Reporting your results for publication according to guidelines

Presented at the 30th Annual EACTS Meeting, Barcelona, Spain (1-5 October 2016)

Transcript

1. How to perform more advanced statistics: basics and pitfalls
   Graeme L. Hickey
   Department of Biostatistics, University of Liverpool
   * No conflicts of interest
   

   View Slide

2. View Slide

3. URL: http://www.oxfordjournals.org/our_journals/ejcts/for_authors/manuscript_instructions.html
   

   View Slide

4. Development and validation of risk prediction models
   Clinical randomized trials
   Systematic reviews and meta-analyses
   Comprehensive database of guidelines
   www.equator-network.org
   

   View Slide

5. URL: http://www.consort-statement.org
   Supplemental file for manuscript submission Manuscript figure
   

   View Slide

6. View Slide

7. + corrigendum: Eur J Cardio-Thoracic Surg 2016;49:1024.
   

   View Slide

8. • Above all
   

   View Slide

9. Examples* of common ‘issues’
   (which are covered in the EJCTS and ICVTS
   guidelines)
   * A somewhat haphazard collection
   

   View Slide

10. View Slide

11. * Austin, P. C. (2007). Propensity-score matching in the cardiovascular surgery literature from 2004 to 2006: a systematic review and suggestions for
    improvement. The Journal of Thoracic and Cardiovascular Surgery, 134(5), 1128–35.
    

    View Slide

12. à
    

    View Slide

13. Table 1. Patient and operative characteristics data by CPB technique with statistical
    comparison.
    518
    Overall On-pump Off-pump Δ (%) P
    Total number n=3402 n=1173 n=2229
    Logistic EuroSCORE (%) 2.4 ± 2.5 2.4 ± 2.8 2.3 ± 2.3 1.8 0.965
    Age (years) 61.7 ±10.6 61.1 ± 10.3 61.9 ± 10.7 -8.1 0.026
    BMI (kg/m2) 28.5 ± 4.6 28.7 ± 4.7 28.4 ± 4.5 6.1 0.090
    N % N % N %
    Female 880 25.9% 325 27.7% 555 24.9% 6.4 0.083
    Preoperative AF 69 2.0% 28 2.4% 41 1.8% 3.8 0.343
    Urgent 733 21.5% 271 23.1% 462 20.7% 5.7 0.119
    NYHA III/IV 645 19.0% 225 19.2% 420 18.8% 0.9 0.846
    History of neurological
    dysfunction 53 1.6% 25 2.1% 28 1.3% 6.8
    0.070
    Diabetes (insulin or diet
    controlled) 600 17.6% 207 17.6% 393 17.6% 0.0
    >0.999
    * Continuous data reported as mean ± standard deviation
    Percentages correctly
    rounded
    Appropriate and
    consistent precision
    Columns clearly labeled
    Notation defined
    Units included
    Number of subjects
    sum correctly
    

    View Slide

14. 0 200 400 600 800 1000
    0.0 0.2 0.4 0.6 0.8 1.0
    Time
    CumSum
    An unacceptably presented Kaplan−Meier graph
    P0.0
    0.2
    0.4
    0.6
    0.8
    1.0
    0 6 12 18 24 30
    Time from diagnosis (months)
    Survival probability
    Male
    Female
    138 86 35 17 7 2
    90 70 30 15 6 1
    No. at risk
    +
    +
    +
    +
    +
    +
    +
    +
    +
    ++
    +
    +
    +
    +
    +
    +
    +
    +
    ++
    + + +
    +
    +
    +
    +
    +
    ++
    +
    +
    +
    +
    +
    +
    +
    ++
    +
    +
    ++
    +
    ++
    +
    +
    +
    +
    +
    +
    ++
    +
    + +
    +
    + +
    An acceptably presented Kaplan−Meier graph
    Log−rank test P = 0.001
    

    View Slide

15. [1] https://www.biostat.wisc.edu/~kbroman/topten_worstgraphs/
    [2] https://garstats.wordpress.com/2016/05/27/the-percentile-bootstrap/
    • Statistical figures are for summarizing complex data
    • Readers will be drawn to them, so they need to be clearly presented
    • Label all axes – and report what is being shown
    

    View Slide

16. Interpreting randomized trial data: Tuesday 4th October 14:15 - 15:45
    

    View Slide

17. View Slide

18. Editorial Board
    Friedhelm Beyersdorf (Editor-in-
    Chief)
    Joel Dunning (Associate Editor)
    Judy Gaillard (Managing Editor)
    Franziska Lueder (Editorial
    Manager)
    Assistant Editors
    (Statistical Consultants)
    Burkhardt Seifert
    Gottfried Sodeck
    Matthew J. Carr
    Hans Ulrich Burger
    Graeme L. Hickey
    + all other editorial members
    

    View Slide