INTRODUCTION
A scientific paper is a piece of original research carried out by a researcher or a group of researchers with an aim to provide new information to fellow researchers, students, and the general public. Scientific papers are critical for the advancement of modern science. The utmost important goal of a scientific paper is to inform the medical fraternity about the results of a study. The information intended to be passed on should be clear, lucid, readable, concise and accurate, as it is most likely to be cited by other researchers when they carry out their own research of similar nature. A good paper should never be self-centered and should highlight both positive and negative findings.
In general, the audience of scientific papers includes two groups of people: The first group is that of the referees/reviewers who help the editor of the journal to decide whether the paper is suitable for publication or not in a particular journal, and the second group is the innumerable readers who intend to benefit from the paper by gaining knowledge about the topic it addresses. A good paper should be able to convince both these groups that the research is authentic, valid, and relevant to the scientific community. The characteristics of a good paper are summarized in Box 1.
41It has now become quite essential for the clinicians, researchers, and students to read articles from scientific journals. Common reasons for reading a paper are: (1) to update oneself with the progress in a particular specialty of study, (2) to find out a solution for a specific problem which could be diagnostic (tests/methods) or therapeutic, (3) to know about causation, clinical features and course of a disorder/disease, (4) to understand certain fundamental aspects like pathophysiology, (5) to get an idea for carrying out a research work, (6) if the article has been assigned to be read (e.g., by an instructor to a postgraduate student), (7) to find support for one's views while preparing a base for his/her research, and (8) to write a good review of literature for his/her dissertation, thesis, etc.
When one looks into PubMed, the widely used database of scientific papers, one can find that, as of 11th July 2017, it has more than 27.3 million records, and about 5,00,000 new records are added each year (Wikipedia.org). Scientific papers can be grouped into two major types: primary literature and secondary literature. While original research articles, surveys, case report/series, conference proceedings, and letter to the editor are usually categorized under primary literature; narrative reviews, systematic reviews, meta-analysis, book reviews, practice/treatment guidelines and commentaries are categorized under secondary literature. All these types of papers have different ways of representation of findings/data. Hence, it is utmost essential to understand how to read a paper efficiently.
Depending on the study design, available evidence can be divided into a hierarchy in which randomized controlled trials (RCTs) are placed at the top, followed by controlled trials without randomization, and other prospective experimental studies. This is followed by prospective cohort studies, case-control studies, and case series. However, one should always remember that this hierarchy is only a guide to evaluate the strength of the evidence, and is not a substitute for the critical appraisal of various types of studies.
In this chapter, we discuss how to read and analyze an original research paper, as it is an essential part of postgraduate training.42
ANATOMY OF A SCIENTIFIC PAPER
The basic anatomy or structure of a scientific paper includes the following parts: title, author names with their affiliations, abstract, introduction, aims and objectives, methodology, statistical analysis, results, discussion, conclusions, limitations, future directions, and references.
Some journals mention that the abstract should comprise 3–5%, introduction 10–15%, aims and methodology 15–20%, results and discussion 45–50%, conclusions 3–5% and references 7–10% of the entire paper. Each part of a paper is described below:
Title of the Paper and Authors
The title of a paper attracts its potential readers in the very first look, like the way a title of a movie/play attracts filmgoers/audience. A good title informs the reader a great deal about the study, and manyatimes helps in deciding whether to go ahead with the paper or dismiss it. Most readers prefer titles that are descriptive and self-explanatory and save the need to have a look at the entire paper to know what it is exactly about. At times, one can find titles which are very descriptive and state their main findings in their titles (e.g. “Risk of oral cancer increases with smoking”). Some titles also provide clear information about the study design and intervention done (e.g., “Double-blind, randomized controlled trial of olanzapine and amisulpride for treatment of negative symptoms of schizophrenia”). Clearer the title, the more readers it attracts.
The order of authors is important to note, along with their affiliations. At times, knowing about the area/region/country in which the study was conducted can provide some additional information about the paper/topic (e.g. the prevalence of a specific disease in a specific community/region). Additionally, one should note the departments/streams of authors involved (statisticians, PhD students, nursing staff, social workers, etc.), as it can help in understanding the methodology of data collection and analysis. It is a common notion that well-known persons in a specific field conduct good quality research and that their research is published in journals with 43high impact factor. However, a good reader should not be carried away by this notion and should try to critically evaluate the paper, as many times many good papers have been found to have trivial mistakes.
Abstract
After the title, it is the abstract which helps the reader to determine whether to read the entire article or not. Most journals provide the abstract free of cost online, allowing the readers to decide whether to purchase the entire article or not. A well-structured abstract should have a background/introduction of the subject, aim/purpose of the study, methods used, results/major findings of the study, and conclusions derived. There is usually a restriction on the number of words, which helps the readers to read the abstract rapidly and decide quickly. A good abstract should be limited to 250–300 words and should be written in a manner to draw the attention of the reader to the study and be able to instigate a desire or interest to read the entire paper. However, many times, an abstract may not be written skilfully and may not exactly convey what the paper is about. So, one should try to note the study design and methodology well.
Introduction
Purpose of the introduction is to provide the rationale for conducting the study. It usually starts with the existing knowledge and previous research available on the topic. A good introduction should provide the proper background for the study and should conclude with the identification of gaps in the literature and how these gaps stimulated the researcher to design the new study. The reader should also determine whether a research hypothesis (study hypothesis) was stated and later check whether it was answered properly under the discussion.
One should approach the introduction part of a paper with good preparation. A good reader should take notes, draw figures and should search for various terminologies from proper sources (for example, can use Wikipedia) to understand the beginning of the paper. It is a good practice to be well-versed 44with the terminologies/definitions used in the topic before reading further, as, many times, the paper does not define all the terminologies used in it due to the word limits imposed by the journals. After reading the introduction, one should be able to say the currently accepted state of knowledge regarding the topic, the data/question that had led directly to the work of the paper, and the exact hypothesis being tested through the paper. Unless one is quite clear about these basic facts, one should not read further, but rather reread the introduction till he/she gets answers to these questions.
Aims and Objectives
The primary focus of the paper is frequently expressed in terms of aims and objectives. Usually, the authors state their aims and objectives towards the end of the introduction; though a few papers can have a separate segment in which the aims and objectives are listed. Many readers often find it difficult to understand the difference between aims and objectives. As per scientific literature, in simple words, aims of a paper suggest what the authors hope to find/achieve from the study; and objectives suggest the actions needed to be taken to achieve the aim. In other words, aims are statements of intent and are usually mentioned in broad terms, while objectives are specific statements that define the outcomes to be measured by the study. As per the requirement of the study, there can be primary objectives and secondary objectives. A good paper should have well-defined objectives which are specific, measurable, realistic, and achievable at the end of the stipulated study period. One should go through the aims and objectives of the study meticulously to interpret whether the paper has attempted to answer a well-defined question/hypothesis or not.
For example, a study can aim to study the risk factors of suicidal ideation in patients with bipolar depression, and the objectives can be (1) to study the suicidal behavior in subjects with bipolar depression, and (2) to compare the clinical profile of subjects of bipolar depression with and without suicidal ideation.45
Methodology/Materials and Methods
This section gives the technical details of how the study/experiment was carried out. It should be detailed enough for another researcher to replicate the work. Studies conducted as a replication of any previous study should have a better methodology in the form of adequate sample size, improved interventions, more stringent assessment procedure, focus on any particular group of population, etc. A reader should be curious about how the new study differs from the previous studies and should try to compare the new study with the previous ones.
All details are rarely included, but there should be enough information to understand how the study was carried out. Information about the number of subjects included, categorization of patients into groups, the sampling methods/technique, and the inclusion and exclusion criteria should be mentioned adequately. While reading the paper, due importance should be given to the following areas:
Study Design
The study design needs to be looked upon; which can be prospective study, retrospective study, double-blind study, single-blind study, randomized controlled trial, etc. The study design should be evaluated to see if it is appropriate or not for the type of intervention used or studied.
Sample, Sample Size and Sampling Technique
The study sample is one of the main determinants of the entire study. Exact details of the study participants about the severity of illness, comorbid illnesses, sociodemographic profile, etc. should be noted, as participants of a study may differ from real-life clinical situations. One should read keenly to detect if there was any selection/recruitment bias. Usually, a good paper mentions its inclusion and exclusion criteria clearly. From the various features of the study sample, one can make out the applicability of the study findings to clinical practice.
The sample size is an important parameter of a study. The power of the study is determined by the sample size, which should be large enough to estimate a particular outcome. 46Usually, the sample size estimation is done based on the available literature data and statistical methods. However, some papers do not mention how the sample size was estimated. In that case, one can try to calculate the sample size by himself/herself or can enquire from the corresponding author of the paper. It should be noted that underpowered studies often lead to Type II error, i.e. the probability of not rejecting the null hypothesis when the given alternative is true. Power of a study is inversely related to the probability of making Type II error and is affected by the sample size.
Sampling technique used, such as purposive/convenient sampling or random selection, should be read carefully, as some consider purposive sampling to be inferior to random selection. If the authors have mentioned about random selection, one should further check the exact procedure of randomization—i.e., use of any randomization table or proper concealment of the randomized information from study subjects, researchers, and raters. One should meticulously check if the blinding method mentioned in the study was appropriately followed or not (For example, in a double-blind study, the study should mention that both the participants and researchers were blind to the intervention provided). In general, double-blind studies are considered superior to single-blind or open-label studies. The sampling technique used also gives an impression of the quality of the methodology used.
Instruments Used
This segment has the mention or brief description of the specific instruments/scales used in the study. The reader should try to get acquainted with the procedures/instruments/scales used for data collection and should critically evaluate whether the procedure/scale used was appropriate or not.
Statistical Analysis
It is one of the important parts of a paper, and one needs to be well-versed in certain basic statistical methods to read this segment. One should at least know which are the best tests to analyze common calculations and comparisons between groups and within groups. One can always refresh one's 47knowledge by going back to statistical methods while reading the paper to know if valid statistics were used. One can even take the help of a statistician or consult someone well-versed in statistics to understand the statistical measures used in the paper. Main points to evaluate in this section are whether the authors have chosen the right test, compared compatible groups; whether they have adjusted for any baseline differences or missing data, provided appropriate references; and whether specific tests were used to estimate correlates, appropriate estimation of the confidence intervals and proper interpretation of p-values were done, etc.
Ethical Aspects
The ethical issues of the studies—like appropriate approval by the ethics committee, etc.—are usually mentioned in this part and needs to be looked upon with interest.
Results
This is a very important segment of a scientific paper. In this part, the authors give the details about the data collected in the form of figures, tables, and graphs. The statistical methods may sometimes be mentioned in this part. One should meticulously go through this segment and check for the reliability and validity of the findings. One should always check if all the subjects present at the beginning of the study were included in the analysis of results. In case of any discrepancy, the reader should check whether if any explanation was given for the same. Results which are statistically significant and those which are not must be identified. One should check if a correct statistical test was employed to get the results and if the level of significance mentioned was appropriate or not. Another important point to remember is that results which are presented as statistically significant may not be clinically significant. Often, if some of the confounding variables are not adequately controlled or if the dropouts are not taken into the analysis intentionally, the results may appear to be statistically significant. Hence, one needs to know the outcome measures studied by the authors and try to examine if there are any actual group differences between the primary 48outcome measures. Additionally, one should look for any secondary outcome measures to understand the interpretations of the study findings. Thus, it is essential to have an in-depth understanding of the statistically significant findings. One should not blindly conclude on the basis of the statistically significant p-value, as many times those differences which appear to be significant may not be relevant when evaluated in day-to-day clinical practice. Some statistical methods like intention to treat analysis, survival analysis, area under the curve, etc. should be understood, as these are very commonly used in intervention studies nowadays.
To scrutinize the results properly, one should read the text thoroughly, examine each figure and table carefully, and take notes of each finding—with all this exercise, one should be able to understand the basic procedure used and the conclusions derived from the results.
Additionally, one should have a good knowledge about different types of bias—like selection bias, exclusion bias, detection bias, systematic bias, etc.—as any type of bias can be introduced in any type of study. A proper understanding of the study design and the steps taken by the authors to minimize bias is essential to interpret the results accurately. A good reader should be able to pinpoint positive and constructive criticisms of the study findings.
Discussion
This is by and large the most important segment of a scientific paper. Here, the authors answer the research questions and present the meaning of each study finding and their interpretation. Usually, the results are compared with those of other studies of similar nature, and a discussion is done on how the findings of the present study are similar or dissimilar to those of the previous studies. Sometimes, there is the use of terms like “trend towards” to explain that the study findings are quite close to the significant difference—but it is advisable to not accept such findings. A good paper has a focused discussion, highlighting the important study findings in a sequence, with the most important findings initially and the least important findings later. In this segment, the authors also mention 49the strengths, limitations, and shortcomings of their paper. A good paper always mentions its limitations in a liberal way. Suggestions about areas for further research are also usually mentioned towards the end of discussion. An important thing to understand is that the discussions are the authors’ interpretations—it is not necessary that the reader must agree with every interpretation of the authors.
To read and analyze the discussion segment clearly, the reader should take notes and try to answer if the conclusions derived by the authors are rational or not. A good reader should be able to separate the facts from the authors’ interpretations/opinions and should be able to relate whether the study hypothesis has been adequately answered or not.
Conclusions
In this segment, the authors highlight the main inferences derived from their study and propose any changes which could have improved their findings. A good conclusion is brief and limited to a few words/sentences. One should tally if the authors have drawn the conclusions from the results described in the results section or not.
References
This forms the last segment of a paper. Here, the studies cited in the text are mentioned in a proper format. Several reference styles are present (e.g., Vancouver style, Harvard style, APA style, etc.), and every journal follows the formatting as per any of these reference styles. A good reader should not stop at the conclusions segment but should take care to check the important references cited and if all the citations were correct or not.
Miscellaneous Segments
There are a few miscellaneous segments in some papers like acknowledgments, conflicts of interest, financial support/sponsorship, etc. In the acknowledgments segment, the authors mention the names of all the people who contributed materials or helped during the study. People who contributed technically but not intellectually are also mentioned in this part. A good paper always mentions its conflicts of interest 50and sponsorship/funding received. The reader should note this segment carefully, as many times, the authors may have been sponsored by the manufacturers of the molecule being studied. One should also very keenly look at the journal in which the article was accepted, as there can be a possibility that the choice of the journal can influence the acceptance of the study findings. Usually, articles published in journals with high impact factor are regarded to be of good quality. However, one should not be influenced by the impact factor of the journal, and should analyze the paper on its methodology, results derived, and inferences drawn, step-by-step.
ESSENTIALS FOR A GOOD READING
After being familiar with the anatomy of a paper, one should understand that one cannot read and understand a paper at a single reading. One must read a paper several times before deriving his/her inferences. Each reading brings out new questions, and one must refer to the paper in more detail to find out the answers. If the answers are not found in the paper being studied, then cross-references should be examined to reach a conclusion. Additionally, one should take the help of his/her teachers, senior colleagues and fellow readers to understand the paper clearly and should not rest until all the queries get answered. One can always write to the corresponding author for clarification on any portions that seem controversial or unclear. Reading a research paper is a good learning exercise and understanding improves with experience and practice. One should not panic at the beginning, as even an impossible-looking paper may be understood with careful and patient reading. A reader gains several skills after reading a paper thoroughly, which includes vocabulary, the way the topic has been presented, the way the data are collected and represented in tables/figures, etc. One should not shy away from asking someone or searching the internet if one cannot understand a specific phrase in the paper. It is always a good habit to prepare an outline or a flowchart of the paper and understand it segment by segment.51
Noting down the questions that arise in one's mind at each segment helps in understanding the topic better. Some questions may be simply technical and can be understood with a thorough reading, while some may be more fundamental and would require a critical analysis of the entire paper. A good reader would be able to summarize the entire paper in his/her own words and should be able to make colleagues understand the paper in a simple language. Finally, one should be able to list out the main take-home messages from the paper. Some tips for a good reading of a paper are provided in Box 2.
SUGGESTED READING
- Alexandrov AV, Hennerici MG. Writing good abstracts. Cerebrovasc Dis Basel Switz. 2007;23:256–9.
- Branson RD. Anatomy of a research paper. Respir Care. 2004;49:1222–8.
- Grover S, Kumar V. How to read a research paper. In: Rao TSS, Tandon, A. (Eds). Psychiatry in India: Training and Training Centres, 2nd edn. Mysuru, Karnataka: Ind J Psychiatr. 2015. pp. 323–32.
- Makela M, Witt K. How to read a paper: critical appraisal of studies for application in healthcare. Singapore Med J. 2005;46:108–15.
- Pwee K. What is this thing called EBM? Singapore Med J. 2004;45:413–7.