Transcript Statistics for MRCGP

Critical Appraisal for MRCGP
Jim McMorran
Coventry GP
GP trainer
Editor GPnotebook (www.gpnotebook.co.uk)
Can all elephants be washed
by children?
Analysing studies






Confounding factors..
Sample size..
Applicability of results..
Placebo controlled..
Blinded…
Primary endpoints…
Critical Appraisal Methodology
(from GPnotebook)
1.
2.
3.
4.
5.
6.
7.
8.
9.
what is the research question - for most studies there
should be a single sentence describing why the study
was done and what hypothesis the authors were
testing
the type of study design used
the study population investigated
comparator drugs and doses studied
sample size and power used in the study
primary versus secondary endpoints and subgroup
analysis
duration of the study
study follow-up
is there a sponsorship or publication bias
Critical Appraisal Methodology
(from GPnotebook)
other considerations
1.
1.
2.
3.
4.
5.
intention to treat analysis
what p-values are used
what confidence intervals are used
comparing statistical and clinical significance of
results
analysis of results e.g absolute risks, relative risks,
odds ratios, hazard ratios, numbers needed to
treat, numbers needed to harm, Kaplan-Meier
survival curves, simple statistical test
Research question

PROACTIVE (PROspective pioglitAzone
Clinical Trial In macroVascular Events)

to find out whether pioglitazone
reduces cardiovascular (CV)
morbidity and mortality in high-risk
patients with type 2 diabetes
Type of Study Design Used




Type of study:
retrospective e.g. case control study
prospective e.g. cohort study, randomised
controlled trial
Other points regarding study design include:



was the study blinded? If so was the study blinded
for the patient and the investigator (double blinded)?
was the study a crossover study?
The findings from a prospective study are
more valid than those from a retrospective
one
Different outcomes Different
studies

Coronary Heart Disease and HRT


CHD and HRT - General Practice Notebook
Management of febrile child – use of tepid sponging
or not?

prevention - General Practice Notebook
Study Population investigated



Age of population
Exclusion criteria used
study population - General Practice Notebook
Study populations investigated

Age when statins are effective

Statins effective for primary prevention up to the age of 70
years


Statins effective for primary prevention up to the age of 85
years


WOSCOPS - General Practice Notebook
MRC / BHF Heart Protection Study - General Practice Notebook
Is the study population comparable to those patients
in primary care


Proactive study - General Practice Notebook
FIELD study - General Practice Notebook
Comparitor drugs and doses
studied

Is there a comparitor Drug

comparator drugs in statistical studies General Practice Notebook
Interesting comparisons

Bupropion versus varenicline for
smoking cessation


varenicline - General Practice Notebook
 Is the indicated length of use of medication the same for
both medications
Exercise versus use of metformin in
prevention of diabetes


exercise in the prevention of type 2 diabetes - General
Practice Notebook
What about the dose of metformin
Sample Size and Power



What is statistical power?
What is the typical value of the power
of the study?
Why is the power of the study
important when considering study
endpoints?

sample size and power of study - General
Practice Notebook
Sample Size and Power
Power = 1 – type II error
Higher Power Occurs with Less Overlap of Distributions
Primary Versus Secondary
Endpoints

The power of a study is based on the primary
endpoint of a study


primary versus secondary endpoints - General
Practice Notebook
Headline studies where the primary endpoint
was not significant include


Proactive study - General Practice Notebook
FIELD study - General Practice Notebook
Duration of study


Sufficient period to allow events to occur
Long enough to relate findings to use in
chronically ill patient

duration of study - General Practice Notebook


Efficacy of topical NSAIDs versus oral NSAIDs
topical non-steroidal anti-inflammatory drugs (NSAIDs) in
osteoarthritis - General Practice Notebook
Study follow-up


Affects the power of the study
study follow-up - General Practice Notebook
Is there sponsorship or
publication bias


Is the study sponsored by the
pharmaceutical industry or government
sponsored
is there sponsorship or publication bias General Practice Notebook
Other considerations – a
selection

Intention to treat analysis


Comparing statistical and clinical significance


intention to treat analysis - General Practice Notebook
statistical versus clinical significance - General Practice
Notebook
NNT


number needed to treat (NNT) - General Practice
Notebook
Consider how many patients are required to be treated
to prevent a single stroke based on the SPARCL trial
(look at CIs)

SPARCL (Stroke Prevention by Aggressive Reduction in
Cholesterol Levels) trial - General Practice Notebook

The End..