Authors unsure of the suitability of their manuscript for publication may save considerable time and effort by sending an Inquiry. An editor will generally respond by email within one week.

INVITED/COMMISSIONED ARTICLES

Certain article types, including reviews and editorials, are usually solicited by the editors in advance of submission. However, authors interested in proposing ideas for these article types may also send Presubmission Inquiries.

Our Statistical Editors recommend the following best practices in preparing manuscripts for submission to the Journal. We recommend that you familiarize yourself with the guidelines and follow them, both in the design of research studies and in reporting research findings to the Journal. The review of a manuscript takes into consideration the methods described in the protocol and statistical analysis plan. However, the scientific integrity of the analysis is of primary importance and, on occasion, may override the prespecified methods; this may lead to requests from the Journal for changes in the analysis as submitted.

1. Guidelines for all studies
   
   1. General requirements and reporting style:
      
      1. A protocol document (or equivalent) and a statistical analysis plan (SAP) should be submitted with each manuscript. The SAP should contain enough detail to enable someone else to replicate the analysis in a similar data set. Original and final versions of these documents are important to include if the documents have evolved. Documents should be dated, and changes in the study plans and accompanying rationale should be described. For clinical trials, formal protocol amendments and corresponding changes to the SAP will typically be available and should be included with the manuscript submission.
      2. The Methods section of the manuscript should contain a brief description of the methods for primary, secondary, and exploratory analyses, as well as a brief description of sample size considerations for the study, including statistical power computations when applicable. More detail on any aspect of the statistical methods can be placed in the supplementary appendix.
      3. To prevent confusion, we recommend that authors draw a clear distinction between statistical significance and clinical (or other nonstatistical) significance. This can be achieved by reserving the adjective “significant” to mean “statistically significant.”
      4. Significance tests should be accompanied by effect estimates with standard errors or confidence intervals.
      5. Confidence intervals for ratio quantities, such as relative risks, odds ratios, and hazard ratios, should be computed in the log scale for inclusion in forest plots.
      6. Unless one-sided tests are required by the study design, such as noninferiority trials, all reported P values should be two-sided.
      7. In general, P values larger than 0.01 should be reported to two decimal places, and those between 0.01 and 0.001 to three decimal places; P values smaller than 0.001 should be reported as P<0.001. Notable exceptions to this policy include P values arising from tests associated with stopping rules in clinical trials or from genome-wide association studies.
      8. Results should be presented with no more precision than is of scientific value and is meaningful, given the available sample size. For example, measures of association, such as odds ratios, should ordinarily be reported to two decimal places. Results derived from models should be limited to the appropriate number of significant digits.
   2. Multiplicity considerations:
      
      1. To enhance the reproducibility of study results, the Journal emphasizes the need to account for the multiplicity of hypothesis testing in the analysis and to control the probability of error. It is essential that methods to address multiplicity be explicitly prespecified in the protocol or SAP.
      2. For confirmatory analyses, such as those typically reported for endpoints in clinical trials, the SAP should prespecify a plan to control the overall type I error (family-wise error rate). For example, this plan may include Bonferroni adjustments or prespecified hierarchical procedures. P values adjusted for multiplicity should be reported when appropriate and labelled as such in the manuscript. In hierarchical testing procedures, P values should be reported only until the last comparison for which the P value was significant. Thus, P values for the first nonsignificant comparison and for all comparisons thereafter should not be reported.
      3. For exploratory analyses, such as genomewide association studies, investigators may use methods for controlling false discovery rates; the methods should be prespecified in the protocol or SAP.
      4. When no method to adjust for a multiplicity of inferences was prespecified in the protocol or SAP, reporting of secondary and exploratory endpoints should be limited to point estimates of effects with 95% confidence intervals. In such cases, the Methods section should state that the widths of the intervals have not been adjusted for multiplicity and that the intervals may not be used in place of hypothesis testing. The interpretation of these confidence intervals should avoid the language of definitive conclusions used to report statistically significant findings as assessed by formal hypothesis testing. These recommendations supersede previous guidance from the Journal.
   3. Missing data:
      
      1. The manuscript should report the number of participants with missing data for baseline variables (e.g., in Table 1) and for all variables used in analyses, including response and predictor variables.
      2. Unless missingness is rare, a complete case analysis is subject to potential bias and is generally not acceptable as the primary analysis. In the Abstract and Results sections of the manuscript, authors should report analyses that address missing data using appropriate methods and assumptions about missing data, such as missing at random.
      3. The Methods section of the manuscript should include a description of the approach used to address missing data, including a statement of assumptions being made about the missing data or the missing data mechanism. Because these assumptions cannot be tested with data, they should be accompanied by a contextual justification.
      4. Multiple imputation, inverse probability case weights, or appropriately specified statistical models can be used when data are assumed to be missing at random. Sensitivity analyses may be necessary to assess the impact of systematic or informative missingness (for example, when the missing-at-random assumption may not be viable). Details on sensitivity analyses that assess the impact of alternative assumptions about missing data should be provided in the supplementary appendix.
      5. The guidelines on missing data provided above apply to all analyses presented in the manuscript, such as those addressing primary and secondary endpoints.
   4. Time-to-event data, and survival analyses:
      
      1. Estimates of the hazard ratio (HR) are not directly interpretable when the assumption of proportional hazards (PH) is not consonant with the data. When reporting HR estimates, the manuscript should address the appropriateness of the PH assumption and provide corroborating evidence. If the PH assumption appears to be inconsistent with the data, HR estimates should not be reported and alternative metrics should be used to report comparisons.
      2. When some censoring may be due to dependent competing events, Kaplan–Meier estimates of event-time distributions may be biased and should be replaced by appropriate estimates of cumulative incidence. HRs estimated with a Cox PH estimate may also be biased and should be replaced by measures of association that account for dependent competing risks.
   5. Bayesian analyses:
      
      1. When a Bayesian approach to the design and analysis of a study has been chosen, the approach should be applied consistently. The design should be based on Bayesian considerations, and the analysis should be consistent with the design.
      2. When a Bayesian analysis is planned for a study, the protocol and SAP should describe the key elements of the approach. In particular, this includes
         
         1. the form of the likelihood function
         2. the choice of prior distributions
         3. the method for deriving the posterior distribution
         4. the method for deriving summaries of the posterior distribution, on which inference will be based.
      3. The results of a Bayesian analysis should be accompanied by an adequate exploration of the sensitivity of the reported inferences to assumptions about the prior distributions.
      4. The presentation of frequentist analyses alongside the primary Bayesian analyses is not encouraged.
2. Additional guidelines for clinical trials
   
   1. The analysis of the primary outcome in manuscripts reporting results of clinical trials should match the analyses prespecified in the final protocol and SAP, except in unusual circumstances. Analyses that do not conform to the protocol should be justified in the Methods section of the manuscript. The editors may ask for additional analyses that are not specified in the protocol.
   2. P values should not be included in the traditional Table 1 of a randomized trial manuscript showing the distribution of baseline variables by treatment group. However, authors should note imbalances in potential confounders that could be due to chance or inconsistencies in randomization.
   3. Forest plots are often used to present results from an analysis of the consistency of a treatment effect across subgroups of factors of interest. Such plots can be a useful display of estimated treatment effects across subgroups, and the editors recommend that they be included for important subgroups. However, multiplicity considerations continue to apply to forest plots. Thus, if these analyses are not covered in the prespecified plan for multiplicity adjustment, the forest plots should not include P values for treatment by subgroup interactions.
   4. When safety outcomes do not constitute primary endpoints in a study, no adjustment for multiplicity is necessary for their analysis. Because the information contained in the safety endpoints may signal problems within specific organ classes, the editors believe that experiment-wide type I error rates larger than 0.05 are acceptable. In particular, the editors may request that P values be reported for comparisons of the frequency of adverse events among treatment groups, regardless of whether such comparisons were prespecified in the SAP.
   5. Per protocol analyses that are based on an analysis dataset constructed by eliminating cases on the basis of post-randomization events (e.g., treatment discontinuation because of side effects or dose changes not specified in the protocol) are generally biased and are generally not allowed.
   6. When possible, the editors prefer that absolute event counts or rates be reported before relative risk or hazard ratio estimates. The goal is to provide the reader with both the actual event frequency and the relative frequency.
   7. In general, relative risk estimates are the preferred quantities to report. If authors plan to rely on odds ratios in the analysis, a justification should be provided in the Methods section addressing the concerns that odds ratios may overestimate the relative risks and may be misinterpreted.
   8. Authors should provide a flow diagram in CONSORT format. The editors also encourage authors to submit all the relevant information included in the CONSORT checklist. Although all this information may not be published in the manuscript, it should be provided in either the manuscript or a supplementary appendix at the time of submission. The CONSORT statement, checklist, and flow diagram are available on the CONSORT website.
   9. In assessing robustness of findings about the primary endpoint, authors should distinguish between sensitivity analyses and other types of analyses (such as analyses of particular subsets). A sensitivity analysis is designed to assess the impact of statistical assumptions, analysis methods, or model choices on inferences about an endpoint.
3. Additional guidelines for observational studies
   
   1. The validity of findings from observational studies depends on several important assumptions, including those relating to sample selection, measured and unmeasured confounding, and the adequacy of methods used to control for confounding. The Methods section of observational studies manuscripts should describe how these and other relevant issues were addressed in study design and analysis.
   2. As noted in the section on general requirements, a protocol document and an SAP document must be submitted with the manuscript. If a protocol document was not compiled for the particular study, alternative sources of information may be used, including such documents as sections of grant applications and study descriptions prepared for local institutional review boards and other review bodies.
   3. In any case, an SAP document must be submitted, even if it was not prepared during the conduct of the study. The SAP should include the aims of the study, identify primary and secondary endpoints, and describe the eligibility criteria for the selection of cases and method of sampling from the data, with a diagram as appropriate. The SAP should also provide a detailed description of the association or causal effect to be estimated and the rationale for this choice, and describe the prespecified methods of analysis to draw inferences about treatment or exposure effect or association and plans to control for the multiplicity of inferences, as discussed earlier in these guidelines. The Journal encourages authors to deposit SAPs for observational studies in one of the online repositories designed for this purpose. The SAP should be dated and should make clear whether it was written before or after preliminary data analyses.
   4. Causal language (e.g., exposure X leads to outcome Y, or changes in exposure X produce changes in outcome Y) should not be used in observational studies where only associations can be estimated.
   5. In studies in which estimating a causal effect is the goal, the analysis should use methods specifically designed for causal inference and should be accompanied by a description of the assumptions required to support a causal interpretation. These methods generally attempt to mimic the benefits of randomization using techniques such as matching, instrumental variables, inverse probability weights, or standardization techniques. The choice of method should be justified and be accompanied by diagnostics.
   6. Studies reporting the association of a treatment or exposure with an outcome should show the distribution of potential confounders and other variables, stratified by exposure or intervention group. When the analysis depends on the confounders being balanced by exposure group, standardized mean differences between groups, calculated after matching, weighting, or other adjustment technique, should be reported.
   7. Complex models and their diagnostics can often be best described in a supplementary appendix. Authors are encouraged to conduct an analysis that quantifies potential sensitivity to bias from unmeasured confounding; in the manuscript, authors must provide a discussion of potential biases induced by unmeasured confounders.
   8. Authors are encouraged to retest findings in similar but independent study or studies to assess the robustness of their findings.

UNITS OF MEASUREMENT

Authors should express all measurements in conventional units, with Système International (SI) units given in parentheses throughout the text. Figures and tables should use conventional units, with conversion factors given in legends or footnotes. In accordance with the Uniform Requirements, however, manuscripts containing only SI units will not be returned for that reason.

ABBREVIATIONS

Except for units of measurement, abbreviations are strongly discouraged; the first time an abbreviation appears, it should be preceded by the words for which it stands.

DRUG NAMES

Generic names should be used. When proprietary brands are used in research, include the brand name and the name of the manufacturer in parentheses after the first mention of the generic name in the Methods section.

COVER LETTER

Though cover letters are not required, the Online Submission system contains a text field through which important information that is not in the metadata, such as a meeting presentation date or a major conflict of interest not in the manuscript, should be communicated with initial manuscript submissions.

MANUSCRIPT TEXT FILE

Compile all text, references, figure legends, and tables into a single double-spaced digital file (preferably an MS Word document). The Journal will also accept text (.txt), or Rich Text Format (.rtf) files.

TITLE PAGE

Create a title page that includes:

• Manuscript title
• Each author’s name, highest degree, and affiliation/institution
• Contact information for one corresponding author

ABSTRACT

Provide an abstract of not more than 250 words with four labelled paragraphs containing the following:

• Background: The problem being addressed in the study
• Methods: How the study was performed
• Results: Salient results
• Conclusions: What the authors conclude from the study results
• Trial registration number

IDENTIFYING DATA

At appropriate places in the manuscript, please provide the following items:

• If applicable, a statement that the research protocol was approved by relevant institutional review boards or ethics committees and that all human participants gave written informed consent
• Identities of those who analyzed the data
• For clinical trials, registration number and registry name.
• For studies containing microarrays, accession numbers and repository name

REFERENCES

References must be double-spaced and numbered consecutively as they are cited. References first cited in a table or figure legend should be numbered so they will be in sequence with references cited in the text at the point where the table or figure is first mentioned. List all citation authors when there are six or fewer; when there are seven or more, list the first three, followed by et al.

Numbered references to personal communications, unpublished data, or manuscripts either “in preparation” or “submitted for publication” are unacceptable. If essential, such materials can be incorporated at appropriate places in the text.

TABLES

All tables should be included at the end of the manuscript text file. Double-space tables (including footnotes) and provide a title for each table. For Original Articles, there is normally a limit of five figures and tables (total) per manuscript. Extensive tables or supplementary materials will be published as supplemental materials with the digital version of the article.

Authors can either insert figures into text files (preferred) or upload figure files separately. Low-resolution images may be submitted for peer review, but be aware that this Journal may, at a later stage, request high-resolution versions.

A manuscript’s Supplementary Appendix should be paginated, with a table of contents, followed by a list of investigators (if there is one), text (such as methods), figures, tables, and then references. Reference citations in the Appendix and the corresponding list of references should be self-contained with respect to the Appendix. The Appendix must be submitted in two formats: PDF and MS Word (or another editable text format). The Appendix will not be edited for style and will be presented online as additional information provided by the authors.

For outcome scales, provide in the figure legend or table footnotes the range, sign, and minimally important difference (if known). There must be an informative reference citation for the scale. Each figure should include a title and a legend, which should appear on the same page as the figure itself. Tables in the Supplementary Appendix should be labelled Table S1, Table S2, etc. Each table should be accompanied by a title and, if necessary, footnotes.

Please include a clinical trial’s protocol and statistical analysis plan with the submission. Protocol and statistical analysis plan documents should not be redacted except for the names and contact information of individuals who may have signed the documents. If the article is selected for publication, the protocol posted with the article may include redactions of proprietary information but must include information on the patient flow and outcomes.