Critical appraisal Treatment Dr. Zen Ahmad, SpPD Departemen Penyakit Dalam RSMH Palembang
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Critical appraisal
Treatment
Dr. Zen Ahmad, SpPDDepartemen Penyakit Dalam RSMH Palembang
Clinical Trials
Many types of design Generally : the simpler the better
Straightforward result Easy to understand No or few assumptions
The complicated ones Not easily understood Frequently uses assumptions
Gold standard : randomized, double blind, placebo controlled clinical trial (Randomized controlled trial, RCT)
Clasification : Pragmatic trial Explanatory trial
Pragmatic Trial
Attempt to determine cause-effect relationship Assuming the result will be applied in actual clinical
practice Preferably : binomial outcome (Yes/No) Analysis :
Intention to treat analysis= All randomized subjects are accounted for the final calculation according to their original allocation
Pragmatic Trial
R
Exp
Ctrl
Ca
b
Y
Y
N
N
A, b, c are accounted as failure of Exp arm
Explanatory Trial
Attempt to explain cause-effect relationship Usually in laboratory investigations
(pharmacology, pharmacodinamic, etc) Analysis : on treatment analysis (only subjects
completed the trial are accounted in analysis) Only minimal drop out is allowed, or
replacement for drop outs
Validity
Randomization; was the randomization list concealed ? Was follow-up of patients sufficiently long and complete ? Were all patients analyzed in the groups to which they were
randomized ? Were patients and clinicians kept blind to treatment ? Equal treatment between groups Were the groups similar at the start of the trial ? Sample size
Importance
1. What is the magnitude the treatment effect ?
2. How precise is this estimate of treatment effect
Importance
E 40 10 50
30 20 50C
Y N
X2 = ; df = 1 ; p = 0.04
Importance
E 40 10 50
30 20 50C
Y N
CER = 20/50 = 0.4; EER = 10/50 = 0.2
RRR = (CER-EER)/ CER = (0.4-0.2)/ 0.4 = 50%
Importance
USA, 1960’sNewspaper : the risk of suffering from deep vein thrombosis in OC users was 2 times compared to that in non OC users ! (this is RRR)
Closer examination :The risk for DVT in non OC : 1/ 100.000 person yearThe risk for DVT in OC : 2/ 100.000 person-year
Thus : to have additional bad outcome, one has to treat 100.000 women for year.
Importance
E 40 10 50
30 20 50C
Y N
CER = 20/50 = 0.4 ; EER = 10/50 = 0.2
ARR = (CER-EER) = 0.4 - 0.2 = 0.2
NNT = 1/ ARR = 1/ 0.2 = 5
NNT = number needed to treat= number of patients should be treated to avoid 1 bad outcome= number of patients should be treated to have 1 additional goodd outcome
NNH = number needed to harm
CI for NNT
NNT = 1/ ARR
First calculate CI for ARR(ARR = diff between proportion)
Then calculate1/ (upper CL of ARR)and 1/ (lower CL of ARR)
Calculating CI for NNT
CER = 20/50 = 0.4; EER = 10/50 = 0.2ARR = (CER – EER) = 0.4 – 0.2 = 0.2NNT = 1/ ARR = 1/0.2 = 5
95 % CI ARR = ARR + 1.96V (p1q1/n1 + p2q2/n2) = 0.2 + 1.96V (0.4 x 0.6)/ 50 + (0.2 x 0.8/ 50) = 0.2 + 0.17 = 0.03 ; 0.37
95% CI = 1/ 0.37 ; 1/0.03= 3 ; 34
1. Is our patient so different from those in the study that its result cannot apply ?
2. Is the treatment feasible in our setting ?
3. What are our patient’s potential benefits and harm from the therapy ?
4. What are our patient’s values and expectations for both the outcome we are trying to prevent and the treatment we are offering ?
Applicability
Applicability
Your own (s)
(Educated guess) – determine f, I. e. a factor reflecting how much severe are your patient compared to the average of the important prognostic factors of the study subjects)
Your NNT = f x NNT
Applicability
LLH : likelihood of being helped vs harmedStep 1. To elicit our patients
Applicability
Determine PEER = patient expected event rate (event rate of your patient if he/ she is not treated with the drug under consideration)
Then :
Your NNT = PEER / (PEER – EER)
ApplicabilityWhat are our patient’s values and expectations for both the outcome we are trying to prevent and the treatment we are offering ?
patient to make his own treatment decision
LLH : likelihood of being helped vs harmed
LHH : likelihood of being helped vs harmed
Step 1. To elicit our patients preferences
• Description (oral or written), discuss (patient; family)• Judgment (outcome; adverse event)
• ie: Relapse 20 times as severe as the side effect• rating scale
• from 0 (worse/ death) to 1 (full health)• Example: 0.05 (out come) and 0.95 (adverse e)• Patient believe that disease progression is 19 times
worse than the adverse event
LHH : likelihood of being helped vs harmed
Step 2. To generate the LHH• Reference
• LHH = 1/NNT vs 1/NNH = 1/9 vs 1/4 = 0.11 vs 0.25 (condition, relapse and adverse e were the same severity) th/ is twice as likely to harm you as to help you
Out come CER EER RRR/ RRI
ARR/ ARI
NNT/NNH
Disability 50 % 39 % 22 % 11 % 9Adverse event 37 % 64 % 73 % 27 % 4
LHH : likelihood of being helped vs harmed
Step 2. To generate the LHH• Our patient LHH = (1/NNT) x f vs (1/NNH) x f
= (1/9) x 3 vs (1/4) x 1 = 1.3 : 1 • Final adjustment LHH = (1/NNT) x f x s vs (1/NHH) x f = (1/9) x 3 x 19 vs (1/4) x 1
= 6.3 : 0.25 = 25 : 1• Patients is 25 times as likely to be helped vs harmed by treatment
Critical appraisal
Report of systematic reviews
Dr. Zen Ahmad, SpPDDepartemen Penyakit Dalam RSMH Palembang
Are the results of this systematic review valid
1. Is this a systematic review of randomized trials2. Does this systematic review have a methods
section that describes: Finding and including all relevant trials How the validity of the individual studies was
assessed3. Were the results consistent from study to study4. Were individual patient data used in the
analysis (or aggregate data)
Are the valid results of this SR Importance
1. What is the magnitude the treatment effect ?
2. How precise is this estimate of treatment effect
Formula to convert OR and RR to NNT
For RR < 1NNT = 1/(1 – RR) x PEER
For RR > 1NNT = 1/(RR - 1) x PEER
For OR < 1NNT = 1 - [PEER x (1- OR)] / (1- PEER) x (PEER) x (1- OR)
For OR > 1NNT = 1 + [PEER x (OR - 1)] / (1- PEER) x (PEER) x (OR- 1)
PEER
OR < 10.9 0.8 0.7 0.6 0.5
0.05 209 104 69 52 410.10 110 54 36 27 2210.20 61 30 20 14 110.30 46 22 14 10 80.40 40 19 12 9 70.50 38 18 11 8 60.70 44 20 13 9 60.90 101 46 27 18 12
1. Is our patient so different from those in the study that its result cannot apply ?
2. Is the treatment feasible in our setting ?
3. What are our patient’s potential benefits and harm from the therapy ?
4. What are our patient’s values and preferences for both the outcome we are trying to prevent and the side effect we may cause ?
Applicability
Meta analysis
Systematic review
Review article
Review article
Non systematic In gathering relevant studies No sufficient appraisals
Prone for severe bias Authors tend to cite studies that support their
opinion Still valuable in some areas of study
Clinical & lab descriptions of diseases
Systematic review and Meta analysis
Systematic review Systematic in:
Gathering relevant articles Appraising the articles
No formal statistical analysis Meta-analysis
Systematic review with formal statistical analysis
Related Documents
