Title: Applying Trials and Systematic Reviews to Individual Patients
1Applying Trials and Systematic Reviews to
Individual Patients
- Paul Glasziou Sharon Sanders
- University of Queensland
- for
- Cochrane Applicability Recommendations Methods
Group
2www.sph.uq.edu.au/CGP/training/CochraneMethodsGrou
p.html
3The problem of applying trial results
4Who does this trial apply to?
- Yes, the trial showed it worked
- But will it work as well in THIS patient?
- A 30 relative risk reduction (RRR) means
- It worked in 30
- It didnt work in 70(and they were at risk of
adverse outcomes) - And what is the importance of it working for
THIS patient?
5Should Mr RM buy an electric toothbrush?
- 72 year old pensioner with Parkinsons Disease
- Has gingivitis and frequent caries
- Trials in young healthy folk showing improvements
in gingivitis scores but not caries. - What should he do?
6(No Transcript)
7Possible approaches to who a trial applies to?
- Effect size (larger will apply more)
- Look at biological mechanisms to predict who it
might or might not work - Is there study evidence to support this?
- How do trial population compare to patient
characteristics - How DOES THE Risk COMPARE?
8Possible approaches to applying reviews and trials
- Clinical significance LHH (NNT)
- Costs of options
- Who were the patients
- Subgroups
- Risks to patients (and are some more at risk)
LHH (NNH) - What is this add-on to??
9Possible approaches to applying reviews and trials
- Inclusion/exclusion criteria
- For reviews overlap or combination?
- Subgroup analysis
- Appropriate methods needed
- Cross-design synthesis
- Combining RCT and database evidence
- 5-Steps of Transferability/Applicability
- Benefits versus harms predicted risk
- Glasziou, Irwig BMJ 1995
- OConnell, Glasziou, Hill. NHMRC How to use the
evidence
10Transferability and applicability of results
- TRANSFERABILITY (across groups)
- What are the benefits and harms?
- Is there predictable variation in the effects?
- How does effect vary with predicted risk?
- APPLICATION (to individual)
- What are the predicted absolute risk reductions
for individuals? - Do the benefits outweigh the harms?
11Example of Toothbrush 5-steps
12Rofecoxib, Celecoxib, and Paracetamol in
Osteoarthritis of the Knee
- Design Randomized, parallel-group, double-blind
trial, - Patients 382 patients 40 yrs with OA of the
knee - Interventions
- Rofecoxib, 12.5 mg/d OR 25 mg/d
- Celecoxib, 200 mg/d OR
- acetaminophen, 4,000 mg/d
- Results Over 6 weeks, rofecoxib, 25 mg/d, better
for - night pain (Plt.002 vs celecoxib and P0.006 vs
acetaminophen and P0.02 vs rofecoxib, 12.5
mg/d), - composite pain subscale (P.03 vs all other
treatments), - stiffness subscale (P0.04 vs celecoxib and
acetaminophen), - physical function subscale (P0.001 vs
acetaminophen). - Global responses over 6 weeks showed a similar
pattern.
Geba GP, et al JAMA. 2002 28764-71.
13All or some responders?
I. Everyone gets small benefit?
14Osteoarthritis N-of-1s
- Comparison of
- 1,000mg paracetamol tds
- 400mg ibuprofen tds
- Two weeks x 6
- Outcome diary of pain and stiffness of target
joint
Paracetamol
NSAID
Pair 1
NSAID
Paracetamol
Pair 2
NSAID
Paracetamol
Pair 3
15N-of-1 overall examples
NSAID responder
NSAID non-responder
16Interventions Levels of Evidence
- N-of-1 Trial
- Systematic review of randomised trials
- A single randomised trial
- Controlled, non-randomised
- Parallel control
- Historical control
- Case-control
- Case-series
Guyatt, JAMA, 2000
17When n-of-1 not possibleThe benefit-harm model
(Lubsen, Tijssen)
- When does benefit outweigh harm?
- Assumptions
- Benefit (rate difference) proportional to event
rate - Harm constant over event rate
- Net benefit benefit - harm
Controlled Clinical Trials 10 151S-160S.
18Transferability and applicability of results
- TRANSFERABILITY (across groups)
- What are the benefits and harms?
- Is there predictable variation in the effects?
- How does effect vary with predicted risk?
- APPLICATION (to individual)
- What are the predicted absolute risk reductions
for individuals? - Do the benefits outweigh the harms?
19Tutorial
- Look at the systematic review
- Do you have any medical questions?
- Dont read it all !!
- Try to extract the information to complete the
first step transferability checklist
20Should I anticoagulate Mr CG?
- 58 year old male with stable angina noted to have
atrial fibrillation - Atrial Fibrillation increases the risk of embolic
stroke - 6 RCTs of warfarin in AF
- Decrease stroke
- Increase bleeding
- Would you anticoagulate him?
211. What are the benefits and harms?
- List all potential benefits
- List all potential harms
- Estimate from (meta-analysis) of best available
of evidence - Relative effect
- Absolute effect
22Step 1 benefits harms
23Safety in systematic reviews
Ernst, Pittler, BMJ 2001 323546
24Transferability and applicability of results
- TRANSFERABILITY (across groups)
- What are the benefits and harms?
- Is there predictable variation in the effects?
- How does effect vary with predicted risk?
- APPLICATION (to individual)
- What are the predicted absolute risk reductions
for individuals? - Do the benefits outweigh the harms?
25The problem of applying trial results
The Trial patients
The actual patients
262. Are there predictable variations in the
effects?
- Does effect vary by (PICO)
- Patient features, e.g., comorbidity or disease
features, e.g., stage - Intervention features e.g., dose/intensity/timing?
- Comparator, e.g., placebo, add-on, or active
- Outcome measures, e.g., reliability, duration
- But beware of artefactual causes
- Differences in followup, compliance, measures ,
27Tutorial
- Look at the systematic review again
- Try to extract the information to complete the
second step transferability checklist - P
- I
- C
- O
28CHD Death Non-fatal MI
LIPID Major Subgroups of Patient Features
29Are the groups different?
DO Test for difference
DONT Test each separately
30Step 2 effect modifiers
313. How does effect vary with predicted risk?
- Is Relative Risk constant across low to high risk
groups? - Relative Risk is most often constant
- Need to check using
- Plots
- Heterogeneity statistics
32When n-of-1 not possibleThe benefit-harm model
(Lubsen, Tijssen)
- When does benefit outweigh harm?
- Assumptions
- Benefit (rate difference) proportional to event
rate - Harm constant over event rate
- Net benefit benefit - harm
Controlled Clinical Trials 10 151S-160S.
33Warfarin in non-valvular Atrial Fibrillation the
trial evidence
343. How does effect vary with predicted risk?
Trials of Warfarin in Atrial Fibrillation
35Rate versus rate plots
LAbbe plot of trials of Warfarin in Atrial
Fibrillation
Treatment group rate
Control group rate
36Rate versus rate plots
LAbbe plot of trials of Warfarin in Atrial
Fibrillation
Treatment group rate
Control group rate
37Which risk measure is most constant?
Analysis of the effect of control rate in 115
meta-analysis Schmid et al Stats in Med 1998
1923-42.
38Step 3 Severity or Risk
394. How do absolute benefits and harms vary with
risk/severity?
40Benefit versus HarmClinical predictors of stroke
Benefit 73 RRR
1 ICH death 4 strokes
Harm 0.01 deaths
1 ICH death 1 stroke
41Trial inclusion/exclusion vs Net Benefit
42Guidelines for Atrial FibrillationEvidence
Recommendations
Recommendations
Risk/Yr 1 3 gt7
NNT 133 50 lt20
No Risk Factors - use aspirin only
One Risk Factor - individualise treatment
Two or more Risk Factors - anticoagulation
strongly recommended
The 5 Risk Factors are hypertension, recent
CCF, previous thromboembolism, LV dysfunction,
atrial size
43Guidelines proportion of patients with AF
needing anti-coagulation
Thomson R BMJ 1998316509-13
44Trial inclusion/exclusion criteriaAn appropriate
basis for transferability?
- Inclusion/exclusion criteria not (usually) aimed
at transferability but at - improving study power
- choose high risk groups
- minimise death from other causes
- ensure good compliance
- maximising safety
- exclude if any possible adverse effects
Some excluded sub-groups may have net
benefit Some included sub-groups may have net harm
45SUMMARY data for applying the results of
controlled trials
Relative Risk Benefits, Harms systematic review
of RCTs
Values of Benefits, Harms Your patient, Trials
patients
Individual Clinical Decision
Patients Expected Event Rate Predictive model
from Cohort study
46sun21.imbi.uni-freiburg.de/mailman/listinfo/applic
abilitygroup
47(No Transcript)
481. What are the benefits and harms?
- For YOUR review
- List all potential benefits
- List all potential harms
- Are they mutually exclusive and exhaustive?
- Will all the data come from trials?
49- GOOD
- Toothbrush example nice
- (but maybe fill out form!!)
- Do DIFFERENTLY
- Do reading before session
- More time more coffee
- Interested in steps 4 5
- Stream groups