Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
GRADE Guidelines detailed version
Transcript of GRADE Guidelines detailed version
or best available evidence for your PICO questions
How confident are you in the effect estimates you have found?
What are the expected benefits and harms?
Formulate with clear language
Which issues do I need to look at?
What is important for my patients?
For every outcome
What are the overall confidence you
have in these estimates across outcomes?
What are patients preferences?
Do they vary? Sociatal values to consider?
Are there resource issues to consider? For whom?
Weak or Strong
Most patients would want, but not all
Nearly all patients would want this
Give a rationale for your decisions
Patient Intervention Comparator Outcomes
Formulate Clinical Questions
Get the right people (knowledge, geography, buy-in)
Leadership and 'who's in charge of what'
Define topics and scope
Methodologists included or assisting
Patients involved? Health oeconomics?
Steering group and Exceutive comittees
Identify and get resources and skills
Define consensus and ledership processes
Formally request participation / support from stakeholders / professional organizations relevant to clinical topic
Appoint chapter editors and guideline panelists
Organize GRADE training sessions for chapter editors
Record overall intellectual and financial conflicts of interest for all guideline panellists
Organize first guideline conference /meeting
Search for best evidence
use your PICO questions
you might need to use different studies for the various outcomes, as not all studies will include your chosen outcomes
for effect estimates a recent published systematic review of randomized trials is what you would want
for baseline risk estimates, a well done observational is want you would want
You do not always get what you want!
you will often have to use the control arm of a study as the basline risk
there might not be a recent systematic review out there, or any at all.
you might have to make your own meta-analysis
you might not find rct's but end up with observational studies
sometimes all you have is expert experience and case series
using GRADE is all about being honest, transparent and explicit
Well, so you didn't find a lot of evidence out there,
but you did look!
And now you have the best available material to base your recommendations on.
you should be proud
Evidence Practice Centers
Benefits & harms
Quality of evidence
Values & Preferences
Use the Evidence profiles
What are the main benefits? Harms?
Look at the absolute difference
Generally you are not more overall confident than the lowest confidence of an critical outcome
We find that clinicians respond very bad to vagueness like:
"you should think about considering...."
We recommend you use active language, like:
"We recommend..... "
"We suggest..... "
You know or expect variability in patients preferences
You have low confidence in the effect estimates, so you don't really know what will happen to your patients
There is a close call between benefits and harms
There are no absolutes
Low confidence in effect estimates can result in Strong recommendations
High confidence in effect estimates and most people would want the intervention, but the resource use it will lead to cannot be justified
... There are always exceptions....
See the www.gradeworkinggroup.org help files for more information
Randomization? Randomization concealment? Blinding (providors, patients,
adjudicators, analyzers)? Unacceptable loss to follow up?
Downgrade when warranted
and decide overall confidence in your effect estimate
Decide on Strength of the recommendations:
GRADE = not just
Quality of the evidence
Take one topic, and all your knowledge about it
With multiple comparisons or combined recommendations
you have to look at more than one Evidence profile
Are the Confidence intervals so large so that you would act different in the different ends of the scale? Use the confidence interval of the absolute difference
Does the evidence match your PICO? Same Population, same intervention, comparison and outcomes? Or are you using indirect evidence?
Are the Studies showing consistent results?
Look at the forrest plots. I2 >50? Overlapping confidence intervals?
Do you suspect publication bias? Are there unpublished studies that might alter your effect estimate? Look at the funnel plots.
Observational studies can be upgraded in special circumstances
5 criteria to downgrade your overall confidence in effect estimates
Why did you recommend what you did?
Each criteria can reduce the confidence by one or, if very serious, by two levels.
4 levels: high, moderate, low, very low
All effect estimates Quality assessed