AMA Manual of Style - Stacy L. Christiansen, Cheryl Iverson 2020

Abstract
Manuscript Preparation for Submission and Publication

In this age of electronic data dissemination and retrieval, in which abstracts are indexed and freely available, a well-written abstract is important in directing readers to articles of potential clinical and research interest and also for discoverability via online searching. The abstract of a research report summarizes the main points of an article: (1) the study objective or importance, (2) the study design and methods, (3) the primary results, and (4) the principal conclusions. Some journals may include funding at the end of the abstract or in the Methods section; others, like the JAMA Network journals, include this information in the Acknowledgment section (see 2.10.10, Funding/Support). For scientific studies and systematic reviews, narrative expressions, such as “X is described,” “Y is discussed,” “Z is also reviewed,” do not add meaning and should be avoided. Results should be presented in quantitative fashion, but authors and editors should be scrupulous in verifying the accuracy of all data and numbers reported and ensuring consistency with the results published in the full article.⁷

2.5.1 Structured Abstracts.

For reports of original research, systematic reviews, and clinical reviews, structured abstracts (abstracts that use section heads) are recommended. Specific advice taken from JAMA’s Instructions for Authors,⁸ adapted from Haynes et al,⁹ is given here. Note that Design, Setting, and Patients or Other Participants subsections may be combined, depending on the description. If no intervention was performed, that heading may be omitted. Many journals limit the number of words in abstracts; the JAMA Network journals, for example, allow 350 words for reports of original research and for systematic reviews.

2.5.1.1 Structured Abstracts for Reports of Original Data (Including Systematic Reviews With Meta-analysis).

In reports of original research, include an abstract of no more than 350 words using the headings suggested below. For brevity, phrases rather than complete sentences may be used. Include the following content in each section:

Importance: Begin the abstract with a sentence or 2 explaining the clinical or research importance of the study question.

Objective: State the precise objective or study question addressed in the report (eg, “To determine whether . . . ”). If more than 1 objective is addressed, indicate the main objective and state only key secondary objectives. If an a priori hypothesis was tested, state that hypothesis.

Design: Describe the basic design of the study (eg, randomized clinical trial, cohort study, cross-sectional study, case-control study, case series, meta-analysis). State the years of the study, the duration of follow-up, and the date of the current analysis if the data are older than 3 years. If applicable, include the name of the study (eg, the Framingham Heart Study). Note: The JAMA Network journals usually combine the following 3 sections as Design, Setting, and Participants during editing for publication. As relevant, indicate whether observers were aware or unaware of patient groups or allocations, particularly for subjective measurements.

Setting: Describe the study setting to assist readers in determining the applicability of the report to other circumstances, for example, general community, primary care or referral center, private or institutional practice, or ambulatory or hospitalized care. If the actual name of the institution is relevant, it may be given here.

Participants: State the clinical disorders, important eligibility criteria, and key sociodemographic features of patients. Provide the numbers of participants and how they were selected (see below), including the number of otherwise eligible individuals who were approached but declined to participate. If matching was used for comparison groups, specify the characteristics that were matched. In follow-up studies, indicate the proportion of participants who completed the study. In intervention studies, provide the number of patients who withdrew from the study because of adverse effects. For selection procedures, use the following terms, if appropriate: random sample (where random refers to a formal, randomized selection in which all eligible individuals have a fixed and usually equal chance of selection), population-based sample, referred sample, consecutive sample, volunteer sample, or convenience sample.

Intervention(s) or Exposure(s): Describe the essential features of any interventions or exposures, including their method and duration. Name the intervention or exposure by its most common clinical name, and use nonproprietary names of drugs or medical devices.

Main Outcome(s) and Measure(s): Indicate the primary study outcome measurement(s) as planned before data collection began. If the manuscript does not report the main planned outcomes of a study, state this fact and indicate the reason. State clearly whether the hypothesis being tested was formulated during or after data collection. Explain outcomes or measurements unfamiliar to a general medical readership.

Results: Report and quantify the main outcomes of the study, including baseline demographic characteristics and final included and analyzed sample. Include absolute numbers and measures of absolute risks (eg, increase, decrease, or absolute differences between groups), along with effect sizes and appropriate measures of uncertainty such as CIs. Use means and SDs for normally distributed data and medians and ranges or interquartile ranges (IQRs) for data that are not normally distributed. Avoid solely reporting the results of statistical hypothesis testing, such as P values, which fail to convey important quantitative information. For most studies, P values should follow the reporting of comparisons of absolute numbers or rates and measures of uncertainty (eg, 0.8%, 95% CI −0.2% to 1.8%; P = .13). P values should never be presented alone without the data that are being compared. Approaches such as number needed to treat to achieve a unit of benefit may be mentioned when appropriate. Measures of relative risk (eg, relative risk, hazard ratios) may also be reported and should include CIs. Studies of screening and diagnostic tests should report sensitivity, specificity, and likelihood ratio. If predictive value or accuracy is reported, prevalence or pretest likelihood should be given as well. All randomized clinical trials should include the results of intention-to-treat analysis and adverse effects. All surveys should include response rates.

Conclusions and Relevance: Provide only conclusions of the study that are directly supported by the results. Give equal emphasis to positive and negative findings of equal scientific merit. In addition, provide a statement of clinical relevance, indicating implications for clinical practice or health policy and avoiding speculation and overgeneralization. The relevance statement may also indicate whether additional study is required before the information should be used in clinical settings.

Trial Registration: For clinical trials, provide the name of the trial registry, the registration number, and the URL of the registry.

2.5.1.2 Structured Abstracts for Meta-analysis

Manuscripts reporting the results of meta-analyses should include an abstract of no more than 350 words using the headings listed below. The text of the manuscript should also include a section describing the methods used for data sources, study selection, data extraction, and data synthesis. Each heading should be followed by a brief description:

Importance: Begin the abstract with a sentence or 2 explaining the importance of the systematic review question that is used to justify the meta-analysis.

Objective: State the precise primary objective of the meta-analysis. Indicate whether the systematic review for the meta-analysis emphasizes factors such as cause, diagnosis, prognosis, therapy, or prevention and include information about the specific population, intervention, exposure, and tests or outcomes that are being analyzed.

Data Sources: Succinctly summarize data sources, including years searched. The search should include the most current information possible, ideally with the search being conducted within several months before the date of manuscript submission. Potential sources include computerized databases and published indexes, registries, meeting abstracts, conference proceedings, references identified from bibliographies of pertinent articles and books, experts or research institutions active in the field, and companies or manufacturers of tests or agents being reviewed. If a bibliographic database is used, state the exact indexing terms used for article retrieval, including any constraints (eg. English language or human study participants). If abstract space does not permit this level of detail, summarize sources in the abstract including databases and years searched, and place the remainder of the information in the Methods section.

Study Selection: Describe inclusion and exclusion criteria used to select studies for detailed review from among studies identified as relevant to the topic. Details of selection should include particular populations, interventions, outcomes, or methodological designs. The method used to apply these criteria should be specified (eg. blinded review, consensus, multiple reviewers). State the proportion of initially identified studies that met selection criteria.

Data Extraction and Synthesis: Describe guidelines (eg, PRISMA, MOOSE) used for abstracting data and assessing data quality and validity (such as criteria for causal inference and whether data were pooled using a fixed-effect or random-effects model). The method by which the guidelines were applied should be stated (eg. independent extraction by multiple observers).

Main Outcome(s) and Measure(s): Indicate the primary study outcome(s) and measurement(s) as planned before data collection began. If the manuscript does not report the main planned outcomes of a study, this fact should be stated and the reason indicated. State clearly if the hypothesis being tested was formulated during or after data collection. Explain outcomes or measurement unfamiliar to a general medical readership.

Results: Provide the number of studies and patients/participants in the analysis and state the main quantitative results of the review. When possible, present numerical results (eg, absolute numbers and/or rates) with appropriate indicators of uncertainty, such as CIs. Use means and SDs for normally distributed data and medians and ranges or interquartile ranges (IQRs) for data that are not normally distributed. Avoid solely reporting the results of statistical hypothesis testing, such as P values, which fail to convey important quantitative information. For most studies, P values should follow the reporting of comparisons of absolute numbers or rates and measures of uncertainty (eg, 0.8%, 95% CI −0.2% to 1.8%; P = .13). P values should never be presented alone without the data that are being compared. Meta-analyses should state the major outcomes that were pooled and include odds ratios or effect sizes and, if possible, sensitivity analyses. Evaluations of screening and diagnostic tests should include sensitivity, specificity, likelihood ratios, receiver operating characteristic curves, and predictive values. Assessments of prognosis should summarize survival characteristics and related variables. Major identified sources of variation between studies should be stated, including differences in treatment protocols, co-interventions, confounders, outcome measures, length of follow-up, and dropout rates.

Conclusions and Relevance: The conclusions and their applications (clinical or otherwise) should be clearly stated, limiting interpretation to the domain of the review.

2.5.1.3 Structured Abstracts for Systematic Reviews (Without Meta-analysis).

In manuscripts that are critical assessments of the literature and data sources pertaining to clinical topics, include an abstract of no more than 350 words, using the headings (which are included in the word count) suggested below. In the text of the manuscript, include a section describing the methods used for searching for data sources, study selection, data extraction, and data synthesis. Follow each heading with a brief description.

Importance: Include 1 or 2 sentences describing the clinical question or issue and its importance in clinical practice or public health.

Objective: State the precise primary objective of the review. Indicate whether the review emphasizes factors such as cause, diagnosis, prognosis, therapy, or prevention and include information about the specific population, intervention, exposure, and tests or outcomes that are being reviewed.

Evidence Review: Describe the information sources used, including the search strategies, years searched, and other sources of material, such as subsequent reference searches of retrieved articles. Methods used for inclusion and exclusion of identified articles and for assessment of quality of included articles should be explained.

Findings: Include a brief summary of the number of articles included, numbers of various types of studies (eg, clinical trials, cohort studies), and numbers of patients or participants represented by these studies. Summarize the major findings of the review of the clinical issue or topic in an evidence-based, objective, and balanced fashion, with the highest-quality evidence available receiving the greatest emphasis. Provide quantitative data.

Conclusions and Relevance: The conclusions should clearly answer the questions posed if applicable, should be based on available evidence, and should emphasize how clinicians might apply current knowledge. Conclusions should be based only on results described in the abstract Findings subsection.

2.5.1.4 Structured Abstracts for Narrative Reviews.

Narrative Reviews on clinical topics provide an up-to-date review for clinicians on a topic of general common interest from the perspective of internationally recognized experts in these disciplines. Include an abstract of no more than 300 words using the headings listed below.

Importance: An overview of the topic and discussion of the main objective or reason for this review.

Observations: The principal observations and findings of the review.

Conclusions and Relevance: The conclusions of the review that are supported by the information, along with clinical applications. How the findings are clinically relevant should be specifically stated.

2.5.2 Unstructured Abstracts.

For other major manuscripts, include an unstructured abstract (a paragraph without headings) of no more than 200 words that summarizes the objective, main points, and conclusions of the article. Abstracts are not required for opinion pieces, letters, and special features, such as news articles, although many journals include short 1-sentence précis summarizing these articles. Consult the journal’s instructions for authors for special requirements in individual publications.

2.5.3 Tables and/or Figures in Abstracts.

Ways to make the structured abstract even more informative continue to evolve. Among these is the inclusion of a small table in the Results subsection in selected abstracts. This approach debuted in the February 6, 2013, issue of JAMA.¹⁰ Other journals have also begun experimenting with inclusion of tables or figures in abstracts (graphical abstracts).¹¹

2.5.4 Clinical Trial Registration Identifier in Abstracts.

The JAMA Network journals require that for all manuscripts that report the results of clinical trials, the trial must be registered at an appropriate online public registry that is owned by a not-for-profit entity, is publicly accessible, and requires the minimum registration data set as described by the ICMJE.¹ The numerical identifier issued by the trial registry should be listed at the end of the abstract, along with the name and URL of the trial registry. The numerical identifier and the trial registry URL should be checked for accuracy by the manuscript editor. Acceptable trial registries include the Australian New Zealand Clinical Trials Registry (http://www.anzctr.org.au), ClinicalTrials.gov (http://www.clinicaltrials.gov), ISRCTN Registry (http://isrctn.org), Nederlands Trial Register (http://www.trialregister.nl/trialreg/index.asp), University Hospital Medical Information Network Clinical Trials Registry (http://www.umin.ac.jp/ctr), and others listed at the Clinical Trial Registry Platform of the World Health Organization (http://www.who.int/ictrp/network/primary/en).

2.5.5 General Guidelines.

A few specific guidelines to consider in preparing a structured or an unstructured abstract follow. Above all, keep in mind that the abstract is meant to stand alone, independent of the main manuscript.

■Consult the journal’s instructions for authors.

■Use the journal’s specific headings when preparing a structured abstract.

■Do not begin the abstract by repeating the title.

■Do not cite references or URLs in the body of the abstract.

■Do not cite figures or tables in the abstract (but see also 2.5.3, Tables and/or Figures in Abstracts).

■Present numerical results (eg, absolute numbers and/or rates) with appropriate indicators of uncertainty, such as CIs. Use means and SDs for normally distributed data and medians and ranges or interquartile ranges (IQRs) for data that are not normally distributed. Avoid solely reporting the results of statistical hypothesis testing, such as P values, which fail to convey important quantitative information. For most studies, P values should follow the reporting of comparisons of absolute numbers or rates and measures of uncertainty (eg, 0.8%, 95% CI −0.2% to 1.8%; P = .13). P values should never be presented alone without the data that are being compared (see 19.1, The Manuscript: Presenting Study Design, Rationale, and Statistical Analysis).

■Include major terms and describe databases and study groups (related to the subject under discussion) in the abstract because the text of the abstract can be searched in many retrieval systems.

■Include the hypothesis or study question, if applicable.

■Ensure that all concepts and data mentioned in the abstract are also included in the text.

■Include the active moiety of a drug at first mention (see 14.4, Drugs).

■Avoid proprietary names or manufacturers’ names unless they are essential to the study (see 14.5, Equipment, Devices, Reagents, and Software).

■Spell out abbreviations at first mention and avoid use of abbreviations unless they appear numerous times (see 13.0, Abbreviations).

■If an isotope is mentioned, spell out the name of the element when first used and provide the isotope number on the line (see 14.9, Isotopes).

■Provide the dates of the study or date ranges for studies and other data included in review articles.

■Verify the numbers provided in the abstract against those provided in the text, tables, and figures to ensure internal consistency.