How To Analyse A Clinical Paper CHRIS DAVIDSON BRIGHTON UK

How To Analyse A Clinical
Paper
CHRIS DAVIDSON
BRIGHTON UK
Why do we read original scientific papers?
2
y Inform clinical practice
y Investigate new drugs/procedures
y Find causes / risk factors for disease
y Part of a research project
y Pass examinations
y Get promotion!
REFERENCE: How to read a paper: The basis of Evidence-Based Medicine.
Trisha Greenhalgh. Blackwell 4th Edition 2010
EFIM Clin. Research Course 2013
How to read a Paper
y What was the
Research Question?
{
Why was the Study
needed?
y What was the Study
Design?
{
EFIM Clin. Research Course 2013
Was the Design
appropriate?
3
What was the Research Design?
y Primary Studies
{ Experiments
{ Clinical Trials
{ Surveys
y Secondary Studies
{ Overviews (metaanalysis etc.)
{ Guidelines
{ Decision Analyses
{ Economic Analyses
EFIM Clin. Research Course 2013
4
Broad Fields of Research
5
y Therapy: Drugs or
Procedures
{
Preferred Design: RCT
y Diagnosis: evaluation
new test
{
Preferred Design:
Cross-section Survey
y Screening
{ Preferred Design:
Cross-section Survey
y Prognosis
{ Preferred Design:
Longitudinal Survey
y Causation
{ Preferred Design:
Cohort / Case-Control
Study
y Psychometric studies
{ Preferred Design:
Qualitative Study
REFERENCE: How to read a paper: The basis of Evidence-Based Medicine.
REFERENCE: How to read a paper: The basis of Evidence-Based Medicine.
Trisha Greenhalgh. Blackwell 3rdrdEdition 2006
Trisha Greenhalgh. Blackwell 3 Edition 2006
EFIM Clin. Research Course 2013
Why are Randomised Controlled
Trials (RCTs) considered so
important?
6
THEY MINIMISE THE EFFECT
OF CONFOUNDING VARIABLES
EFIM Clin. Research Course 2013
RCTs : Statistics for the Amateur…
7
y
Do the patients selected
reflect the ‘Real World’?
{
y
Inclusion/exclusion criteria
{
Compare demography / Rx
in each
Are there significant
numbers of ‘drop-outs’ or
‘cross-over’ patients?
{
Are the statistical tests
appropriate?
{
Are the studied groups
comparable clinically?
{
y
y
Side-effects or
patient/doctor preference
EFIM Clin. Research Course 2013
y
Parametric vs nonParametric data
Is the p-value appropriate
with multiple tests (p < 0.05
can occur every 20 tests by
chance)
Is the difference seen
clinically relevant?
{
RELATIVE and ABSOLUTE
differences
Clinical Relevance of Trial results
8
From: Trial and Error: How to Avoid Commonly Encountered Limitations of Published Clinical Trials
J Am Coll Cardiol. 2010;55(5):415-427. doi:10.1016/j.jacc.2009.06.065
EFIM Clin. Research Course 2013
Invasive compared with non-invasive treatment in unstable
coronary artery disease: FRISC II prospective randomised
multicentre study
FRagmin and Fast Revascularisation during InStability in Coronary artery
disease (FRISC II) Investigators*
THE LANCET • Vol 354 • August 28, 1999
B a c k g r o u n d: In unstable coronary-artery disease early invasive procedures are
common, despite lack of evidence for the superiority of this approach. W e compared
an early invasive with a non-invasive treatment strategy in unstable coronary-artery
disease .
I n t e r p r e t a t i o n: The early invasive approach should be the preferred
strategy in most patients with unstable coronary artery disease who have
signs of ischaemia on electrocardiography or raised biochemical markers of
myocardial damage.
EFIM Clin. Research Course 2013
9
FRISC II Trial - Patients
10
EFIM Clin. Research Course 2013
FRISC II Trial: Results
11
Figure 3: Probability of death or myocardial infarction
in invasive and non-invasive groups
EFIM Clin. Research Course 2013
FRISC II Results
14
P = 0.031
12
RELATIVE
RISK REDUCTION
P = 0.045
10
% Events at 6
months
12.1 - 9.4 /12.1
= 22.3 %
8
6
P = 0.10
4
12.1 - 9.4
= 2.7 %
2
0
Intervention
Control
ABSOLUTE
RISK REDUCTION
Combined
M. Infarct
Death
9.4
12.1
7.8
10.1
1.9
2.9
EFIM Clin. Research Course 2013
12
FRISC II Study
EFIM Clin. Research Course 2013
13
y MACE
{
{
{
Death
Myocardial Infarction
Re-vascularisation
Used as a Quality standard for laboratories and published research in
cardiac intervention
J. Am. Coll. Cardiol. 2008;51;701-707
EFIM Clin. Research Course 2013
14
EFIM Clin. Research Course 2013
15
Composite End-Points
16
Problems with use of composite end points in cardiovascular trials: systematic review
of randomised controlled trials Ferreira-Gonzalez et al BMJ 2007 334; 786
EFIM Clin. Research Course 2013
ARISTOTLE Trial
N Engl J Med 2011; 365:981-992
17
EFIM Clin. Research Course 2013
So how to critically appraise?
18
3 STEPS
EFIM Clin. Research Course 2013
Appraisal Tools
19
y Critical Appraisal Skills Programme
{
http://www.casp-uk.net/homepage/
y Evidence based medicine: Tool kit
University of Alberta
{
http://www.ebm.med.ualberta.ca/
EFIM Clin. Research Course 2013
CASP website: http://www.casp-uk.net
EFIM Clin. Research Course 2013
20
EFIM Clin. Research Course 2013
21
ARISTOTLE Trial
N Engl J Med 2011; 365:981-992
22
y
Group A: study methods and patient selection
y
Group B: study results
y
Group C: study design and conclusions
EFIM Clin. Research Course 2013
ARISTOTLE Study
DISCUSSION
ARISTOTLE Trial: patient characteristics
24
EFIM Clin. Research Course 2013
ARISTOTLE Trial: patient characteristics
25
EFIM Clin. Research Course 2013
CHADS‐2 Score in AF
26
RISK FACTOR
Cardiac Failure
Hypertension
Age >75 years
Diabetes
Stroke / TIA
JAMA 2001;285:2864 – 2870.
Madrid 2012
SCORE
1
1
1
1
2
CHADS‐2 Score in AF
27
ESC recommendation 2010:
Treat if Score 2 or more
Madrid 2012
ARISTOTLE Trial
28
EFIM Clin. Research Course 2013
ARISTOTLE Trial
29
EFIM Clin. Research Course 2013
ARISTOTLE Trial: time course
30
EFIM Clin. Research Course 2013
ARISTOTLE Trial: time course
31
EFIM Clin. Research Course 2013
ARISTOTLE Trial: sub-group analysis
32
EFIM Clin. Research Course 2013
33
EFIM Clin. Research Course 2013
Aristotle Trial: Conclusions
34
y Study Objectives met:
{ Non-inferiority
to standard care warfarin
{ Significantly fewer heamorrhagic CVAs
{ Fewer serious haemorrhages
{ [No increase in MI]
EFIM Clin. Research Course 2013
Aristotle Trial: Conclusions (2013)
35
y METHODS
{ Non_inferiority design
{ Definition haemorrhagic stroke
{ Representative sample
{ Interaction with anti platelet drugs
{ Double blind INR?
{ Control group comparable
EFIM Clin. Research Course 2013
Aristotle Trial: Conclusions (2013)
36
y RESULTS
{ Withdrawal and lost patients low
{ INR 66%
{ Stroke positive but especially haemorrhagic stroke
{ Haemorhagic events significantly lower
{ 81% adverser outcomes
{ 33% serious adverse events
EFIM Clin. Research Course 2013
Aristotle Trial: Conclusions (2013)
37
y DISCUSSION
{ Target INR levels important
{ Absolute vs relative risks
{ No cost effective ananlysis
EFIM Clin. Research Course 2013
Aristotle Trial: Conclusions (2012)
38
y BUT
{ Stroke (primary endpoint) not clearly defined
{ Discrepancy between haemorrhagic stroke and cerebral
haemorrhages
{ Other haemorrhagic complications (60%) not detailed
{ Further details necessary about drop-outs (>25%)
{ Further details on warfarin control (especially at the time of
events)
{ ?heterogeneity from Asian recruits (INR control, % cerebral
haemorhages).
EFIM Clin. Research Course 2013
From: Influence of Global Region on Outcomes in Heart Failure Beta-Blocker Trials
J Am Coll Cardiol. 2011;58(9):915-922. doi:10.1016/j.jacc.2011.03.057
Figure Legend:
Major β-Blocker Trials (A) Overall results. (B) The United States versus the rest of the world (ROW). BEST = β-Blocker Evaluation of
Survival Trial; CIBIS-II = Cardiac Insufficiency Bisoprolol Study; COPERNICUS = Carvedilol Prospective Randomized Cumulative
Survival trial; MERIT-HF = Metoprolol Controlled-Release Randomized Intervention Trial in Heart Failure.
Date of download:
6/13/2013
Copyright © The American College of Cardiology.
All rights reserved.
ARISTOTLE Study
IS THIS A LANDMARK
STUDY?