How to Write an Article Reporting Clinical Research Tom Lang, MA Tom Lang Communications 1
Science cannot exist without writing! Writing allows science to be distinct from authority, intuition, and tradition as a way of establishing truth
2
Science, to be science, must be public, objective, predictive, reproducible, systematic, and cumulative The printed article is the only medium that allows these characteristics to exist
3
Publication is the final stage of research If the results are not published, from a scientific standpoint, the research never took place Publication advances both science and scientists 4
Milestones in the Development of the Scientific Article 1658 First scientific journals, Journal des scavans and the Proceedings of the Royal Society 1779 First dedicated medical journal is published in England; Foreign Medical Review
5
1797 First American medical journal, The Medical Repository, is published in New York 1812 The New England Journal of Medicine and Surgery and the Collateral Branches of Science begins publication in Boston; oldest continuously published medical journal in the world 6
1858 Pasteur introduces the methods sections, creating the IMRAD format (Introduction, Materials and Methods, Results, and Discussion) 1957 JAMA begins to include abstracts with articles reporting original research
7
1972 American National Standards Institute standard Z39 establishes IMRD as a norm for reporting scientific information 1978 Uniform Requirements for Manuscripts Submitted to Biomedical Journals is released by the Vancouver Group: the International Committee of Medical Journal Editors (ICMJE) 8
1987 Structured abstracts for reporting clinical trials are introduced by Annals of Internal Medicine
1997 CONSORT Statement for reporting randomized trials adopted by JAMA.
9
What is the most important part of a scientific article?
Title
Results
Abstract
Discussion
Introduction
References
Methods 10
What is the most important part of a scientific article?
The Title • The part most often read • Often the only part read • The key link between the research and the reader 11
What is the second most important part of a scientific article?
Title
Results
Abstract
Discussion
Introduction
References
Methods 12
What is the second most important part of a scientific article?
The Abstract • After the title, the part most often read • After the title, often the only part read
13
Which part usually has the most errors?
Title
Results
Abstract
Discussion
Introduction
References
Methods 14
Which part usually has the most errors?
The References • Tedious to do well • Minor errors delay identification • Major errors preclude identification • Error rates of 50% are common • Quotation error rates also common 15
In theory, and aside from the title, which part could be the shortest?
Title
Results
Abstract
Discussion
Introduction
References
Methods 16
In theory, and aside from the title, which part could be the shortest?
The Results • ”The results are in the figures.” • Prefer figures and tables to text when reporting results
17
Which part is usually the easiest to write?
Title
Results
Abstract
Discussion
Introduction
References
Methods 18
Which part is usually the easiest to write?
The Methods • Most authors remember what they did
19
Which part is usually the weakest?
Title
Results
Abstract
Discussion
Introduction
References
Methods 20
Which part is usually the weakest?
The Discussion • Have to interpret the results • Have to put the results in context of the literature • If the research was weak to begin with, the discussion will be weak 21
The typical scientific manuscript contains how many: • Double-spaced pages of text? • References? • Tables? • Figures? • Total double-spaced pages of the submitted manuscript? 22
The typical scientific manuscript contains how many: median (IQR) • Text pages 12 (10-15) • References 23 (15-32) • Tables 3 (1-4) • Figures 4 (2-6.5) • Total pages 20 (16-27) 23
How many scientific journals are published in any given year?
24
How many health science journals are published in any given year?
No one knows • What constitutes a journal? • What constitutes health science?
25
Peer review is little better than chance in determining which articles will be selected for publication. True or False?
26
Peer review is little better than chance in determining which articles will be published.
True • Larger journals do better than smaller ones
27
It is impossible to write a scientific article so badly that it cannot be published. True or False?
28
It is impossible to write a scientific article so badly that it cannot be published.
True • Peer review determines where an article will be published, not if.
29
“There seems to be no study too fragmented, no hypothesis too trivial, no literature citation too biased or too egotistical, no design too warped, no methodology too bungled, no presentation of results too inaccurate, no argument too circular, no conclusions too trifling or too unjustified, and no grammar and syntax too offensive for a paper to end up in print.”
Drummond Rennie, MD, 1986 30
When a well-done trial or experiment or observational study is fairly, honestly, and thoroughly reported, it will have so many warts, footnotes, and exceptions that it may be hard for the uninitiated to believe that the work was of high quality. Frederick Mosteller, PhD
31
Publication is the final stage of research
32
What Journal Editors Want
Manuscripts prepared according to the Instructions for Authors! 33
Website of the Mulford Library of the Medical College of Ohio, Toledo; links to Instructions for Authors Uniform Requirements for Manuscripts Submitted to Biomedical Journals (The Vancouver Style)
34
AMA Manual of Style, American Medical Association Scientific Style and Format, Council of Science Editors
35
Prohibition against prior publication: the Ingelfinger-Relman rule Prohibition against duplicate and redundant publication Prohibition against "salami science"
36
Readership: what the journal sells to advertisers Manuscripts that are accurate and complete Manuscripts that are short and clearly written
37
Research that is:
New True Important Clearly reported 38
Organization The “IMRAD” Format Introduction: Why did you start? Methods: What did you do? Results: What did you find? Discussion: What does it mean? 39
Typical Order of Sections Title Page Abstract Introduction Methods Results Discussion Acknowledgments References Tables Figure Captions Figures 40
General Writing Guidelines • First-person pronouns (I, we) are acceptable throughout the text “The authors decided . . .” “We decided . . .”
41
• Prefer the active voice but use the passive voice when appropriate (Methods section) “We concluded that . . .” “It was concluded that . . .” “The nurse reviewed the histories . . .” “The histories were reviewed . . .”
42
• Use subheads within IMRAD format to make information easier to find Statistical Methods Limitations of the Study Conclusions
43
Title Page Purpose • Identifies the manuscript as it moves through the journal’s publication system • Contains information about the manuscript needed for review 44
Title Authors, degrees, institutions Previous presentations Financial sponsors Competing interests “Running head” (short title) Trial registration number Corresponding author’s name and contact information • Key words • • • • • • • •
45
Titles Purpose • To help readers find and decide whether to read the full article • To help readers NOT read the article if it will NOT be of interest
46
Titles • Most important part of the article - The part most often read - Often the only part read • Must stand alone: no abbreviations • Put important nouns first • Keep short 47
Avoid Declarative Titles (headline, or sentence titles) Reports the results of the study rather than identifying what was studied: Lack of Fiber Cell Induction Stops Normal Growth of Rat Lenses in Organ Culture 48
Avoid Interrogative Titles Titles in the form of questions: Measurement of colonic polyps by radiologists and endoscopists: Who is most accurate? Usually reserved for editorials 49
Avoid “Our Experience” Titles Titles citing the authors’ institution: Outcomes of Heart Transplantation: The Cleveland Clinic Experience
50
Would you read this article? Outcomes of Heart Transplantation: The Tinytown Community Hospital Experience
51
Instead, say what matters
Outcomes of Heart Transplantation: A Review of 250 Cases with 3-Year Followup
52
Informative Titles
Identify the relationship under study Cardiac Effects of Antiretroviral Therapy in HIV-Negative Infants Born to HIV-Positive Mothers
53
Titles for Clinical Research Articles Setting (location) Patients (what was studied) Intervention (treatment) Comparator (control group) Endpoint (outcome of interest) Design (study design) 54
Original A Randomized Trial of Low-Air-Loss Beds for Treatment of Pressure Ulcers [72 characters and spaces] Revised Low-Air-Loss Beds vs. Foam Mattresses for Treating Pressure Ulcers in Nursing Home Patients: A Randomized Trial [111 characters and spaces] 55
Low-Air-Loss Beds vs. Foam Mattresses for Treating Pressure Ulcers: A Randomized Trial [86 characters and spaces] Low-Air-Loss Beds vs. Foam Mattresses for Treating Pressure Ulcers [66 characters and spaces] 56
Subtitles can be Useful A Cost-Benefit Analysis of Low-AirLoss Beds vs. Foam Mattresses for Treating Pressure Ulcers in Nursing Home Patients Low-Air-Loss Beds vs. Foam Mattresses for Treating Pressure Ulcers in Nursing Home Patients: A Cost-Benefit Analysis 57
Subtitles can be Important Low-Air-Loss Beds vs. Foam Mattresses for Treating Pressure Ulcers in Nursing Home Patients: A Cost-Benefit Analysis Low-Air-Loss Beds vs. Foam Mattresses for Treating Pressure Ulcers in Nursing Home Patients: A Systematic Review
58
Authorship Purpose • To award credit where it is due • To assign responsibility for the research
59
Authorship credit should be based only on: 1) Substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data 2) Drafting the article or revising it critically for important intellectual content; 60
3) final approval of the version to be published. . . Conditions 1, 2, and 3 must all be met. Acquisition of funding, the collection of data, or general supervision of the research group, by themselves, do not justify authorship. Uniform Requirements, ICMJE 61
• List authors in order of greatest contribution to the work, from most to least • Do not allow guest (“gift”) authors who do not meet the criteria for authorship: - Senior researchers; directors - Those referring patients - Famous people 62
Contributorship: identifying the contribution of each author to the research and its reporting Guarantorship: naming one or two authors who personally and publicly guarantee the authenticity of the research and its report
63
Abstracts Purpose • To help readers find and decide whether to read the full article • To help readers NOT read the article if it will not be of interest
64
• After the title, the second most important part of the article • Often, the only other part read
65
Types • Descriptive (indicative) • Informative (IMRAD format) • Structured (uses headings) • Meeting (different requirements) • Promissory (research is not yet complete) 66
• Must stand alone: no abbreviations or citations to references, figures, or tables • Limit to one paragraph - usually 150 words for unstructured - usually 250 for structured • Key words are suggestions only 67
Structured Abstracts • • • • • • • • •
Background or Context Purpose or Objective Patients or Participants Setting Intervention or Methods Main Outcome Measure Results Limitations Conclusions 68
Problems with Abstracts • Often do not contain all information needed to determine readability • Often inconsistent with information in the full report • Often not followed by publication of the full report of the research 69
Introduction Purpose • To establish the need for the study • To establish the importance of the study • Write a four-part introduction 70
1. Background statement: provides the context for understanding the problem and approach 2. Problem statement: describes the nature, scope, severity, or importance of the problem that stimulated the research
71
3. Activity statement: indicates the research question, hypothesis, approach, or activities undertaken to investigate the problem 4. Forecasting statement: tells readers what they will find if they continue to read
72
Part 1: Background Statement
“In patients with atherosclerotic vascular disease, aspirin is recommended to prevent myocardial infarction and graft occlusion.”
73
Part 2: Problem Statement
“However, aspirin is also associated with bleeding. Patients are often asked to stop taking aspirin before bronchoscopy, to reduce the risk of bleeding. The effectiveness of this practice has never been tested.”’ 74
Part 3: Activity Statement
“Thus, we sought to determine whether aspirin really does increase the risk of bleeding after bronchoscopy.”
75
Part 4: Forecasting Statement
“In this trial, we compared the number and severity of bleeding events in those taking aspirin with those who were not and determined that aspirin does not increase the risk of bleeding.” 76
Problems with Introductions • Weak background and problem statements • Many authors assume readers will know 1) what problem was studied 2) why the problem is important 77
Methods Purpose • To permit readers to judge the validity of the study • To permit others to replicate the study (nice thought, but . . . )
78
• Assurance that the study was reviewed - Institutional review board (IRB) - Institutional Animal Care and Use Committee • Informed consent from adults and assent from children
79
General Information on Experimental Methods • Tell how the sample was selected; if applicable, how sample size was determined • Identify the materials used as well as the suppliers’ or manufacturers’ location 80
• Avoid leaving gaps in the logic of the methods • Use as many subheadings as reasonable to organize the section and to help readers find information
81
General Information on Measurements • Science is measurement • Environmental or experimental conditions during measurements • Unit of measurement or unit of observation 82
• indicators, biomarkers, or surrogate endpoints • Operational definitions of variables • Level of measurement of variables • Characteristics and qualification of any judges 83
• Precision of the measurements • Calibration and settings of equipment • Validity of the measurements • Reliability of the measurements
84
General Information on Statistical Methods • How the data will be reported • The primary statistical comparisons • Details of any statistical power calculation 85
Details of group assignment For randomized controlled trials - source of random numbers - allocation concealment - blocking and stratification - blinding The alpha level (0.05) The statistical software package 86
Features of Clinical Research Articles
87
• Study Design (e.g., cohort, casecontrol) • Patient Selection (sampling or recruitment; eligibility criteria) • Patient Assignment (how groups were formed, e.g., case definitions, random assignment procedures) 88
• Interventions (drugs, therapeutic procedures, exposures) • Measurements and data collection • Primary and secondary endpoints • Statistical Methods (statistical analyses, sample size calculation) 89
Study Designs • • • •
Retrospective (case-control) study Cross-sectional (survey) study Prospective (cohort) study Randomized controlled clinical trial (RCT)
90
Variables Explanatory variables; interventions Response variables; outcomes - Identify the unit of analysis (e.g., eyes, patients) - Provide measurable definitions (hypertension, depression, Chinese-American) 91
Measurements • Science is measurement! Report the who, what, when, where, and why • Precision, reliability, and validity of measurements
92
Patients • Describe the population studied • Give the eligibility criteria • Give the dates of the data collection period 93
Assignment • Case-control trials: identify case definition and control recruitment • Cohort trials: identify exposures or diagnoses of interest • Randomized trials: report assignment process and nature of control group(s) 94
Randomized Controlled Trials (RCT) • How patients were assigned to groups Who made assignments? • Source of random numbers • Allocation concealment • Blinding of treatment • Data collection or interpretation, analysis • CONSORT Statement 95
Statistical Methods • Identify the comparisons • Say how sample size was determined • Intention-to-treat vs. on-protocol analysis • Specify the alpha level (e.g., 0.05; 0.01) • Identify the statistical software package 96
Treatment
Control
100
No. Assigned
100
40
No. Dropouts
10
60
No. Completed
90
40
No. Cured
40
40/60 66%
On-Protocol Analysis
40/90 44%
40/100 40%
Intent-to-Treat Analysis
40/100 40%
97
Results Purpose • To tell what happened during the study • To present the findings of the study • Explain any deviations from the study as planned • Prefer figures to tables to text 98
• Report data with the appropriate descriptive statistics • The standard error of the mean (SEM) is not a descriptive statistic • Report the standard deviation (SD) only for normally distributed data • Report measurements in Systems Internationale (SI) units [SI] 99
Useful Techniques Poor: Figure 2 shows the decline in blood values. Preferred: Blood values declined (Figure 2).
100
Useful Techniques • Give numerators and denominators of all percentages (“In all, 33% of the rats lived, 33% died, and the last one got away.”) • “Significant” is always statistical significance; use “marked” or “important” instead 101
Results of Clinical Studies • Provide a schematic summary of the study to: - Show the study design - Indicate the flow of subjects through the study - Account for all subjects or observations 102
Patients Approached n = 89 Patients Exclud ed n=5 Patients Assigned n = 84
Low-Air Loss Bed n = 43
Foam Mattre ss n = 41
Complet e hea lers n = 21
Complet e hea lers n=5
• Focus on the primary comparisons: - The actual change or difference between groups (the “estimated treatment effect”) - The 95% confidence interval - Exact P value (until P < 0.001) - The test used in the analysis
104
• Report the results of the primary comparisons first • Give exact P values: P = 0.02 vs P < 0.05 • The smallest P value that need be reported is P < 0.001 • Many journals require 95% CIs 105
Discussion (Comment) Purpose • To explain the nature and importance of the findings • (To “cuss and dis-cuss!”)
106
1. Summarize the study and the main results. 2. Interpret the results and suggest an explanation for them 3. Describe how the results compare with what else is known about the problem; review the literature
107
4. Suggest how the results might be generalized 5. Discuss the implications of the results 6. State the limitations of the study 7. List the conclusions 108
Conclusions We conclude that Tamoxifen reduced the incidence of DMH-induced colon cancer in rats. We also showed that DMH induced the expression of estrogen receptors in colonic mucosa, but that the number of estrogen receptors in the colonic mucosa was not correlated with blood levels of estradiol, polyamine, or ornithine decarboxylase. Finally, we found no relationship between blood levels of estradiol and tumor incidence. 109
In conclusion, we found that: • Tamoxifen reduced the incidence of DMH-induced colon cancer in rats. • DMH stimulated the expression of estrogen receptors in colonic mucosa. • Blood levels of estradiol, polyamine, or ornithine decarboxylase were not correlated with the number of estrogen receptors in the mucosa. • Tumor incidence was not related to blood levels of estradiol. 110
Common Problems • Not answering the research question • Repeating the results rather than discussing their implications • Confusing statistical significance with biological importance • Confusing fact with speculation 111
Acknowledgments Purpose • To thank contributors who are not authors - must have approval of those listed
112
References Purpose • To allow readers to verify authors’ claims and arguments
113
• Do not cite: - personal communications - abstracts - unpublished observations - submitted manuscripts - secondary sources - unread sources(!)
114
Visuals (Tables and Figures) Purpose • To communicate information more effectively and efficiently than is possible in text
115
• One visual per page • With titles and captions, should be self-explanatory • Cite all figures and tables in the text in numerical order
116
Tables • Order the data to show patterns; order rows to aid in finding values • Put data to be compared side-by-side • Double-space the entire table • Use no vertical lines • Highlight important values 117
Figures • Have figures drawn professionally • Redraw figures from other media • Avoid using 3D figures to display two-dimensional data • Images and labels must remain legible after reduction 118
Photographs • Submit camera-ready prints or digital files in the format specified by the journal • Include a scale line; give the magnification, if necessary • Obtain permission from identifiable persons • Add arrows and labels to indicate areas of interest 119
Clinical and Laboratory Images • Identify the subject of the image • Explain how the image was acquired • Explain what the image shows and where it shows it • Interpret the image 120
Submitting the Manuscript
Follow the journal's Instructions for Authors!! 121
• In the cover letter: - Describe the value of the article - Declare conflicts of interest - Disclose previous publication or presentations - If appropriate, request waiver of submittal fee or page charges,
122
• On-line submittal - Read instructions before starting - Record how to access your account - Collect needed data before starting - Be prepared to spend a long time: file formats, size, and names; word counts, author information
123
• Assurances: - Copyright transfer - Conflict of interest statement - Copyright permission - Signatures of authors
124
• Payments: - Submittal or reviewing charge - Page charges - Charges for color photographs - Payments for reprints • Expect a response within 3 months • Track progress on the journal’s website 125
The secret to scientific publishing:
Have something to say.
Say it.
Stop! 126
Tom Lang, MA Tom Lang Communications and Training
[email protected] 530-758-8716 1925 Donner Ave., #3, Davis CA 95618 USA www.TomLangCommunications.Com 127
Books by Tom Lang
How to Report Statistics in Medicine Annotated Guidelines for Authors, Editors, and Reviewers
How to Write, Publish, and Present in the Health Sciences A Guide for Clinicians and Laboratory Researchers Both published by the American College of Physicians 128
