23.1 Guidance on how to write a scientific paper reporting the results of a trial
Planning the publication strategy
It is important that the results of an intervention trial are published as soon as possible after the trial data have been analysed and the results are available. Generally, the sponsor will require a comprehensive report covering all aspects of the trial. Once such a report has been prepared, papers for publication in scientific journals can be prepared, based on the full report. It is good practice to try to include all of the important findings from the trial in one main paper and to avoid so-called ‘confetti’ publishing where the results are distributed among multiple different papers. While the trial is ongoing, it may be worth publishing a paper on the design and methods used in the trial (some journals specialize in publishing summaries of trial protocols, for example, Trials <http://www.trialsjournal.com>), as then reference can be made to this paper when the main results are published, without having to repeat details of the methodology.
The choice of which journal to submit a paper to will depend on the topic under study, and unfortunately on the results. Some journals are more likely to publish papers with ‘positive’ findings than those showing no effect of an intervention. Most authors will seek to publish their results in a journal with high ‘impact’ (i.e. likely to be read by many people), but it is important to think about who the target audience for the paper is and which journals that audience is most likely to read. It is a good idea to scan past issues of the journal to see the sorts of paper they publish to judge whether there is likely to be interest in publication of the results of a specific trial.
Once the decision has been made of which journal to submit a paper to, it is important to read the instructions to authors, as these vary from journal to journal. Links to websites, which provide instructions to authors for over 6000 journals in the health and life sciences, are given at <http://mulford.utoledo.edu/instr/>.
Drafting a paper
Shown in Box A23.1 is the general structure that most scientific papers have if they are presenting original study results. Approaches to writing papers vary from author to author, but one that we have found useful is outlined here. Also shown in the box is the order in which we suggest different sections of the paper might be drafted.
Box A23.1 Structure of paper and suggested order in which to write the sections
|8||Materials and methods|
|5||Legends to figures|
What parts of a paper are read and by whom?
The vast majority of readers of a journal will scan the title of a paper, and they may look at the list of authors. It is important therefore to highlight, in so far as is possible, the subject of the research and the ‘headline’ finding in the title, in order to provoke interest in reading further. A much smaller proportion of readers will read the abstract/summary than the title, but it is important to try to get all of the messages you want to convey into the summary, as a very small proportion of readers will go beyond that point and read the main body of a paper. A small number of readers will scan the tables and figures, so these should be made as comprehensible as possible, without having to read the paper. Unfortunately, in most instances, a miniscule proportion of those who access the journal will read the whole paper, but these may be the people who really matter!
A good place to start the writing of a paper is to decide on the title! It is suggested that this is revisited, once the drafting of the paper is finished, to consider whether any revision is appropriate. Thus, it is listed as both 1 and 13 in Box A23.1.
An issue which is frequently contentious is who should be included as an author in a paper and in which order the authors should appear. Journals give guidelines as to what contributions are sufficient to merit authorship. Also many journals require that an account is given of the contribution that each author made to the research reported. There is no simple answer as to who should, and who should not, be included as an author, but it is good practice to plan the publications that are likely to come out of a specific trial well in advance of the final analysis of the results and to agree who will be included as an author in different publications. It should also be decided who will be the ‘lead’ authors with the primary responsibility of producing the first draft of specific papers. However, all authors share responsibility for the contents of the paper. It is important to remember this, even if you are only one of many authors in the middle of the publication list. Errors in a publication are usually permanent, and, even if corrections are made in a subsequent communication, these are often missed by readers.
The most critical component in constructing a paper is deciding on, and designing, the tables (or figures) that are needed to describe the study and to summarize the results. Once the tables and figures have been constructed, writing the paper around them should be relatively straightforward. There are four aspects of a trial to which the tables will generally relate:
- description of the characteristics of the study population
- main results
- secondary findings
- your findings in the context of other studies (though a table on these is not always needed).
Ensure that the title of each table is adequate to inform the reader of its content. Try to work out a complete description of the trial results through tables (and figures), even if later the content of smaller tables might be incorporated into the text. Avoid duplication of data in tables and figures. Plan the tables and figures, such that the paper can be largely ‘read’, based on these alone. Keep tables as simple as possible, and avoid unnecessary data, especially data that are not referred to in the text. Two simple tables are better than one complicated table. Label the rows and columns of each table very clearly, and, to the extent possible, avoid abbreviations. Avoid too many significant figures after the decimal point in numbers. For example, an OR of 4.7 is probably sufficient, rather than 4.735. In general, relate the number of decimal places included to the width of the CIs. For example, OR = 1.2 (95% CI: 0.1, 9.7) is more appropriate than OR = 1.23 (95% CI: 0.13, 9.68), whereas OR = 1.48 (95% CI: 1.41, 1.55) is more appropriate than OR = 1.5 (95% CI: 1.4, 1.6). When the tables (and figures) have been drafted, it is a good idea to give them to a colleague who is unfamiliar with the trial for them to tell you how they interpret them.
Figures may be a very powerful way of illustrating findings in a trial. They should be kept as simple as possible, but, if they are too simple, question whether they are really necessary. Consider whether a specific point is better made with a figure or table, and use one or the other, but not both. Label all axes of a graph very clearly, and give the units of measurement either in the figure or in the legend to the figure. For maps and similar diagrams, give a key to all of the symbols used, and show the scale diagrammatically (not 1 cm = 1 km, as the journal may shrink the figure). Have an arrow pointing north on all maps. Avoid using multiple colours, unless really necessary, as many journals are either only printed in black and white or charge extra for colour figures; and, anyway, many readers will print or photocopy a colour figure in black and white.
The section of a paper describing the results of the trial should follow directly from the tables. Summarize what is shown in the tables, with appropriate reference to them. Start with the simplest analysis, for example, simple description of differences, without adjustment for confounding factors, etc. Then develop and describe more sophisticated analyses, as appropriate. Comment on all data shown in each table. If data are not com- mented upon, question the need to include them in tables. When estimates of effect are given (for example, vaccine efficacy), also include the CIs (usually 95%) and the ‘p-value’, but only if this contributes information beyond the CI.
In the initial part of the discussion, focus on the key result(s) of the trial being pre- sented, and summarize the overall findings. Discuss the strengths and limitations of the trial, for example, possible biases that could have influenced the results, and dis- cuss the additional analyses that have been performed to control for potential biases, as appropriate. Then, put the findings of the trial in the context of other such studies, summarizing those studies as necessary, possibly in tabular or figure form. Then, draw overall conclusions derivable from the present study and other similar studies. Finally, make any recommendations for public health action or further research.
Materials and methods
Much of the materials and methods section may have been included in a previous paper, and it may be sufficient merely to summarize them and make reference back to that paper. However, this section of the paper must provide sufficient information for the reader to understand what was done, without having to go back to any previously published paper. The kinds of information that a reader will hope to glean from this section (or the earlier paper) are summarized in Box A23.2.
The introductory section of the paper should be kept as brief as possible, giving the minimum necessary background information to explain any current controversies and why the trial was conducted. Make reference to any recent review papers, as appropri- ate. Specify the hypotheses that the study was designed to evaluate in quantitative terms.
Most journals will give specific instructions of how the summary should be formatted and the maximum number of words allowed. The reasons for doing the trial and why it is important should be summarized in one or two sentences. There should then be a concise summary of results, using the maximum number of words allowed by the journal. Include as many of the key findings as possible, including summary estimates of the effect, with CIs and p-values. Finally, in a sentence or two, summarize the impli- cations of the results and their public health relevance.
Box A23.2 Information that should be included in the Materials and Methods section of a paper
- study area (relevant features)
- study design adopted (for example, cluster randomized trial)
- study population
- sample size determination
- methods of selection/exclusion of participants
- randomization methods and blinding
- informed consent procedures
- measurement methods
- laboratory assays
- follow-up methods
- computing and statistical packages used
- statistical methods employed
- ethical approval (and data and safety monitoring arrangements).
Funding agencies for a trial will often require their contribution to be referred to in a specific way (for example, including the grant reference). There should also be acknowledgment of the contributions of all those who facilitated the conduct of the trial who are not included as authors. These will usually include local health authorities, study participants, fieldworkers, laboratory workers, other study staff, including key administrative staff, local medical staff, and any advisors or consultants. If in doubt as to whether someone should be acknowledged or not, it is generally diplomatic to include them!
Authors should avoid trying to impress with how widely read they are and should only include references to papers which are key to the content of the current paper. Use recent review articles, and select from more accessible journals (for example, open access), wherever possible. Make sure that all of the references are complete (for example, check using PubMed at <http://www.ncbi.nlm.nih.gov/pubmed>), and it is bad practice to include references to articles you have not read! Pay strict attention to the instructions that the journal gives for the formatting of references. For this purpose, it is useful to have invested in a good reference manager system (for example, Reference Manager, Endnote, or Mendeley (<www.mendeley.com>)—which is free).
23.2 Checklist of information to include when reporting a randomized trial
The Consolidated Standards of Reporting Trials (CONSORT) Group have produced several very useful documents (see <http://www.consort-statement.org/>) about how to report trials. These include a very useful checklist (Schulz et al., 2010) which is reproduced with permission in Table A23.1 (abstracted from <www.consort- statement.org>).
Table A23.1 Consort 2010—checklist of information to include when reporting a randomized trial
Adapted from Schulz, K. f. et al., CONSORT 2010 statement: updated guidelines for reporting parallel group
randomised trials, PLoS Medicine, Volume 7, Issue 3, Copyright © Shulz et al. 2010. Reproduced under the Creative Commons Attribution (CC BY) licence. This table is adapted from an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
23.3 A communication action plan for a trial (Annabelle South, Aoife Doyle, David Ross, personal communication)
These extracts are from the aims and objectives and then two key tables (Tables A23.2 and A23.3) and a box (Box A23.3) within the initial communication action plan for the MEMA kwa Vijana (MkV) Trial’s Long-term Evaluation (Doyle et al., 2010). This was a cluster randomized trial of an adolescent sexual and reproductive health (ASRH) intervention in rural Tanzania. The intervention had four main components (Obasi et al., 2006):
- in-school sexual and reproductive health education through teacher-led, peer- assisted participatory lessons that included the use of drama, stories, and games
- youth-friendly reproductive health services, education of health workers about the needs, and methods of providing sexual and reproductive health (SRH) services to youth
- community-based condom promotion and distribution, for and by youth
- community activities to create a supportive environment for the adolescent sexual health interventions.
MEMA kwa Vijana (MkV) Communication Strategy (excerpts)
- Inform ASRH policy and programme design in Tanzania and internationally.
- Increase national and international awareness and uptake of relevant MkV find- ings, materials, and activities.
- Increase stakeholder awareness of, and commitment to, the importance of evidence- based ASRH policy making.
- Improve awareness of availability and policy relevance, and increase uptake of MkV findings, materials, and activities.
- Strengthen ASRH programming and implementation within non-governmental organizations and other civil society organizations through their involvement and partnership in networks and capacity-building activities.
Table A23.2 Example of extracts from a communication action plan for a trial: target audiences
Table A23.3 Example of extracts from a communication action plan for a trial: list of activities
Box A23.3 Example messages for different audiences (drafted after the trial results were known)
Messages about the MkV interventions
- MkV aims to help young people to protect themselves from STIs and un- wanted pregnancies.
- MkV is an innovative adolescent health programme, including teacher-led, peer-assisted sessions in school classrooms. It uses carefully designed and tested education materials and provides youth-friendly health services.
General information on SRH education in schools
- Half of all students in primary schools in rural Mwanza Region have had sex by the time they are 15 years old.
- ASRH education in schools has previously been shown not to increase stu- dents’ sexual activity in many studies around the world.
- ASRH interventions in schools and health units need to be supported by sus- tained interventions in the wider community.
Messages for international technical agencies (WHO, UNAIDS, UNESCO, UNFPA, etc.)
- The MkV trial in Tanzania rigorously evaluated the impact of an innova- tive, multi-component package of interventions delivered by government departments.
- It demonstrated that the package of MkV interventions substantially improved participants’ sexual health-related knowledge, reported attitudes, and some re- ported sexual risk behaviours, but there was no evidence that it reduced HIV, other STIs, or pregnancies.
Message for government department of primary education
- After a pilot project in 60 schools, the MkV sexual health education pro- gramme has been successfully scaled up to over 600 schools through existing government systems and has been shown to improve students’ knowledge.
Source: data courtesy of Annabelle South, Aoife Doyle, and David Ross (personal communication).