Biostatistics is one of the most unavoidable area in the modern day practice of evidence based medicine . In the ppt , trying to give a glimpse on how a clinician should approach Biostatistics
2. Points to cover
• How to read and analyse a scientific article critically
• How to have a literature search on relevant topics
• P value- level of evidence
• Meta-analysis –what is it?
• How to interpret a forest plot
3. Why is evidence in health care
important?
Value for
money
Patient
safety
Clinical
effectiven
ess
Can we
afford it?
Evidence
Is it safe?
Does it
work?
4. Failure to translate research finding
into clinical practice….
30-40% patients do not get treatments of proven
effectiveness
20-25% patients get care that is not needed or
potentially harmful
Schuster et al 1998 Milbank Memorial Quarterly
R Grol (2001) Med Care
5. The principles of evidence based
medicine
Asking good clinical questions
Putting patients first
Using evidence from well designed studies
Being systematic
7. A research finding is less likely to be true when:
• the studies conducted in a field are smaller
• Homogeneity of population studied is poor
• effect sizes are smaller
• SAMPLE SIZE
• there is greater flexibility in designs, definitions, outcomes, and analytical
modes
• there is greater financial and other interest and prejudice
• CONFLICTS OF INTEREST
Ioannidis JP, Plos Med 2005
Why Most Published Research Findings Are
False - John P.A. Ioannidis
8. Top cited articles are not ALWAYS telling
the truth
Why you should question the results
in all papers
Data fabrication can occur even in highly
esteemed journals
Why you should question the results in all papers
How you should question the results
9. • Initial findings…might prove wrong in the future
• Highly cited studies (>1000 citations) with efficacy claims
• 16% were contradicted by subsequent research
• 16% were found to have initially stronger effects
• 44% were replicated also with a larger sample size in subsequent
research compared with the original highly cited study)
• 24% had remained largely unchallenged
Top cited articles are not always telling the truth
Ioannidis JP, JAMA 2005
10. How you should question the results
Use a standard critical appraisal approach when
evaluating a scientific article
11. The typical anatomy of a paper
In most scientific journals, scientific papers follow a standard format.
Most journals use a conventional IMRD structure.
12. • What is the rationale for performing the study
• Is/are the research question(s) clearly defined and if not,
should they be?
Framework for How to Read and Critique a Research Study
American Nurses Association
http://www.nursingworld.org
Use a standard critical approach when
evaluating a scientific article
13. Critically evaluate the design of the study
• Is the design appropriate for the study?
• Does the sample fit with the research design and is the size sufficient?
• How were data collected?
• Is the analytical approach consistent with the study questions and research
design?
Use a standard critical approach when
evaluating a scientific article
Framework for How to Read and Critique a Research Study
American Nurses Association
http://www.nursingworld.org
14. Examine whether the study has been registered in a public
registry
• Has the trial been registered prior starting the trial?
• Are the primary endpoints the same with the registered version of the trial?
Use a standard critical approach when
evaluating a scientific article
15. Always review the existing literature
• Is the literature review relevant to the study, comprehensive, and include
recent research?
• Does the literature review support the need for the study?
• Are the findings consistent with existing literature?
Use a standard critical approach when
evaluating a scientific article
Framework for How to Read and Critique a Research Study
American Nurses Association
http://www.nursingworld.org
16. Critically evaluate the results and conclusions
• Are the results presented clearly in the text, tables and figures?
• Are the statistics clearly explained?
• Are the results explained in relationship to the theoretical framework, research
questions?
• Are the limitations presented and their implications discussed?
• Are there recommendations for clinical practice
Use a standard critical approach when
evaluating a scientific article
Framework for How to Read and Critique a Research Study
American Nurses Association
http://www.nursingworld.org
17. Generally...
you first read the Abstract in order to understand the major points of
the work.
• It clarifies whether you in fact know enough background to appreciate
the paper.
• It refreshes your memory about the topic.
• It helps you as the reader integrate the new information into your
previous knowledge about the topic.
18. Approach
• Introduction can be skimmed.
• What is the accepted state of knowledge in the field (take notes and even draw your own
figures)?
• What data led directly to this work?
• What question are they answering? (Is there a clear hypothesis?)
• Look at the major headings (do they follow the “anatomy” we just described?)
• The logical flow of papers goes straight from the Introduction to Results.
• Then to Discussion for interpretation of the findings.
• What is the conclusion of the paper?
• (It may not make sense to you at the moment, but note what it is.)
19. How to approach the introduction…
Grab a blank piece of paper:
Take notes.
Draw figures.
Define vocabulary.
Answer these questions:
What is the hypothesis being tested?
What are the basic conclusions?
20. Read the methods first or read them as you read the results.
• With each experiment/figure you should be able to explain
1) the basic procedure
2) the question it sought to answer
3) the results
4) the conclusion
• You should be able to explain all of these (1-4) to another classmate
clearly!
Materials and Methods, Results :
21. How to read a discussion
Take notes and answer these questions:
What conclusions do the authors draw?
Opinion/ interpretation?
Describe for yourself why these data significant.
Does it contribute to knowledge or correct errors?
22. Do you agree with the authors’ rationale for setting up the
experiments as they did?
Did they perform the experiments appropriately?
Were there enough experiments to support the major finding?
Do you see trends/patterns in their data?
Do you agree with the author’s conclusions?
What further questions do you have?
What might you suggest they do next?
Reflection and Criticism
23. Objective: The objective of the study was (1) “to evaluate the therapeutic efficacy of endometrial scratching in repeated controlled ovarian stimulation (COS) failure
cycles.” And (2) “to compare differences in pregnancy outcome by endometrial scratching in early (D2–D4) and late follicular phases (D7–D9) of the same stimulation
cycle.”
Materials and Methods: Women attending infertility clinic in a tertiary care center and who have two or more repeated COS failure cycles and planned for COS with
intrauterine insemination (IUI) were included in the study which is a prospective parallel, interventional, single-blinded, randomized control study, in 1:1 allocation ratio.
A total of 165 patients were recruited and randomly allocated into three groups: Group A (n = 55) underwent endometrial scratching on D2–D4 of the same COS cycle,
Group B (n = 55) on D7–D9, and Group C (n = 55) no intervention done. All the patients underwent COS according to standard protocol followed by IUI.
Results: Clinical pregnancy rate was 12.73% (odds ratio [OR] =0.87 95% confidence interval [CI] =0.288–2.55, P = 1), 16.36% (OR = 1.15; 95% CI = 0.40–3.23, P = 1), and
14.54%, respectively, in Group A, B, and C, respectively (P = 0.86), as per intention to treat analysis. Using Chi-square test, P value between Group A and B was 0.787,
between Group A and C was 1.000, and between Group B and C was 1.000. As per protocol analysis, clinical pregnancy rate was 13.46% (OR = 0.83; 95% CI = 0.27–
2.5, P = 0.74), 19.57% (OR = 1.3 95%; CI = 0.45–3.73, P = 0.41), and 15.69%. Using Chi-square test, P value between Group A and B was 0.588, between Group A and C
was 0.967, and between Group B and C was 0.815. No abortions and multiple pregnancies occurred in either of the groups.
Conclusion: The effect found was of good quantum in Group B as per protocol analysis which could be of clinical relevance if larger sample size would have been taken.
Endometrial scratching is a cost-effective and easy technique which may improve clinical pregnancy rates in previous COS failure cycles, but more trials are needed to be
conducted using larger sample size to achieve the improved and significant outcome.
Year : 2018 | Volume : 11 | Issue : 1 | Page : 59-71
Therapeutic efficacy of endometrial scratching in repeated Controlled Ovarian Stimulation
(COS) failure cycles
I
M
R
D
24. Introduction to evaluate the therapeutic efficacy of endometrial
scratching in repeated controlled ovarian
stimulation (COS) failure cycles.”
“to compare differences in pregnancy outcome by
endometrial scratching in early (D2–D4) and late
follicular phases (D7–D9) of the same stimulation
cycle.”
Materials Women attending infertility clinic in a tertiary care
center and who have two or more repeated COS
failure cycles and planned for COS with intrauterine
insemination (IUI) were included in the study which is
a prospective parallel, interventional, single-blinded,
randomized control study
Results Clinical pregnancy rate
A: 12.73% (odds ratio [OR] =0.87 95%
confidence interval [CI] =0.288–2.55, P = 1),
B:16.36% (OR = 1.15; 95% CI = 0.40–3.23, P = 1),
C: 14.54%,
Discussion Endometrial scratching is a cost-
effective and easy technique which
may improve clinical pregnancy
rates in previous COS failure cycles
The effect found was of good
quantum in Group B as per
protocol analysis which could be of
clinical relevance if larger sample
size would have been taken
33. • literature search is an methodical search for all of the literature
published on a topic. An effective search of the literature can be done
quickly, but demands an organised and systematic approach, so it is
important to keep records of the searches made and the information
found.
34. Literature
search
• Remember... research lit review is: systematic, explicit and
reproducible
Select appropriate research question
Identify appropriate databases
Break your question into concepts
Identify synonyms and subject headings for each concept
Combine synonym searches with OR
Combine concept searches with AND
Apply practical and methodological screens
Send search results to a citation manager
Remove duplicates
Use your lit review to summarize knowledge, assess research and
support new research initiatives
35.
36. Literature search
Define your topic /
research question:
1
Develop initial
keywords to
represent concepts
in your search:
2
Identify what type
of literature you
are searching for:
3
Identify sources to
search:
4
Identify scope /
limits of search:
5
37. To prepare for your search:
• Define your topic – write down your research question
• Identify what type of literature you are looking for e.g. primary research in
journal articles, systematic reviews, research reports, policy documents, books,
etc.
• Identify sources to search – databases, Google Scholar, individual organisations’
websites, library catalogues etc. (see more on sources below)
• Develop keywords / search terms that are logical and relevant to your search
• Think about scope of topic / search restrictions – anything related to your topic
that you wish to exclude
• Design a means of recording what you find (keep records)
38. Identify
keywords
Identify keywords that represent the main concepts in
the research topic.
• Consider:
• Synonyms (e.g. “young people” / adolescents)
• Differences in European and American terminology
(e.g. Accident and Emergency / Emergency Room)
• Differences in spelling (e.g. anaemia / anemia)
• Old and new terminology (e.g. mongolism / down
syndrome)
• Lay and medical terminology (e.g. stroke /
cerebrovascular accident)
• Acronyms (e.g. AIDS)
39. Sources of
literature
• Databases (references to journal articles)
• Medline, EMBASE, CINAHL, Cochrane Library, Web of Science,
Scopus
• Peer-reviewed journals
• Theses and dissertations
• Books, but note, tend to be published some time after research on
which they are based, and more recent research is often available.
• Internet
• Google Scholar
• Government websites: funded research reports, discussion
papers, government policies, and enquiry results
• Organisations and professional bodies’ websites
• Conference literature
• Newspapers
• Statistics
• Library catalogues
40. Search
techniques
• When searching databases or the Internet, use the advanced
options whenever possible. When you enter the advanced
searching option of the database, this is where you can make
use of AND/NOT/OR commands – Boolean operators. These
are used for combining terms and allow precision in finding
information.
• AND – allows you to narrow your search by combining words using
AND, e.g. “hospital managers” AND upskilling
• OR – lets you broaden your search to include similar or other
information connected by OR, e.g. retina OR eye;
• NOT – allows you to exclude specific information from the search, e.g.
anxiety NOT depression
47. 0 1
Impossible
Absolutely certain is
due to chance
How certain are you that the observed
difference in outcomes is due to chance?
Acknowledgement: Amanda Burls ; CEBM, Oxford
P < 0.05?
48.
49. Take home message
• Any observed difference between two
groups, no matter how small, can be
made to be “statistically significant” -
at any level of significance - by taking a
sufficiently large sample.
• Small p values only exclude the effects
of differences being due to chance
(statistical significance)
• P values do not tell you how effective
the intervention is and whether this is
clinically important (vs. clinical
significance)
50. Systematic reviews provide more information than single trials
More comprehensive
More reliable
Do you agree?
51. Types of questions and Levels of evidence
Level Intervention Prognosis Diagnosis Etiology
Least biased
I
SR of level II studies SR of Level II
studies
SR of Level II
studies
SR of Level II
studies
II RCT Inception cohort
study
Cross sectional
study among
consecutive
patients
Prospective cohort
study
III •Non-randomized
controlled clinical
trial
•Controlled before
and after study
•Cohort study
•Case control study
•Untreated controls
in an RCT
•Retrospectively
assembled cohort
study
•Cross sectional
study among non-
consecutive
patients
•Case control study
•Retrospective
cohort study
•Case control study
Most biased
IV
Case series Case series Case series
Cohort of patients
at different stages
of disease
Cross sectional
study
52. What is a systematic review ?
• “The application of scientific strategies that limit
bias to the systematic assembly, critical appraisal,
and synthesis of all relevant studies on a specific
topic”
Moher et al, Lancet 1999
• Not all systematic reviews use meta analysis to
synthesize data
• Systematic reviews can answer questions that
individual trials cannot
53. What is meta-analysis ?
• The statistical method that combines the results of
several independent studies
• A meta-analysis attempts to provide an average result
across the studies that are combined
• Can combine studies evaluating prevalence & incidence;
diagnosis; prognosis; and the effects of interventions
• The main advantage is of increased power compared to
the individual studies
• However…….meta-analyses can seriously mislead
54. Meta-analyses can mislead if done outside a systematic review
• It may not include all relevant studies but only a biased subset of
studies done
• Due to reporting biases (Publication bias; Language bias; Retrieval bias;
Citation bias)
• Are more likely to show statistically significant results in favor of
intervention, be in English, and easily retrieved
• Including only positive studies will give you positive (but wrong?) results
• When poor quality studies are included or when quality issues are
ignored
• Studies at high risk of bias yield misleading (and variable) results
• When inadequate attention is given to heterogeneity;
• Indiscriminate data aggregation can lead to inaccurate conclusions
• Important to know the overall effect, the dispersion of results, and the
reasons for important differences in results
Egger M et al. Uses and abuses of meta-analysis. Clinical Medicine 2001;1:478-84
55. Narrative Review Systematic Review
No Methods section;
not reproducible
Clearly described protocol with detailed
methods
Limited searching for trials (often limited to
Medline); leads to ‘publication bias’
Comprehensive searching for published and
unpublished trials with no language
restrictions
Include different study designs, often do not
evaluate validity
Mostly include only RCTs; evaluates validity
Over-reliance on p values Estimates size of effect with confidence
intervals (precision)
Uses ‘vote counting’; each trial given same
weight
Differentially weights trials so that larger trials
with more information and precise results are
given more weight
Descriptive Meta-analysis pools results of similar trials;
provides a ‘tower of power’
Subjective; Biased Objective ( two or more authors who
independently undertake the review)
56.
57. P- Who is the patient or what problem is
being addressed?
I- What is the intervention or exposure?
C– What is the comparison group?
O- What is the outcome or endpoint?
Architecture of a focused 4-part review question
+ study design
Richardson et al. The well-built clinical question: a key to evidence-based decisions. ACP Journal Club 1995;A-12
Counsell C. Formulating questions and locating primary studies for inclusion in systematic reviews. Ann Intern Med 1997;127:380-7.
58. Study Characteristics Eligibility criteria
(Insert eligibility criteria for each
characteristic as defined in the
Protocol)
Y
es No Unclear
Location in
text
(pg &
¶/fig/table)
Type of study Randomised Controlled Trial
Controlled Clinical Trial
(quasi-randomised trial)
Participants
Types of intervention
Types of comparison
Types of outcome measures
PICO
Intervention review – RCTs only
•Study Eligibility
59. Objectives
• To compare the effectiveness of a GnRH agonist with HCG
for triggering final oocyte maturation in IVF and ICSI patients
undergoing controlled ovarian hyperstimulation in a GnRH
antagonist protocol followed by embryo transfer.
Patients: IVF patients undergoing controlled ovarian
hyperstimulation in a GnRH antagonist protocol followed by
embryo transfer.
Intervention: a GnRH agonist
Comparison: HCG
Outcome: Triggering final oocyte maturation
60. Types of studies
• Randomised controlled studies (RCTs) comparing
GnRH
agonist versus HCG for final oocyte maturation
triggering in
GnRH antagonist IVF and ICSI treatment cycles
• Quasi-randomised controlled studies (e.g. allocation
by
alternation, reference to case record number or date
of birth);
cross-over studies were excluded
61. Types of participants
• Inclusion criteria
• Couples undergoing in vitro fertilisation (IVF) or intracytoplasmic
sperm injection (ICSI) for therapeutic reasons or for oocyte
donation and randomised to receive either a GnRH agonist or
HCG for final oocyte maturation triggering
• Exclusion criteria
• Women who were not IVF or ICSI patients, i.e. intrauterine
insemination (IUI) patients
62. Types of interventions
• GnRH agonists in comparison with HCG for final
oocyte maturation triggering in GnRH-antagonist
controlled hyperstimulation cycles, IVF or ICSI
followed by embryo transfer (ET) with or without
luteal phase support, in both autologous and donor
cycles
63. Types of outcome measures
Primary outcomes
• Live birth rate (LBR) per woman randomised: live birth
defined as delivery of a live fetus after 20 completed weeks of
gestation
• Ovarian hyperstimulation syndrome (OHSS) incidence per
woman randomised: detected by clinical grading of OHSS;
laboratory investigations as haematocrit, haemoglobin, renal
function; and imaging techniques including ovarian and
abdominal ultrasound and chest X-ray
64. Types of outcome measures
Secondary outcomes
• Ongoing pregnancy rate (OPR) per woman randomised:
ongoing pregnancy defined as a pregnancy beyond 12 weeks
• Clinical pregnancy rate (CPR) per woman randomised:
clinical pregnancy defined as the presence of a fetal heart rate with
transvaginal ultrasound
• Miscarriage rate per woman randomised
• Multiple pregnancy rate per woman randomised
65. How a focused question helps in
searching for studies
Patient
or Problem
Intervention &
comparison
Outcome
Study design
filters
+
PICO + STUDY DESIGN FILTER
Studies most likely to
address the question
66. Search methods for identification of studies
All published and unpublished RCTs of GnRH agonist
versus HCG for final oocyte maturation triggering were
sought, without language restriction and in consultation
with the Menstrual Disorders and Subfertility Group
(MDSG) Trials Search Co-ordinator, using the following
search strategy.
67. Search methods for identification of studies
• The following databases were searched from their
inception:
• Cochrane Central Register of Controlled Trials (CENTRAL)
(The Cochrane Library, 1st Quarter 2009)
• Ovid MEDLINE (1950 to 21 October 2010)
• EMBASE (to 21 October 2010);
• Menstrual Disorders and Subfertility Group Specialised
Register (to 21 October 2010)
68. Searching other resources
• Reference lists of all known primary studies and review
articles,
• Citation lists of relevant publications, abstracts of major
scientific meetings (for example ESHRE, ASRM)
• Known experts and personal contacts were contacted
regarding any unpublished materials.
• In addition, we searched the following.
• Trial registers for on-going and registered trials
• Citation indexes:
• LILACS database
• ClinicalStudyResults
• PubMed
• Open SIGLE database for grey literature in Europe
69. Data collection
• Two review authors independently assessed trial
selection, trial quality and extracted data.
• Sought additional information on trial
methodology or actual original trial data from
the principal author of studies that appeared to
meet eligibility criteria but were unclear in
aspects of methodology, or where the data were
in a form unsuitable for meta-analysis.
73. Why assess study quality?
• if poor quality trials are the building blocks of the
review, the review may follow high quality methods,
but the quality of evidence may still be poor
74. Describe Drop-
out
Not all RCTs are made of 24-carat gold
Intervention
Control
Outcome
Outcome
Randomisation
Equal
treatment
Blind participants
and experimenters
Concealment of
allocation
Blind
assessment
Report all
outcomes
Selection Bias Performance
Bias
Attrition bias Bias
Detection Bias
Reporting Bias
75. The Cochrane approach
• describe the following for each study in detail:
• random sequence generation
• allocation concealment
• blinding
• incomplete outcome data
• selective outcome reporting
• any other risks
• empirical research shows that these components
can have a significant effect on results, often
leading to exaggerated effects
76. The COCHRANE RISK OF BIAS TOOL
Risk of bias table
Risk of bias graph
Risk of bias summary figure
79. When can you do a meta-
analysis?
• When more than one study has estimated an effect
• When there are no differences in the study
characteristics (patients, interventions) that are likely
to substantially affect outcome, so that combing
data will produce a a clinically useful and meaningful
result
• When the outcome has been measured in similar
ways
• When the data are available
• REMEMBER, you do not need to statistically pool
results to do a systematic review
80. Steps in doing a meta-analysis
• Define comparisons [GnRH agonists versus HCG]
• Decide on appropriate study results (outcomes) for
each comparison [live birth rate]
• Select an appropriate summary statistic for each
comparison [risk ratio / odds ratio]
• Pool summary statistics from each study (if
appropriate)
• Assess the similarity of study results within each
comparison
• Consider the reliability of the summaries
81. Averaging studies
• A simple average gives each study equal weight
• This seems intuitively wrong
• Some studies are more likely to give an answer closer to the ‘true’
effect than others
82. Weighting studies
• More weight to the studies which give us more
information
• More participants
• More events
• Lower variance
• Weight is proportional to inverse variance
• What is variance?
Cochrane Collaboration Glossary