24
Structure of a Scientific Paper in Biomedical Sciences
Introduction
Writing of a scientific paper signifies the logical culmination of research—the first step towards communicating the new information and data generated to be an integral part of the formal domain of scholarly communication system—the learned journal. Unless the research, carried out with considerable effort and often with public funding, becomes readily accessible, it does not serve its intended purpose of application of the new information for the advancement of science and technology. Traditionally, publication in a learned journal is considered the most appropriate and acceptable means of informing the peer community. Publication of scientific papers is thus an intrinsic and inevitable facet of doing science, and therefore scientists and physicians should communicate their research for publication in learned journals as quickly as possible.
Why Write Well?
Publication in a scientific journal envisages writing or putting on to paper the new information/data generated in a formal and set format. This style of presentation of research papers is to help the reader understand quickly the underlying messages. In addition, a well-prepared manuscript gives an indication of clear thinking on the conceptualisation and execution of the research topic that surely raises the image of the author, besides of course, helping the reader. A well-prepared paper goes through the various stages prior to publication much faster than a poorly written one. The referees, who scrutinise the paper for its scientific validity, are also likely to review the manuscript quickly, thus, helping the editor expedite the publication process. A badly written paper, on the other hand, often puts off the referee from immediate scrutiny, as the key message(s) of the paper are not clear; moreover, no one likes to read a poorly drafted paper. Because of its easy comprehensibility, a well-prepared paper also has a greater chance of getting rapidly assimilated into the common pool of knowledge, considerably enhancing chances of it being used and cited. At the other extreme, a poorly prepared paper may even get returned from the editor’s desk without undergoing peer review, even though the intrinsic data may be good. It is not very uncommon for editors to hear from an irate reviewer returning a very badly written paper requesting resubmission after rewriting for clarity. This is more so for papers submitted to widely circulated prestigious international journals which receive a large number of submissions.
Is the Paper ‘New’ or ‘Original’?
The quality of a scientific paper and its eventual publication often depends upon the planning that goes before the paper is actually written. The following are some of the key elements that should be kept in mind.1 First and foremost is the message itself. Decide on the crucial points being conveyed through the paper. What is the key message(s) that is/are being communicated? Are these messages important enough to warrant the writing of a research paper? If so, what kind of paper would be most appropriate for this new information, for example a full paper, a short communication, a letter to editor or a case report? If there is justification for the writing of a paper, the next issue that needs to be ascertained is the ‘newness’ or ‘originality’ of the information being reported. Are you the first to report these findings? Or, is it just the first report from a province or country, say, North India? If a similar work has been reported earlier, are your data more convincing and credible or sample size more large and appropriate for the research question? Or, at least, is it ‘new’ to the target journal? A thorough literature search at this stage is strongly recommended to check and reconfirm the originality of the content proposed to be reported. Often, no fresh literature search has been done since planning of the study and therefore what is considered ‘new’ by the authors at the time of conception of the study may already have been reported. Also, a chance discovery, as it often happens in case reports, needs to be checked to ensure that the case being reported is indeed unique and uncommon. In addition, even if a literature search has been done earlier during the conduct of the study, with the explosion in new literature it is possible that something new may have been published in the area in the last few months. More importantly, a fresh, comprehensive literature search can save potential embarrassment to the authors should the referees locate and point out that similar findings have been reported elsewhere. Moreover, such a paper would obviously be considered a repetitive study, and would certainly not enhance the reputation of the authors or the journal. Also, it would be difficult to justify non-inclusion of key recent references, even if these were ‘inadvertently’ left out.
One of the common mistakes committed by many authors is sending papers for publication without the mandatory last-minute updating of the literature search. Many papers are turned down by journals merely because the authors are not well informed about their own area. It is clear from our experience that current awareness is among the weakest links in medical writing in general, especially publication from scientists and physicians from the developing countries. This was perhaps understandable in the past as access to current literature was a serious concern for researchers from developing countries due to concerns of cost and non-availability of new information technology. Now, with the current availability of user-friendly, internet-based databases, especially the free-to-access PubMed and plethora of open access journals, conducting a new literature search should not be a problem. More importantly, a fresh search may throw up some new paper(s) that may provide clues to a critical question that has been a persistent loose end in the arguments in your paper.
Choice of Journal
Another crucial decision that the authors need to make is to decide where their research should be submitted for publication. Editors advocate application of what is called the ‘so what’ principle. When confronted with too many papers (In the editors’ world, they always have more papers than they can possibly publish—the rejection rates of good journals average >75%), decisions about acceptance are nearly always taken by editors on the simple question whether the paper under consideration adds value to their journal. What happens if this paper is not published? Will the journal and readers lose any critical information? A positive answer would enhance the confidence of the authors for submitting the paper to a particular journal and possibly the chances of eventual publication.
Another logical question is the intended audience of the paper. Is it meant for a small group of super-specialists (say, paediatric gastroenterologists), a larger group of specialists (gastroenterologists) or all practicing physicians? Often, an incorrect choice of journal will mean avoidable delay, as the editor will return the paper with a polite note that your paper is not of interest to their readership. It would therefore be helpful if some thinking goes into choosing the appropriate journal. One quick way is to scan some recent issues of a short list of journals that have recently published similar articles. Or just check the journals listed in the bibliography of your paper to get an idea of potential list of journals. The chosen journals could be arranged according to ‘quality’ viz., their peer assessment, readership, international outreach, impact factor, etc. In addition, going through the journal’s editorial policy would help understand its scope, the kind of papers entertained and other similar details necessary for the preparation of a manuscript. One should try and find out the best match of the audience with the topic at hand, as the primary objective of publication of a research paper is to ensure right readership for eventual application of the new knowledge generated and not merely appearance in a high-impact journal that the target audience does not read. To my mind, the most important criterion is to try and ensure the information reaches the intended group of readers who could apply in their pursuits—either research or practice.
Authorship
An important issue that merits attention at the time of writing a research paper concerns the people who are going to be authors. It is strongly recommended that names of all potential persons who are going to be authors, if not the sequence of their appearance in the paper, should be decided before the paper is written, or much better at the time of conception of the study. There are international guidelines on authorship for biomedical journals formulated by the International Committee of Medical Journal Editors (ICMJE).2 These Guidelines have been revised several times to address new issues that regularly crop up in publication ethics. Many universities, public-funded research institutes and agencies have authorship policies that can provide overall guidance. But at a very broad level, it is recommended that all the scientific and technical personnel who have significantly contributed to the ‘intellectual content’ of the paper should be included as authors. In other words, they could include people who have had a major role in the study, either in the study design, collection of data, execution of the study, analysis and/or interpretation of data, help in writing parts of the paper, in the revision of the intellectual content of the paper and/or final approval of the content. Many journals now insist on a declaration that is to be submitted with the manuscript that all the authors listed on the masthead have participated in the study, have seen the final version of the paper and agree to be authors in the order listed. In addition, all the authors should agree on who among them would be corresponding with the editor and would be responsible for settling disputes, if any, on authorship. Such an ‘undertaking’ on authorship is mandatory for journals like the Indian Journal of Medical Research. A new concept ‘contributorship’3,4 is emerging as an alternative to authorship primarily to address some unethical practices like ‘guest authorship’ (giving away credit to those who did not participate in the study) and ‘ghost authorship’ (written by someone else who was not part of the original project team). Unlike authorship, contributorship emphasises the issue of accountability along with credit. Now some international journals like Journal of American Medical Association, British Medical Journal, The Lancet, etc., routinely seek information from the authors to spell out their relative contribution in the research work being reported. This is increasingly being adapted by many Indian journals as well.
Getting Started
Once the abovementioned issues have been settled, the actual process of reporting the research and the long journey of communicating the new information generated in the laboratory to the user community through a learned journal as a research article, begins. Before the process of writing a paper is initiated, it is necessary to keep all the relevant documents at hand. These include the data (evidence) collected, analysed and arranged in the form of tables and/or figures/charts, etc., reprints of publications likely to be cited in the paper collected and systematically organised in a folder. Simultaneously, one should check of any statutory copyright and other permissions that have to be obtained (relating to material used for research or any unpublished information being used). If so, the process of obtaining the necessary permissions should be initiated at this stage.
Basically, all scientific papers endeavour to report something new and original, present arguments as to how these data are original and novel and in what way they differ from similar reports published earlier. And if these data significantly differ from those already reported, the authors provide logical reasons for the difference and attempt to convince the editor and referees of the credibility and originality of the research being reported. A scientific paper is thus a piece of scientific writing with a set of ideas and critical arguments arranged in a logical sequence. The authors are expected to do this writing exercise in a set format as prescribed by the journal. First, the problem studied is described either by posing a question or through a hypothesis (see Chapters 3 and 5). Evidence is then presented on the main points and then the subsidiary points, keeping the same sequence throughout the text. The credibility of data (evidence) gathered through the meticulously planned experiments is explained in the light of similar published findings, if any. The implications of the research are discussed in the light of available knowledge, especially if there is any conflicting evidence. Arguments are presented defending and/or justifying the new points made in the paper. Finally, a verdict is made in terms of conclusions, with implications outlined with suggested recommendations, if any, for future work. There could be a take-home message suggesting future lines of work for other researchers, new treatment protocol or a suggested new strategy/recommendation for policy makers.
Structure of a Scientific Paper
Until the early part of this century, scientific papers were written mostly in the first person singular form describing the findings in an informal, loosely structured way. Not many papers were published as the scientific enterprise was small and it was easy for scientists to read nearly all the papers in their area of scientific interest. When the publication of papers witnessed a steep rise since the post-War era, especially from the early 1960s, it became increasingly difficult for researchers to cope with the information overload. Several innovations were attempted to address the issue providing access to new information rapidly through computerised databases and online searches and other new technologies (see Chapter 21 for details). Simultaneously, serious steps were initiated to streamline and standardise the scholarly communication system to help the readers quickly access all the relevant information in a user-friendly way. It was felt that it would be a great advantage if the scientific paper was written in a form that would help the reader assimilate its contents at a glance. In addition, having a standard structure of reporting new research would help library and information science professionals to index and computerise the papers to facilitate quick and easy retrieval of archived information.
In 1965, Sir A. Bradford Hill, a British statistician, proposed a structure that attempted to address these issues.5 He proposed the following structure with clear-cut demarcations on the information to be presented in an original research paper.
Why did you start the work?
Clear statement of the hypothesis stated or implied. Where the question came from would form the Introduction.
What did you do to get these data?
How the hypothesis/research question was tested/verified with the selected subjects and methods would form the Material and Methods section.
What are the answers obtained?
The salient findings and supporting evidence would form the Results section.
What does it mean or the implications of your study?
The main answers to the question, supporting and counter-evidence and assessment of evidence would form the Discussions and Conclusions section.
To sum up, this narrative sequence would form the structure of the scientific paper, as we know now.
The IMRaD format.
Introduction
Materials and Methods
Results and
Discussion
The IMRaD format is now accepted in most scientific journals, especially the biomedical journals all over the world. The ICMJE also recommends it as the standard format for writing and publishing original research papers.
Title
Title is an important component of a research paper as it is a kind of sign post that tells the reader about the content and helps them decide if they want to read any further. Readers looking for new information typically scan contents pages of journals from information alerting sources like the Current Contents, contents pages sent by journals through email, secondary information databases like the PubMed and IndMed, or the new web-based search tools like Google Scholar (for more information on this topic see Chapter 21). When these databases retrieve thousands of articles, it is the title that usually attracts a reader to go any further. Therefore, if the title does not sufficiently describe the main content of the paper, the researcher is unlikely to be interested in going beyond the title. Thus, the very purpose of writing the paper, that is, reaching the intended audience, is defeated.
The title is expected to give a broad idea of the contents of the paper, even when read independently, which is nearly always the case. The title should be descriptive enough to convey the main content of the paper and also concise and precise. For this purpose, the title must carry a sufficient number of keywords that include all major concepts studied. The title could include the type of study (in vivo, in vitro), animal/human, study design (case-control, randomised controlled trials, etc.). A title must also be specific and can be made interesting enough to grab the attention of a busy reader; it should never be clever or misleading.
There are no specific rules on the length of a title of a paper. Most journals suggest that the title should neither be too long (risk losing the main message due to verbiage) nor too short (will surely miss the key points of the paper). About 15–20 words (about 100–150 characters) are ideal. The title should use specific words, familiar words and short words to make the title more effective and meaningful. Inclusion of waste and/or empty words like ‘Studies on…’, ‘Investigations on…’ should be avoided or at least minimised. The use of uncommon abbreviations, special notations and reference numbers should be avoided in the title. Some journals encourage declarative titles that indicate the main outcome of the study. But most journals prefer only descriptive titles as giving out the key message in the title may discourage the reader from going beyond the title. It is, however, advisable to consult the journal’s policy outlined in information to authors before the title is finalised.
It is worthy to remember that errors often occur while rewriting the title to make it concise and precise but not descriptive, etc. A common mistake that occurs in titles relates to faulty syntax (arrangement of words); for example the titles ‘Glutathione and its redox system in obese polymorphonuclear leucocytes’ or ‘Changing profile of aspartate and alanine aminotransferases in normal and pregnant sera’.
To sum up, a title should be short, specific and interesting to attract and retain the interest of the reader. At the same time, the title should include all the crucial points addressed in the paper. In some ways the title could be considered a condensed version of the abstract. So it makes sense to write it last. While the paper is being written, a tentative title could be given which could even be two to three sentences. It is a good practice to write the final title only after the entire paper has been written and edit the longer version to make sure that the final title contains all the key concepts of the paper and removing redundant words. In view of its importance, therefore, time spent on rewriting and polishing the title is well worth it.
Name(s) and Address(s) of the Author(s)
The names of all the authors should be spelt out completely and clearly and in the same way as done in all the earlier publications. Often, authors change the way their name is presented which happens often when Indians move abroad (especially to the United States) when the first name finds prominence and a middle name (abbreviated) suddenly makes an appearance! Presenting the first name and surname in different ways makes the retrieval of a paper difficult for the readers and indexers. It is more frustrating for citation-based retrieval of papers like the Science Citation Index or the Web of Science. Researchers should decide the way they want to spell their name early in their career and then maintain that style throughout their career. If different authors of the paper are located in different Departments in the same Institute or are from different institutions/countries, complete details should be provided with appropriate symbols, usually given as superscripts, on the masthead. If an author moves out of the research site after the paper was communicated, the current address could be provided separately as ‘present address’. It should be ensured that the address given is the place where the work was carried out. If part of the work was outsourced, it should be indicated in the ‘methods’ section. Finally, it is important that a corresponding author should be identified and complete contact details provided and any changes promptly communicated to the editorial office. It is ideal to provide an email for correspondence.
Abstract
After the title, the abstract is the most important component that decides whether a reader will go further to consult the full contents of the paper. In fact, the title and abstract are the only components that are made available by all journals free to the readers on their websites and through the secondary information services like the PubMed. The abstract is a crucial part of the scientific paper as the essence of the paper is summarised for the benefit of a busy reader. Understandably, a well-written abstract with a clear message is likely to encourage the reader to seek more details from the paper. The abstract is a mini version of the paper and should aim to sum up the contents of various sections of the paper—IMRaD. Abstracts are usually expected to be written in about 250–300 words containing information about the aim and purpose of the study, how the study was carried out, key new data/information obtained and main conclusions and implications of the study including key arguments, any new theory, treatment, apparatus, technique, etc., described in the paper. The abstract should contain names of new compounds, species and new numerical data, such as physical or biological constants. If details cannot be given due to space constraint, the abstract should draw attention to them.
With the emergence of the structured abstract, writing abstracts has become very easy. Structured abstracts not only help the author(s) write informative pieces that can be easily understood but also help editors and peer reviewers evaluate the paper. The structured abstract is based on the standard IMRaD format with more details on the design of the study, participants, interventions, etc. Many journals now insist on the ‘structured abstract’ which would take care of some of the mistakes that commonly occur. The structured abstract consists of headings such as Background, Objectives, Design, Setting, Participants, Measurements, Results and Conclusions. Each heading can have one to two sentences containing essential information.
The following should be avoided in the abstract: (i) repetition of the title; (ii) references; (iii) uncommon abbreviations; (iv) structural formulae and figures; (v) trivial results and experimental details and (vi) excessive speculation. The abstract should ideally be written after the paper has been finalised. It should be ensured that the numerical data quoted in the abstract match those given in the text and tables in the main text and that it does not contain any information and conclusions that are not stated in the main body of the paper.
Introduction
Introduction is the first part of the main text of the paper. As mentioned above, the ‘introduction’ essentially justifies why the study being reported was undertaken. The ‘introduction’ also briefly outlines the main objectives of the study. Besides justifying the study, it is essential to put the research work being reported into perspective. This is done by quoting earlier studies that have formed the basis of the research being reported and the gap being bridged. While citing published literature, only the most essential papers should be referred to in strict relevance to the study being reported. Historical survey of the earlier work should be avoided. Quite often citing a good contemporary review article is preferable to citing a long list of original papers, as most of the important work would have been cited in the review.
A long-winded introduction is undesirable for most journals. Editors expect a brief, crisp background about the research question with most relevant references in support of the statements. Has the research problem not been addressed adequately in earlier work? If so, what are the gaps that are being filled? Are the existing gaps worthy of being addressed? Minimising trivial and already known information will keep the word count of the introduction and also help retain the reader’s interest. There is no point in explaining the study in too much detail with information that has already found its way into textbooks or monographs. The referee and the reader are specialists in that area and would have enough knowledge, if not more than the authors, to grasp the content without such historical detail. Citing the most recent references would give a positive impression of the knowledge-base of the authors. It might help if a statement or two are given about the study sample/population included, study design, especially if there have been earlier studies with different methodologies and how you have addressed the question. To sum up, the introduction should be a clear statement of the research question. Some journals expect the introduction to conclude with a brief statement of what has been achieved. It would therefore help consult Guidelines to Contributors before submitting the paper.
Materials and Methods
This is a very important section of the paper as this section describes how the research being reported was carried out. In other words, how the answer for the research question/hypothesis proposed in the ‘introduction’ was sought. Primarily, the ‘materials and methods’ section describes and defends the methodology used to conduct the study including the experimental design employed and the way the data were collected, collated and analysed. The ‘methods’ section is important to establish the credibility of the evidence gathered for study and for substantiating the conclusions.
The major components of the section include (i) subjects/material; (ii) interventions (treatments, etc.); (iii) measurements/parameters (their units/expression); (iv) statistical analyses/procedures and (v) ethics. The ‘methods’ section could vary in its structure depending upon whether the study being reported is an experimental study, an observational study or a clinical trial.
For an experimental study some items that need description could include (i) description of experiment or design; (ii) specific description of experimental methods; (iii) animals—species, strains, age, sex, weight, physiologic or pathologic state, nutritional state, supplier, etc.; (iv) animal diets, maintenance, treatment diets, constituents, source; (v) microorganisms—genus, species, strain, serotype, other characteristics, details of standard/reference strains; (vi) drugs and other interventional agents; generic and trade names; chemical names, if non-standard drugs; suppliers; (vii) drug administration schedules, forms, doses; (viii) procedures for other kinds of interventions—surgical procedures; (ix) tissues, tissue cultures: source, prior treatment; (x) cell lines, DNA; (xi) reagents; (xii) materials not commercially available and (xiii) identification of equipment used—with specifications and the manufacturer’s name and location.
If a standard method was used, just citing the authentic reference will suffice and there is no need for an elaborate description of the method itself. If several alternative common methods were used, the methods may be briefly identified with appropriate authentication. If a standard method has been modified, details of modification should be given. For methods published but not well known, a brief description and a citation are required. For new or substantially modified methods, reasons for using them have to be provided with their critical evaluation, especially any limitations and how the current method is superior to previous methods and relevant to the study being reported. If any new equipment is used in the experiments, the ranges covered must be provided with sufficient details to justify such usage and list its advantages so that other scientists working in the field can repeat/confirm the work.
For materials used in the study, relevant and exact technical specifications must be provided with quality of the chemical reagents obtained with source (manufacturers). Generic and chemical names of drugs used would suffice in most instances, and trade names should be avoided. Equipments used for analysis, especially those used for cell separation and isolation of sub-cellular components, should be spelt out with model number, manufacturer’s name and location, etc.
For clinical trials, the following are some key components that should be listed: (i) drug/intervention, etc., used; (ii) source; (iii) trial design (actual phase, randomisation, blinding and use of placebo); (iv) trial organisation and procedure; (v) study participants; (vi) inclusion/exclusion criteria; (vii) measurements; (viii) methods used for monitoring and (ix) data collection and assessment. The measures of outcomes of the clinical trial should be clearly spelt out, for example symptomatic recovery, decrease in severity of disease and fall in blood pressure. The parameters used in the trial (clinical, biochemical, histochemical, imaging, etc.) must be defined unambiguously. If the parameters of outcome are standard, for example blood glucose, cholesterol, a reference would suffice without any further details. If the endpoint is, for example, healing of a wound, the endpoint of the study in terms of size of lesion (e.g. size < 2 mm) should be defined.
In studies using drugs, information on permission of the drug regulatory agencies (e.g. Drugs Controller-General of India, US Food and Drug Administration), for conducting the appropriate stage of trial (Phase I, II, etc.), should be provided. If the study envisages the transfer of biological samples from the institute to a foreign country or vice versa, it should be mentioned and appropriate details of permissions obtained from the competent national authority should be described. For drug trials in India, it is now mandatory to register trials started on or after July 2008 with the Clinical Trials Registry-India (CTRI; www.ctri.in) located at the National Institute of Medical Statistics, New Delhi, set up by the Indian Council of Medical Research, New Delhi, in 2007. In the United States, it is a public law to register clinical trials and they can be registered and periodically updated until results are posted on Clinicaltrials.gov.
There are national guidelines available for conducting research in animals and in human subjects.† In all studies with animals, it is necessary for the institute conducting and reporting studies to have animal houses, maintenance, feeding facilities, etc., which conforms to the guidelines. There should be a functional Animal Care and Use Committee in each institute. While reporting research with animals, it is mandatory to mention approval of the ethics committee.
For studies involving human participants, the number of subjects, age profile, gender, socio-economic status and disease state (criteria of diagnosis) should be described. There should be a functional Institutional Ethics Committee (IEC) or review board (IRB) in each institute.‡ While reporting research, mentioning of ethical approval of the IEC/IRB is mandatory and should be indicated in the manuscript. It is also necessary to give details on how informed consent (written/verbal) was obtained from participants. To the extent possible, patients’ names, initials or hospital numbers, especially in illustrative material, should be avoided. Any information that would impinge upon the privacy and confidentiality of subjects should be avoided. If it is necessary to use such details or pictures, written permission should be sought from the subjects involved.
Another important component of the ‘methods’ section is data analysis through appropriate statistical methods. Details as to how data were collected, collated and analysed should be clearly described. Information on how the sample size was estimated and, if the study participants were randomised into various groups, how the segregation was done should be described. The statistical tests used, with reasons why a particular test was chosen, should be described with justification. The level of statistical significance should be clearly stated. If computer software is used for data analysis, details of the software package with version should be indicated.
In summary, the ‘materials and methods’ section forms a very critical component of the paper as the conclusions and outcomes are based on how sound and rigorous is the methodology employed. Also, this is the only section in a research paper where readers seek adequate detail, so that they may be able to replicate the experiments and hence there is less constraint for space.
Results
In the ‘results’ section all the key data and findings of the study are described. A common temptation to include every scrap of collected data should be avoided and only the most important and relevant data, which are described and argued in the ‘discussion’ section, should be included in a clear and concise form. In other words, this section should only contain the most relevant, important and representative data. Results is the section where the methods adopted to test the hypothesis—study design, sampling, statistical methods, etc.—are justified to substantiate the claim that the data collected are credible and the conclusions are reliable. See also Chapter 26 for additional information.
There are three crucial elements in the presentation of results: evidence, efficiency and emphasis. Evidence is the data collected to support the outcome of the study and that defends the conclusions. But unless the maze of data collected is presented efficiently, the reader will not understand the numbers and outcome of the study, much less get convinced. This is done through the appropriate use of tables or figures depending upon the evidence being presented and the message being communicated. Finally, emphasis should be on the core data and issues of the data collected and presented in the ‘results’ section. Material that is primarily of a supportive nature to the main research question should either be downplayed or could even be excluded.
The ‘results’ section varies with the type of study being reported, that is, a basic research paper where animals, cell lines, etc., are used or clinical trials involving human participants with designs as cohort, case-control, randomised controlled trials, meta-analysis, etc.
The writing of the ‘results’ section starts with the collating and analysing the data with the construction of tables and figures. At this stage the data that are to be used in the paper (i.e. what to include) should be clear in the author’s mind. For example, is there a need to include all the collected data to answer the research question being addressed? Can some data be deleted or put aside for subsequent use in another paper? Only the most relevant data that are closely interrelated should be presented in the results.
Whereas writing the ‘results’ section is rather easy, a little bit of planning aids the clarity of presentation. One can start with the arrangement of relevant tables and figures in some order. It could be in the same order of description in the ‘methods’ section or depend upon the importance of various parameters studied and the focus of the study. One should meticulously check data under each column and row for any missing information in the master tables which could be used to start writing. It is generally observed that there are two ways of writing the ‘results’ section: (i) describing each bit of data in the text and also referring to the tables and figures or (ii) cryptically dismissing the reader with statements like ‘Data are shown in Tables I to III and Figures 1 to 2’. Neither of these methods is appropriate or acceptable to journal editors. It is necessary to strike a balance between the description in text and the reference to tables. While all major points could be listed in the beginning, only the most salient points that the author would like to draw reader’s attention could be very briefly described. These could include demographic data, description of subject characteristics, uniqueness of the design, etc. If many data are being reported, it makes sense to follow a specific and logical sequence with appropriate subheadings.
While characteristics of the animals or subjects are being described, terms like normal/abnormal should not be used without providing a definition. Also, in clinical trials, the inclusion/exclusion criteria should be clearly spelt out. Typically, positive findings are described first in a chronological order. To the extent possible, actual numbers should be used rather than percentages alone. Descriptive data like the number of animals/subjects who died during the study should be provided. Negative association, unexpected outcome and their statistical significance should be provided in the results. Explanations for unexpected findings have to be provided only in the ‘discussion’ section.
To the extent possible, the text should be arranged into small paragraphs with each parameter logically described. Repetition of information that is clear and apparent in the tables is not necessary in text. Also, inclusion of data and description of parameters that are not originally sought to be addressed in the ‘introduction’ should be avoided.
The use of statistics in terms of appropriate tests used should be mentioned with exact ‘P’ values indicated. Statistical terms like ‘significant’ have to be used carefully only when the data are statistically significant (P < 0.05). If a table does not convey anything more than the absence of a statistically significant difference between two or more groups being compared, it could well be deleted.
Details such as methods employed, data for which methods are not described, discussion/interpretation of data should be strictly avoided in the ‘results’. In addition, references should not be cited in the ‘results’ section.
Data Presentation
There are three main ways of presenting data: (i) in text, (ii) in tables and (iii) in illustrations. A particular set of data should be presented in only one of these forms and duplication should be avoided as far as possible. Text is normally used for simple descriptive data. When precise values are required (for data needed by the readers with a high degree of precision), tabular presentation should be preferred, as also when actual comparisons with other similar studies in the paper are based on the values given. Figures/charts can be used when trends are to be highlighted and the actual data sets are not important. It is always advisable to look at the journal’s policy in respect of the use of tables/figures. Depending on the kind of article, many journals limit the number of tables/figures, in view of the high cost involved in printing. Some journals may even charge the authors for pictures, especially if they are coloured photographs. Therefore, one needs to exercise restraint in the selection of illustrative materials.
Tabular Presentation of Data
Tables should be preferred when data on two variables, one dependent and one independent, are compared or when comparing data on one or more variables that change with time or when data that varies with related but unknown factors are compared. In short, tables should be used when description in text cannot alone fulfill the requirement in view of the large number of data needed to present complex details. Tables should be self-sufficient and be able to independently convey the desired message. Each table should have a clear self-explanatory title. Column headings should be brief and can be abbreviated, if needed. For units of measurement, standard abbreviations should be used and these should be placed below the column headings.
While constructing tables, dependent variables should be placed in the column headings, while the independent variables should be placed in the column of the extreme left hand side (the stub). Tables should not be made complex by including too many details in too many columns. Rather than making a cumbersome table, the material can be divided and presented in two or more tables. It is often possible to simplify tables by removing common data and putting them as running matter below the title as footnotes. For example, for presenting statistical data such as ‘P’ values comparing the level of significance between various groups.
Another important point that should be kept in mind is the ‘table-text’ sequence in the paper. The sequence of tables in the paper should form a logical order throughout the paper. Occasionally, some tables are included in the first draft but are deleted or combined at the final stage while sending for publication or tables deleted or combined while revising a manuscript. While combining the tables, it should be ensured that critical data get included and appropriate corrections in numbering the tables are made in the text.
Graphical Presentation of Data
Illustrations can be of various kinds, such as continuous tone pictures like photomicrographs, electron micrographs, simple graphs, flow charts, family pedigrees computer printouts and Illustrative material should be used only where evidence bearing on the conclusions of a paper cannot be adequately and convincingly described in the text or presented as a table. Illustrations should not duplicate data already provided in text or when simple trend is to be shown. For example, a figure showing a linear relationship can be safely dispensed with by just making a statement to that effect in the text.
Various types of illustrations can be used, such as (i) graphs, (ii) histograms (bar diagrams) and (iii) pie charts. When there is continuity of variation between two parameters (e.g. age versus weight), graphs should be used. For data collected over periodic intervals (e.g. change in blood pressure with different doses of a drug over a period of time), bar diagrams would be ideal. When the purpose of an illustration is to show relative proportions of components of an entity (e.g. percentages of different genetic disorders occurring in a country), a pie chart could be preferred. However, pie charts are rarely used in original research papers.
When preparing illustrations, attention should be paid to the font size of the letters and the size of the illustration itself as, nearly always, figures appear in a smaller size in print (depending on the column size in a single- or double-column width). Guidelines for these measurements are provided in the instructions to contributors of the journal. The font size of letters, numbers, dots, etc., should be such that on reduction they neither become illegible nor remain too big. It is good to have the lettering done in a size that closely matches the text size (10 points in printer’s language) on reduction. Consulting a recent issue of the journal can help plan the size and number of illustrations. Considerable economy in space and cost can be achieved in the following ways: (i) combining several simple graphs or photographs (continuous tone illustrations) into a composite illustration when the parameters are common or related and (ii) making judicious alterations in scales to reduce the size of the illustrations. Editors tend to appreciate such exercise efforts from the authors.
Along with the original drawings and pictures, one or two sets of good quality photocopies should be supplied to the editor for use at the peer-review stage and other prior to final printing operations. Sufficient attention should be given to the figure legends. The legends should be ‘stand-alone’ and should explain all details given in the picture including parameters used in the X- and Y-axes, statistical parameters used (mean, standard deviation (SD) or standard error (SE)), stains used, magnifications in photomicrographs, abbreviations used in the figures, etc. Figure legends should be given on a separate sheet and not on the back of the figure. A brief title of the figure, the figure number and an arrow marking the ‘top’ of the figure should be written on the back of the figure using a soft pencil and minimal pressure.
Discussion
The ‘discussion’ section is the most important and perhaps the most difficult part of the paper to write. In fact, this is the weakest section in most papers submitted for publication and often a major reason why a paper is returned to authors for revision or returned. Many authors are not able to interpret their data properly and extract the true meaning and implications of their research clearly and effectively. Often the ‘discussion’ section is too long and rambling, without clearly providing the answers for which the study was planned and conducted. The primary message is often lost in the verbiage.
What are key components of a good ‘discussion’? The ‘discussion’ essentially presents the principles, relationships and generalisations that come out of the ‘results’ section. Thus, there is no need to recapitulate the descriptions already provided in the ‘results’ section. The primary aim of ‘discussion’ should be to show the relationships among observed facts in the ‘results’ section. The significance of the paper should be discussed with sharp focus on the salient outcomes of the study.
The main components of a good discussion include ‘what does all this (the research being reported) mean?’ Briefly, the results obtained have to be clearly explained with clear implications of the study with, if needed, a clear ‘take home’ message. The ‘discussion’ could start with the principal findings of the study very briefly recapitulated. This could be followed by discussing the data obtained vis-à-vis known information. It is important to organise the ‘discussion’ with the most important points being placed upfront followed by other minor points. The study being reported could be compared with results of other studies reporting similar or dissimilar data.
There are several problems normally encountered in the ‘discussion’ section in addition to verbosity. Often the ‘discussion’ does not follow the logical sequence in terms of addressing the main data collected and presented in the ‘results’ but instead focuses on new issues not warranted by the study. Often, negative and difficult to explain data are glossed over.
Data being reported in the paper should always be compared with landmark studies and if the study being reported differs markedly from earlier ones, a plausible explanation should be provided. Confounding variables should be discussed and while criticising the earlier work, logical and rational arguments should be used. Being overly critical should be avoided. If anything unusual is being reported in the study, such as sample size, definitions of outcome, methodology and statistical tests, it should be highlighted. If there is a selection bias in terms of inclusion/exclusion criteria, drop out, etc., they should also be mentioned as a limitation of the study. One should be very careful about prediction as, often, mere correlation does not indicate causation. Careful evaluation of risk factors should be done in the ‘discussion’ and sweeping conclusions and/or generalisations should be scrupulously avoided. When discussing the interrelationships between key variables, any unusual method(s) used should be spelt out. If any new statistical methodology has been used or sample size is quite different from the earlier studies, it should be clearly stated with adequate explanations. Wherever possible and relevant, implications and speculations should not be extended beyond scientific credibility and acceptability. Alternative explanations should be considered in the discussion to address the possibility of errors in methods employed in the study with suitable implications mentioned in the conclusions. For controversial findings, it is worthwhile to provide extra evidence as support. It is not always possible or necessary to resolve conflicts. For any unresolved conflict, where no adequate or tangible explanation is possible, one can just say ‘we currently have no explanation for these disparate results’.
Discussion is the section to write about the strengths and limitations of the study in terms of design, technique, results and how the study is better than those reported earlier. The implications of findings for future research can be mentioned as also the new questions the study raises. If it is relevant, one or two take home messages can be mentioned. Also read Chapter 27 for additional information on this section.
References
Referencing is important and imperative for reporting all science. Citation of earlier work is essentially done to acknowledge the intellectual debt to earlier scientists on whose work the present research is largely based. References indicate the sources of information, quotations and borrowed ideas. The references are cited in research papers primarily to substantiate statements made in the paper with published work to provide credibility. It is essential to understand that the citation of a reference implies that the original paper has been seen, read, understood and verified. Therefore, while choosing a references for citation (typically papers from learned journals), cross-citation from other papers should be avoided. Referencing should be done only when it is absolutely essential and relevant to the statements made. Therefore, there is no need to provide references for well-known facts. All the published sources cited should be accurate and consistent and indicated in the body of the text. Care should be taken to ensure that all the cited references are checked and verified. Otherwise, editors, referees and readers will infer (often rightly) that sloppy referencing is reflective of sloppy work. A common mistake made by many authors is that they fail to check references during the drafting and revising stages of the manuscript, that is, stages when references are likely to be added, deleted or shifted within the paper. As a result, the numbers cited in the final manuscript (as only numbers are given in the Vancouver style) do not match the paper in the bibliography. There is need to thoroughly recheck the citations with the authors’ names of the actual paper being referred to in the final version.
Only the papers which are readily accessible to a reader, wherever he/she is located, should be cited. Citations should ideally be restricted to primary research papers, review articles, textbooks and such reports/documents that are available through the worldwide web or other sources to enable an interested reader locate the paper. Therefore, conference abstracts/proceedings, obscure journal articles, local textbooks/monographs, Ph.D./M.D. theses, newspaper clippings, popular science journals, articles ‘in press’, personal communications, etc., should be avoided due to their limited global availability. In addition, most of such sources are non-peer reviewed and of questionable quality. Although there is no general restriction on the number of references that can be cited, some journals do suggest a limit. However, it makes sense to restrict citations to the bare minimum, that is, citing only the relevant and recent references.
There are three major styles of citing references: ‘name-and-year’ system, also called the Harvard system; alphabet-date-number-year system, also called CBE (CSE) style; and the citation order system, the Vancouver/ICMJE system. But for biomedical journals the ICMJE system of numbering references, as they are cited, numbered and listed in the same order in which they are cited in the text, is the most popular style and is accepted by most journals. Some journals may use other systems like the ‘name-and-year’ or the ‘alphabet-number’ systems which could be ascertained from the ‘instructions to authors’ section. The number of authors to be cited in the references should also be checked as the Vancouver system demands names of the first six authors before an ‘et al.’, if the number of authors are seven or more. For details, see Chapter 29.
Acknowledgements
This is the last portion of the paper and has its own importance. In the current scenario, it is impossible to do research without collaboration and help from others. Assistance received could include material help in terms of chemicals, reagents, use of laboratory facilities, equipment, help in running experiments, referring patients, providing clinical material, etc. All such support should be clearly acknowledged in the paper. Often senior scientists may be requested to go through the paper for comments. Such intellectual help should also be acknowledged, of course with their consent. Financial support from all sources, Government, philanthropic and from the industry, must be mentioned. Many journals now insist on disclosure of any direct or indirect relationship between the authors and the industry. These should be clearly spelt out as ‘conflicts of interest’ is emerging as a very serious ethical issue in medical research and publication. It is not necessary to acknowledge trivial help such as secretarial support or for providing access to routine facilities by the Head/Director of the Institute, etc.
Non-IMRAD Structure
There are many prestigious journals, like Nature and Science, which have a manuscript format that does not conform to the conventional IMRaD structure. Yet it makes sense to write the manuscript in the IMRaD structure and rearrange the various sections of the paper in the way that the target journal demands.
Language and Style
English is the lingua franca of science and, unlike native English speakers, many scientists from countries where English is not the primary spoken language find it difficult to write research papers in English, even those from India where most schooling is in English. But with some effort, it should be possible to write acceptable quality research papers primarily because there is a structure and format in scientific communications and it is not very difficult to write the paper under various sections of the IMRaD structure. The only caution to be exercised is clarity in communication. As with any other form of writing, scientific writing needs to be understood by the intended reader. Experts in language strongly recommend that the language of scientific paper should be simple, devoid of metaphors and idioms. Scientific papers are primarily read for their content and message and not for language. Simple, easy-to-understand English is the best, as science reported in many research papers is difficult to understand. Writing is tough, which is why one should write, rewrite and rewrite, until the message being communicated is clear and understandable to the reader.
