Statements of
Good Biostatistical Practice:
A Biostatistical Guide for Creating Evidence
Release1.2, 22 Feb 2010
Relaese 1.1, 23 Dec 2009
MedicReS Guide for Creating Evidence was named as Good Biostatistical Practice at the 2nd Meeting of MedicReS Board Members in Ankara, Turkey in 2007, November 24th
‘MedicReS The Good Biostatistical Practice’ has been announced for the first time in VII. International Congress on Headache in Children and Adolescents in Istanbul, Turkey in 2008, May 17–21 [1]
This guide endorses CONSORT, STROBE, STARD, PRISMA, MOOSE Statements from EQUATOR Network [2]
The Three Main Parts of GBP with their attributed weights
Design– Good Planning (50%)
Analysis– Good Executing (30%)
Publication – Good Reporting (20%)
MedicRS GBRS (Good Biostatistical Review Standards Release 1.2)
Editor
E. Arzu KANIK, PhD
Assoc. Professor of Biostatistics
MedicReS Program Coordinator
Outline of Good Biostatistical Practice (full text of GBP is given at MedicReS Courses)
Part I. Design- Good Planning
This part consists of 20 subtitles in the form of Expanded PICOS (E-PICOS); E-PICOS
includes the following From Purpose to Submitting
P: Purpose & Population & Patients & Participants & Power & P value & Protocol (7 subtitles)
I: Intervention & Interpretation (2 subtitles)
C: Comparators & Controls & Covariate & Confounding (4 subtitles)
O: Outcomes & Outputs (2 subtitles)
S: Study Design, Sample Size, Summary Statistics, Statistical Software and Submitting (5 subtitles)
Part II. Analyze- Good Executing
This part of GBP consists of 12 subtitles of 4D (Data Collecting, Data Control, Data Analyze, and Data Interpretation)
Data Collection (2 subtitles)
• Validity
• Reliability
Data Control (2 subtitles)
• Missing values
• Outliers
Data Analyze (4 subtitles)
• Preparing data to analyze
• Calculating summary statistics
• Controlling assumptions
• Deciding statistical methods for testing hypothesis
Data Interpretation (4 subtitles)
• Parameter estimating
• Interpretation of p values
• Clinical significance vs. statistical significance
• Small sample size vs. large sample size
Part III. Publication- Good Reporting
This Part consist of 8 subtitles for Good Biostatistical Reporting
(MedicReS Good Biostatistical Review Standards-GBRS)
|
MedicReS Good Biostatistical Review Standards (GBRS) Release1.2, 22 Feb 2010 [1]
|
|
GBRS Subtitles
|
GBRS Questions
|
|
Study Design
|
Was a suitable design used to achieve the objective?
|
|
Was/Were reporting guideline(s) suitable for study design used? [2]
|
|
Was appropriate control group used?
|
|
Were any efforts made to avoid potential sources of bias?
|
|
Sample Size
|
Was minimum sample size calculated?
|
|
If calculated, was pre-study calculation of sample size reported?
|
|
Participants
|
Were socio-demographic characteristics of participants and those who didn’t wish to participate reported in the study?
|
|
Were withdrawals from study independent of study groups and/or doses?
|
|
Was a flow diagram of participants given for all stages of study?
|
|
Summary Statistics
|
Were the validity and reliability of measurement methods used reported in the study? (Gold standard, inter-rater agreement)
|
|
Were the analysis of randomness of missing values and outliers made?
|
|
Were appropriate summary statistics used?
|
|
Were there any misuses of standard error?
|
|
Were confidence intervals calculated for all of the summary statistics used in the study?
|
|
Were the terms ‘relation’, ‘correlation’, ‘difference’ and risk terms used correctly?
|
|
Statistical Analysis
|
Were statistical methods compatible with study design and variables which are used in this study?
|
|
Were any assumptions of statistical methods violated?
|
|
Was the choice of parametric or nonparametric test correct?
|
|
Were all statistical methods used in the study reported in methods section the paper?
|
|
Was it stated in the study which statistical method was used for which hypothesis?
|
|
Were covariant and effect mixing variables considered during analysis? If necessary, was multiple data analysis conducted?
|
|
Was multivariate analysis necessary? If necessary was it used correctly?
|
|
Were subgroups constituted during data analysis?
|
|
Were method and aim of subgroup analysis correct?
|
|
Were Cutoffs used for quantitative tests? if used, how were they determined ?
|
|
Were subjective criteria used for qualitative tests? if used, how they were determined?
|
|
Was sample size sufficient for subgroup analysis?
|
|
Tables and Graphs
|
Was the number of significant digits in tables used correctly?
|
|
Were the graphics selected compatible with data analysis?
|
|
Were indications and interpretations of ratios and percentages in tables in accordance with the content?
|
|
Were both significant and non significant p values given in tables?( to avoid publication bias)
|
|
Interpretation
|
Were indication, interpretation of p-values in study and generalizations made correctly?
|
|
Were both statistical and clinical significance values of results discussed?
|
|
Ethics
|
Was the statistical software used commercial? If commercial, was this usage legal?
|
|
Was there a statistical expert contribution in the material? If so, was this contribution valued?
|
[2] Simera I, Moher D, Hoey J, Schulz KF, Altman DG. A catalogue of reporting guidelines for health research. Eur J Clin Invest 2010 Jan;40(1):35-53. http://www.equator-network.org |