How to Read a Paper(如何阅读研究论文)

545T. Athanasiou (eds.), Key Topics in Surgical Research and Methodology, DOI:10.1007/978-3-540-71915-1_40, © Springer-Verlag Berlin Heidelberg 2010 How to Read a Paper Hutan Ashrafi an and Thanos Athanasiou 40 Abstract To adequately equip surgeons with the required skills necessary to successfully read a paper, one needs not only time and energy but also a core level of experience. This chapter aims to classify the components that make up a scientifi c paper with the goal of presenting the audience to some of the analyti- cal concepts that will enable the successful reading of a surgical paper. 40.1 Introduction Scientific papers are the most favoured vehicles through which research is communicated. Each manu- script has been specifi cally designed to allow the reader to understand why a research question was addressed, how this was done and what the implica- tions are for the newly discovered results. As a result, unlike the text of a novel, wherein a story develops in sequential order, the text of a scientifi c manuscript is totally different, wherein it objectively states a prob- lem that needs solving, and states how the authors went about solving it. Thus, the process of reading a scientifi c text varies signifi cantly from normal prose, and requires a consistent application of both analytical and critical faculties. Although surgical papers in print would have under- gone a process of peer-review, the ultimate responsibil- ity of assessing published material lies with the reader. To adequately equip surgeons with the required skills necessary to successfully read a paper, one needs not only time and energy but also a core level of experi- ence. This chapter aims to classify the components that make up a scientifi c paper with the goal of presenting the audience to some of the analytical concepts that will enable the successful reading of a surgical paper. H. Ashrafi an (�) Department of Biosurgery and Surgical Technology, Imperial College London, Queen Elizabeth the Queen Mother (QEQM) Building, Imperial College Healthcare NHS Trust, 10th Floor, St. Mary’s Hospital Campus, Praed Street, London W2 1NY, UK e-mail: h.ashrafi an@imperial.ac.uk Contents 40.1 Introduction ............................................................ 545 40.2 The Conceptual Basis of A Scientifi c Paper ......... 546 40.3 Reasons to Read a Research Journal .................... 546 40.4 The Psychology of Reading a Paper ..................... 546 40.5 Originality ............................................................... 547 40.6 Types of Paper and Quality of Evidence .............. 548 40.7 Core Components ................................................... 549 40.8 Title .......................................................................... 550 40.9 Authorship and Ancillary Information ................ 550 40.10 Abstract/Summary ................................................. 551 40.11 Introduction/Backround ........................................ 551 40.12 Materials and Methods .......................................... 551 40.13 Results, Tables, Figures .......................................... 552 40.14 Discussion ................................................................ 552 40.15 Acknowledgements and Declarations ................... 553 40.16 References and Bibliography ................................ 553 40.17 Supplementary fi les ................................................ 553 40.18 Conference Discussion ........................................... 553 40.19 Editorial .................................................................. 553 40.20 The importance of assessing a paper .................... 553 40.21 Conclusions ............................................................. 554 References ........................................................................... 554 546 H. Ashrafi an and T. Athanasiou 40.2 The Conceptual Basis of A Scientifi c Paper Some authors suggest that not only are research papers a simple method by which to communicate scientifi c discoveries, but they themselves can be considered an innovation that permitted scientifi c progress to occur through the transmission and interaction of information [8]. This concept is not unlike the theory that human thought capacity increased following a developmental improvement in our speech ability (vocalisation) through the evolutionary migration of our larynx further down in the neck when compared with other ape species [15]. Thus, the more we develop and read our scientifi c papers, the more advances we can make in research. It can be considered that “all knowledge is the result of imposing some kind of order upon the reactions of the psychic system as they fl ow into our consciousness” [13]. Therefore, an imposition of order and decrease in chaos on our sensory perception is what leads to mean- ingful information. In keeping with this concept, we read scientifi c literature in order to decrease the uncer- tainty and chaos inherent in our current state of knowl- edge and therefore increase our personal information. Claude E. Shannon in his 1948 paper “A Mathematical Theory of Communication” [19] introduced the con- cept of information entropy (a measure of chaos) sum- marised in the equation below: 2 1 log n s j j j H K p p = = - å where H s provides a mathematical measure of disorder that may exist in a quantity of given information. K is a constant, and pj is the probability of fi nding one par- ticular piece of information from among a subset of data. The role of a research paper is to alter the con- stant K in order to minimise and decrease the uncer- tainty and chaos in our current state of understanding to a state of better “more meaningful” understanding. 40.3 Reasons to Read a Research Journal Although research journals are a means of communi- cating scientifi c papers, and thereby information, we read these in surgery not for academic interest alone, but more importantly for how we can apply these fi nd- ings to treat our patients. These reasons can be classi- fi ed into the following (Fig. 40.1): 40.4 The Psychology of Reading a Paper In his seminal work on studying how scientists read sci- entifi c papers, educationalist Charles Bazerman closely studied and interviewed seven physicists on how they discerned information from the academic papers that they read [4]. His work was based on the premise that scientifi c reading habits are affected by psychological and Fig. 40 .1 Reasons for reading a surgical journal Translational medicine Other Academic vanity As a news source within subspeciality Advertising/applying for employment Learning Comparing Technical Improving patient managenent Identifying ‘Best Care’ Identifying ‘Gold Standard’ procedures New surgical procedures New diagnostic procedures Treatments Diagnostic Modalities Individuals and Units from top units from experienced individuals from the mistakes of others Reading a Surgical Journal Molecular biology Pathophysiology Disease aetiology To Understand Disease Mechanism To improve Clinical Treatment 40 How to Read a Paper 547 sociological variables, and further came up with the concept that all scientists have a dynamic knowledge- based mind-map (or schema) that can be built upon and expanded by information and data from new papers. Here, he analysed each individual’s choice of paper (Fig. 40.2), identifying that these are picked by per- sonal research needs and the necessary self-updating for each scientist’s own particular speciality. “Must reading” was found to be proportional to the amount of research available in the relevant fi eld. Understanding a paper (Fig. 40.3) relies on whether the manuscript’s subject is close to that of the reader’s own speciality. Increased familiarity of a subject to a reader will allow faster information gathering and a more complete understanding of the paper. However, if the paper is poorly written, then it is obvious that an increased effort for reading will be required, and an increased time requirement to assimilate the informa- tion presented. 40.5 Originality Important scientifi c discoveries are frequently a result of their originality, although this is a notoriously diffi cult concept to measure. To help quantify the degree of originality of the data from a new publication, Lynn Dirk, a specialist in science communication, has pro- posed a method to measure and score originality in a scientifi c manuscript [9]. Using this technique, each paper is broken down into three component units of hypothesis–methods–results. Each of the three compo- nents is assigned a value of originality of “P” – Previously reported or “N” – New. This then allows each paper to have an originality score for each of the hypothesis–methods–results subsections that put together can concisely refl ect the originality of the paper subsections. For example, if all three components of hypothesis–methods–results were new, then the paper would be scored as N–N–N, whereas if all three compo- nents were previously known, then they would be scored as P–P–P. Using this typology, eight combinations of original- ity can be assigned to a scientifi c paper. Dirk went on to perform a mail survey on 301 scientists, 68% of whom responded. They rated papers selected from the “Citation Classics” in Current Contents® – Life Sciences over a 5 year period (Table 40.1), demonstrating that this tech- nique can be used to attain useful insights into assessing the originality of scientifi c papers. Fig. 40.3 Mechanisms of understanding a scientifi c paper In-depth reading of whole paper In-depth reading of whole paperUnderstanding a Paper Research close to researcher’s own field If poorly written Research not directly related to researcher’s own Increased effort of reading Numerous re-reading of manuscript Unexpected finding triggers more in-depth reading Speedy selective reading to reveal new facts Reading relies heavily on personal methodological experience Fig. 40.2 Choosing a scientifi c paper directed by scientist’s own research needs Necessary periodic scanning of relevant sources Trigger words noticed during ‘scanning’ of manuscripts Word of mouth Reasons to read manuscripts Mechanism of choice Choice of Paper 25% from single word in title 75% from other triggers in the manuscript Directly found in relevant database (e.g. PubMed) Individuals or Institution who carried out research 548 H. Ashrafi an and T. Athanasiou 40.6 Types of Paper and Quality of Evidence Scientifi c research is not a homogeneous entity and can be broadly categorised into four main types (Fig. 40.4): Many papers combine these for research elements to varying extents, and as a result, a number of research paper types are used to try and communicate this varied data. Types of surgical research include topics that can: Assess or improve upon surgical treatments• Assess or improve upon surgical disease diagnosis • and screening Elucidate underlying surgical disease aetiology and • pathophysiology Assess or improve upon surgical skills and training• Assess or reduce surgical errors• Typical types of surgical paper are catalogued below in Fig. 40.5. In clinical research, scientifi c papers can be assessed by their “quality of evidence”, which can improve with increased subject numbers and randomisation of both patients and treatments (thereby decreasing the likeli- hood of false results). The traditional hierarchy of evi- dence in clinical papers has been (in descending order, with the most important fi rst) [20]: Fig. 40.4 Types of scientifi c research Testing a specific Hypothesis Analytical Methodological Descriptive Comparative Scientific Research Improving research techniques Introducing research techniques Listing the findings of a study Formulating a hypothesis Statistical analysis Defining an ‘effect size’ Breaks down a postulation into its component parts Fig. 40.5 Catalogue of surgical research papers Local National International Qualitative Analysis Surgical Papers Reviews Prospective (Clinical Trial) Retrospective (Survey) Guidelines Experimental Clinical safety Failure Mode and Effect Analysis (FMEA) Other safety assessment tool studies Interview-based studies Decision Observational study Behavioural/Psychological studies Error Economical Clinical Skills Assessment Simulator-based assessment Behavioural/Psychological studies Quantitative Systematic Molecular biology Physiology New surgical technique Novel/New surgical technology (e.g. Robotics) Animal experimentation Non-Comparative Case seriesCross-sectional survey Randomised control trials (RCTs) Cohort studies Case-control studies Comparative (with 2 or more groups) Cross-sectional survey Case series Case report Non-Comparative Comparative (with 2 or more groups) Case-control studies Parallel group comparison Matched comparison Within-participant comparison Single Blinded Double Blinded Crossover Placebo control Factorial design Non-systematic Meta-analysis Cohort studies (most) Originality score: hypothesis + methods + results Frequency (%) N + N + N 15 N + N + P 1 N + P + P 4 N + P + N 43 P + N + P 3 P + N + N 11 P + P + N 11 P + P + P 13 N, new; P, previous (Adapted from Dirk [9]). Table 40.1 Originality scoring and frequency of highly cited papers from Current Contents® – life sciences 40 How to Read a Paper 549 1. Systematic Reviews and Meta-analyses 2. Randomised Controlled Trials (RCTs) with Defi n- itive Results (confi dence intervals that do not over- lap the threshold clinically signifi cant effect) 3. RCTs with Non-Defi nitive Results (confi dence inter- vals that do overlap the threshold clinically signifi - cant effect) 4. Cohort Studies 5. Case-controlled studies 6. Cross-sectional surveys 7. Case reports (only one or two patients) To further reveal the level of quality of a scientifi c paper, the Centre for Evidence-Based Medicine at Oxford has come up wisth a classifi cation for papers that grades them according to the level of evidence (Table 40.2) and subsequent grade of recommendation. 40.7 Core Components Before starting to read a paper, it is important not to miss vital “self-evident” information. This includes in which journal or internet site is the paper published, and what audience is it aimed for? Is for instance in a purely a surgical journal where the contents are intended Table 40.2 Levels of evidence and grades of recommendation modifi ed from the Oxford Centre for Evidence-based Medicine (May 2001) [17] Level Therapy/prevention, aetiology/harm Prognosis Diagnosis 1a Systematic review (with homogeneity) of RCTs Systematic review (with homogeneity) of inception cohort studies Systematic review (with homogene- ity) of level 1 diagnostic studies 1b Individual RCT (with narrow confi dence interval) Individual inception cohort study with > 80% follow-up Validating cohort study with good reference standards 1c All or none studies All or none case-series Absolute SpPins and SnNouts 2a Systematic review (with homogeneity) of cohort studies Systematic review (with homogeneity) of either retrospective cohort studies or untreated control groups in RCTs Systematic review (with homogene- ity) of level greater than two diagnostic studies 2b Individual cohort study (including low-quality RCT; e.g. < 80% follow-up) Retrospective cohort study or follow-up of untreated control patients in an RCT Exploratory cohort study with good reference standards 2c “Outcomes” Research; Ecological studies “Outcomes” Research 3a Systematic Review (with homogeneity) of case-control studies Systematic Review (with homoge- neity) of 3b and better studies 3b Individual Case-Control Study Non-consecutive study; or without consistently applied reference standards 4 Case-series (and poor quality cohort and case-control studies) Case-series (and poor-quality prognostic cohort studies) Case-control study, poor or non-independent reference standard 5 Expert opinion without explicit critical appraisal, or based on physiology, bench research or “fi rst principles” Expert opinion without explicit critical appraisal, or based on physiology, bench research or “fi rst principles” Expert opinion without explicit critical appraisal, or based on physiology, bench research or “fi rst principles” Grades of Recommendation A Consistent level 1 studies B Consistent level 2 or 3 studies or extrapolations from level 1 studies C Level 4 studies or extrapolations from level 2 or 3 studies D Level 5 evidence or troublingly inconsistent or inconclusive studies of any level 550 H. Ashrafi an and T. Athanasiou to be read only by one specialist group, or is it pub- lished in one of the internationally renowned medical journals such as The Lancet or The New England Journal of Medicine, whereby it might deliver a research message that carries a broader scope of interest. Furthermore, the section under which the paper is published alludes to type of research being presented. Common sections include case reports, clinical trials, reviews and meta-analyses. There are, however, some exceptions, and thus, for example, in the journal Nature, original research papers are divided into the categories of Articles or Letters, although the latter should not be con- fused with the totally separate Correspondence section. Scientifi c papers reporting empirical fi ndings are traditionally structured by the IMRD system: Intro- duction, Methods, Results and Discussion [1]. This system is still currently in use, but has been expanded on, to add a variety of extra information for the scien- tifi c reader. To assess international reading strategies of IMRD articles, delegates at the 6th General Assembly and Conference of the European Association of Science Editors (EASE) were surveyed on their reading-order of paper subsections [7]. It was demonstrated that peo- ple rarely followed IMRD when reading as scientists (15%), but were more likely to use it if reading as reviewers (42%), and even more likely when reading as editors (56%). “Hard” scientists (physicists and chem- ists) used IMRD the most, incorporating this sequence in 48% of their reading strategies, biomedical scientists in 33.3% and social scientists the least at 17.8%. Although native-speakerhood can affect reading strat- egy, age does not seem to be a signifi cant factor. A typical surgical paper is broken down into the fol- lowing subsections: Title• Author(s)• List of Departments and Institutions involved in the • project Abstract• Introduction or Background• Materials and Methods• Results• Figures and tables• Conclusion or Discussion• Acknowledgements• References or Bibliography• Declaration of confl icts of interest or sources of • funding Supplementary data/documents/fi les• Conference Discussion (at a meeting where the • paper may have been presented) The majority of surgical journals use the above format in the sequence listed, though some variation does exist. Thus, for example, the Results and Conclusion sections are sometimes combined, or the declaration of the confl icts of interest may appear earlier in the manu- script, or an overall summary may appear at the end of the paper. Furthermore, each journal is characterised by its own unique use of fonts, printing style and refer- ence format. Once the identifying details of a paper have been elicited, one can discern a fair amount regarding the information that can be derived from that paper. For example, is it original research, has it been by invita- tion only, is it in a high-impact journal and what is the reputation of the author or the institution writing the man