Transcript Document

Clinical research original article:
How to write an article and get it published
in European Urology
Shahrokh F. Shariat, MD
Chair and Professor, Medical University of Vienna, Vienna, AUT
Adjunct Professor, Weill Cornell Medical University and
New York Presbyterian Hospital, New York, USA
Adjunct Professor of Urology, UT Southwestern, Dallas, Texas, USA
Remember the
Introduction
purpose of your
Materials & Methods
manuscript
Results
…is to inform
Discussion
Conclusions
Introduction
Materials & Methods
A single piece of work
Results
• Keep it joined together
• One section should flow
into the next
Discussion
• NO SURPRISES
Conclusions
Introduction section
• Clearly and simply explain
– what the research question is
– why it is relevant, why it is original
– very briefly how it will be answered
Introduction section
• Summarize other relevant papers in orderly
fashion (logic) to set the background
– Not be an in-depth literature review
Introduction section
• An excellent study has an obviously important and
original question, and therefore needs only brief
introduction.
– Make it short: 250-500 words,
~3-4 paragraphs,
1 page
Introduction section
• 1st paragraph: brief background in present tense to
establish context, relevance, or nature of the problem,
question, or purpose (what we know)
• 2nd paragraph: importance of the problem and unclear
issues (what we do not know - gap in knowledge - why it is
important to fill that gap)
• 3rd paragraph: rationale, hypothesis, main objective, or
purpose (why the study was done - hypothesis for how you
will fill that gap in knowledge).
Introduction section
1st paragraph:
brief background on
radical cystectomy and
role of lymph node
dissection
Introduction section
2nd paragraph:
importance of the
problem and unclear
issues (minimal number of
LN to remove, anatomical
extent of LND, role of pT
stage as predictor of LN
involvement)
3rd paragraph:
rationale, hypothesis,
main objective
Introduction section
• Usually not easy for inexperienced authors
• Writing intro last can prevent writer’s block and is
easier
Methods section
• M&M section is the core of each paper
• Describes - study design,
- how it was performed, and
- data analysis
• Provides all elements to allow others to reproduce the study
• Easiest part of manuscript
• Writing this section of the paper in the most complete way
before starting the study can help to discover methodological
biases in a moment when they are easily fixed!
What methods section should include
• Period of enrolment/evaluation
• Type of study (retrospective; prospective; controlled;
randomized)
• Inclusion criteria / patient selection
• Exclusion criteria / reasons to exclude patients
• Details on used materials or technique
- report in detail original methods/techniques
- cite (and reference) known methods
•
Ethical issues
http://www.equator-network.org/home/
http://www.equator-network.org/home/
http://www.equator-network.org/home/
Methods section in RCT – CONSORT statement
www.consort-statement.org
Methods section
IRB approval
Quality control
Clinical setting
Inclusion/exclusion criteria
Description of pathology procedure
TNM staging system
What methods section should NOT include
• Study results
- by definition, they go in the “Results” section
• Comments
- on patients characteristics, indications, inclusion and
exclusion criteria  they go in the “Discussion” section
Results
• Describe final population
• Patient details
• Describe intervention
• Treatments
• Observations
• Describe outcomes
• Survival
• Disease related
• Toxicities
• Others
Discussion
• Summarize your main findings
• Compare your data with previous reports
• Describe & explain discrepancies
• Where do your findings sit in our world
• Where are they going to take us
• What next ….
• Limitations and concerns
• Final conclusions
The discussion should place your findings into the scientific literature
Results: General design
• Patient population(s)
• Treatment(s) received
• Outcomes in each group
• Good and Bad findings
• Comparisons between groups
• Analysis of each factor/predictive features
Results – Using figures and tables
Text
Figures
Tables
Results – Using figures and tables
Need all three components
Text
to understand the paper
But … a reader should be
able to follow paper using
just tables and figures
Tables
Figures
So … make the legends
descriptive and include all
necessary findings
They should integrate together
Results
• Be clear about your study ….. what is it?
 Observational
 Retrospective or prospective
 Single or Multi-institutional
 Data quality
 Screening
 Randomized or other
• Whilst Result organization is broadly similar ... each
may require different reporting details
Results
CONSORT Guidelines = A good place to start
- Participant flow
Initial participants
Why excluded?
Final participants
Typically
Table 1. = Details of participants
Figure 1. = Flow chart
- What time period?
Consecutive or selected patients
Filtered cases or “all comers”
- Institutions –How many from each?
http://www.consort-statement.org/
Table 1. Details of patients and treatments
Population totals
Population descriptives
Number
Percentage Mean (St. Dev) Median (St. Error)
What is your population
Gender
Age
Years recruited
Number from each hospital
Important risk factors
Smoking
Occupation
Coffee consumption
Previous gynaecological surgery
Previous Neurology
Previous TURP
Important disease related factors
PSA
Describe disease extent
Grade
Stage
Severity scores - IPSS
Describe treatments
Monitoring
Surgery
Radiotherapy
Medical
ESWL
TVT
Describe follow up
Follow up (mean (ST. dev))
Number to reach 2 years
Number to reach 5 years
Dead of disease
Dead of other causes
The purpose of this table is to allow a reader to use
your data:
Clinicians:
1. Do the cases represent the disease?
2. Are the patients the same as mine?
3. Can I identify how to treat a patient in clinic?
Researchers:
1. Can I use your data to compare with my results
2. Can I incorporate these into a meta-analysis
3. Can I use these data for new research?
Patients:
1. Have I been treated correctly?
2. Is there a better doctor for me?
How to report data?
Continuous data
e.g. Age, PSA ….
Show range, distribution and variability
Mean and standard deviation, if normally distributed
Median and ranges (25th and 75th centiles), if not
Ordinal or nominal data
e.g. stage or grade
Treat as individuals (not as continuous data)
Shown number and proportions for each
Compare populations, if relevant
Are the patients or the disease the same in each arm?
Figure 1. Flow chart of
patients/studies/interventions
Keep it as simple as possible
Results: Outcomes - reporting and analysis
In general:
- The results should match the primary and
secondary outcomes (as stated in M&M)
- Use paragraphs/sections for each outcome
- Intention to treat analysis is more robust (as it
avoids bias of losses or excluded cases)
Methods
Results
Methods
Results
Results: Outcomes - reporting and analysis
• Report outcomes for each
arm/population
• Table 2. typically report appropriate
details (mean/median & St. Dev or
Ranges)
• Use confidence intervals (95% CI)
Results – Survival outcomes
Be precise ….
1.
Are you describing death or survival
Disease specific survival vs. Disease specific mortality
2.
What are you describing?
Overall survival (death all causes) vs. Non-disease
specific survival (death from other causes) or competing
mortality vs. disease specific survival
Results – Treatment outcomes
Be precise …. in your terminology
1. Bladder cancer behavior
Recurrence, Progression, Relapse ….
2. Prostate cancer treatment
PSA recurrence – Post surgery, Post ADT, Post
radiotherapy
3. Incontinence outcomes
Pads, further treatment, QOL outcomes
4. etc.
Results – Treatment outcomes
Be precise ….
Use Objective measures where possible
use of another treatment (salvage radiotherapy)
worsening radiology or pathology
Try to avoid clinician based outcomes “progression” or
“failure” with no supporting data
Use blinded outcomes where possible
Be precise ….
Use Important outcomes (to the patient, clinician,
researcher, society, healthcare provider … )
Results – Treatment outcomes
Results – Treatment outcomes
Be precise ….
Use Realistic outcomes
Results – Compare outcomes:
Specifics
Use the appropriate test
For the Data:
Distribution: Normal or not
Data: Continuous or interval outcomes
Binary or many intervals
Nominal data
Beware of too few cases
Results – Compare outcomes: Specifics
Use the appropriate test
For the Outcome:
Inter group comparisons
Relative and absolute outcomes
Survival analysis
Competing mortality
Numbers needed to treat
Adjustments for trial design
Results – Compare outcomes: Specifics
Use the appropriate test
For the Context:
Univarbiable vs. Multivariable
Prognostic vs. Predictive
Results – Compare outcomes: General concepts
Results – Compare outcomes: General
concepts
Statistical vs. clinical significance
Results: Further analysis
Typical to look at sub-group analysis or analysis of
variables in later sections
Maybe the most interesting findings
Again … be realistic with your data
Present analyses clearly and logically
Do not look for analysis where the data are missing
Do not perform analysis that are not justified
Keep the reader with you
Discussion
• Briefly summarize your findings
• Answer your initial question
• Briefly mention previous reports and compare your data these reports
• Explain differences – you may both be right!
• Place your findings into the field
• How important are they?
• How strong can your conclusions be?
• How does this change patient care/treatments etc.?
• Where next with this field/work ….
• Limitations and shortcomings: Be critical and open with limitations … it makes
the data more believable and the paper stronger
Discussion: Common issues and faults
•
Too long
•
Too many points addressed
•
Too many comparisons and in too much detail
•
Incoherent or illogical organization
• Poor flow, changing directions, back and forth ….
• Write on paper a plan and stick to it
•
Discussion that is not supported by the data
•
Not addressing major inconstancies with the field
•
Too limited discussion of weaknesses etc.
The final check
1. Do you clearly state/explain and answer your main question?
2. Have you explained the novelty and importance of this work? Is it
clear for someone outside the field?
3. Rigor and presentation:
We are trusting that you have done this work/reported the outcomes
If you can not be bothered to present/edit/check the paper properly
…. then have you really treated 2,500 men consistently ?
4. Consistency:
Abstract vs. main body
Tables & figures vs. main text
5. Are your conclusions justified?