottom of this page and click on History. We can easily return to earlier versions of the document.)

Principles for Biostatistical Reviewers of Methods Papers: (Checklist)

1. When it comes to applied papers in biostatistical methods, it is sufficient for the paper to (a) make an original contribution; (b) the contribution will be useful for at least a segment of potential users; (c) the contribution is technically correct; and (d) the assumptions under which the application is to be used are completely specified. If it meets all four criteria, it should be recommended for acceptance, possibly with revisions.

2. It should be up to the authors to set the entire research scope of the paper. Reviewers can recommend further research, but the authors should have the final say as to whether to conduct the research, suggest the research for further consideration by others, or respond privately to the reviewers and editors on the matter.

3. All reviewer and editor comments must be responded to by the authors.

4. Do not accept the role of reviewer if you have the slightest perception of a conflict of interest.

Principles for Biostatistical Reviewers of Biomedical Research Papers: (Checklist)

1. If the paper follows a technically correct and peer reviewed protocol prescribed analysis, that analysis should be accepted as the primary one for the paper.

2. If the protocol prescribed analysis was either incorrectly conceived or incorrectly executed, the paper might still be acceptable if they replace the analysis by another, describe the nature of the plan and execution of the original analysis, and state why it was replaced.

3. Effect sizes and standard errors should accompany all inferential analyses. Prospectively obtained power analyses for major questions are strongly recommended (mandatory for randomized clinical trials), or a reason for their absence stated.

4. Sample sizes should be disclosed as well as the original planned sample sizes.

5. If multiple looks at the data occurred, this must be disclosed.

6. The paper should disclose the name of the individuals who designed and analyzed the study from a statistical perspective, whether or not that individual was a co-author.

7. It is appropriate to request a copy of the original protocol, if the reviewer needs it to properly review the article.

8. Do not accept the role of reviewer if you have the slightest perception of a conflict of interest.

Principles for Biostatistics Reviewers of Biostatistical Grants: (Checklist)

1. Reviewers must be knowledgeable of the mission statement for the specific committee (e.g. study section), and apply these mission criteria rigorously.

2. The grant should be reviewed according to its specific aims. It is totally up to the applicants, not the reviewers, to set them.

3. If a reviewer likes a grant, s/he must object to a decision not to discuss it “UNSCORED” in the NIH definition.

4. Reviewers should be encouraged to insert some positive comments in their review, especially when it comes to those meeting the NIH definition of “Young Investigator”. Comments such as “Strengths: None Noted” are inappropriate when applied to young investigators.

5. Reviewers should consider themselves, at least initially, as advocates for the project, unless or until the project is deemed infeasible.

6. Conflicts of Interest: Although most review groups have definitions of conflicts of interest, reviewers need to be proactive, and make all close decisions in favor of having a conflict.

Principles for Biostatistics Reviewers of Biomedical Research Grants: (Checklist)

1. Recognize that there are valid alternatives to planning and analyzing a study. Avoid insisting on your personal philosophy when it comes to the proposed statistical design and analysis plan.

2. If possible, a biostatistical reviewer should conduct a quality assurance of the power analysis or sample size justification.

3. The biostatistical reviewer may be the only reviewer knowledgeable in data management, and therefore s/he should evaluate the applicants’ plans.

4. Conflicts of Interest: Although most review groups have definitions of conflicts of interest, reviewers need to be proactive, and make all close decisions in favor of having a conflict.

This topic: CTSpedia > WebHome > EthicsBiostat > PrinciplesEthicsReview
Topic revision: 16 Oct 2013, MaryBanach

---