Transcript How to write a scientific paper - the Cardiovascular Journal of Africa

How to write a scientific paper
Grace Townshend
CardioVascular Journal of Africa:
Workshop at Pan African Society of Cardiology
(PASCAR) Conference
26 May 2011
Programme



Why publish?
Journal considerations
IMRAD structure




Parts of the paper
Statistics – brief points
When you think you’ve finished…
Appendix: Figures and tables
Why publish?



Share your results
Establish credibility of
the work
Further your career
Who is your reader?




Other scientists
Funding agencies
Government officials
Other?
Value of unpublished work

Who benefits?



Possibly, person who did the work (?)
Nobody
Who is disadvantaged?



Other researchers
Funders of the work
Patients
Programme



Why publish?
Journal considerations
IMRAD structure




Parts of the paper
Statistics – brief points
When you think you’ve finished…
Appendix: Figures and tables
Why do journals exist?




To publish research that is important
to their audience
To educate readers
To interpret and synthesise research
findings
To encourage debate
What do journals want?
A paper that:
 Adds to knowledge
 Shows a well-designed study
 Is well written
 Is clinically significant
 Provides an enlightening discussion
 Conforms with journal requirements and
international codes
The life of a scientific
manuscript
Data
Writing
Re-writing
My
doc
Co-authors,
Rewriting
Submission
Editor’s
decision
Reviewers’
reports
Reviewer 1
Peer review
Reviewer 2
Revisions
needed
Revise and
resubmit
Reject
Accept
Reject
Production
Publication
Peer review



‘Quality controllers’ of the scientific
literature
Unpaid: done in the spirit of ‘giving
back’
Can greatly improve articles
(by addressing issues in the research
itself or the reporting of the research)
CONSORT guidelines for
clinical trial reports


Outlines a
standard way for
authors to report
trial findings
Freely accessible
online
Schulz et al. CONSORT 2010 Statement: Updated guidelines for Reporting Parallel Group Randomised
Trials. PLOS Medicine 7(3):1-7. http://www.consort-statement.org/?o=1001
Ethics: research


Get ethical approval before you start
working with humans or animals
Obtain written informed consent for
any interventions
Ethics: publishing





Plagiarism is unacceptable
Cite references
Do not copy work directly unless it is shown
as a quote
Permission is needed to re-use tables and
figures
Can phrases ever be re-used?
Choosing a journal







Published topics
Medline indexed
Peer reviewed
Impact factor
Geographic coverage
Rejection rate
Length of time from submission to
publication
Why choose the journal before
the manuscript is finished?

Style for journal when writing




instructions for authors
previously published manuscripts
Content specific to journal audience
Enquire about suitability prior to
submission?
Before you start




Check journal requirements!
- Instructions to authors
Have a copy of a sample article
Assemble all your data
Assemble references
Programme



Why publish?
Journal considerations
IMRAD structure




Parts of the paper
Statistics – brief points
When you think you’ve finished…
Appendix: Figures and tables
Structuring any kind of writing






‘One big idea’
A few key sections
Use helpful subheadings
Provide supporting facts
Group logically within sections
Don’t assume everything will be read
The IMRAD structure







Title
Abstract
Introduction
Methods
Results
and
Discussion
References
 Acknowledgements
 Appendices
IMRAD




Introduction
What did I do, and why?
Methods (including Materials/Patients)
How did I do it?
Results
What did I find?
Discussion
What does it mean?
IMRAD - benefits



Based on the principles of critical
argument
Reader knows at the beginning what
the paper is about, and where to find
items
N.B. This is the order in the published
paper, but not necessarily the order in
which to write!
The IMRAD structure: possible
writing order







Outline
Methods
Results
Introduction
Discussion
Title
Abstract
 Acknowledgements
 Appendices
Work out what suits
you best!
Outline
Write an outline based on IMRAD:
 Introduction – why did you do the study?
 Results – which results answer the
question?
 Methods – which ones need to be included?
 Discussion – key points
Select which references you need and order
any missing ones
Function of the title





Contains the essence of your ‘story’
Helps you focus
Should be a true description of the
contents!
Used for scanning
Used by indexing and abstracting services
Types of title

Passive: gives topic
Concurrent resistance and aerobic training as
protection against heart disease

Active: gives main conclusion
The vascular endothelium masks the persistent
inhibition of rat thoracic arterial tone induced by
S-nitrosoglutathione

Less common
Tips for writing the Title




Look at journal for examples of usual
style
Be as specific as possible
Don’t waste words
Avoid non-standard abbreviations
Titles: exercise
In groups of two, look at these titles of articles which
all report research studies.
 Is each one specific enough?
 Would it ‘hook’ you into reading the study?
Introduction: function



What’s the problem?
Why should anyone care?
How do we propose to solve it?
Introduction: organisation (1)
Start with the known
Move to the unknown
Tips for writing the
Introduction

Give brief background
Provide the context that shows why your
question is important
State aim of your study
Keep it short!

Consider the journal’s readers:



 What do they need to know?
 What do they know already?
 What terminology is appropriate?
Introduction: references



The references cited here help the
reader understand the question
Choose the most important and bestconducted studies
Minimum number necessary
Including everything doesn’t
prove that you know everything; it
proves that you lack discrimination
Introduction: Exercise
The exercise consists of two introductions,
one short, one long. Read the two
introductions quickly (don’t try to read in
depth) and then discuss (in groups of two):
 Do they present what is known?
 Do they clearly state what they set out to
study?
 Is the length right?
Methods and patients



A ‘recipe’ for the study
In basic research papers: repeatability
In clinical papers: comparability
Q: Why do methods matter?
A: Validity of your results will be judged
on the methods
Structuring methods and
results: a mirror image
Standard sequence of methods:
Study design
Patient description, interventions
Definition of endpoints
Primary endpoint
Secondary efficacy endpoints
Safety endpoints
Statistics
Standard sequence of results:
Study flow
Patient baseline characteristics
Results of each endpoint
Primary endpoint
Secondary efficacy endpoints
Safety endpoints
Methods: items on CONSORT
checklist








Trial design
Participants
Intervention
Outcomes
Sample size
Randomisation
Blinding
Statistical methods
Tips for writing the Methods




Avoid fussy detail
Use past tense
Use subheadings
Give references for established
techniques; describe only novel
techniques in detail
Results: function
Beware! Do not mix Results and Discussion
The Results section DOES
 state the results
 present the supporting data
(including tables and figures)
It does NOT
 compare the work with that of others
 relate the findings to other work
Tips for writing the Results




Logical order, following methods
Use subheadings
Report only relevant results
For every result reported, there must
be a method, and vice versa
Structuring methods and
results: a mirror image
Standard sequence of methods:
Study design
Patient description, interventions
Definition of endpoints
Primary endpoint
Secondary efficacy endpoints
Safety endpoints
Statistics
Standard sequence of results:
Study flow
Patient baseline characteristics
Results of each endpoint
Primary endpoint
Secondary efficacy endpoints
Safety endpoints
Results: items on CONSORT
checklist







Participant flow
Recruitment
Baseline data
Numbers analysed
Outcomes and estimation of precision
Ancillary analyses
Harms (i.e. adverse events)
Use of figures and tables
Evidence: To summarise data
that supports your findings
Efficiency: To give precise data that
cannot easily be presented in text
Emphasis: To highlight key points
A picture can be worth 500 words!
Programme



Why publish?
Journal considerations
IMRAD structure




Parts of the paper
Statistics – brief points
When you think you’ve finished…
Appendix: Figures and tables
Statistical methods


Get expert statistical advice from the
beginning
Report:



Alpha level (value defining statistical
significance)
Statistical test(s) used and type of data
Statistical software program used
Statistics: exercise
1.
In a drug trial, 4.1% of patients in the control group had a heart
attack, compared with 2.7% of those in the drug group.
What was the absolute reduction in risk?
What was the relative reduction in risk?
2.
A trial of a diabetes drug in 6,000 overweight patients showed
the following:
Mean weight loss with Drug A = 1.2 kg
Mean weight loss with Drug B = 0.9 kg (p <0.005).
What can we say about the statistical significance and clinical
significance of these results?
Relative/absolute benefits




In a drug trial, 4.1% of patients in the control
group had a heart attack, compared with 2.7% of
those in the drug group.
What was the absolute reduction in risk?
Absolute risk reduction is 4.1%2.7% = 1.4%
What was the relative reduction in risk?
Relative risk reduction is (4.12.7)/4.1 = 34%
Report absolute risk reductions
Percentages

Report the number of patients/cases as well
as percentages

33% of patients has very different value in a
sample of 3 or 50!
Pitfalls of probability





Do not use the word ‘significant’ unless you
mean statistically significant
Statistical significance ≠ clinical significance
p-values <0.05 are significant
No significant difference ≠ equal
p-values do not show magnitude of
advantage or difference
Measures of precision




Results of most biomedical studies are
estimates
Estimates require a measure of precision
Standard deviation (SD) – measures how
spread out the data are
95% confidence interval (CI) –
the true result will fall within this interval
95% of the time
Measures of precision
ECG measurement
Group A
Group B
p value
Mean IVSDd (cm)
1.11 ± 0.22
1.10 ± 0.23
0.90
Mean PWDd (cm)
1.15 ± 0.32
1.08 ± 0.24
0.49
Ojji 2009
Rate of VTE
Group A
(n=240)
(%)
Group B
(n=240)
(%)
Absolute difference
AB
(%) (95% CI)
p value
0-84 days
2.1
4.6
2.5 (5.7 to 0.7)
0.20
85-365 days
10.4
8.3
2.1 (3.1 to 7.3)
0.43
Hull 2009
Programme



Why publish?
Journal considerations
IMRAD structure





Parts of the paper
Statistics – brief points
Parts of the paper, continued
When you think you’ve finished…
Appendix: Figures and tables
Discussion: contents





Answer the main question and give the
supporting evidence briefly
Do your results fit with existing
information?
If not, provide possible reasons
Explain strengths and limitations of the
study
Explain the implications
Tips for writing the Discussion





Restate the main question with the answer
Provide supporting evidence by restating
results briefly
Focus on what the results mean,
don’t repeat detailed values
Do not introduce new results
Use subheads in a long discussion
(remove from final draft)
Discussion: possible limitations
to discuss
Study design
Methods/
Data collection
Patient groups
studied
Unusual
conditions
Analysis
Conclusion



Often merged with Discussion
Possibly restate main finding
State implications/next step
Discussion: key points
 Summarises key
findings
 Projects clinical
relevance of study
 May speculate on
reasons for the findings
 May provide opinion
 Conclusion: key ‘take
home’ message
Abstract: function




Stands alone
Often the only part that’s read
Summarises the paper
(don’t just ‘cut and paste’)
Attracts the reader
Abstract: organisation

Write in IMRAD order
(some journals require a structured
abstract)
Context/question
Minimal methods
Most important results
Main conclusion

Key words




Tips for writing the abstract





Write it last
Keep to required length
Focus on results
Includes a ‘take home message’
Abstracts do not contain:
 References
 Figures or tables
 Information that is not in the paper
Abstracts: exercise
With a partner, please read and discuss
the provided abstracts, focusing on:
 What information has been included
 ‘Take home message’
 Which abstract is the most
informative?
References



Follow Instructions to Authors!
Cite only relevant references
Reference software, e.g. EndNote,
RefMan can be useful
References: which to cite




First
Most important
Most recent
Most readily available:


Journal articles, websites
Book chapters
Articles submitted – can only be cited once
they are accepted
Authorship and
acknowledgements
Authorship



Decide on authorship at the beginning
Consult journal instructions
Follow international recommendations1
Acknowledgements


Acknowledge funding sources
Disclose conflicts of interest
1. International Committee of Medical Journal Editors. http://www.icmje.org/ethical_1author.html
The IMRAD structure: writing
order







Outline
Methods
Results
Introduction
Discussion
Title
Abstract
 Acknowledgements
 Appendices
Work out what suits you
best!
Programme



Why publish?
Journal considerations
IMRAD structure




Parts of the paper
Statistics/probability
When you think you’ve finished…
Appendix: Figures and tables
When you think you’ve
finished…
Revising the draft



Sleep on it
Be prepared to make several drafts...
Get someone to read it
Check logic and story-flow
first ...
Does the paper say all that it should?
Are there any unnecessary items?
Does each part follow logically?
..then check for accuracy...




Data check for accuracy
Check references
Check units and terminology for
consistency
Figures and tables:



are these all mentioned in the text?
do they appear in numerical order?
do all the percentages and numbers add
up?
Top tips for writing any
document…
1. Understand the purpose of the document
2. Know what message(s) you have to
communicate
3. Write a logical outline
4. Make your text active
5. Use concise, uncomplicated language
6. Avoid repetition and padding
7. Keep to length
Writing is hard work!
“When something can be read without
effort, great effort has gone into its
writing.”
Enrique Jardiel Poncela
But ...
It gets easier with practice!
Celebrate when your work is
published!
Who benefits?

You




Career; future research (clarified
thinking)
Other researchers
Funders of the work
Patients
Acknowledgements
In this talk I have used ideas from:
 Dr Jane Fraser, UK
 Dr Joan Marsh, UK
 Watermeadow Medical, UK
[email protected]
Useful resources
Guidelines for Authors and Translators of Scientific
Articles to be Published in English
http://www.ease.org.uk/pdfguidelines/Author
GuidelinesHighRes.pdf
AuthorAID is a global research community that
provides networking, mentoring, resources and
training for researchers in developing countries.
You can search their resource library at:
http://www.authoraid.info/resource-library
Going forward

What do you need in order to help
you publish your research?
APPENDIX:
FIGURES AND TABLES
Figures and tables
Group A
(n=100)
Group B
(n=102)
Group C
(n=99)
Age (years)
63.2
64.1
63.9
Female (%)
40
38
37
Systolic BP
(mmHg)
180
175
178
Study
No. of
infusions
No. of
patients
Local pain or
redness
Systemic
reactions (%)
Gardulf 1991
3232
25
20% of patients
0.93
Gardulf 1995
33,168
158
19% of patients
0.3
Abrahamsen
1996
1100
8
2.1% of
infusions
0
Chapel 2000
1222
30
10.4% of
infusions
3.3
Berger 2003
1500
11
Not reported
0.15
When to use tables
DO NOT use tables
 when data can be summarised briefly in the text
 when relationships of the data or trends can be
shown more clearly in a graph
DO use tables
 when readers want exact values of data and this
would need several sentences of text
Tables: layout (1)




Give each type of information in its own
column
Sometimes it’s obvious which way round
the rows and columns have to be (with
rows, there is space for more items)
It’s generally easier to read the same item
across a row
Study groups will usually be shown as
column headings
Table: example
Age
Female
(years) (%)
SBP
(mmHg)
Group A
(n=100)
63.2
40
180
37
Group B
(n=102)
64.1
38
175
178
Group C
(n=99)
63.9
37
178
Group A
(n=100)
Group B Group C
(n=102) (n=99)
Age
(years)
63.2
64.1
63.9
Female
(%)
40
38
175
SBP
180
(mmHg)
Tables: layout





Follow journal instructions
Blank cells: leave blank or use -, …,
ND, as required by journal
Footnotes: use symbols as required by
journal (e.g. 1, 2, 3; a, b, c)
Footnotes follow sequence: across
columns first, then down rows
Statistical significance: indicate with *
Table: example
Group A Group B Group C
(n=100) (n=102) (n=99)
Age (years)
63.2
64.1
63.9a
Female (%)
40
38
37
Systolic BP
(mmHg)
180b
175
178
Based on 92 patients
b Based on 95 patients
a
Tables: checking contents



Check data against text!
Check calculations (e.g. percentages)
Check percentages add up to 100
When to use figures
To portray reality
e.g. electron micrographs
To convey patterns or complex data
Trends, relationships
between data sets
To visualise concepts or structures
e.g. flow charts, organigrams
Types of graph: curve
Percent of ideal body weight
Shows trend of a value that varies continuously
with some other value
110
108
106
Conventional group
Intensive group
104
102
100
0
3
6
Months
9
12
DCCT. Diabetes Care 1988;
11:567–73.
Types of graph: bar chart

Useful for showing sets of values under different conditions
Usually discrete (‘single points’) data relative to other data
Adrenaline
Glucagon
350
120
300
100
250
80
200
60
150
40
100
50
20
0
0
Non-diabetic
1 month
1–5 years
14–31 years
Peak hormone responses to insulin induced hypoglycaemia (2.5 mM)
Glucagon (pg/ml)
Adrenaline (pg/ml)

Types of graph: pie chart
Useful for showing the relative shares of a whole,
e.g. proportions of a population or of cost
6.7
13.3
NFA
Regular needle
Nurses’ preference for safety
needles (NFA) versus regular
needles (%)
No preference
80.0
Figures: format (1)



Keep it simple
Present similar results in a similar way
Use symbols consistently
Figures: format (2)





Label the axes
Remember to state the units
Use an appropriate scale and suitable
intervals
Mark any breaks in the axis
Mean values: give vertical bars for SD
or SEM, state which
Figures: photographs


Crop if necessary
Include scale bar
Figures and tables


Don’t show the same data in figures
and tables
Don’t duplicate the text
(OK to quote selected data to make a
point)
Figure/table titles



What does it show?
Figure/table and title must be selfcontained
Figure/table number
- use correct format
- check numbers against text at end
Figures/tables:
additional information


Definition of symbols/abbreviations
(in title, on figure, in footnote)
Reproduced with permission from …
(necessary if using existing
figure/table)